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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1567 - (show annotations)
Fri Jun 12 16:24:58 2015 UTC (4 years, 5 months ago) by ph10
File size: 280072 byte(s)
Fix \a and \e in pcretest, and \a in pcre_compile, on EBCDIC platforms.
1 ChangeLog for PCRE
2 ------------------
3
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
6
7 Version 8.38 xx-xxx-xxxx
8 ------------------------
9
10 1. If a group that contained a recursive back reference also contained a
11 forward reference subroutine call followed by a non-forward-reference
12 subroutine call, for example /.((?2)(?R)\1)()/, pcre2_compile() failed to
13 compile correct code, leading to undefined behaviour or an internally
14 detected error. This bug was discovered by the LLVM fuzzer.
15
16 2. Quantification of certain items (e.g. atomic back references) could cause
17 incorrect code to be compiled when recursive forward references were
18 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
19 This bug was discovered by the LLVM fuzzer.
20
21 3. A repeated conditional group whose condition was a reference by name caused
22 a buffer overflow if there was more than one group with the given name.
23 This bug was discovered by the LLVM fuzzer.
24
25 4. A recursive back reference by name within a group that had the same name as
26 another group caused a buffer overflow. For example:
27 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
28
29 5. A forward reference by name to a group whose number is the same as the
30 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
31 a buffer overflow at compile time. This bug was discovered by the LLVM
32 fuzzer.
33
34 6. A lookbehind assertion within a set of mutually recursive subpatterns could
35 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
36
37 7. Another buffer overflow bug involved duplicate named groups with a
38 reference between their definition, with a group that reset capture
39 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
40 fixed by always allowing for more memory, even if not needed. (A proper fix
41 is implemented in PCRE2, but it involves more refactoring.)
42
43 8. There was no check for integer overflow in subroutine calls such as (?123).
44
45 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
46 being treated as a literal 'l' instead of causing an error.
47
48 10. There was a buffer overflow if pcre_exec() was called with an ovector of
49 size 1. This bug was found by american fuzzy lop.
50
51 11. If a non-capturing group containing a conditional group that could match
52 an empty string was repeated, it was not identified as matching an empty
53 string itself. For example: /^(?:(?(1)x|)+)+$()/.
54
55 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
56 \a and \e in test subject lines.
57
58 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
59 instead of the EBCDIC value.
60
61
62 Version 8.37 28-April-2015
63 --------------------------
64
65 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
66 for those parentheses to be closed with whatever has been captured so far.
67 However, it was failing to mark any other groups between the hightest
68 capture so far and the currrent group as "unset". Thus, the ovector for
69 those groups contained whatever was previously there. An example is the
70 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
71
72 2. If an assertion condition was quantified with a minimum of zero (an odd
73 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
74
75 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
76 unrecognized modifier, a crash could occur.
77
78 4. An attempt to do global matching in pcretest with a zero-length ovector
79 caused a crash.
80
81 5. Fixed a memory leak during matching that could occur for a subpattern
82 subroutine call (recursive or otherwise) if the number of captured groups
83 that had to be saved was greater than ten.
84
85 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
86 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
87 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
88
89 7. A UTF pattern containing a "not" match of a non-ASCII character and a
90 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
91
92 8. When a pattern is compiled, it remembers the highest back reference so that
93 when matching, if the ovector is too small, extra memory can be obtained to
94 use instead. A conditional subpattern whose condition is a check on a
95 capture having happened, such as, for example in the pattern
96 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
97 setting the highest backreference number. This mattered only if pcre_exec()
98 was called with an ovector that was too small to hold the capture, and there
99 was no other kind of back reference (a situation which is probably quite
100 rare). The effect of the bug was that the condition was always treated as
101 FALSE when the capture could not be consulted, leading to a incorrect
102 behaviour by pcre_exec(). This bug has been fixed.
103
104 9. A reference to a duplicated named group (either a back reference or a test
105 for being set in a conditional) that occurred in a part of the pattern where
106 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
107 to be incorrectly calculated, leading to overwriting.
108
109 10. A mutually recursive set of back references such as (\2)(\1) caused a
110 segfault at study time (while trying to find the minimum matching length).
111 The infinite loop is now broken (with the minimum length unset, that is,
112 zero).
113
114 11. If an assertion that was used as a condition was quantified with a minimum
115 of zero, matching went wrong. In particular, if the whole group had
116 unlimited repetition and could match an empty string, a segfault was
117 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
118 assertions to be quantified, but not if they are being used as conditions,
119 so the above pattern is faulted by Perl. PCRE has now been changed so that
120 it also rejects such patterns.
121
122 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
123 failed to allow the zero-repeat case if pcre2_exec() was called with an
124 ovector too small to capture the group.
125
126 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
127 Red Hat Product Security:
128
129 (a) A crash if /K and /F were both set with the option to save the compiled
130 pattern.
131
132 (b) Another crash if the option to print captured substrings in a callout
133 was combined with setting a null ovector, for example \O\C+ as a subject
134 string.
135
136 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
137 forward reference repeated a large (but limited) number of times within a
138 repeated outer group that has a zero minimum quantifier, caused incorrect
139 code to be compiled, leading to the error "internal error:
140 previously-checked referenced subpattern not found" when an incorrect
141 memory address was read. This bug was reported as "heap overflow",
142 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
143 CVE-2015-2325.
144
145 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
146 call within a group that also contained a recursive back reference caused
147 incorrect code to be compiled. This bug was reported as "heap overflow",
148 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
149 number CVE-2015-2326.
150
151 24. Computing the size of the JIT read-only data in advance has been a source
152 of various issues, and new ones are still appear unfortunately. To fix
153 existing and future issues, size computation is eliminated from the code,
154 and replaced by on-demand memory allocation.
155
156 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
157 adjacent to the end of the range, and the range contained characters with
158 more than one other case, caused incorrect behaviour when compiled in UTF
159 mode. In that example, the range a-j was left out of the class.
160
161 26. Fix JIT compilation of conditional blocks, which assertion
162 is converted to (*FAIL). E.g: /(?(?!))/.
163
164 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
165 discovered by the LLVM fuzzer.
166
167 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
168 when this assertion was used as a condition, for example (?(?!)a|b). In
169 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
170 error about an unsupported item.
171
172 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
173 possessification code could take exponential time to complete. A recursion
174 depth limit of 1000 has been imposed to limit the resources used by this
175 optimization.
176
177 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
178 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
179 because \S ensures they are all in the class. The code for doing this was
180 interacting badly with the code for computing the amount of space needed to
181 compile the pattern, leading to a buffer overflow. This bug was discovered
182 by the LLVM fuzzer.
183
184 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
185 other kinds of group caused stack overflow at compile time. This bug was
186 discovered by the LLVM fuzzer.
187
188 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
189 between a subroutine call and its quantifier was incorrectly compiled,
190 leading to buffer overflow or other errors. This bug was discovered by the
191 LLVM fuzzer.
192
193 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
194 assertion after (?(. The code was failing to check the character after
195 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
196 was discovered by the LLVM fuzzer.
197
198 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
199 a fixed maximum following a group that contains a subroutine reference was
200 incorrectly compiled and could trigger buffer overflow. This bug was
201 discovered by the LLVM fuzzer.
202
203 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
204 caused a stack overflow instead of the diagnosis of a non-fixed length
205 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
206
207 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
208 (e.g. /(?<=\Ka)/) could make pcregrep loop.
209
210 37. There was a similar problem to 36 in pcretest for global matches.
211
212 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
213 and a subsequent item in the pattern caused a non-match, backtracking over
214 the repeated \X did not stop, but carried on past the start of the subject,
215 causing reference to random memory and/or a segfault. There were also some
216 other cases where backtracking after \C could crash. This set of bugs was
217 discovered by the LLVM fuzzer.
218
219 39. The function for finding the minimum length of a matching string could take
220 a very long time if mutual recursion was present many times in a pattern,
221 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
222 method has been implemented. This infelicity was discovered by the LLVM
223 fuzzer.
224
225 40. Static linking against the PCRE library using the pkg-config module was
226 failing on missing pthread symbols.
227
228
229 Version 8.36 26-September-2014
230 ------------------------------
231
232 1. Got rid of some compiler warnings in the C++ modules that were shown up by
233 -Wmissing-field-initializers and -Wunused-parameter.
234
235 2. The tests for quantifiers being too big (greater than 65535) were being
236 applied after reading the number, and stupidly assuming that integer
237 overflow would give a negative number. The tests are now applied as the
238 numbers are read.
239
240 3. Tidy code in pcre_exec.c where two branches that used to be different are
241 now the same.
242
243 4. The JIT compiler did not generate match limit checks for certain
244 bracketed expressions with quantifiers. This may lead to exponential
245 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
246 issue should be resolved now.
247
248 5. Fixed an issue, which occures when nested alternatives are optimized
249 with table jumps.
250
251 6. Inserted two casts and changed some ints to size_t in the light of some
252 reported 64-bit compiler warnings (Bugzilla 1477).
253
254 7. Fixed a bug concerned with zero-minimum possessive groups that could match
255 an empty string, which sometimes were behaving incorrectly in the
256 interpreter (though correctly in the JIT matcher). This pcretest input is
257 an example:
258
259 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
260 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
261
262 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
263 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
264 for an empty string was breaking the inner loop and carrying on at a lower
265 level, when possessive repeated groups should always return to a higher
266 level as they have no backtrack points in them. The empty string test now
267 occurs at the outer level.
268
269 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
270 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
271
272 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
273 doesn't).
274
275 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
276 to hold a pattern to be incorrectly computed (too small) when there were
277 named back references to duplicated names. This could cause "internal
278 error: code overflow" or "double free or corruption" or other memory
279 handling errors.
280
281 11. When named subpatterns had the same prefixes, back references could be
282 confused. For example, in this pattern:
283
284 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
285
286 the reference to 'Name' was incorrectly treated as a reference to a
287 duplicate name.
288
289 12. A pattern such as /^s?c/mi8 where the optional character has more than
290 one "other case" was incorrectly compiled such that it would only try to
291 match starting at "c".
292
293 13. When a pattern starting with \s was studied, VT was not included in the
294 list of possible starting characters; this should have been part of the
295 8.34/18 patch.
296
297 14. If a character class started [\Qx]... where x is any character, the class
298 was incorrectly terminated at the ].
299
300 15. If a pattern that started with a caseless match for a character with more
301 than one "other case" was studied, PCRE did not set up the starting code
302 unit bit map for the list of possible characters. Now it does. This is an
303 optimization improvement, not a bug fix.
304
305 16. The Unicode data tables have been updated to Unicode 7.0.0.
306
307 17. Fixed a number of memory leaks in pcregrep.
308
309 18. Avoid a compiler warning (from some compilers) for a function call with
310 a cast that removes "const" from an lvalue by using an intermediate
311 variable (to which the compiler does not object).
312
313 19. Incorrect code was compiled if a group that contained an internal recursive
314 back reference was optional (had quantifier with a minimum of zero). This
315 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
316 caused segmentation faults because of stack overflows at compile time.
317
318 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
319 group that is quantified with an indefinite repeat, caused a compile-time
320 loop which used up all the system stack and provoked a segmentation fault.
321 This was not the same bug as 19 above.
322
323 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
324 Patch by Mike Frysinger.
325
326
327 Version 8.35 04-April-2014
328 --------------------------
329
330 1. A new flag is set, when property checks are present in an XCLASS.
331 When this flag is not set, PCRE can perform certain optimizations
332 such as studying these XCLASS-es.
333
334 2. The auto-possessification of character sets were improved: a normal
335 and an extended character set can be compared now. Furthermore
336 the JIT compiler optimizes more character set checks.
337
338 3. Got rid of some compiler warnings for potentially uninitialized variables
339 that show up only when compiled with -O2.
340
341 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
342 of a match later then the end of the match. The pcretest program was not
343 handling the case sensibly - it was outputting from the start to the next
344 binary zero. It now reports this situation in a message, and outputs the
345 text from the end to the start.
346
347 5. Fast forward search is improved in JIT. Instead of the first three
348 characters, any three characters with fixed position can be searched.
349 Search order: first, last, middle.
350
351 6. Improve character range checks in JIT. Characters are read by an inprecise
352 function now, which returns with an unknown value if the character code is
353 above a certain threshold (e.g: 256). The only limitation is that the value
354 must be bigger than the threshold as well. This function is useful when
355 the characters above the threshold are handled in the same way.
356
357 7. The macros whose names start with RAWUCHAR are placeholders for a future
358 mode in which only the bottom 21 bits of 32-bit data items are used. To
359 make this more memorable for those maintaining the code, the names have
360 been changed to start with UCHAR21, and an extensive comment has been added
361 to their definition.
362
363 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
364 to the export list in Makefile.am (they were accidentally omitted from the
365 8.34 tarball).
366
367 9. The informational output from pcretest used the phrase "starting byte set"
368 which is inappropriate for the 16-bit and 32-bit libraries. As the output
369 for "first char" and "need char" really means "non-UTF-char", I've changed
370 "byte" to "char", and slightly reworded the output. The documentation about
371 these values has also been (I hope) clarified.
372
373 10. Another JIT related optimization: use table jumps for selecting the correct
374 backtracking path, when more than four alternatives are present inside a
375 bracket.
376
377 11. Empty match is not possible, when the minimum length is greater than zero,
378 and there is no \K in the pattern. JIT should avoid empty match checks in
379 such cases.
380
381 12. In a caseless character class with UCP support, when a character with more
382 than one alternative case was not the first character of a range, not all
383 the alternative cases were added to the class. For example, s and \x{17f}
384 are both alternative cases for S: the class [RST] was handled correctly,
385 but [R-T] was not.
386
387 13. The configure.ac file always checked for pthread support when JIT was
388 enabled. This is not used in Windows, so I have put this test inside a
389 check for the presence of windows.h (which was already tested for).
390
391 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
392 The algorithm provides a way to skip certain starting offsets, and usually
393 faster than linear prefix searches.
394
395 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
396 as for 'fr_FR' and 'french'. For some reason, however, it then used the
397 Windows-specific input and output files, which have 'french' screwed in.
398 So this could never have worked. One of the problems with locales is that
399 they aren't always the same. I have now updated RunTest so that it checks
400 the output of the locale test (test 3) against three different output
401 files, and it allows the test to pass if any one of them matches. With luck
402 this should make the test pass on some versions of Solaris where it was
403 failing. Because of the uncertainty, the script did not used to stop if
404 test 3 failed; it now does. If further versions of a French locale ever
405 come to light, they can now easily be added.
406
407 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
408 there was a message during ./configure, but it did not stop. This now
409 provokes an error. The invalid example in README has been corrected.
410 If a value less than the minimum is given, the minimum value has always
411 been used, but now a warning is given.
412
413 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
414 was a bug in the test system, which is now fixed. Also, the list of various
415 configurations that are tested for each release did not have one with both
416 16/32 bits and --enable-bar-anycrlf. It now does.
417
418 18. pcretest was missing "-C bsr" for displaying the \R default setting.
419
420 19. Little endian PowerPC systems are supported now by the JIT compiler.
421
422 20. The fast forward newline mechanism could enter to an infinite loop on
423 certain invalid UTF-8 input. Although we don't support these cases
424 this issue can be fixed by a performance optimization.
425
426 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
427 not take account if existing stack usage. There is now a new global
428 variable called pcre_stack_guard that can be set to point to an external
429 function to check stack availability. It is called at the start of
430 processing every parenthesized group.
431
432 22. A typo in the code meant that in ungreedy mode the max/min qualifier
433 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
434 not match "ab".
435
436 23. When UTF was disabled, the JIT program reported some incorrect compile
437 errors. These messages are silenced now.
438
439 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
440 compiler.
441
442 25. Change all the temporary files used in RunGrepTest to be different to those
443 used by RunTest so that the tests can be run simultaneously, for example by
444 "make -j check".
445
446
447 Version 8.34 15-December-2013
448 -----------------------------
449
450 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
451 executable memory. Patch inspired by Carsten Klein.
452
453 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
454 this macro is never tested and has no effect, because the work to support
455 coverage involves only compiling and linking options and special targets in
456 the Makefile. The comment in config.h implied that defining the macro would
457 enable coverage support, which is totally false. There was also support for
458 setting this macro in the CMake files (my fault, I just copied it from
459 configure). SUPPORT_GCOV has now been removed.
460
461 3. Make a small performance improvement in strlen16() and strlen32() in
462 pcretest.
463
464 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
465 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
466 these cases). There was also one in pcretest.c.
467
468 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
469
470 6. In UTF mode, the code for checking whether a group could match an empty
471 string (which is used for indefinitely repeated groups to allow for
472 breaking an infinite loop) was broken when the group contained a repeated
473 negated single-character class with a character that occupied more than one
474 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
475 UTF-8 mode). The effect was undefined: the group might or might not be
476 deemed as matching an empty string, or the program might have crashed.
477
478 7. The code for checking whether a group could match an empty string was not
479 recognizing that \h, \H, \v, \V, and \R must match a character.
480
481 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
482 an empty string. If it can, pcretest shows this in its information output.
483
484 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
485 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
486 matched by pcre_exec() without using JIT:
487
488 (a) If the rest of the pattern did not match after a maximal run of
489 grapheme clusters, the code for backing up to try with fewer of them
490 did not always back up over a full grapheme when characters that do not
491 have the modifier quality were involved, e.g. Hangul syllables.
492
493 (b) If the match point in a subject started with modifier character, and
494 there was no match, the code could incorrectly back up beyond the match
495 point, and potentially beyond the first character in the subject,
496 leading to a segfault or an incorrect match result.
497
498 10. A conditional group with an assertion condition could lead to PCRE
499 recording an incorrect first data item for a match if no other first data
500 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
501 first data item, and therefore matched "ca" after "c" instead of at the
502 start.
503
504 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
505 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
506
507 12. The source of pcregrep now includes z/OS-specific code so that it can be
508 compiled for z/OS as part of the special z/OS distribution.
509
510 13. Added the -T and -TM options to pcretest.
511
512 14. The code in pcre_compile.c for creating the table of named capturing groups
513 has been refactored. Instead of creating the table dynamically during the
514 actual compiling pass, the information is remembered during the pre-compile
515 pass (on the stack unless there are more than 20 named groups, in which
516 case malloc() is used) and the whole table is created before the actual
517 compile happens. This has simplified the code (it is now nearly 150 lines
518 shorter) and prepared the way for better handling of references to groups
519 with duplicate names.
520
521 15. A back reference to a named subpattern when there is more than one of the
522 same name now checks them in the order in which they appear in the pattern.
523 The first one that is set is used for the reference. Previously only the
524 first one was inspected. This change makes PCRE more compatible with Perl.
525
526 16. Unicode character properties were updated from Unicode 6.3.0.
527
528 17. The compile-time code for auto-possessification has been refactored, based
529 on a patch by Zoltan Herczeg. It now happens after instead of during
530 compilation. The code is cleaner, and more cases are handled. The option
531 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
532 options in pcretest are provided to set it. It can also be set by
533 (*NO_AUTO_POSSESS) at the start of a pattern.
534
535 18. The character VT has been added to the default ("C" locale) set of
536 characters that match \s and are generally treated as white space,
537 following this same change in Perl 5.18. There is now no difference between
538 "Perl space" and "POSIX space". Whether VT is treated as white space in
539 other locales depends on the locale.
540
541 19. The code for checking named groups as conditions, either for being set or
542 for being recursed, has been refactored (this is related to 14 and 15
543 above). Processing unduplicated named groups should now be as fast at
544 numerical groups, and processing duplicated groups should be faster than
545 before.
546
547 20. Two patches to the CMake build system, by Alexander Barkov:
548
549 (1) Replace the "source" command by "." in CMakeLists.txt because
550 "source" is a bash-ism.
551
552 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
553 without these the CMake build does not work on Solaris.
554
555 21. Perl has changed its handling of \8 and \9. If there is no previously
556 encountered capturing group of those numbers, they are treated as the
557 literal characters 8 and 9 instead of a binary zero followed by the
558 literals. PCRE now does the same.
559
560 22. Following Perl, added \o{} to specify codepoints in octal, making it
561 possible to specify values greater than 0777 and also making them
562 unambiguous.
563
564 23. Perl now gives an error for missing closing braces after \x{... instead of
565 treating the string as literal. PCRE now does the same.
566
567 24. RunTest used to grumble if an inappropriate test was selected explicitly,
568 but just skip it when running all tests. This make it awkward to run ranges
569 of tests when one of them was inappropriate. Now it just skips any
570 inappropriate tests, as it always did when running all tests.
571
572 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
573 character types such as \d or \w, too many callouts were inserted, and the
574 data that they returned was rubbish.
575
576 26. In UCP mode, \s was not matching two of the characters that Perl matches,
577 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
578 were matched by \h. The code has now been refactored so that the lists of
579 the horizontal and vertical whitespace characters used for \h and \v (which
580 are defined only in one place) are now also used for \s.
581
582 27. Add JIT support for the 64 bit TileGX architecture.
583 Patch by Jiong Wang (Tilera Corporation).
584
585 28. Possessive quantifiers for classes (both explicit and automatically
586 generated) now use special opcodes instead of wrapping in ONCE brackets.
587
588 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
589 (because it's meaningless), this was not happening when PCRE_CASELESS was
590 set. Not wrong, but inefficient.
591
592 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
593 properties for \w, \d, etc) is present in a test regex. Otherwise if the
594 test contains no characters greater than 255, Perl doesn't realise it
595 should be using Unicode semantics.
596
597 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
598 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
599 does in Unicode mode.
600
601 32. Added the "forbid" facility to pcretest so that putting tests into the
602 wrong test files can sometimes be quickly detected.
603
604 33. There is now a limit (default 250) on the depth of nesting of parentheses.
605 This limit is imposed to control the amount of system stack used at compile
606 time. It can be changed at build time by --with-parens-nest-limit=xxx or
607 the equivalent in CMake.
608
609 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
610 errors. Perl warns for these when in warning mode, but PCRE has no facility
611 for giving warnings.
612
613 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
614 However, this was not working for (?!) because it is optimized to (*FAIL),
615 for which PCRE does not allow quantifiers. The optimization is now disabled
616 when a quantifier follows (?!). I can't see any use for this, but it makes
617 things uniform.
618
619 36. Perl no longer allows group names to start with digits, so I have made this
620 change also in PCRE. It simplifies the code a bit.
621
622 37. In extended mode, Perl ignores spaces before a + that indicates a
623 possessive quantifier. PCRE allowed a space before the quantifier, but not
624 before the possessive +. It now does.
625
626 38. The use of \K (reset reported match start) within a repeated possessive
627 group such as (a\Kb)*+ was not working.
628
629 40. Document that the same character tables must be used at compile time and
630 run time, and that the facility to pass tables to pcre_exec() and
631 pcre_dfa_exec() is for use only with saved/restored patterns.
632
633 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
634 features for Builds with MSVC:
635
636 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
637 builds). The .rc files can be used to set FileDescription and many other
638 attributes.
639
640 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
641 This allows higher-level build scripts which want .pdb files to avoid
642 hard-coding the exact files needed."
643
644 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
645 mean "start of word" and "end of word", respectively, as a transition aid.
646
647 43. A minimizing repeat of a class containing codepoints greater than 255 in
648 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
649 compiled to use the heap for recursion.
650
651 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
652 is configured.
653
654
655 Version 8.33 28-May-2013
656 ------------------------
657
658 1. Added 'U' to some constants that are compared to unsigned integers, to
659 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
660 variables that are added to signed variables, to ensure the result is
661 signed and can be negated.
662
663 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
664 CMake config files.
665
666 3. Revise the creation of config.h.generic so that all boolean macros are
667 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
668 overriding via -D on the command line possible.
669
670 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
671 to unsigned int is reported to make a quite noticeable speed difference in
672 a specific Windows environment. Testing on Linux did also appear to show
673 some benefit (and it is clearly not harmful). Also fixed the definition of
674 Xop which should be unsigned.
675
676 5. Related to (4), changing the definition of the intermediate variable cc
677 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
678 improvements.
679
680 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
681 unnecessary spaces.
682
683 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
684 and later.
685
686 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
687
688 9. Optimizing fast_forward_start_bits in JIT.
689
690 10. Adding support for callouts in JIT, and fixing some issues revealed
691 during this work. Namely:
692
693 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
694
695 (b) Minimum length was not checked before the matching is started.
696
697 11. The value of capture_last that is passed to callouts was incorrect in some
698 cases when there was a capture on one path that was subsequently abandoned
699 after a backtrack. Also, the capture_last value is now reset after a
700 recursion, since all captures are also reset in this case.
701
702 12. The interpreter no longer returns the "too many substrings" error in the
703 case when an overflowing capture is in a branch that is subsequently
704 abandoned after a backtrack.
705
706 13. In the pathological case when an offset vector of size 2 is used, pcretest
707 now prints out the matched string after a yield of 0 or 1.
708
709 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
710 Only supported by the JIT compiler at the moment.
711
712 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
713
714 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
715 offset of the starting point of the matching process, provided the offsets
716 vector is large enough.
717
718 17. The \A escape now records a lookbehind value of 1, though its execution
719 does not actually inspect the previous character. This is to ensure that,
720 in partial multi-segment matching, at least one character from the old
721 segment is retained when a new segment is processed. Otherwise, if there
722 are no lookbehinds in the pattern, \A might match incorrectly at the start
723 of a new segment.
724
725 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
726
727 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
728 gives some modest performance improvement in 8-bit mode.
729
730 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
731 be expressed in certain programming languages using Universal Character
732 Names.
733
734 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
735 which points out that "non characters" are not "characters that may not
736 appear in Unicode strings" but rather "characters that are reserved for
737 internal use and have only local meaning".
738
739 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
740 there was a conditional group that depended on an assertion, if the
741 assertion was false, the callout that immediately followed the alternation
742 in the condition was skipped when pcre_exec() was used for matching.
743
744 23. Allow an explicit callout to be inserted before an assertion that is the
745 condition for a conditional group, for compatibility with automatic
746 callouts, which always insert a callout at this point.
747
748 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
749 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
750
751 25. (*PRUNE) is now supported by the JIT compiler.
752
753 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
754
755 27. Fix the case where there are two or more SKIPs with arguments that may be
756 ignored.
757
758 28. (*SKIP) is now supported by the JIT compiler.
759
760 29. (*THEN) is now supported by the JIT compiler.
761
762 30. Update RunTest with additional test selector options.
763
764 31. The way PCRE handles backtracking verbs has been changed in two ways.
765
766 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
767 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
768 backtracking. In some cases this makes it more Perl-compatible, but Perl's
769 rather obscure rules do not always do the same thing.
770
771 (2) Previously, backtracking verbs were confined within assertions. This is
772 no longer the case for positive assertions, except for (*ACCEPT). Again,
773 this sometimes improves Perl compatibility, and sometimes does not.
774
775 32. A number of tests that were in test 2 because Perl did things differently
776 have been moved to test 1, because either Perl or PCRE has changed, and
777 these tests are now compatible.
778
779 32. Backtracking control verbs are now handled in the same way in JIT and
780 interpreter.
781
782 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
783 contained a forward subroutine reference caused a compile error.
784
785 34. Auto-detect and optimize limited repetitions in JIT.
786
787 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
788 blocking (*UTF) etc.
789
790 36. In the interpreter, maximizing pattern repetitions for characters and
791 character types now use tail recursion, which reduces stack usage.
792
793 37. The value of the max lookbehind was not correctly preserved if a compiled
794 and saved regex was reloaded on a host of different endianness.
795
796 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
797 of the compiled pattern block, expand the flags field from 16 to 32 bits
798 because it was almost full.
799
800 39. Try madvise first before posix_madvise.
801
802 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
803 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
804 original change locked out all matches of empty strings. This has been
805 changed so that one match of an empty string per line is recognized.
806 Subsequent searches on the same line (for colouring or for --only-matching,
807 for example) do not recognize empty strings.
808
809 41. Applied a user patch to fix a number of spelling mistakes in comments.
810
811 42. Data lines longer than 65536 caused pcretest to crash.
812
813 43. Clarified the data type for length and startoffset arguments for pcre_exec
814 and pcre_dfa_exec in the function-specific man pages, where they were
815 explicitly stated to be in bytes, never having been updated. I also added
816 some clarification to the pcreapi man page.
817
818 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
819 a segmentation fault.
820
821
822 Version 8.32 30-November-2012
823 -----------------------------
824
825 1. Improved JIT compiler optimizations for first character search and single
826 character iterators.
827
828 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
829 Patch by Daniel Richard G.
830
831 3. Single character iterator optimizations in the JIT compiler.
832
833 4. Improved JIT compiler optimizations for character ranges.
834
835 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
836 Reported by Giuseppe D'Angelo.
837
838 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
839 of a line, was being set incorrectly in cases where .* appeared inside
840 atomic brackets at the start of a pattern, or where there was a subsequent
841 *PRUNE or *SKIP.
842
843 7. Improved instruction cache flush for POWER/PowerPC.
844 Patch by Daniel Richard G.
845
846 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
847 grep:
848
849 (a) There is now no limit to the number of patterns to be matched.
850
851 (b) An error is given if a pattern is too long.
852
853 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
854 are now supported.
855
856 (d) --exclude-from and --include-from (multiple use) have been added.
857
858 (e) Exclusions and inclusions now apply to all files and directories, not
859 just to those obtained from scanning a directory recursively.
860
861 (f) Multiple uses of -f and --file-list are now supported.
862
863 (g) In a Windows environment, the default for -d has been changed from
864 "read" (the GNU grep default) to "skip", because otherwise the presence
865 of a directory in the file list provokes an error.
866
867 (h) The documentation has been revised and clarified in places.
868
869 9. Improve the matching speed of capturing brackets.
870
871 10. Changed the meaning of \X so that it now matches a Unicode extended
872 grapheme cluster.
873
874 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
875 out POSIX threads when JIT support is configured.
876
877 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
878
879 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
880 structure could go wrong in environments where size_t is not the same size
881 as int.
882
883 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
884 set.
885
886 15. The EBCDIC support had decayed; later updates to the code had included
887 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
888 general tidy up of EBCDIC-related issues, and the documentation was also
889 not quite right. There is now a test that can be run on ASCII systems to
890 check some of the EBCDIC-related things (but is it not a full test).
891
892 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
893 in a small tidy to the code.
894
895 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
896
897 18. If the --only-matching (-o) option in pcregrep is specified multiple
898 times, each one causes appropriate output. For example, -o1 -o2 outputs the
899 substrings matched by the 1st and 2nd capturing parentheses. A separating
900 string can be specified by --om-separator (default empty).
901
902 19. Improving the first n character searches.
903
904 20. Turn case lists for horizontal and vertical white space into macros so that
905 they are defined only once.
906
907 21. This set of changes together give more compatible Unicode case-folding
908 behaviour for characters that have more than one other case when UCP
909 support is available.
910
911 (a) The Unicode property table now has offsets into a new table of sets of
912 three or more characters that are case-equivalent. The MultiStage2.py
913 script that generates these tables (the pcre_ucd.c file) now scans
914 CaseFolding.txt instead of UnicodeData.txt for character case
915 information.
916
917 (b) The code for adding characters or ranges of characters to a character
918 class has been abstracted into a generalized function that also handles
919 case-independence. In UTF-mode with UCP support, this uses the new data
920 to handle characters with more than one other case.
921
922 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
923 whose other case is greater than 256 are now correctly matched
924 caselessly. Previously, the high codepoint matched the low one, but not
925 vice versa.
926
927 (d) The processing of \h, \H, \v, and \ in character classes now makes use
928 of the new class addition function, using character lists defined as
929 macros alongside the case definitions of 20 above.
930
931 (e) Caseless back references now work with characters that have more than
932 one other case.
933
934 (f) General caseless matching of characters with more than one other case
935 is supported.
936
937 22. Unicode character properties were updated from Unicode 6.2.0
938
939 23. Improved CMake support under Windows. Patch by Daniel Richard G.
940
941 24. Add support for 32-bit character strings, and UTF-32
942
943 25. Major JIT compiler update (code refactoring and bugfixing).
944 Experimental Sparc 32 support is added.
945
946 26. Applied a modified version of Daniel Richard G's patch to create
947 pcre.h.generic and config.h.generic by "make" instead of in the
948 PrepareRelease script.
949
950 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
951 pcre_compile.c when checking for a zero character.
952
953 28. Introducing a native interface for JIT. Through this interface, the compiled
954 machine code can be directly executed. The purpose of this interface is to
955 provide fast pattern matching, so several sanity checks are not performed.
956 However, feature tests are still performed. The new interface provides
957 1.4x speedup compared to the old one.
958
959 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
960 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
961 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
962
963 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
964 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
965 the "old" RFC 2279). Instead, it ended up passing a negative length to
966 pcre_exec().
967
968 31. Add support for GCC's visibility feature to hide internal functions.
969
970 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
971 "unknown -C option" after outputting 0 or 1.
972
973 33. There is now support for generating a code coverage report for the test
974 suite in environments where gcc is the compiler and lcov is installed. This
975 is mainly for the benefit of the developers.
976
977 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
978 unaddressable using valgrind annotations, allowing valgrind to detect
979 invalid memory accesses. This is mainly for the benefit of the developers.
980
981 25. (*UTF) can now be used to start a pattern in any of the three libraries.
982
983 26. Give configure error if --enable-cpp but no C++ compiler found.
984
985
986 Version 8.31 06-July-2012
987 -------------------------
988
989 1. Fixing a wrong JIT test case and some compiler warnings.
990
991 2. Removed a bashism from the RunTest script.
992
993 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
994 to unsigned type, result still unsigned" that was given by an MS compiler
995 on encountering the code "-sizeof(xxx)".
996
997 4. Partial matching support is added to the JIT compiler.
998
999 5. Fixed several bugs concerned with partial matching of items that consist
1000 of more than one character:
1001
1002 (a) /^(..)\1/ did not partially match "aba" because checking references was
1003 done on an "all or nothing" basis. This also applied to repeated
1004 references.
1005
1006 (b) \R did not give a hard partial match if \r was found at the end of the
1007 subject.
1008
1009 (c) \X did not give a hard partial match after matching one or more
1010 characters at the end of the subject.
1011
1012 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1013 a partial match for the string "\r".
1014
1015 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1016 a partial match for a CR character at the end of the subject string.
1017
1018 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1019 running pcretest, the text "(JIT)" added to the output whenever JIT is
1020 actually used to run the match.
1021
1022 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1023 or /S+[+] with a digit between 1 and 7.
1024
1025 8. OP_NOT now supports any UTF character not just single-byte ones.
1026
1027 9. (*MARK) control verb is now supported by the JIT compiler.
1028
1029 10. The command "./RunTest list" lists the available tests without actually
1030 running any of them. (Because I keep forgetting what they all are.)
1031
1032 11. Add PCRE_INFO_MAXLOOKBEHIND.
1033
1034 12. Applied a (slightly modified) user-supplied patch that improves performance
1035 when the heap is used for recursion (compiled with --disable-stack-for-
1036 recursion). Instead of malloc and free for each heap frame each time a
1037 logical recursion happens, frames are retained on a chain and re-used where
1038 possible. This sometimes gives as much as 30% improvement.
1039
1040 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1041 call.
1042
1043 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1044
1045 15. It is now possible to link pcretest with libedit as an alternative to
1046 libreadline.
1047
1048 16. (*COMMIT) control verb is now supported by the JIT compiler.
1049
1050 17. The Unicode data tables have been updated to Unicode 6.1.0.
1051
1052 18. Added --file-list option to pcregrep.
1053
1054 19. Added binary file support to pcregrep, including the -a, --binary-files,
1055 -I, and --text options.
1056
1057 20. The madvise function is renamed for posix_madvise for QNX compatibility
1058 reasons. Fixed by Giuseppe D'Angelo.
1059
1060 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1061
1062 22. Changed the option for creating symbolic links for 16-bit man pages from
1063 -s to -sf so that re-installing does not cause issues.
1064
1065 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1066
1067 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1068 matches in certain environments (the workspace was not being correctly
1069 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1070 some of the workspace data at the beginning of a restart.
1071
1072 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1073 was not doing so when it should - probably a typo introduced by SVN 528
1074 (change 8.10/14).
1075
1076 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1077 \w+ when the character tables indicated that \x{c4} was a word character.
1078 There were several related cases, all because the tests for doing a table
1079 lookup were testing for characters less than 127 instead of 255.
1080
1081 27. If a pattern contains capturing parentheses that are not used in a match,
1082 their slots in the ovector are set to -1. For those that are higher than
1083 any matched groups, this happens at the end of processing. In the case when
1084 there were back references that the ovector was too small to contain
1085 (causing temporary malloc'd memory to be used during matching), and the
1086 highest capturing number was not used, memory off the end of the ovector
1087 was incorrectly being set to -1. (It was using the size of the temporary
1088 memory instead of the true size.)
1089
1090 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1091 ovector size, it uses memory at the end of the block that it has got.
1092
1093 29. Check for an overlong MARK name and give an error at compile time. The
1094 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1095
1096 30. JIT compiler update.
1097
1098 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1099 Rill for the patch.
1100
1101 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1102
1103 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1104
1105 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1106 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1107 was enabled.
1108
1109 35. Improve JIT code generation for greedy plus quantifier.
1110
1111 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1112 1 to "aa" instead of to an empty string. The bug affected repeated groups
1113 that could potentially match an empty string.
1114
1115 37. Optimizing single character iterators in JIT.
1116
1117 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1118 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1119 codepoints that are too big for the mode are faulted, and in a UTF mode,
1120 disallowed codepoints are also faulted.
1121
1122 39. If PCRE was compiled with UTF support, in three places in the DFA
1123 matcher there was code that should only have been obeyed in UTF mode, but
1124 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1125 processing when bytes with values greater than 127 were present. In 16-bit
1126 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1127 both cases the values are those that cannot be the first data item in a UTF
1128 character. The three items that might have provoked this were recursions,
1129 possessively repeated groups, and atomic groups.
1130
1131 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1132 and pcregrep, because some OS require shared objects to be explicitly
1133 passed to ld, causing the link step to fail if they are not.
1134
1135 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1136 mode, patterns that started with \h* or \R* might be incorrectly matched.
1137
1138
1139 Version 8.30 04-February-2012
1140 -----------------------------
1141
1142 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1143 name is defined in ctype.h.
1144
1145 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1146 only in quite long subpatterns.
1147
1148 3. Removed the function pcre_info(), which has been obsolete and deprecated
1149 since it was replaced by pcre_fullinfo() in February 2000.
1150
1151 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1152 match a (*MARK), and the match failed at the start of the subject, a
1153 reference to memory before the start of the subject could occur. This bug
1154 was introduced by fix 17 of release 8.21.
1155
1156 5. A reference to an unset group with zero minimum repetition was giving
1157 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1158 /(another)?(\1?)test/ matched against "hello world test". This bug was
1159 introduced in release 8.13.
1160
1161 6. Add support for 16-bit character strings (a large amount of work involving
1162 many changes and refactorings).
1163
1164 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1165 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1166 from a file.
1167
1168 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1169 rounding is not applied in this particular case).
1170
1171 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1172 if they appear, or are escaped, in patterns.
1173
1174 10. Get rid of a number of -Wunused-but-set-variable warnings.
1175
1176 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1177 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1178 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1179 also returns the mark "x". This bug applied to capturing parentheses,
1180 non-capturing parentheses, and atomic parentheses. It also applied to some
1181 assertions.
1182
1183 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1184 information out of configure.ac instead of relying on pcre.h.generic, which
1185 is not stored in the repository.
1186
1187 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1188 -lreadline.
1189
1190 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1191
1192 15. Applied Graycode's patch to put the top-level frame on the stack rather
1193 than the heap when not using the stack for recursion. This gives a
1194 performance improvement in many cases when recursion is not deep.
1195
1196 16. Experimental code added to "pcretest -C" to output the stack frame size.
1197
1198
1199 Version 8.21 12-Dec-2011
1200 ------------------------
1201
1202 1. Updating the JIT compiler.
1203
1204 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1205 are added as well.
1206
1207 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1208 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1209 calling _pcre_jit_exec. Some extra comments are added.
1210
1211 4. (*MARK) settings inside atomic groups that do not contain any capturing
1212 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1213 was introduced by change 18 for 8.20.
1214
1215 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1216 ECMA-262 standard.
1217
1218 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1219 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1220 This bug was probably introduced by change 9 of 8.13.
1221
1222 7. While fixing 6 above, I noticed that a number of other items were being
1223 incorrectly rejected as "not fixed length". This arose partly because newer
1224 opcodes had not been added to the fixed-length checking code. I have (a)
1225 corrected the bug and added tests for these items, and (b) arranged for an
1226 error to occur if an unknown opcode is encountered while checking for fixed
1227 length instead of just assuming "not fixed length". The items that were
1228 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1229 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1230 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1231
1232 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1233 being incorrectly compiled and would have given unpredicatble results.
1234
1235 9. A possessively repeated subpattern with minimum repeat count greater than
1236 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1237 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1238 the first (A) could occur when it should not.
1239
1240 10. Add a cast and remove a redundant test from the code.
1241
1242 11. JIT should use pcre_malloc/pcre_free for allocation.
1243
1244 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1245 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1246 is anything other than /usr, -L is still shown).
1247
1248 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1249
1250 14. Perl does not support \N without a following name in a [] class; PCRE now
1251 also gives an error.
1252
1253 15. If a forward reference was repeated with an upper limit of around 2000,
1254 it caused the error "internal error: overran compiling workspace". The
1255 maximum number of forward references (including repeats) was limited by the
1256 internal workspace, and dependent on the LINK_SIZE. The code has been
1257 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1258 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1259 of around 200,000 forward references. While doing this, I also speeded up
1260 the filling in of repeated forward references.
1261
1262 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1263 incorrectly expecting the subject to contain another "a" after the start.
1264
1265 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1266 in the match, the SKIP should be ignored. This was not happening; instead
1267 the SKIP was being treated as NOMATCH. For patterns such as
1268 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1269 tested.
1270
1271 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1272 now much more compatible with Perl, in particular in cases where the result
1273 is a non-match for a non-anchored pattern. For example, if
1274 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1275 "m", where previously it did not return a name. A side effect of this
1276 change is that for partial matches, the last encountered mark name is
1277 returned, as for non matches. A number of tests that were previously not
1278 Perl-compatible have been moved into the Perl-compatible test files. The
1279 refactoring has had the pleasing side effect of removing one argument from
1280 the match() function, thus reducing its stack requirements.
1281
1282 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1283 subsequent uses of /S (without +) incorrectly behaved like /S+.
1284
1285 21. Retrieve executable code size support for the JIT compiler and fixing
1286 some warnings.
1287
1288 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1289 not work when the shorter character appeared right at the end of the
1290 subject string.
1291
1292 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1293 systems.
1294
1295 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1296 output it when the /M option is used in pcretest.
1297
1298 25. The CheckMan script was not being included in the distribution. Also, added
1299 an explicit "perl" to run Perl scripts from the PrepareRelease script
1300 because this is reportedly needed in Windows.
1301
1302 26. If study data was being save in a file and studying had not found a set of
1303 "starts with" bytes for the pattern, the data written to the file (though
1304 never used) was taken from uninitialized memory and so caused valgrind to
1305 complain.
1306
1307 27. Updated RunTest.bat as provided by Sheri Pierce.
1308
1309 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1310
1311 29. Computation of memory usage for the table of capturing group names was
1312 giving an unnecessarily large value.
1313
1314
1315 Version 8.20 21-Oct-2011
1316 ------------------------
1317
1318 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1319 a POSIX class. After further experiments with Perl, which convinced me that
1320 Perl has bugs and confusions, a closing square bracket is no longer allowed
1321 in a POSIX name. This bug also affected patterns with classes that started
1322 with full stops.
1323
1324 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1325 captured substring, but while checking the failing first alternative,
1326 substring 1 is temporarily captured. If the output vector supplied to
1327 pcre_exec() was not big enough for this capture, the yield of the function
1328 was still zero ("insufficient space for captured substrings"). This cannot
1329 be totally fixed without adding another stack variable, which seems a lot
1330 of expense for a edge case. However, I have improved the situation in cases
1331 such as /(a)(b)x|abc/ matched against "abc", where the return code
1332 indicates that fewer than the maximum number of slots in the ovector have
1333 been set.
1334
1335 3. Related to (2) above: when there are more back references in a pattern than
1336 slots in the output vector, pcre_exec() uses temporary memory during
1337 matching, and copies in the captures as far as possible afterwards. It was
1338 using the entire output vector, but this conflicts with the specification
1339 that only 2/3 is used for passing back captured substrings. Now it uses
1340 only the first 2/3, for compatibility. This is, of course, another edge
1341 case.
1342
1343 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1344 main code base, and can be used by building with --enable-jit. When this is
1345 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1346 runtime --no-jit option is given.
1347
1348 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1349 ovector, the return from the function was zero, implying that there were
1350 other matches that did not fit. The correct "exactly full" value is now
1351 returned.
1352
1353 6. If a subpattern that was called recursively or as a subroutine contained
1354 (*PRUNE) or any other control that caused it to give a non-standard return,
1355 invalid errors such as "Error -26 (nested recursion at the same subject
1356 position)" or even infinite loops could occur.
1357
1358 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1359 computing the minimum length on reaching *ACCEPT, and so ended up with the
1360 wrong value of 1 rather than 0. Further investigation indicates that
1361 computing a minimum subject length in the presence of *ACCEPT is difficult
1362 (think back references, subroutine calls), and so I have changed the code
1363 so that no minimum is registered for a pattern that contains *ACCEPT.
1364
1365 8. If (*THEN) was present in the first (true) branch of a conditional group,
1366 it was not handled as intended. [But see 16 below.]
1367
1368 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1369 Sheri Pierce.
1370
1371 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1372 the first byte in a match must be "a".
1373
1374 11. Change 17 for 8.13 increased the recursion depth for patterns like
1375 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1376 pattern contains any instances of (*THEN). If it does not, the old
1377 optimizations are restored. It would be nice to do this on a per-group
1378 basis, but at the moment that is not feasible.
1379
1380 12. In some environments, the output of pcretest -C is CRLF terminated. This
1381 broke RunTest's code that checks for the link size. A single white space
1382 character after the value is now allowed for.
1383
1384 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1385 For "fr", it uses the Windows-specific input and output files.
1386
1387 14. If (*THEN) appeared in a group that was called recursively or as a
1388 subroutine, it did not work as intended. [But see next item.]
1389
1390 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1391 pattern fragments (but not containing any | characters). If A and B are
1392 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1393 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1394 D. In other words, Perl considers parentheses that do not contain any |
1395 characters to be part of a surrounding alternative, whereas PCRE was
1396 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1397 differently. PCRE now behaves in the same way as Perl, except in the case
1398 of subroutine/recursion calls such as (?1) which have in any case always
1399 been different (but PCRE had them first :-).
1400
1401 16. Related to 15 above: Perl does not treat the | in a conditional group as
1402 creating alternatives. Such a group is treated in the same way as an
1403 ordinary group without any | characters when processing (*THEN). PCRE has
1404 been changed to match Perl's behaviour.
1405
1406 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1407 RunGrepTest script failed.
1408
1409 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1410 inevitable for groups that contain captures, but it can lead to a lot of
1411 stack use in large patterns. The old behaviour has been restored for atomic
1412 groups that do not contain any capturing parentheses.
1413
1414 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1415 suppress the check for a minimum subject length at run time. (If it was
1416 given to pcre_exec() or pcre_dfa_exec() it did work.)
1417
1418 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1419 fail to work when decoding hex characters in data strings in EBCDIC
1420 environments.
1421
1422 21. It appears that in at least one Mac OS environment, the isxdigit() function
1423 is implemented as a macro that evaluates to its argument more than once,
1424 contravening the C 90 Standard (I haven't checked a later standard). There
1425 was an instance in pcretest which caused it to go wrong when processing
1426 \x{...} escapes in subject strings. The has been rewritten to avoid using
1427 things like p++ in the argument of isxdigit().
1428
1429
1430 Version 8.13 16-Aug-2011
1431 ------------------------
1432
1433 1. The Unicode data tables have been updated to Unicode 6.0.0.
1434
1435 2. Two minor typos in pcre_internal.h have been fixed.
1436
1437 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1438 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1439 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1440
1441 4. There were a number of related bugs in the code for matching backrefences
1442 caselessly in UTF-8 mode when codes for the characters concerned were
1443 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1444 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1445 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1446 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1447 2-byte code at the end of the subject (it thought there wasn't enough data
1448 left).
1449
1450 5. Comprehensive information about what went wrong is now returned by
1451 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1452 as the output vector has at least 2 elements. The offset of the start of
1453 the failing character and a reason code are placed in the vector.
1454
1455 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1456 now returned is for the first byte of the failing character, instead of the
1457 last byte inspected. This is an incompatible change, but I hope it is small
1458 enough not to be a problem. It makes the returned offset consistent with
1459 pcre_exec() and pcre_dfa_exec().
1460
1461 7. pcretest now gives a text phrase as well as the error number when
1462 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1463 failure, the offset and reason code are output.
1464
1465 8. When \R was used with a maximizing quantifier it failed to skip backwards
1466 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1467 back over a single character (\n). This seems wrong (because it treated the
1468 two characters as a single entity when going forwards), conflicts with the
1469 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1470 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1471 has been changed.
1472
1473 9. Some internal refactoring has changed the processing so that the handling
1474 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1475 time (the PCRE_DOTALL option was changed this way some time ago: version
1476 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1477 which was always a bit of a fudge. It also means that there is one less
1478 argument for the match() function, which reduces its stack requirements
1479 slightly. This change also fixes an incompatibility with Perl: the pattern
1480 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1481
1482 10. More internal refactoring has drastically reduced the number of recursive
1483 calls to match() for possessively repeated groups such as (abc)++ when
1484 using pcre_exec().
1485
1486 11. While implementing 10, a number of bugs in the handling of groups were
1487 discovered and fixed:
1488
1489 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1490 (a|)*(?1) gave a compile-time internal error.
1491 ((a|)+)+ did not notice that the outer group could match an empty string.
1492 (^a|^)+ was not marked as anchored.
1493 (.*a|.*)+ was not marked as matching at start or after a newline.
1494
1495 12. Yet more internal refactoring has removed another argument from the match()
1496 function. Special calls to this function are now indicated by setting a
1497 value in a variable in the "match data" data block.
1498
1499 13. Be more explicit in pcre_study() instead of relying on "default" for
1500 opcodes that mean there is no starting character; this means that when new
1501 ones are added and accidentally left out of pcre_study(), testing should
1502 pick them up.
1503
1504 14. The -s option of pcretest has been documented for ages as being an old
1505 synonym of -m (show memory usage). I have changed it to mean "force study
1506 for every regex", that is, assume /S for every regex. This is similar to -i
1507 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1508 using it. It makes it easier to run collections of tests with and without
1509 study enabled, and thereby test pcre_study() more easily. All the standard
1510 tests are now run with and without -s (but some patterns can be marked as
1511 "never study" - see 20 below).
1512
1513 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1514 restoration of the capturing data to the outer values was not happening
1515 correctly.
1516
1517 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1518 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1519 pattern had matched an empty string, and so incorrectly returned a no
1520 match.
1521
1522 17. There was optimizing code for the last branch of non-capturing parentheses,
1523 and also for the obeyed branch of a conditional subexpression, which used
1524 tail recursion to cut down on stack usage. Unfortunately, now that there is
1525 the possibility of (*THEN) occurring in these branches, tail recursion is
1526 no longer possible because the return has to be checked for (*THEN). These
1527 two optimizations have therefore been removed. [But see 8.20/11 above.]
1528
1529 18. If a pattern containing \R was studied, it was assumed that \R always
1530 matched two bytes, thus causing the minimum subject length to be
1531 incorrectly computed because \R can also match just one byte.
1532
1533 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1534 was incorrectly computed.
1535
1536 20. If /S is present twice on a test pattern in pcretest input, it now
1537 *disables* studying, thereby overriding the use of -s on the command line
1538 (see 14 above). This is necessary for one or two tests to keep the output
1539 identical in both cases.
1540
1541 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1542 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1543
1544 22. When an atomic group that contained a capturing parenthesis was
1545 successfully matched, but the branch in which it appeared failed, the
1546 capturing was not being forgotten if a higher numbered group was later
1547 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1548 group 1 to "a", when in fact it should be unset. This applied to multi-
1549 branched capturing and non-capturing groups, repeated or not, and also to
1550 positive assertions (capturing in negative assertions does not happen
1551 in PCRE) and also to nested atomic groups.
1552
1553 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1554 subject after a captured substring, to make it easier to tell which of a
1555 number of identical substrings has been captured.
1556
1557 24. The way atomic groups are processed by pcre_exec() has been changed so that
1558 if they are repeated, backtracking one repetition now resets captured
1559 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1560 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1561 "aaa". Previously, it would have been "a". As part of this code
1562 refactoring, the way recursive calls are handled has also been changed.
1563
1564 25. If an assertion condition captured any substrings, they were not passed
1565 back unless some other capturing happened later. For example, if
1566 (?(?=(a))a) was matched against "a", no capturing was returned.
1567
1568 26. When studying a pattern that contained subroutine calls or assertions,
1569 the code for finding the minimum length of a possible match was handling
1570 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1571 group 1 called group 2 while simultaneously a separate group 2 called group
1572 1). A stack overflow occurred in this case. I have fixed this by limiting
1573 the recursion depth to 10.
1574
1575 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1576 Fortmann. This supports explicit test numbers on the command line, and has
1577 argument validation and error reporting.
1578
1579 28. An instance of \X with an unlimited repeat could fail if at any point the
1580 first character it looked at was a mark character.
1581
1582 29. Some minor code refactoring concerning Unicode properties and scripts
1583 should reduce the stack requirement of match() slightly.
1584
1585 30. Added the '=' option to pcretest to check the setting of unused capturing
1586 slots at the end of the pattern, which are documented as being -1, but are
1587 not included in the return count.
1588
1589 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1590 compiled something random. Now it gives a compile-time error (as does
1591 Perl).
1592
1593 32. A *MARK encountered during the processing of a positive assertion is now
1594 recorded and passed back (compatible with Perl).
1595
1596 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1597 had alternative anchored branches, the search for a second match in a line
1598 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1599 matched the line "0102" twice. The same bug affected patterns that started
1600 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1601 twice.
1602
1603 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1604 does, and because of capturing effects, quantifying parenthesized
1605 assertions may at times be useful. Quantifiers are now allowed for
1606 parenthesized assertions.
1607
1608 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1609
1610 36. \g was being checked for fancy things in a character class, when it should
1611 just be a literal "g".
1612
1613 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1614 appearance of a nested POSIX class supersedes an apparent external class.
1615 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1616 unescaped square brackets may also appear as part of class names. For
1617 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1618 more like Perl. (But see 8.20/1 above.)
1619
1620 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1621 was because it thought it was \N{name}, which is not supported).
1622
1623 39. Add minix to OS list not supporting the -S option in pcretest.
1624
1625 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1626 cannot analyze patterns in sufficient detail to catch mutual recursions
1627 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1628 subgroup is called recursively as a subpattern for a second time at the
1629 same position in the subject string. In previous releases this might have
1630 been caught by the recursion limit, or it might have run out of stack.
1631
1632 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1633 happen only once. PCRE was, however incorrectly giving a compile time error
1634 "recursive call could loop indefinitely" because it cannot analyze the
1635 pattern in sufficient detail. The compile time test no longer happens when
1636 PCRE is compiling a conditional subpattern, but actual runaway loops are
1637 now caught at runtime (see 40 above).
1638
1639 42. It seems that Perl allows any characters other than a closing parenthesis
1640 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1641 has been changed to be the same.
1642
1643 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1644 as not to get warnings when autogen.sh is called. Also changed
1645 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1646
1647 44. To help people who use pcregrep to scan files containing exceedingly long
1648 lines, the following changes have been made:
1649
1650 (a) The default value of the buffer size parameter has been increased from
1651 8K to 20K. (The actual buffer used is three times this size.)
1652
1653 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1654 PCRE is built.
1655
1656 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1657 to be set at run time.
1658
1659 (d) Numerical values in pcregrep options can be followed by K or M, for
1660 example --buffer-size=50K.
1661
1662 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1663 given and the return code is set to 2.
1664
1665 45. Add a pointer to the latest mark to the callout data block.
1666
1667 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1668 partial match of an empty string instead of no match. This was specific to
1669 the use of ".".
1670
1671 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1672 complete match instead of a partial match. This bug was dependent on both
1673 the PCRE_UTF8 and PCRE_DOTALL options being set.
1674
1675 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1676 starting byte set, because \b was not being ignored.
1677
1678
1679 Version 8.12 15-Jan-2011
1680 ------------------------
1681
1682 1. Fixed some typos in the markup of the man pages, and wrote a script that
1683 checks for such things as part of the documentation building process.
1684
1685 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1686 --match-limit and --recursion-limit options (added for 8.11). In
1687 particular, this made one of the standard tests fail. (The integer value
1688 went into the wrong half of a long int.)
1689
1690 3. If the --colour option was given to pcregrep with -v (invert match), it
1691 did strange things, either producing crazy output, or crashing. It should,
1692 of course, ignore a request for colour when reporting lines that do not
1693 match.
1694
1695 4. Another pcregrep bug caused similar problems if --colour was specified with
1696 -M (multiline) and the pattern match finished with a line ending.
1697
1698 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1699 matched in multiline mode, the following line was shown as part of the
1700 match. This seems wrong, so I have changed it.
1701
1702 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1703 the check for further matches in the same line (so they could be coloured)
1704 to overrun the end of the current line. If another match was found, it was
1705 incorrectly shown (and then shown again when found in the next line).
1706
1707 7. If pcregrep was compiled under Windows, there was a reference to the
1708 function pcregrep_exit() before it was defined. I am assuming this was
1709 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1710 reported by a user. I've moved the definition above the reference.
1711
1712
1713 Version 8.11 10-Dec-2010
1714 ------------------------
1715
1716 1. (*THEN) was not working properly if there were untried alternatives prior
1717 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1718 backtracked to try for "b" instead of moving to the next alternative branch
1719 at the same level (in this case, to look for "c"). The Perl documentation
1720 is clear that when (*THEN) is backtracked onto, it goes to the "next
1721 alternative in the innermost enclosing group".
1722
1723 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1724 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1725 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1726 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1727 (*THEN).
1728
1729 3. If \s appeared in a character class, it removed the VT character from
1730 the class, even if it had been included by some previous item, for example
1731 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1732 of \s, but is part of the POSIX "space" class.)
1733
1734 4. A partial match never returns an empty string (because you can always
1735 match an empty string at the end of the subject); however the checking for
1736 an empty string was starting at the "start of match" point. This has been
1737 changed to the "earliest inspected character" point, because the returned
1738 data for a partial match starts at this character. This means that, for
1739 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1740 (previously it gave "no match").
1741
1742 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1743 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1744 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1745 has an implication that the given string is incomplete (because a partial
1746 match is preferred over a full match). For this reason, these items now
1747 give a partial match in this situation. [Aside: previously, the one case
1748 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1749 match rather than a full match, which was wrong by the old rules, but is
1750 now correct.]
1751
1752 6. There was a bug in the handling of #-introduced comments, recognized when
1753 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1754 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1755 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1756 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1757 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1758 places in pcre_compile().
1759
1760 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1761 comments when looking ahead for named forward references to subpatterns,
1762 the only newline sequence it recognized was NL. It now handles newlines
1763 according to the set newline convention.
1764
1765 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1766 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1767 cater for a lack of strerror(). These oversights have been fixed.
1768
1769 9. Added --match-limit and --recursion-limit to pcregrep.
1770
1771 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1772
1773 11. When the -o option was used, pcregrep was setting a return code of 1, even
1774 when matches were found, and --line-buffered was not being honoured.
1775
1776 12. Added an optional parentheses number to the -o and --only-matching options
1777 of pcregrep.
1778
1779 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1780 can match an empty string. The code to do it in pcretest and pcredemo
1781 needed fixing:
1782
1783 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1784 only one byte after an empty string match just before CRLF (this case
1785 just got forgotten; "any" and "anycrlf" were OK).
1786
1787 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1788 mode when an empty string match preceded an ASCII character followed by
1789 a non-ASCII character. (The code for advancing by one character rather
1790 than one byte was nonsense.)
1791
1792 (c) The pcredemo.c sample program did not have any code at all to handle
1793 the cases when CRLF is a valid newline sequence.
1794
1795 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1796 as a starting offset was within the subject string. There is now a new
1797 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1798 negative or greater than the length of the string. In order to test this,
1799 pcretest is extended to allow the setting of negative starting offsets.
1800
1801 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1802 starting offset points to the beginning of a UTF-8 character was
1803 unnecessarily clumsy. I tidied it up.
1804
1805 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1806 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1807
1808 17. Nobody had reported that the --include_dir option, which was added in
1809 release 7.7 should have been called --include-dir (hyphen, not underscore)
1810 for compatibility with GNU grep. I have changed it to --include-dir, but
1811 left --include_dir as an undocumented synonym, and the same for
1812 --exclude-dir, though that is not available in GNU grep, at least as of
1813 release 2.5.4.
1814
1815 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1816 characters from a string of bytes) have been redefined so as not to use
1817 loops, in order to improve performance in some environments. At the same
1818 time, I abstracted some of the common code into auxiliary macros to save
1819 repetition (this should not affect the compiled code).
1820
1821 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1822 compile-time error is now given if \c is not followed by an ASCII
1823 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1824 different, and any byte value is allowed.)
1825
1826 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1827 START_OPTIMIZE option, which is now allowed at compile time - but just
1828 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1829 to pcregrep and other applications that have no direct access to PCRE
1830 options. The new /Y option in pcretest sets this option when calling
1831 pcre_compile().
1832
1833 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1834 back references. Groups containing recursive back references were forced to
1835 be atomic by that change, but in the case of named groups, the amount of
1836 memory required was incorrectly computed, leading to "Failed: internal
1837 error: code overflow". This has been fixed.
1838
1839 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1840 pcretest.c, to avoid build problems in some Borland environments.
1841
1842
1843 Version 8.10 25-Jun-2010
1844 ------------------------
1845
1846 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1847 THEN.
1848
1849 2. (*ACCEPT) was not working when inside an atomic group.
1850
1851 3. Inside a character class, \B is treated as a literal by default, but
1852 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1853 causes the error). The code is unchanged, but I tidied the documentation.
1854
1855 4. Inside a character class, PCRE always treated \R and \X as literals,
1856 whereas Perl faults them if its -w option is set. I have changed PCRE so
1857 that it faults them when PCRE_EXTRA is set.
1858
1859 5. Added support for \N, which always matches any character other than
1860 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1861
1862 6. When compiling pcregrep with newer versions of gcc which may have
1863 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1864 declared with attribute warn_unused_result" were given. Just casting the
1865 result to (void) does not stop the warnings; a more elaborate fudge is
1866 needed. I've used a macro to implement this.
1867
1868 7. Minor change to pcretest.c to avoid a compiler warning.
1869
1870 8. Added four artifical Unicode properties to help with an option to make
1871 \s etc use properties (see next item). The new properties are: Xan
1872 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1873
1874 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1875 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1876 this option. Modified pcretest to add /W to test this facility. Added
1877 REG_UCP to make it available via the POSIX interface.
1878
1879 10. Added --line-buffered to pcregrep.
1880
1881 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1882 studied, and the match started with a letter with a code point greater than
1883 127 whose first byte was different to the first byte of the other case of
1884 the letter, the other case of this starting letter was not recognized
1885 (#976).
1886
1887 12. If a pattern that was studied started with a repeated Unicode property
1888 test, for example, \p{Nd}+, there was the theoretical possibility of
1889 setting up an incorrect bitmap of starting bytes, but fortunately it could
1890 not have actually happened in practice until change 8 above was made (it
1891 added property types that matched character-matching opcodes).
1892
1893 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1894 possible starting bytes for non-anchored patterns.
1895
1896 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1897 \R, and also a number of cases that involve Unicode properties, both
1898 explicit and implicit when PCRE_UCP is set.
1899
1900 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1901 input, it could crash or give wrong results if characters with values
1902 greater than 0xc0 were present in the subject string. (Detail: it assumed
1903 UTF-8 input when processing these items.)
1904
1905 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1906 size_t is 64-bit (#991).
1907
1908 17. Added a check for running out of memory when PCRE is compiled with
1909 --disable-stack-for-recursion (#990).
1910
1911 18. If the last data line in a file for pcretest does not have a newline on
1912 the end, a newline was missing in the output.
1913
1914 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1915 less than 128) in its various bitmaps. However, there is a facility for
1916 generating tables according to the current locale when PCRE is compiled. It
1917 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1918 characters, are recognized by isspace() and therefore were getting set in
1919 these tables, and indeed these tables seem to approximate to ISO 8859. This
1920 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1921 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1922 which of course cannot start UTF-8 characters. I have changed the code so
1923 that only real ASCII characters (less than 128) and the correct starting
1924 bytes for UTF-8 encodings are set for characters greater than 127 when in
1925 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1926 altogether.)
1927
1928 20. Added the /T option to pcretest so as to be able to run tests with non-
1929 standard character tables, thus making it possible to include the tests
1930 used for 19 above in the standard set of tests.
1931
1932 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1933 reference to a subpattern the other side of a comment that contains an
1934 opening parenthesis caused either an internal compiling error, or a
1935 reference to the wrong subpattern.
1936
1937
1938 Version 8.02 19-Mar-2010
1939 ------------------------
1940
1941 1. The Unicode data tables have been updated to Unicode 5.2.0.
1942
1943 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1944 configured.
1945
1946 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1947 original author of that file, following a query about its status.
1948
1949 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1950 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1951
1952 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1953 quantifier applied to a forward-referencing subroutine call, could compile
1954 incorrect code or give the error "internal error: previously-checked
1955 referenced subpattern not found".
1956
1957 6. Both MS Visual Studio and Symbian OS have problems with initializing
1958 variables to point to external functions. For these systems, therefore,
1959 pcre_malloc etc. are now initialized to local functions that call the
1960 relevant global functions.
1961
1962 7. There were two entries missing in the vectors called coptable and poptable
1963 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1964 I've fixed the data, and added a kludgy way of testing at compile time that
1965 the lengths are correct (equal to the number of opcodes).
1966
1967 8. Following on from 7, I added a similar kludge to check the length of the
1968 eint vector in pcreposix.c.
1969
1970 9. Error texts for pcre_compile() are held as one long string to avoid too
1971 much relocation at load time. To find a text, the string is searched,
1972 counting zeros. There was no check for running off the end of the string,
1973 which could happen if a new error number was added without updating the
1974 string.
1975
1976 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1977
1978 11. \K was not working if it appeared in an atomic group or in a group that
1979 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1980 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1981 the assertion is positive, but not if it is negative.
1982
1983 12. Change 11 fortuitously reduced the size of the stack frame used in the
1984 "match()" function of pcre_exec.c by one pointer. Forthcoming
1985 implementation of support for (*MARK) will need an extra pointer on the
1986 stack; I have reserved it now, so that the stack frame size does not
1987 decrease.
1988
1989 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1990 item in branch that calls a recursion is a subroutine call - as in the
1991 second branch in the above example - was incorrectly given the compile-
1992 time error "recursive call could loop indefinitely" because pcre_compile()
1993 was not correctly checking the subroutine for matching a non-empty string.
1994
1995 14. The checks for overrunning compiling workspace could trigger after an
1996 overrun had occurred. This is a "should never occur" error, but it can be
1997 triggered by pathological patterns such as hundreds of nested parentheses.
1998 The checks now trigger 100 bytes before the end of the workspace.
1999
2000 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2001
2002
2003 Version 8.01 19-Jan-2010
2004 ------------------------
2005
2006 1. If a pattern contained a conditional subpattern with only one branch (in
2007 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2008 computed the wrong minimum data length (which is of course zero for such
2009 subpatterns). This could cause incorrect "no match" results.
2010
2011 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2012 the pattern is reset in the first branch, pcre_compile() failed with
2013 "internal error: code overflow at offset...". This happened only when
2014 the reset was to the original external option setting. (An optimization
2015 abstracts leading options settings into an external setting, which was the
2016 cause of this.)
2017
2018 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2019 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2020 assertion pattern did not match (meaning that the assertion was true), it
2021 was incorrectly treated as false if the SKIP had been reached during the
2022 matching. This also applied to assertions used as conditions.
2023
2024 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2025 assertion subpattern, including such a pattern used as a condition,
2026 unpredictable results occurred, instead of the error return
2027 PCRE_ERROR_DFA_UITEM.
2028
2029 5. The C++ GlobalReplace function was not working like Perl for the special
2030 situation when an empty string is matched. It now does the fancy magic
2031 stuff that is necessary.
2032
2033 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2034 removed. (These were left over from very, very early versions of PCRE.)
2035
2036 7. Some cosmetic changes to the code to make life easier when compiling it
2037 as part of something else:
2038
2039 (a) Change DEBUG to PCRE_DEBUG.
2040
2041 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2042 called "current" as "current_branch", to prevent a collision with the
2043 Linux macro when compiled as a kernel module.
2044
2045 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2046 prevent a collision with the Linux macro when compiled as a kernel
2047 module.
2048
2049 8. In pcre_compile() there are some checks for integer overflows that used to
2050 cast potentially large values to (double). This has been changed to that
2051 when building, a check for int64_t is made, and if it is found, it is used
2052 instead, thus avoiding the use of floating point arithmetic. (There is no
2053 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2054 double.
2055
2056 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2057 2005 (difference between two addresses compared to an unsigned value).
2058
2059 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2060 custom one, because of the following reported problem in Windows:
2061
2062 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2063 under Win32.
2064 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2065 therefore missing the function definition.
2066 - The compiler thus generates a "C" signature for the test function.
2067 - The linker fails to find the "C" function.
2068 - PCRE fails to configure if asked to do so against libbz2.
2069
2070 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2071 messages were output:
2072
2073 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2074 rerunning libtoolize, to keep the correct libtool macros in-tree.
2075 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2076
2077 I have done both of these things.
2078
2079 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2080 most of the time, it *can* run out if it is given a pattern that contains a
2081 runaway infinite recursion. I updated the discussion in the pcrestack man
2082 page.
2083
2084 13. Now that we have gone to the x.xx style of version numbers, the minor
2085 version may start with zero. Using 08 or 09 is a bad idea because users
2086 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2087 interpreted as invalid octal numbers. I've updated the previous comment in
2088 configure.ac, and also added a check that gives an error if 08 or 09 are
2089 used.
2090
2091 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2092 causing partial matching to fail when the end of the subject matched \W
2093 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2094
2095 15. There were some discrepancies between the declarations in pcre_internal.h
2096 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2097 their definitions. The declarations used "const uschar *" and the
2098 definitions used USPTR. Even though USPTR is normally defined as "const
2099 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2100 reported that: "This difference in casting confuses some C++ compilers, for
2101 example, SunCC recognizes above declarations as different functions and
2102 generates broken code for hbpcre." I have changed the declarations to use
2103 USPTR.
2104
2105 16. GNU libtool is named differently on some systems. The autogen.sh script now
2106 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2107 (FreeBSD).
2108
2109 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2110 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2111 comment: "Figure out how to create a longlong from a string: strtoll and
2112 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2113 instance, but it only takes 2 args instead of 3!"
2114
2115 18. A subtle bug concerned with back references has been fixed by a change of
2116 specification, with a corresponding code fix. A pattern such as
2117 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2118 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2119 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2120 same bug. Such groups have to be quantified to be useful, or contained
2121 inside another quantified group. (If there's no repetition, the reference
2122 can never match.) The problem arises because, having left the group and
2123 moved on to the rest of the pattern, a later failure that backtracks into
2124 the group uses the captured value from the final iteration of the group
2125 rather than the correct earlier one. I have fixed this in PCRE by forcing
2126 any group that contains a reference to itself to be an atomic group; that
2127 is, there cannot be any backtracking into it once it has completed. This is
2128 similar to recursive and subroutine calls.
2129
2130
2131 Version 8.00 19-Oct-09
2132 ----------------------
2133
2134 1. The table for translating pcre_compile() error codes into POSIX error codes
2135 was out-of-date, and there was no check on the pcre_compile() error code
2136 being within the table. This could lead to an OK return being given in
2137 error.
2138
2139 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2140 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2141 in a Windows environment.
2142
2143 3. The pcregrep --count option prints the count for each file even when it is
2144 zero, as does GNU grep. However, pcregrep was also printing all files when
2145 --files-with-matches was added. Now, when both options are given, it prints
2146 counts only for those files that have at least one match. (GNU grep just
2147 prints the file name in this circumstance, but including the count seems
2148 more useful - otherwise, why use --count?) Also ensured that the
2149 combination -clh just lists non-zero counts, with no names.
2150
2151 4. The long form of the pcregrep -F option was incorrectly implemented as
2152 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2153 but it seems right to fix it, and I didn't think it was worth preserving
2154 the old behaviour.
2155
2156 5. The command line items --regex=pattern and --regexp=pattern were not
2157 recognized by pcregrep, which required --regex pattern or --regexp pattern
2158 (with a space rather than an '='). The man page documented the '=' forms,
2159 which are compatible with GNU grep; these now work.
2160
2161 6. No libpcreposix.pc file was created for pkg-config; there was just
2162 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2163
2164 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2165 when UCP support is not needed, by modifying the Python script that
2166 generates it from Unicode data files. This should not matter if the module
2167 is correctly used as a library, but I received one complaint about 50K of
2168 unwanted data. My guess is that the person linked everything into his
2169 program rather than using a library. Anyway, it does no harm.
2170
2171 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2172 was a minimum greater than 1 for a wide character in a possessive
2173 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2174 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2175 character. Chaos in the form of incorrect output or a compiling loop could
2176 result.
2177
2178 9. The restrictions on what a pattern can contain when partial matching is
2179 requested for pcre_exec() have been removed. All patterns can now be
2180 partially matched by this function. In addition, if there are at least two
2181 slots in the offset vector, the offset of the earliest inspected character
2182 for the match and the offset of the end of the subject are set in them when
2183 PCRE_ERROR_PARTIAL is returned.
2184
2185 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2186 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2187 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2188 and may be more useful for multi-segment matching.
2189
2190 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2191 used to be given if ever the end of the subject was reached; now it is
2192 given only if matching could not proceed because another character was
2193 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2194 string "Z", which now yields "no match" instead of "partial match". In the
2195 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2196 final character ended with (*FAIL).
2197
2198 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2199 if the pattern had a "must contain" character that was already found in the
2200 earlier partial match, unless partial matching was again requested. For
2201 example, with the pattern /dog.(body)?/, the "must contain" character is
2202 "g". If the first part-match was for the string "dog", restarting with
2203 "sbody" failed. This bug has been fixed.
2204
2205 13. The string returned by pcre_dfa_exec() after a partial match has been
2206 changed so that it starts at the first inspected character rather than the
2207 first character of the match. This makes a difference only if the pattern
2208 starts with a lookbehind assertion or \b or \B (\K is not supported by
2209 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2210 matching functions compatible, and I think it's the right thing to do.
2211
2212 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2213 so that the demonstration program is easily available in environments where
2214 PCRE has not been installed from source.
2215
2216 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2217 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2218 library.
2219
2220 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2221 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2222 is not the first non-POSIX option to be added. Clearly some people find
2223 these options useful.
2224
2225 17. If a caller to the POSIX matching function regexec() passes a non-zero
2226 value for nmatch with a NULL value for pmatch, the value of
2227 nmatch is forced to zero.
2228
2229 18. RunGrepTest did not have a test for the availability of the -u option of
2230 the diff command, as RunTest does. It now checks in the same way as
2231 RunTest, and also checks for the -b option.
2232
2233 19. If an odd number of negated classes containing just a single character
2234 interposed, within parentheses, between a forward reference to a named
2235 subpattern and the definition of the subpattern, compilation crashed with
2236 an internal error, complaining that it could not find the referenced
2237 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2238 [The bug was that it was starting one character too far in when skipping
2239 over the character class, thus treating the ] as data rather than
2240 terminating the class. This meant it could skip too much.]
2241
2242 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2243 /g option in pcretest when the pattern contains \K, which makes it possible
2244 to have an empty string match not at the start, even when the pattern is
2245 anchored. Updated pcretest and pcredemo to use this option.
2246
2247 21. If the maximum number of capturing subpatterns in a recursion was greater
2248 than the maximum at the outer level, the higher number was returned, but
2249 with unset values at the outer level. The correct (outer level) value is
2250 now given.
2251
2252 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2253 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2254 make it do so. The string so far is captured, making this feature
2255 compatible with Perl.
2256
2257 23. The tests have been re-organized, adding tests 11 and 12, to make it
2258 possible to check the Perl 5.10 features against Perl 5.10.
2259
2260 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2261 pattern matches a fixed length string. PCRE did not allow this; now it
2262 does. Neither allows recursion.
2263
2264 25. I finally figured out how to implement a request to provide the minimum
2265 length of subject string that was needed in order to match a given pattern.
2266 (It was back references and recursion that I had previously got hung up
2267 on.) This code has now been added to pcre_study(); it finds a lower bound
2268 to the length of subject needed. It is not necessarily the greatest lower
2269 bound, but using it to avoid searching strings that are too short does give
2270 some useful speed-ups. The value is available to calling programs via
2271 pcre_fullinfo().
2272
2273 26. While implementing 25, I discovered to my embarrassment that pcretest had
2274 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2275 study optimizations had never been tested with that matching function.
2276 Oops. What is worse, even when it was passed study data, there was a bug in
2277 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2278 were also very few tests of studied patterns with pcre_dfa_exec().
2279
2280 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2281 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2282 on the other side of the coin, they are no longer allowed to have different
2283 names, because these cannot be distinguished in PCRE, and this has caused
2284 confusion. (This is a difference from Perl.)
2285
2286 28. When duplicate subpattern names are present (necessarily with different
2287 numbers, as required by 27 above), and a test is made by name in a
2288 conditional pattern, either for a subpattern having been matched, or for
2289 recursion in such a pattern, all the associated numbered subpatterns are
2290 tested, and the overall condition is true if the condition is true for any
2291 one of them. This is the way Perl works, and is also more like the way
2292 testing by number works.
2293
2294
2295 Version 7.9 11-Apr-09
2296 ---------------------
2297
2298 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2299 (pcretest), all targets were linked against these libraries. This included
2300 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2301 libraries. This caused unwanted dependencies to be created. This problem
2302 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2303 pcretest is linked with readline.
2304
2305 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2306 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2307 moved outside it again, because FALSE and TRUE are already defined in AIX,
2308 but BOOL is not.
2309
2310 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2311 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2312
2313 4. The pcregrep documentation said spaces were inserted as well as colons (or
2314 hyphens) following file names and line numbers when outputting matching
2315 lines. This is not true; no spaces are inserted. I have also clarified the
2316 wording for the --colour (or --color) option.
2317
2318 5. In pcregrep, when --colour was used with -o, the list of matching strings
2319 was not coloured; this is different to GNU grep, so I have changed it to be
2320 the same.
2321
2322 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2323 each matching line was coloured. Now it goes on to look for further matches
2324 of any of the test patterns, which is the same behaviour as GNU grep.
2325
2326 7. A pattern that could match an empty string could cause pcregrep to loop; it
2327 doesn't make sense to accept an empty string match in pcregrep, so I have
2328 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2329 seems to be how GNU grep behaves. [But see later change 40 for release
2330 8.33.]
2331
2332 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2333 start or after a newline", because the conditional assertion was not being
2334 correctly handled. The rule now is that both the assertion and what follows
2335 in the first alternative must satisfy the test.
2336
2337 9. If auto-callout was enabled in a pattern with a conditional group whose
2338 condition was an assertion, PCRE could crash during matching, both with
2339 pcre_exec() and pcre_dfa_exec().
2340
2341 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2342 used for matching.
2343
2344 11. Unicode property support in character classes was not working for
2345 characters (bytes) greater than 127 when not in UTF-8 mode.
2346
2347 12. Added the -M command line option to pcretest.
2348
2349 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2350
2351 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2352
2353 16. Added comments and documentation about mis-use of no_arg in the C++
2354 wrapper.
2355
2356 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2357 from Martin Jerabek that uses macro names for all relevant character and
2358 string constants.
2359
2360 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2361 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2362 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2363 these, but not everybody uses configure.
2364
2365 19. A conditional group that had only one branch was not being correctly
2366 recognized as an item that could match an empty string. This meant that an
2367 enclosing group might also not be so recognized, causing infinite looping
2368 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2369 with the subject "ab", where knowledge that the repeated group can match
2370 nothing is needed in order to break the loop.
2371
2372 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2373 exec(), but without supplying a callout function, matching went wrong.
2374
2375 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2376 leak if the size of the offset vector was greater than 30. When the vector
2377 is smaller, the saved offsets during recursion go onto a local stack
2378 vector, but for larger vectors malloc() is used. It was failing to free
2379 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2380 error, in fact).
2381
2382 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2383 heapframe that is used only when UTF-8 support is enabled. This caused no
2384 problem, but was untidy.
2385
2386 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2387 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2388 included within another project.
2389
2390 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2391 slightly modified by me:
2392
2393 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2394 not building pcregrep.
2395
2396 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2397 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2398
2399 25. Forward references, both numeric and by name, in patterns that made use of
2400 duplicate group numbers, could behave incorrectly or give incorrect errors,
2401 because when scanning forward to find the reference group, PCRE was not
2402 taking into account the duplicate group numbers. A pattern such as
2403 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2404
2405 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2406 the feature of a custom pointer more persuasive (as requested by a user).
2407
2408 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2409 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2410 pre-defined.
2411
2412 28. Added support for (*UTF8) at the start of a pattern.
2413
2414 29. Arrange for flags added by the "release type" setting in CMake to be shown
2415 in the configuration summary.
2416
2417
2418 Version 7.8 05-Sep-08
2419 ---------------------
2420
2421 1. Replaced UCP searching code with optimized version as implemented for Ad
2422 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2423 stage table and inline lookup instead of a function, giving speed ups of 2
2424 to 5 times on some simple patterns that I tested. Permission was given to
2425 distribute the MultiStage2.py script that generates the tables (it's not in
2426 the tarball, but is in the Subversion repository).
2427
2428 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2429 scripts.
2430
2431 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2432 a group with a zero qualifier. The result of the study could be incorrect,
2433 or the function might crash, depending on the pattern.
2434
2435 4. Caseless matching was not working for non-ASCII characters in back
2436 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2437 It now works when Unicode Property Support is available.
2438
2439 5. In pcretest, an escape such as \x{de} in the data was always generating
2440 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2441 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2442 truncation.
2443
2444 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2445
2446 7. Added two (int) casts to pcregrep when printing the difference of two
2447 pointers, in case they are 64-bit values.
2448
2449 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2450 test 2 if it fails.
2451
2452 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2453 and a #define of that name to empty if it is not externally set. This is to
2454 allow users of MSVC to set it if necessary.
2455
2456 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2457 the convenience functions in the pcre_get.c source file.
2458
2459 11. An option change at the start of a pattern that had top-level alternatives
2460 could cause overwriting and/or a crash. This command provoked a crash in
2461 some environments:
2462
2463 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2464
2465 This potential security problem was recorded as CVE-2008-2371.
2466
2467 12. For a pattern where the match had to start at the beginning or immediately
2468 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2469 pcre_dfa_exec() could read past the end of the passed subject if there was
2470 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2471 pcretest so that it places the subject at the end of its malloc-ed buffer.
2472
2473 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2474 exec() might read past the end of the data buffer in UTF-8 mode.
2475
2476 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2477 the data contained the byte 0x85 as part of a UTF-8 character within its
2478 first line. This applied both to normal and DFA matching.
2479
2480 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2481 /^[^d]*?$/8 failed to match "abc".
2482
2483 16. Added a missing copyright notice to pcrecpp_internal.h.
2484
2485 17. Make it more clear in the documentation that values returned from
2486 pcre_exec() in ovector are byte offsets, not character counts.
2487
2488 18. Tidied a few places to stop certain compilers from issuing warnings.
2489
2490 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2491 supplied by Stefan Weber. I made a further small update for 7.8 because
2492 there is a change of source arrangements: the pcre_searchfuncs.c module is
2493 replaced by pcre_ucd.c.
2494
2495
2496 Version 7.7 07-May-08
2497 ---------------------
2498
2499 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2500 a string to a long long, pretend we don't even have a long long." This is
2501 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2502
2503 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2504 pre-7.6 versions, which defined a global no_arg variable instead of putting
2505 it in the RE class. (See also #8 below.)
2506
2507 3. Remove a line of dead code, identified by coverity and reported by Nuno
2508 Lopes.
2509
2510 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2511
2512 (1) The include/exclude patterns were being applied to the whole pathnames
2513 of files, instead of just to the final components.
2514
2515 (2) If there was more than one level of directory, the subdirectories were
2516 skipped unless they satisfied the include/exclude conditions. This is
2517 inconsistent with GNU grep (and could even be seen as contrary to the
2518 pcregrep specification - which I improved to make it absolutely clear).
2519 The action now is always to scan all levels of directory, and just
2520 apply the include/exclude patterns to regular files.
2521
2522 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2523 --exclude_dir in the tests to avoid scanning .svn directories.
2524
2525 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2526 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2527 doesn't support NULs in patterns.
2528
2529 7. Added some missing "const"s to declarations of static tables in
2530 pcre_compile.c and pcre_dfa_exec.c.
2531
2532 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2533 caused by fix #2 above. (Subsequently also a second patch to fix the
2534 first patch. And a third patch - this was a messy problem.)
2535
2536 9. Applied Craig's patch to remove the use of push_back().
2537
2538 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2539 matching function regexec().
2540
2541 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2542 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2543 references. PCRE supports relative numbers with this syntax (I don't think
2544 Oniguruma does).
2545
2546 12. Previously, a group with a zero repeat such as (...){0} was completely
2547 omitted from the compiled regex. However, this means that if the group
2548 was called as a subroutine from elsewhere in the pattern, things went wrong
2549 (an internal error was given). Such groups are now left in the compiled
2550 pattern, with a new opcode that causes them to be skipped at execution
2551 time.
2552
2553 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2554 to the way PCRE behaves:
2555
2556 (a) A lone ] character is dis-allowed (Perl treats it as data).
2557
2558 (b) A back reference to an unmatched subpattern matches an empty string
2559 (Perl fails the current match path).
2560
2561 (c) A data ] in a character class must be notated as \] because if the
2562 first data character in a class is ], it defines an empty class. (In
2563 Perl it is not possible to have an empty class.) The empty class []
2564 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2565 The negative empty class [^] matches any one character, independently
2566 of the DOTALL setting.
2567
2568 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2569 non-existent subpattern following a character class starting with ']' and
2570 containing () gave an internal compiling error instead of "reference to
2571 non-existent subpattern". Fortunately, when the pattern did exist, the
2572 compiled code was correct. (When scanning forwards to check for the
2573 existence of the subpattern, it was treating the data ']' as terminating
2574 the class, so got the count wrong. When actually compiling, the reference
2575 was subsequently set up correctly.)
2576
2577 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2578 it was being rejected as not supported by pcre_dfa_exec(), even though
2579 other assertions are supported. I have made pcre_dfa_exec() support
2580 (*FAIL).
2581
2582 16. The implementation of 13c above involved the invention of a new opcode,
2583 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2584 cannot be changed at match time, I realized I could make a small
2585 improvement to matching performance by compiling OP_ALLANY instead of
2586 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2587 on the OP_ANY path.
2588
2589 17. Compiling pcretest on Windows with readline support failed without the
2590 following two fixes: (1) Make the unistd.h include conditional on
2591 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2592
2593 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2594 ncurses library to be included for pcretest when ReadLine support is
2595 requested, but also to allow for it to be overridden. This patch came from
2596 Daniel Bergström.
2597
2598 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2599 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2600 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2601 spotting this.
2602
2603
2604 Version 7.6 28-Jan-08
2605 ---------------------
2606
2607 1. A character class containing a very large number of characters with
2608 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2609 overflow.
2610
2611 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2612 HAVE_LONG_LONG is not defined.
2613
2614 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2615 bring them up to date and include new features. This patch includes:
2616
2617 - Fixed PH's badly added libz and libbz2 support.
2618 - Fixed a problem with static linking.
2619 - Added pcredemo. [But later removed - see 7 below.]
2620 - Fixed dftables problem and added an option.
2621 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2622 HAVE_LONG_LONG.
2623 - Added readline support for pcretest.
2624 - Added an listing of the option settings after cmake has run.
2625
2626 4. A user submitted a patch to Makefile that makes it easy to create
2627 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2628 Makefile.am that cause it to include this special target, without
2629 affecting anything else. Note that the same mingw target plus all
2630 the other distribution libraries and programs are now supported
2631 when configuring with CMake (see 6 below) instead of with
2632 Configure/Make.
2633
2634 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2635 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2636 exported in the Windows port". It has not yet been confirmed that the patch
2637 solves the problem, but it does no harm.
2638
2639 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2640 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2641 with CMake, and also correct the comment about stack recursion.
2642
2643 7. Remove the automatic building of pcredemo from the ./configure system and
2644 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2645 of a program that users should build themselves after PCRE is installed, so
2646 building it automatically is not really right. What is more, it gave
2647 trouble in some build environments.
2648
2649 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2650
2651
2652 Version 7.5 10-Jan-08
2653 ---------------------
2654
2655 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2656 values in parens when parsing an RE using the C++ wrapper."
2657
2658 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2659 Characters greater than 255 were excluded from the class instead of being
2660 included.
2661
2662 3. The same bug as (2) above applied to negated POSIX classes such as
2663 [:^space:].
2664
2665 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2666 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2667 I have changed it.
2668
2669 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2670 first named subpattern) and a construct such as (?&a) would reference the
2671 first named subpattern whose name started with "a" (in other words, the
2672 length check was missing). Both these problems are fixed. "Subpattern name
2673 expected" is now given for (?&) (a zero-length name), and this patch also
2674 makes it give the same error for \k'' (previously it complained that that
2675 was a reference to a non-existent subpattern).
2676
2677 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2678 this is right because (?- can be followed by option settings as well as by
2679 digits. I have, however, made the messages clearer.
2680
2681 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2682 than the number used in the conditional) now cause a compile-time error.
2683 This is actually not compatible with Perl, which accepts such patterns, but
2684 treats the conditional as always being FALSE (as PCRE used to), but it
2685 seems to me that giving a diagnostic is better.
2686
2687 8. Change "alphameric" to the more common word "alphanumeric" in comments
2688 and messages.
2689
2690 9. Fix two occurrences of "backslash" in comments that should have been
2691 "backspace".
2692
2693 10. Remove two redundant lines of code that can never be obeyed (their function
2694 was moved elsewhere).
2695
2696 11. The program that makes PCRE's Unicode character property table had a bug
2697 which caused it to generate incorrect table entries for sequences of
2698 characters that have the same character type, but are in different scripts.
2699 It amalgamated them into a single range, with the script of the first of
2700 them. In other words, some characters were in the wrong script. There were
2701 thirteen such cases, affecting characters in the following ranges:
2702
2703 U+002b0 - U+002c1
2704 U+0060c - U+0060d
2705 U+0061e - U+00612
2706 U+0064b - U+0065e
2707 U+0074d - U+0076d
2708 U+01800 - U+01805
2709 U+01d00 - U+01d77
2710 U+01d9b - U+01dbf
2711 U+0200b - U+0200f
2712 U+030fc - U+030fe
2713 U+03260 - U+0327f
2714 U+0fb46 - U+0fbb1
2715 U+10450 - U+1049d
2716
2717 12. The -o option (show only the matching part of a line) for pcregrep was not
2718 compatible with GNU grep in that, if there was more than one match in a
2719 line, it showed only the first of them. It now behaves in the same way as
2720 GNU grep.
2721
2722 13. If the -o and -v options were combined for pcregrep, it printed a blank
2723 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2724 does the same. The return code can be used to tell if there were any
2725 non-matching lines.
2726
2727 14. Added --file-offsets and --line-offsets to pcregrep.
2728
2729 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2730 infinitely looping recursion. The bug was that positive lookaheads were not
2731 being skipped when checking for a possible empty match (negative lookaheads
2732 and both kinds of lookbehind were skipped).
2733
2734 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2735 inclusion of <windows.h> to before rather than after the definition of
2736 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2737
2738 17. Specifying a possessive quantifier with a specific limit for a Unicode
2739 character property caused pcre_compile() to compile bad code, which led at
2740 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2741 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2742 caused the error; without that there was no problem.
2743
2744 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2745
2746 19. Added --enable-pcretest-libreadline.
2747
2748 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2749 RE::GlobalReplace(). As a result, the number of replacements returned was
2750 double what it should be. I removed one of the increments, but Craig sent a
2751 later patch that removed the other one (the right fix) and added unit tests
2752 that check the return values (which was not done before).
2753
2754 21. Several CMake things:
2755
2756 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2757 the names libpcre and libpcreposix, not just pcre and pcreposix.
2758
2759 (2) The above change means that pcretest and pcregrep are now correctly
2760 linked with the newly-built libraries, not previously installed ones.
2761
2762 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2763
2764 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2765 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2766 UTF-8 newline character). The key issue is that the pattern starts .*;
2767 this means that the match must be either at the beginning, or after a
2768 newline. The bug was in the code for advancing after a failed match and
2769 checking that the new position followed a newline. It was not taking
2770 account of UTF-8 characters correctly.
2771
2772 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2773 character classes. PCRE was not treating the sequence [:...:] as a
2774 character class unless the ... were all letters. Perl, however, seems to
2775 allow any characters between [: and :], though of course it rejects as
2776 unknown any "names" that contain non-letters, because all the known class
2777 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2778 for example, whereas PCRE did not - it did not recognize a POSIX character
2779 class. This seemed a bit dangerous, so the code has been changed to be
2780 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2781 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2782 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2783 Perl does, and where it didn't before.
2784
2785 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2786 Windows environments %n is disabled by default.
2787
2788
2789 Version 7.4 21-Sep-07
2790 ---------------------
2791
2792 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2793 means that a class such as [\s] counted as "explicit reference to CR or
2794 LF". That isn't really right - the whole point of the change was to try to
2795 help when there was an actual mention of one of the two characters. So now
2796 the change happens only if \r or \n (or a literal CR or LF) character is
2797 encountered.
2798
2799 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2800 of both had grown to the point where there were only 3 bits left.
2801 Fortunately, there was spare space in the data structure, and so I have
2802 moved the internal flags into a new 16-bit field to free up more option
2803 bits.
2804
2805 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2806 but did not set the internal JCHANGED flag - either of these is enough to
2807 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2808 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2809 start sets both bits.
2810
2811 4. Added options (at build time, compile time, exec time) to change \R from
2812 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2813
2814 5. doc/pcresyntax.html was missing from the distribution.
2815
2816 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2817 compatibility, even though it is no longer used.
2818
2819 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2820 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2821 windows.h file is present (where different names are used). [This was
2822 reversed later after testing - see 16 below.]
2823
2824 8. Changed all #include <config.h> to #include "config.h". There were also
2825 some further <pcre.h> cases that I changed to "pcre.h".
2826
2827 9. When pcregrep was used with the --colour option, it missed the line ending
2828 sequence off the lines that it output.
2829
2830 10. It was pointed out to me that arrays of string pointers cause lots of
2831 relocations when a shared library is dynamically loaded. A technique of
2832 using a single long string with a table of offsets can drastically reduce
2833 these. I have refactored PCRE in four places to do this. The result is
2834 dramatic:
2835
2836 Originally: 290
2837 After changing UCP table: 187
2838 After changing error message table: 43
2839 After changing table of "verbs" 36
2840 After changing table of Posix names 22
2841
2842 Thanks to the folks working on Gregex for glib for this insight.
2843
2844 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2845 unicode-properties was also set.
2846
2847 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2848
2849 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2850 checked only for CRLF.
2851
2852 14. Added casts to pcretest.c to avoid compiler warnings.
2853
2854 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2855
2856 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2857 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2858 entirely. This removes changes made in 7 above.
2859
2860 17. The CMake files have been updated, and there is now more information about
2861 building with CMake in the NON-UNIX-USE document.
2862
2863
2864 Version 7.3 28-Aug-07
2865 ---------------------
2866
2867 1. In the rejigging of the build system that eventually resulted in 7.1, the
2868 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2869 brackets there is not right, since it causes compilers to look for an
2870 installed pcre.h, not the version that is in the source that is being
2871 compiled (which of course may be different). I have changed it back to:
2872
2873 #include "pcre.h"
2874
2875 I have a vague recollection that the change was concerned with compiling in
2876 different directories, but in the new build system, that is taken care of
2877 by the VPATH setting the Makefile.
2878
2879 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2880 when the subject happened to end in the byte 0x85 (e.g. if the last
2881 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2882 characters but of course it shouldn't be taken as a newline when it is part
2883 of another character. The bug was that, for an unlimited repeat of . in
2884 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2885 characters when looking for a newline.
2886
2887 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2888
2889 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2890 in debug output.
2891
2892 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2893 long printing in the pcrecpp unittest when running under MinGW.
2894
2895 6. ESC_K was left out of the EBCDIC table.
2896
2897 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2898 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2899 limit also applies to "virtual nesting" when a pattern is recursive, and in
2900 this case 1000 isn't so big. I have been able to remove this limit at the
2901 expense of backing off one optimization in certain circumstances. Normally,
2902 when pcre_exec() would call its internal match() function recursively and
2903 immediately return the result unconditionally, it uses a "tail recursion"
2904 feature to save stack. However, when a subpattern that can match an empty
2905 string has an unlimited repetition quantifier, it no longer makes this
2906 optimization. That gives it a stack frame in which to save the data for
2907 checking that an empty string has been matched. Previously this was taken
2908 from the 1000-entry workspace that had been reserved. So now there is no
2909 explicit limit, but more stack is used.
2910
2911 8. Applied Daniel's patches to solve problems with the import/export magic
2912 syntax that is required for Windows, and which was going wrong for the
2913 pcreposix and pcrecpp parts of the library. These were overlooked when this
2914 problem was solved for the main library.
2915
2916 9. There were some crude static tests to avoid integer overflow when computing
2917 the size of patterns that contain repeated groups with explicit upper
2918 limits. As the maximum quantifier is 65535, the maximum group length was
2919 set at 30,000 so that the product of these two numbers did not overflow a
2920 32-bit integer. However, it turns out that people want to use groups that
2921 are longer than 30,000 bytes (though not repeat them that many times).
2922 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2923 made it possible to implement the integer overflow checks in a much more
2924 dynamic way, which I have now done. The artificial limitation on group
2925 length has been removed - we now have only the limit on the total length of
2926 the compiled pattern, which depends on the LINK_SIZE setting.
2927
2928 10. Fixed a bug in the documentation for get/copy named substring when
2929 duplicate names are permitted. If none of the named substrings are set, the
2930 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2931 empty string.
2932
2933 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2934 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2935 because the ] is interpreted as the first data character and the
2936 terminating ] is not found. PCRE has been made compatible with Perl in this
2937 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2938 cause memory overwriting.
2939
2940 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2941 string has been matched (to stop an infinite loop). It was not recognizing
2942 a conditional subpattern that could match an empty string if that
2943 subpattern was within another subpattern. For example, it looped when
2944 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2945 condition was not nested. This bug has been fixed.
2946
2947 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2948 past the start of the subject in the presence of bytes with the top bit
2949 set, for example "\x8aBCD".
2950
2951 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2952 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2953
2954 14. Optimized (?!) to (*FAIL).
2955
2956 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2957 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2958 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2959 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2960 does: it's just the validity check that is more restrictive.
2961
2962 16. Inserted checks for integer overflows during escape sequence (backslash)
2963 processing, and also fixed erroneous offset values for syntax errors during
2964 backslash processing.
2965
2966 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2967 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2968
2969 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2970 caused an overrun.
2971
2972 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2973 something other than just ASCII characters) inside a group that had an
2974 unlimited repeat caused a loop at compile time (while checking to see
2975 whether the group could match an empty string).
2976
2977 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2978 [\P{Any}] did so. (Error in the code for printing property names.)
2979
2980 21. An orphan \E inside a character class could cause a crash.
2981
2982 22. A repeated capturing bracket such as (A)? could cause a wild memory
2983 reference during compilation.
2984
2985 23. There are several functions in pcre_compile() that scan along a compiled
2986 expression for various reasons (e.g. to see if it's fixed length for look
2987 behind). There were bugs in these functions when a repeated \p or \P was
2988 present in the pattern. These operators have additional parameters compared
2989 with \d, etc, and these were not being taken into account when moving along
2990 the compiled data. Specifically:
2991
2992 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2993 length.
2994
2995 (b) An item such as \pL+ within a repeated group could cause crashes or
2996 loops.
2997
2998 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2999 "reference to non-existent subpattern" error.
3000
3001 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3002
3003 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3004 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3005
3006 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3007
3008 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3009 character were causing crashes (broken optimization).
3010
3011 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3012 \p or \P) caused a compile-time loop.
3013
3014 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3015 break. For example, the unstudied pattern [\r\n]A does not match the string
3016 "\r\nA" because change 7.0/46 below moves the current point on by two
3017 characters after failing to match at the start. However, the pattern \nA
3018 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3019 the same is true. There doesn't seem any very clean way out of this, but
3020 what I have chosen to do makes the common cases work: PCRE now takes note
3021 of whether there can be an explicit match for \r or \n anywhere in the
3022 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3023 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3024 pattern has explicit CR or LF references.
3025
3026 29. Added (*CR) etc for changing newline setting at start of pattern.
3027
3028
3029 Version 7.2 19-Jun-07
3030 ---------------------
3031
3032 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3033 which is apparently normally available under Windows.
3034
3035 2. Re-jig the pcregrep tests with different newline settings in an attempt
3036 to make them independent of the local environment's newline setting.
3037
3038 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3039
3040 4. Some of the "internals" tests were previously cut out when the link size
3041 was not 2, because the output contained actual offsets. The recent new
3042 "Z" feature of pcretest means that these can be cut out, making the tests
3043 usable with all link sizes.
3044
3045 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3046 stack recursion. This gives a massive performance boost under BSD, but just
3047 a small improvement under Linux. However, it saves one field in the frame
3048 in all cases.
3049
3050 6. Added more features from the forthcoming Perl 5.10:
3051
3052 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3053 recursion call. It refers to the nth most recently opened parentheses.
3054
3055 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3056 to be opened parentheses.
3057
3058 (c) Conditions that refer to capturing parentheses can be specified
3059 relatively, for example, (?(-2)... or (?(+3)...
3060
3061 (d) \K resets the start of the current match so that everything before
3062 is not part of it.
3063
3064 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3065
3066 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3067 reference syntax.
3068
3069 (g) (?| introduces a group in which the numbering of parentheses in each
3070 alternative starts with the same number.
3071
3072 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3073
3074 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3075 PCRE_INFO_JCHANGED.
3076
3077 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3078 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3079 for detecting groups that can match an empty string.
3080
3081 9. A pattern with a very large number of alternatives (more than several
3082 hundred) was running out of internal workspace during the pre-compile
3083 phase, where pcre_compile() figures out how much memory will be needed. A
3084 bit of new cunning has reduced the workspace needed for groups with
3085 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3086 workspace instead of running out of the 4096 that are available.
3087
3088 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3089
3090 11. Applied patch from Google to remove an optimization that didn't quite work.
3091 The report of the bug said:
3092
3093 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3094 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3095 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3096
3097 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3098 it matched the wrong number of bytes.
3099
3100
3101 Version 7.1 24-Apr-07
3102 ---------------------
3103
3104 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3105 that is more "standard", making use of automake and other Autotools. There
3106 is some re-arrangement of the files and adjustment of comments consequent
3107 on this.
3108
3109 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3110 for recursive directory scanning broke on some systems because the files
3111 are not scanned in any specific order and on different systems the order
3112 was different. A call to "sort" has been inserted into RunGrepTest for the
3113 approprate test as a short-term fix. In the longer term there may be an
3114 alternative.
3115
3116 3. I had an email from Eric Raymond about problems translating some of PCRE's
3117 man pages to HTML (despite the fact that I distribute HTML pages, some
3118 people do their own conversions for various reasons). The problems
3119 concerned the use of low-level troff macros .br and .in. I have therefore
3120 removed all such uses from the man pages (some were redundant, some could
3121 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3122 HTML has been updated to handle .nf/.fi and to complain if it encounters
3123 .br or .in.
3124
3125 4. Updated comments in configure.ac that get placed in config.h.in and also
3126 arranged for config.h to be included in the distribution, with the name
3127 config.h.generic, for the benefit of those who have to compile without
3128 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3129
3130 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3131 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3132 makevp.bat for the current PCRE, using the additional files
3133 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3134
3135 6. A Windows user reported a minor discrepancy with test 2, which turned out
3136 to be caused by a trailing space on an input line that had got lost in his
3137 copy. The trailing space was an accident, so I've just removed it.
3138
3139 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3140 that is needed.
3141
3142 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3143 as "const" (a) because they are and (b) because it helps the PHP
3144 maintainers who have recently made a script to detect big data structures
3145 in the php code that should be moved to the .rodata section. I remembered
3146 to update Builducptable as well, so it won't revert if ucptable.h is ever
3147 re-created.
3148
3149 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3150 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3151 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3152 support is not required. This saves 1.5-2K of code, which is important in
3153 some applications.
3154
3155 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3156 so as not to refer to the tables, even though these functions will never be
3157 called when UTF-8 support is disabled. Otherwise there are problems with a
3158 shared library.
3159
3160 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3161
3162 (a) It was defining its arguments as char * instead of void *.
3163
3164 (b) It was assuming that all moves were upwards in memory; this was true
3165 a long time ago when I wrote it, but is no longer the case.
3166
3167 The emulated memove() is provided for those environments that have neither
3168 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3169 is clearly not the case, as these two bugs were recently reported.
3170
3171 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3172 and Detrail to create the HTML documentation, the .txt form of the man
3173 pages, and it removes trailing spaces from listed files. It also creates
3174 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3175 case, it wraps all the #defines with #ifndefs. This script should be run
3176 before "make dist".
3177
3178 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3179 with Unicode property support.
3180
3181 (a) For a maximizing quantifier, if the two different cases of the
3182 character were of different lengths in their UTF-8 codings (there are
3183 some cases like this - I found 11), and the matching function had to
3184 back up over a mixture of the two cases, it incorrectly assumed they
3185 were both the same length.
3186
3187 (b) When PCRE was configured to use the heap rather than the stack for
3188 recursion during matching, it was not correctly preserving the data for
3189 the other case of a UTF-8 character when checking ahead for a match
3190 while processing a minimizing repeat. If the check also involved
3191 matching a wide character, but failed, corruption could cause an
3192 erroneous result when trying to check for a repeat of the original
3193 character.
3194
3195 13. Some tidying changes to the testing mechanism:
3196
3197 (a) The RunTest script now detects the internal link size and whether there
3198 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3199 values substituted by "configure". (The RunGrepTest script already did
3200 this for UTF-8.) The configure.ac script no longer substitutes the
3201 relevant variables.
3202
3203 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3204 with length and offset values. This means that the output is different
3205 for different internal link sizes. Test 2 is skipped for link sizes
3206 other than 2 because of this, bypassing the problem. Unfortunately,
3207 there was also a test in test 3 (the locale tests) that used /B and
3208 failed for link sizes other than 2. Rather than cut the whole test out,
3209 I have added a new /Z option to pcretest that replaces the length and
3210 offset values with spaces. This is now used to make test 3 independent
3211 of link size. (Test 2 will be tidied up later.)
3212
3213 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3214 segmentation fault instead of returning the appropriate error message.
3215
3216 15. In multiline mode when the newline sequence was set to "any", the pattern
3217 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3218 This doesn't seem right; it now treats the CRLF combination as the line
3219 ending, and so does not match in that case. It's only a pattern such as ^$
3220 that would hit this one: something like ^ABC$ would have failed after \r
3221 and then tried again after \r\n.
3222
3223 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3224 in an attempt to make files that differ only in their line terminators
3225 compare equal. This works on Linux.
3226
3227 17. Under certain error circumstances pcregrep might try to free random memory
3228 as it exited. This is now fixed, thanks to valgrind.
3229
3230 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3231 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3232 was because its rules for how to advance for /g after matching an empty
3233 string at the end of a line did not allow for this case. They now check for
3234 it specially.
3235
3236 20. pcretest is supposed to handle patterns and data of any length, by
3237 extending its buffers when necessary. It was getting this wrong when the
3238 buffer for a data line had to be extended.
3239
3240 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3241 CRLF as a newline sequence.
3242
3243 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3244 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3245 I have nevertheless tidied it up.
3246
3247 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3248
3249 24. Added a man page for pcre-config.
3250
3251
3252 Version 7.0 19-Dec-06
3253 ---------------------
3254
3255 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3256 moving to gcc 4.1.1.
3257
3258 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3259 sys/time.h, which is documented as needed for this function. It doesn't
3260 seem to matter on Linux, but it showed up on some releases of OS X.
3261
3262 3. It seems that there are systems where bytes whose values are greater than
3263 127 match isprint() in the "C" locale. The "C" locale should be the
3264 default when a C program starts up. In most systems, only ASCII printing
3265 characters match isprint(). This difference caused the output from pcretest
3266 to vary, making some of the tests fail. I have changed pcretest so that:
3267
3268 (a) When it is outputting text in the compiled version of a pattern, bytes
3269 other than 32-126 are always shown as hex escapes.
3270
3271 (b) When it is outputting text that is a matched part of a subject string,
3272 it does the same, unless a different locale has been set for the match
3273 (using the /L modifier). In this case, it uses isprint() to decide.
3274
3275 4. Fixed a major bug that caused incorrect computation of the amount of memory
3276 required for a compiled pattern when options that changed within the
3277 pattern affected the logic of the preliminary scan that determines the
3278 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3279 that the computed length was too small. The symptoms of this bug were
3280 either the PCRE error "internal error: code overflow" from pcre_compile(),
3281 or a glibc crash with a message such as "pcretest: free(): invalid next
3282 size (fast)". Examples of patterns that provoked this bug (shown in
3283 pcretest format) are:
3284
3285 /(?-x: )/x
3286 /(?x)(?-x: \s*#\s*)/
3287 /((?i)[\x{c0}])/8
3288 /(?i:[\x{c0}])/8
3289
3290 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3291 is now done differently.
3292
3293 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3294 wrapper classes; (b) implement a new function in the C++ scanner that is
3295 more efficient than the old way of doing things because it avoids levels of
3296 recursion in the regex matching; (c) add a paragraph to the documentation
3297 for the FullMatch() function.
3298
3299 6. The escape sequence \n was being treated as whatever was defined as
3300 "newline". Not only was this contrary to the documentation, which states
3301 that \n is character 10 (hex 0A), but it also went horribly wrong when
3302 "newline" was defined as CRLF. This has been fixed.
3303
3304 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3305 was being set to -1 for the "end of line" case (supposedly a value that no
3306 character can have). Though this value is never used (the check for end of
3307 line is "zero bytes in current character"), it caused compiler complaints.
3308 I've changed it to 0xffffffff.
3309
3310 8. In pcre_version.c, the version string was being built by a sequence of
3311 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3312 string (as it is for production releases) called a macro with an empty
3313 argument. The C standard says the result of this is undefined. The gcc
3314 compiler treats it as an empty string (which was what was wanted) but it is
3315 reported that Visual C gives an error. The source has been hacked around to
3316 avoid this problem.
3317
3318 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3319 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3320 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3321 of them did).
3322
3323 10. Originally, pcretest opened its input and output without "b"; then I was
3324 told that "b" was needed in some environments, so it was added for release
3325 5.0 to both the input and output. (It makes no difference on Unix-like
3326 systems.) Later I was told that it is wrong for the input on Windows. I've
3327 now abstracted the modes into two macros, to make it easier to fiddle with
3328 them, and removed "b" from the input mode under Windows.
3329
3330 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3331
3332 12. Added -help and --help to pcretest as an official way of being reminded
3333 of the options.
3334
3335 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3336 and pcrecpp.cc because they annoy compilers at high warning levels.
3337
3338 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3339
3340 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3341 scripts are not C programs :-) and which was not noticed because it works
3342 on Linux.
3343
3344 16. pcretest is supposed to handle any length of pattern and data line (as one
3345 line or as a continued sequence of lines) by extending its input buffer if
3346 necessary. This feature was broken for very long pattern lines, leading to
3347 a string of junk being passed to pcre_compile() if the pattern was longer
3348 than about 50K.
3349
3350 17. I have done a major re-factoring of the way pcre_compile() computes the
3351 amount of memory needed for a compiled pattern. Previously, there was code
3352 that made a preliminary scan of the pattern in order to do this. That was
3353 OK when PCRE was new, but as the facilities have expanded, it has become
3354 harder and harder to keep it in step with the real compile phase, and there
3355 have been a number of bugs (see for example, 4 above). I have now found a
3356 cunning way of running the real compile function in a "fake" mode that
3357 enables it to compute how much memory it would need, while actually only
3358 ever using a few hundred bytes of working memory and without too many
3359 tests of the mode. This should make future maintenance and development
3360 easier. A side effect of this work is that the limit of 200 on the nesting
3361 depth of parentheses has been removed (though this was never a serious
3362 limitation, I suspect). However, there is a downside: pcre_compile() now
3363 runs more slowly than before (30% or more, depending on the pattern). I
3364 hope this isn't a big issue. There is no effect on runtime performance.
3365
3366 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3367 newline (only possible for the last line of a file) and it was a
3368 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3369
3370 19. Added additional timing features to pcretest. (1) The -tm option now times
3371 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3372 separate command line item, by a number that specifies the number of
3373 repeats to use when timing. The default is 50000; this gives better
3374 precision, but takes uncomfortably long for very large patterns.
3375
3376 20. Extended pcre_study() to be more clever in cases where a branch of a
3377 subpattern has no definite first character. For example, (a*|b*)[cd] would
3378 previously give no result from pcre_study(). Now it recognizes that the
3379 first character must be a, b, c, or d.
3380
3381 21. There was an incorrect error "recursive call could loop indefinitely" if
3382 a subpattern (or the entire pattern) that was being tested for matching an
3383 empty string contained only one non-empty item after a nested subpattern.
3384 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3385 incorrectly, because the \d was being skipped in the check.
3386
3387 22. The pcretest program now has a new pattern option /B and a command line
3388 option -b, which is equivalent to adding /B to every pattern. This causes
3389 it to show the compiled bytecode, without the additional information that
3390 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3391 is the same as /B/I).
3392
3393 23. A new optimization is now able automatically to treat some sequences such
3394 as a*b as a*+b. More specifically, if something simple (such as a character
3395 or a simple class like \d) has an unlimited quantifier, and is followed by
3396 something that cannot possibly match the quantified thing, the quantifier
3397 is automatically "possessified".
3398
3399 24. A recursive reference to a subpattern whose number was greater than 39
3400 went wrong under certain circumstances in UTF-8 mode. This bug could also
3401 have affected the operation of pcre_study().
3402
3403 25. Realized that a little bit of performance could be had by replacing
3404 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3405
3406 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3407
3408 27. Possessive quantifiers such as a++ were previously implemented by turning
3409 them into atomic groups such as ($>a+). Now they have their own opcodes,
3410 which improves performance. This includes the automatically created ones
3411 from 23 above.
3412
3413 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3414 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3415 the first matched character to be a colon. This applied both to named and
3416 numbered groups.
3417
3418 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3419
3420 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3421 building PCRE on Windows easier, so I have included it in the distribution.
3422
3423 31. There is now a check in pcretest against a ridiculously large number being
3424 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3425 loop, the loop is abandoned.
3426
3427 32. Forward references to subpatterns in conditions such as (?(2)...) where
3428 subpattern 2 is defined later cause pcre_compile() to search forwards in
3429 the pattern for the relevant set of parentheses. This search went wrong
3430 when there were unescaped parentheses in a character class, parentheses
3431 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3432
3433 33. "Subroutine" calls and backreferences were previously restricted to
3434 referencing subpatterns earlier in the regex. This restriction has now
3435 been removed.
3436
3437 34. Added a number of extra features that are going to be in Perl 5.10. On the
3438 whole, these are just syntactic alternatives for features that PCRE had
3439 previously implemented using the Python syntax or my own invention. The
3440 other formats are all retained for compatibility.
3441
3442 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3443 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3444 also .NET compatible.
3445
3446 (b) A recursion or subroutine call to a named group can now be defined as
3447 (?&name) as well as (?P>name).
3448
3449 (c) A backreference to a named group can now be defined as \k<name> or
3450 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3451 5.10, are also .NET compatible.
3452
3453 (d) A conditional reference to a named group can now use the syntax
3454 (?(<name>) or (?('name') as well as (?(name).
3455
3456 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3457 groups (named and numbered) that are never evaluated inline, but can be
3458 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3459 is always false. There may be only one alternative in such a group.
3460
3461 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3462 as the simple (?(R). The condition is true only if the most recent
3463 recursion is that of the given number or name. It does not search out
3464 through the entire recursion stack.
3465
3466 (g) The escape \gN or \g{N} has been added, where N is a positive or
3467 negative number, specifying an absolute or relative reference.
3468
3469 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3470 some "unreachable code" warnings.
3471
3472 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3473 things, this adds five new scripts.
3474
3475 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3476 There were also incompatibilities regarding the handling of \Q..\E inside
3477 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3478 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3479
3480 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3481 matches an empty string, and forcibly breaks the loop. There were bugs in
3482 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3483 against aaaa the result was just "a" rather than "aaaa", for example. Two
3484 separate and independent bugs (that affected different cases) have been
3485 fixed.
3486
3487 39. Refactored the code to abolish the use of different opcodes for small
3488 capturing bracket numbers. This is a tidy that I avoided doing when I
3489 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3490 The new approach is not only tidier, it makes it possible to reduce the
3491 memory needed to fix the previous bug (38).
3492
3493 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3494 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3495 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3496 mode.
3497
3498 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3499 report.
3500
3501 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3502 copy construction and assignment in the C++ wrapper.
3503
3504 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3505 couple of bugs that could have given wrong results in the "--newline=crlf"
3506 case.
3507
3508 44. Added a number of casts and did some reorganization of signed/unsigned int
3509 variables following suggestions from Dair Grant. Also renamed the variable
3510 "this" as "item" because it is a C++ keyword.
3511
3512 45. Arranged for dftables to add
3513
3514 #include "pcre_internal.h"
3515
3516 to pcre_chartables.c because without it, gcc 4.x may remove the array
3517 definition from the final binary if PCRE is built into a static library and
3518 dead code stripping is activated.
3519
3520 46. For an unanchored pattern, if a match attempt fails at the start of a
3521 newline sequence, and the newline setting is CRLF or ANY, and the next two
3522 characters are CRLF, advance by two characters instead of one.
3523
3524
3525 Version 6.7 04-Jul-06
3526 ---------------------
3527
3528 1. In order to handle tests when input lines are enormously long, pcretest has
3529 been re-factored so that it automatically extends its buffers when
3530 necessary. The code is crude, but this _is_ just a test program. The
3531 default size has been increased from 32K to 50K.
3532
3533 2. The code in pcre_study() was using the value of the re argument before
3534 testing it for NULL. (Of course, in any sensible call of the function, it
3535 won't be NULL.)
3536
3537 3. The memmove() emulation function in pcre_internal.h, which is used on
3538 systems that lack both memmove() and bcopy() - that is, hardly ever -
3539 was missing a "static" storage class specifier.
3540
3541 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3542 containing an extended class (one that cannot be represented by a bitmap
3543 because it contains high-valued characters or Unicode property items, e.g.
3544 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3545 pattern, but PCRE should not loop if you do not (it no longer does).
3546 [Detail: two cases were found: (a) a repeated subpattern containing an
3547 extended class; (b) a recursive reference to a subpattern that followed a
3548 previous extended class. It wasn't skipping over the extended class
3549 correctly when UTF-8 mode was not set.]
3550
3551 5. A negated single-character class was not being recognized as fixed-length
3552 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3553 compile error "lookbehind assertion is not fixed length".
3554
3555 6. The RunPerlTest auxiliary script was showing an unexpected difference
3556 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3557 write a Perl script that can interpret lines of an input file either as
3558 byte characters or as UTF-8, which is what "perltest" was being required to
3559 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3560 can't do is switch easily at run time between having the "use utf8;" pragma
3561 or not. In the end, I fudged it by using the RunPerlTest script to insert
3562 "use utf8;" explicitly for the UTF-8 tests.
3563
3564 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3565 the end of the subject string, contrary to the documentation and to what
3566 Perl does. This was true of both matching functions. Now it matches only at
3567 the start of the subject and immediately after *internal* newlines.
3568
3569 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3570 a pointer to an int instead of a pointer to an unsigned long int. This
3571 caused problems on 64-bit systems.
3572
3573 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3574 instance of the 'standard' template library not being so standard".
3575
3576 10. There was no check on the number of named subpatterns nor the maximum
3577 length of a subpattern name. The product of these values is used to compute
3578 the size of the memory block for a compiled pattern. By supplying a very
3579 long subpattern name and a large number of named subpatterns, the size
3580 computation could be caused to overflow. This is now prevented by limiting
3581 the length of names to 32 characters, and the number of named subpatterns
3582 to 10,000.
3583
3584 11. Subpatterns that are repeated with specific counts have to be replicated in
3585 the compiled pattern. The size of memory for this was computed from the
3586 length of the subpattern and the repeat count. The latter is limited to
3587 65535, but there was no limit on the former, meaning that integer overflow
3588 could in principle occur. The compiled length of a repeated subpattern is
3589 now limited to 30,000 bytes in order to prevent this.
3590
3591 12. Added the optional facility to have named substrings with the same name.
3592
3593 13. Added the ability to use a named substring as a condition, using the
3594 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3595 are numbers (not recommended). Forward references are permitted.
3596
3597 14. Added forward references in named backreferences (if you see what I mean).
3598
3599 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3600 pattern could run off the end of the subject. For example, the pattern
3601 "(?s)(.{1,5})"8 did this with the subject "ab".
3602
3603 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3604 PCRE_CASELESS was set when matching characters that were quantified with ?
3605 or *.
3606
3607 17. A character class other than a single negated character that had a minimum
3608 but no maximum quantifier - for example [ab]{6,} - was not handled
3609 correctly by pce_dfa_exec(). It would match only one character.
3610
3611 18. A valid (though odd) pattern that looked like a POSIX character
3612 class but used an invalid character after [ (for example [[,abc,]]) caused
3613 pcre_compile() to give the error "Failed: internal error: code overflow" or
3614 in some cases to crash with a glibc free() error. This could even happen if
3615 the pattern terminated after [[ but there just happened to be a sequence of
3616 letters, a binary zero, and a closing ] in the memory that followed.
3617
3618 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3619 over the years. Originally (before any Unicode support), just the bottom 8
3620 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3621 output from "man perlunicode" includes this:
3622
3623 The regular expression compiler produces polymorphic opcodes. That
3624 is, the pattern adapts to the data and automatically switches to
3625 the Unicode character scheme when presented with Unicode data--or
3626 instead uses a traditional byte scheme when presented with byte
3627 data.
3628
3629 Sadly, a wide octal escape does not cause a switch, and in a string with
3630 no other multibyte characters, these octal escapes are treated as before.
3631 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3632 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3633 Unicode string.
3634
3635 I have not perpetrated such confusion in PCRE. Up till now, it took just
3636 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3637 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3638 translate to the appropriate multibyte character.
3639
3640 29. Applied some refactoring to reduce the number of warnings from Microsoft
3641 and Borland compilers. This has included removing the fudge introduced
3642 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3643 a warning about an unused variable.
3644
3645 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3646 characters since release 4.0, because Perl (from release 5.004) does not.
3647 [Or at least, is documented not to: some releases seem to be in conflict
3648 with the documentation.] However, when a pattern was studied with
3649 pcre_study() and all its branches started with \s, PCRE still included VT
3650 as a possible starting character. Of course, this did no harm; it just
3651 caused an unnecessary match attempt.
3652
3653 22. Removed a now-redundant internal flag bit that recorded the fact that case
3654 dependency changed within the pattern. This was once needed for "required
3655 byte" processing, but is no longer used. This recovers a now-scarce options
3656 bit. Also moved the least significant internal flag bit to the most-
3657 significant bit of the word, which was not previously used (hangover from
3658 the days when it was an int rather than a uint) to free up another bit for
3659 the future.
3660
3661 23. Added support for CRLF line endings as well as CR and LF. As well as the
3662 default being selectable at build time, it can now be changed at runtime
3663 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3664 specify that it is scanning data with non-default line endings.
3665
3666 24. Changed the definition of CXXLINK to make it agree with the definition of
3667 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3668
3669 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3670 recursions. This makes a big different to stack usage for some patterns.
3671
3672 26. If a subpattern containing a named recursion or subroutine reference such
3673 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3674 the space required for the compiled pattern went wrong and gave too small a
3675 value. Depending on the environment, this could lead to "Failed: internal
3676 error: code overflow at offset 49" or "glibc detected double free or
3677 corruption" errors.
3678
3679 27. Applied patches from Google (a) to support the new newline modes and (b) to
3680 advance over multibyte UTF-8 characters in GlobalReplace.
3681
3682 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3683 difference for some implementation of PCRE in some Windows version.
3684
3685 29. Added some extra testing facilities to pcretest:
3686
3687 \q<number> in a data line sets the "match limit" value
3688 \Q<number> in a data line sets the "match recursion limt" value
3689 -S <number> sets the stack size, where <number> is in megabytes
3690
3691 The -S option isn't available for Windows.
3692
3693
3694 Version 6.6 06-Feb-06
3695 ---------------------
3696
3697 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3698 in pcreposix.h. I have copied the definition from pcre.h.
3699
3700 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3701 because pcre.h is no longer a built file.
3702
3703 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3704 not normally included in the compiled code.
3705
3706
3707 Version 6.5 01-Feb-06
3708 ---------------------
3709
3710 1. When using the partial match feature with pcre_dfa_exec(), it was not
3711 anchoring the second and subsequent partial matches at the new starting
3712 point. This could lead to incorrect results. For example, with the pattern
3713 /1234/, partially matching against "123" and then "a4" gave a match.
3714
3715 2. Changes to pcregrep:
3716
3717 (a) All non-match returns from pcre_exec() were being treated as failures
3718 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3719 error message is output. Some extra information is given for the
3720 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3721 probably the only errors that are likely to be caused by users (by
3722 specifying a regex that has nested indefinite repeats, for instance).
3723 If there are more than 20 of these errors, pcregrep is abandoned.
3724
3725 (b) A binary zero was treated as data while matching, but terminated the
3726 output line if it was written out. This has been fixed: binary zeroes
3727 are now no different to any other data bytes.
3728
3729 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3730 used to set a locale for matching. The --locale=xxxx long option has
3731 been added (no short equivalent) to specify a locale explicitly on the
3732 pcregrep command, overriding the environment variables.
3733
3734 (d) When -B was used with -n, some line numbers in the output were one less
3735 than they should have been.
3736
3737 (e) Added the -o (--only-matching) option.
3738
3739 (f) If -A or -C was used with -c (count only), some lines of context were
3740 accidentally printed for the final match.
3741
3742 (g) Added the -H (--with-filename) option.
3743
3744 (h) The combination of options -rh failed to suppress file names for files
3745 that were found from directory arguments.
3746
3747 (i) Added the -D (--devices) and -d (--directories) options.
3748
3749 (j) Added the -F (--fixed-strings) option.
3750
3751 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3752
3753 (l) Added the --colo(u)r option.
3754
3755 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3756 is not present by default.
3757
3758 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3759 items such as (?R) or (?1), when the recursion could match a number of
3760 alternatives. If it matched one of the alternatives, but subsequently,
3761 outside the recursion, there was a failure, the code tried to back up into
3762 the recursion. However, because of the way PCRE is implemented, this is not
3763 possible, and the result was an incorrect result from the match.
3764
3765 In order to prevent this happening, the specification of recursion has
3766 been changed so that all such subpatterns are automatically treated as
3767 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3768
3769 4. I had overlooked the fact that, in some locales, there are characters for
3770 which isalpha() is true but neither isupper() nor islower() are true. In
3771 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3772 and ordfeminine) are like this. This affected the treatment of \w and \W
3773 when they appeared in character classes, but not when they appeared outside
3774 a character class. The bit map for "word" characters is now created
3775 separately from the results of isalnum() instead of just taking it from the
3776 upper, lower, and digit maps. (Plus the underscore character, of course.)
3777
3778 5. The above bug also affected the handling of POSIX character classes such as
3779 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3780 permanent tables. Instead, the bit maps for such a class were previously
3781 created as the appropriate unions of the upper, lower, and digit bitmaps.
3782 Now they are created by subtraction from the [[:word:]] class, which has
3783 its own bitmap.
3784
3785 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3786 It is created by subtracting the vertical space characters (\x09, \x0a,
3787 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3788 subtraction was done in the overall bitmap for a character class, meaning
3789 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3790 be recognized. This bug has been fixed.
3791
3792 7. Patches from the folks at Google:
3793
3794 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3795 real life, but is still worth protecting against".
3796
3797 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3798 regular expressions".
3799
3800 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3801 have it.
3802
3803 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3804 "configure" and the latter not, in order to fix a problem somebody had
3805 with compiling the Arg class on HP-UX.
3806
3807 (e) Improve the error-handling of the C++ wrapper a little bit.
3808
3809 (f) New tests for checking recursion limiting.
3810
3811 8. The pcre_memmove() function, which is used only if the environment does not
3812 have a standard memmove() function (and is therefore rarely compiled),
3813 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3814 returning a result (though PCRE never actually uses the result).
3815
3816 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3817 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3818 returned instead of calling malloc() with an overflowing number that would
3819 most likely cause subsequent chaos.
3820
3821 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3822
3823 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3824 with this option is matched, the nmatch and pmatch options of regexec() are
3825 ignored.
3826
3827 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3828 provided in case anyone wants to the the POSIX interface with UTF-8
3829 strings.
3830
3831 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3832 C++ linking (needed for some HP-UX environments).
3833
3834 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3835 (unused parameter) and in the pcre_printint() function (omitted "default"
3836 switch label when the default is to do nothing).
3837
3838 15. Added some code to make it possible, when PCRE is compiled as a C++
3839 library, to replace subject pointers for pcre_exec() with a smart pointer
3840 class, thus making it possible to process discontinuous strings.
3841
3842 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3843 much the same function. They were added by different people who were trying
3844 to make PCRE easy to compile on non-Unix systems. It has been suggested
3845 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3846 for compiling on Windows systems. I have therefore replaced it with
3847 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3848 defaults to "extern" for C or "extern C" for C++, which works fine on
3849 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3850 SCOPE with something explicit in config.h. In addition:
3851
3852 (a) pcreposix.h still had just "extern" instead of either of these macros;
3853 I have replaced it with PCRE_DATA_SCOPE.
3854
3855 (b) Functions such as _pcre_xclass(), which are internal to the library,
3856 but external in the C sense, all had PCRE_EXPORT in their definitions.
3857 This is apparently wrong for the Windows case, so I have removed it.
3858 (It makes no difference on Unix-like systems.)
3859
3860 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3861 of recursive calls to match(). This is different to MATCH_LIMIT because
3862 that limits the total number of calls to match(), not all of which increase
3863 the depth of recursion. Limiting the recursion depth limits the amount of
3864 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3865 when PCRE is compiled, and changed at run time. A patch from Google adds
3866 this functionality to the C++ interface.
3867
3868 18. Changes to the handling of Unicode character properties:
3869
3870 (a) Updated the table to Unicode 4.1.0.
3871
3872 (b) Recognize characters that are not in the table as "Cn" (undefined).
3873
3874 (c) I revised the way the table is implemented to a much improved format
3875 which includes recognition of ranges. It now supports the ranges that
3876 are defined in UnicodeData.txt, and it also amalgamates other
3877 characters into ranges. This has reduced the number of entries in the
3878 table from around 16,000 to around 3,000, thus reducing its size
3879 considerably. I realized I did not need to use a tree structure after
3880 all - a binary chop search is just as efficient. Having reduced the
3881 number of entries, I extended their size from 6 bytes to 8 bytes to
3882 allow for more data.
3883
3884 (d) Added support for Unicode script names via properties such as \p{Han}.
3885
3886 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3887 matching that character.
3888
3889 20. When matching a repeated Unicode property with a minimum greater than zero,
3890 (for example \pL{2,}), PCRE could look past the end of the subject if it
3891 reached it while seeking the minimum number of characters. This could
3892 happen only if some of the characters were more than one byte long, because
3893 there is a check for at least the minimum number of bytes.
3894
3895 21. Refactored the implementation of \p and \P so as to be more general, to
3896 allow for more different types of property in future. This has changed the
3897 compiled form incompatibly. Anybody with saved compiled patterns that use
3898 \p or \P will have to recompile them.
3899
3900 22. Added "Any" and "L&" to the supported property types.
3901
3902 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3903 but give a compile time error if the value is greater than 0xff.
3904
3905 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3906 accidentally not being installed or uninstalled.
3907
3908 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3909 made were to insert the current release number. This seemed silly, because
3910 it made things harder for people building PCRE on systems that don't run
3911 "configure". I have turned pcre.h into a distributed file, no longer built
3912 by "configure", with the version identification directly included. There is
3913 no longer a pcre.h.in file.
3914
3915 However, this change necessitated a change to the pcre-config script as
3916 well. It is built from pcre-config.in, and one of the substitutions was the
3917 release number. I have updated configure.ac so that ./configure now finds
3918 the release number by grepping pcre.h.
3919
3920 26. Added the ability to run the tests under valgrind.
3921
3922
3923 Version 6.4 05-Sep-05
3924 ---------------------
3925
3926 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3927 "--" to be printed when multiple files were scanned, even when none of the
3928 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3929 consider it to be a bug, and have restored the previous behaviour.
3930
3931 2. A couple of code tidies to get rid of compiler warnings.
3932
3933 3. The pcretest program used to cheat by referring to symbols in the library
3934 whose names begin with _pcre_. These are internal symbols that are not
3935 really supposed to be visible externally, and in some environments it is
3936 possible to suppress them. The cheating is now confined to including
3937 certain files from the library's source, which is a bit cleaner.
3938
3939 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3940 file's purpose clearer.
3941
3942 5. Reorganized pcre_ucp_findchar().
3943
3944
3945 Version 6.3 15-Aug-05
3946 ---------------------
3947
3948 1. The file libpcre.pc.in did not have general read permission in the tarball.
3949
3950 2. There were some problems when building without C++ support:
3951
3952 (a) If C++ support was not built, "make install" and "make test" still
3953 tried to test it.
3954
3955 (b) There were problems when the value of CXX was explicitly set. Some
3956 changes have been made to try to fix these, and ...
3957
3958 (c) --disable-cpp can now be used to explicitly disable C++ support.
3959
3960 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3961 backslash in a target when C++ was disabled. This confuses some
3962 versions of "make", apparently. Using an intermediate variable solves
3963 this. (Same for CPP_LOBJ.)
3964
3965 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3966 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3967 necessary on certain architectures.
3968
3969 4. Added a setting of -export-symbols-regex to the link command to remove
3970 those symbols that are exported in the C sense, but actually are local
3971 within the library, and not documented. Their names all begin with
3972 "_pcre_". This is not a perfect job, because (a) we have to except some
3973 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3974 available (and never for static libraries). I have made a note to try to
3975 find a way round (a) in the future.
3976
3977
3978 Version 6.2 01-Aug-05
3979 ---------------------
3980
3981 1. There was no test for integer overflow of quantifier values. A construction
3982 such as {1111111111111111} would give undefined results. What is worse, if
3983 a minimum quantifier for a parenthesized subpattern overflowed and became
3984 negative, the calculation of the memory size went wrong. This could have
3985 led to memory overwriting.
3986
3987 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3988
3989 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3990 operating environments where this matters.
3991
3992 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3993 PCRE options from within the C++ wrapper.
3994
3995 5. Named capturing subpatterns were not being correctly counted when a pattern
3996 was compiled. This caused two problems: (a) If there were more than 100
3997 such subpatterns, the calculation of the memory needed for the whole
3998 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3999 back references of the form \12, where the number was greater than 9, were
4000 not recognized as back references, even though there were sufficient
4001 previous subpatterns.
4002
4003 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4004 versions of gcc, e.g. 2.95.4.
4005
4006
4007 Version 6.1 21-Jun-05
4008 ---------------------
4009
4010 1. There was one reference to the variable "posix" in pcretest.c that was not
4011 surrounded by "#if !defined NOPOSIX".
4012
4013 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4014 the cross-check on the old pcre_info() function, for the benefit of the
4015 cut-down version of PCRE that is currently imported into Exim.
4016
4017 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4018 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4019 patterns that start (?i) but I don't think that matters. The use of (?i) is
4020 just an example; this all applies to the other options as well.
4021
4022 4. Since libtool seems to echo the compile commands it is issuing, the output
4023 from "make" can be reduced a bit by putting "@" in front of each libtool
4024 compile command.
4025
4026 5. Patch from the folks at Google for configure.in to be a bit more thorough
4027 in checking for a suitable C++ installation before trying to compile the
4028 C++ stuff. This should fix a reported problem when a compiler was present,
4029 but no suitable headers.
4030
4031 6. The man pages all had just "PCRE" as their title. I have changed them to
4032 be the relevant file name. I have also arranged that these names are
4033 retained in the file doc/pcre.txt, which is a concatenation in text format
4034 of all the man pages except the little individual ones for each function.
4035
4036 7. The NON-UNIX-USE file had not been updated for the different set of source
4037 files that come with release 6. I also added a few comments about the C++
4038 wrapper.
4039
4040
4041 Version 6.0 07-Jun-05
4042 ---------------------
4043
4044 1. Some minor internal re-organization to help with my DFA experiments.
4045
4046 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4047 didn't matter for the library itself when fully configured, but did matter
4048 when compiling without UCP support, or within Exim, where the ucp files are
4049 not imported.
4050
4051 3. Refactoring of the library code to split up the various functions into
4052 different source modules. The addition of the new DFA matching code (see
4053 below) to a single monolithic source would have made it really too
4054 unwieldy, quite apart from causing all the code to be include in a
4055 statically linked application, when only some functions are used. This is
4056 relevant even without the DFA addition now that patterns can be compiled in
4057 one application and matched in another.
4058
4059 The downside of splitting up is that there have to be some external
4060 functions and data tables that are used internally in different modules of
4061 the library but which are not part of the API. These have all had their
4062 names changed to start with "_pcre_" so that they are unlikely to clash
4063 with other external names.
4064
4065 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4066 a different (DFA) algorithm. Although it is slower than the original
4067 function, it does have some advantages for certain types of matching
4068 problem.
4069
4070 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4071 including restarting after a partial match.
4072
4073 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4074 defined when compiling for Windows was sent to me. I have put it into the
4075 code, though I have no means of testing or verifying it.
4076
4077 7. Added the pcre_refcount() auxiliary function.
4078
4079 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4080 match before or at the first newline in the subject string. In pcretest,
4081 the /f option on a pattern can be used to set this.
4082
4083 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4084 would behave wrongly. This has been present in PCRE since release 4.0.
4085
4086 10. A number of changes to the pcregrep command:
4087
4088 (a) Refactored how -x works; insert ^(...)$ instead of setting
4089 PCRE_ANCHORED and checking the length, in preparation for adding
4090 something similar for -w.
4091
4092 (b) Added the -w (match as a word) option.
4093
4094 (c) Refactored the way lines are read and buffered so as to have more
4095 than one at a time available.
4096
4097 (d) Implemented a pcregrep test script.
4098
4099 (e) Added the -M (multiline match) option. This allows patterns to match
4100 over several lines of the subject. The buffering ensures that at least
4101 8K, or the rest of the document (whichever is the shorter) is available
4102 for matching (and similarly the previous 8K for lookbehind assertions).
4103
4104 (f) Changed the --help output so that it now says
4105
4106 -w, --word-regex(p)
4107
4108 instead of two lines, one with "regex" and the other with "regexp"
4109 because that confused at least one person since the short forms are the
4110 same. (This required a bit of code, as the output is generated
4111 automatically from a table. It wasn't just a text change.)
4112
4113 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4114 option but starts with a hyphen. Could be a pattern or a path name
4115 starting with a hyphen, for instance.
4116
4117 (h) "-" can be given as a file name to represent stdin.
4118
4119 (i) When file names are being printed, "(standard input)" is used for
4120 the standard input, for compatibility with GNU grep. Previously
4121 "<stdin>" was used.
4122
4123 (j) The option --label=xxx can be used to supply a name to be used for
4124 stdin when file names are being printed. There is no short form.
4125
4126 (k) Re-factored the options decoding logic because we are going to add
4127 two more options that take data. Such options can now be given in four
4128 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4129
4130 (l) Added the -A, -B, and -C options for requesting that lines of context
4131 around matches be printed.
4132
4133 (m) Added the -L option to print the names of files that do not contain
4134 any matching lines, that is, the complement of -l.
4135
4136 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4137 continue to scan other files.
4138
4139 (o) The -s option was incorrectly implemented. For compatibility with other
4140 greps, it now suppresses the error message for a non-existent or non-
4141 accessible file (but not the return code). There is a new option called
4142 -q that suppresses the output of matching lines, which was what -s was
4143 previously doing.
4144
4145 (p) Added --include and --exclude options to specify files for inclusion
4146 and exclusion when recursing.
4147
4148 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4149 Hopefully, it now does.
4150
4151 12. Missing cast in pcre_study().
4152
4153 13. Added an "uninstall" target to the makefile.
4154
4155 14. Replaced "extern" in the function prototypes in Makefile.in with
4156 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4157 world, but is set differently for Windows.
4158
4159 15. Added a second compiling function called pcre_compile2(). The only
4160 difference is that it has an extra argument, which is a pointer to an
4161 integer error code. When there is a compile-time failure, this is set
4162 non-zero, in addition to the error test pointer being set to point to an
4163 error message. The new argument may be NULL if no error number is required
4164 (but then you may as well call pcre_compile(), which is now just a
4165 wrapper). This facility is provided because some applications need a
4166 numeric error indication, but it has also enabled me to tidy up the way
4167 compile-time errors are handled in the POSIX wrapper.
4168
4169 16. Added VPATH=.libs to the makefile; this should help when building with one
4170 prefix path and installing with another. (Or so I'm told by someone who
4171 knows more about this stuff than I do.)
4172
4173 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4174 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4175 match everything, including newlines. This is not POSIX-compatible, but
4176 somebody wanted the feature. From pcretest it can be activated by using
4177 both the P and the s flags.
4178
4179 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4180
4181 19. libpcre.pc was being incorrectly installed as executable.
4182
4183 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4184 it now also looks for '\r' so that it will work unmodified on Windows.
4185
4186 21. Added Google's contributed C++ wrapper to the distribution.
4187
4188 22. Added some untidy missing memory free() calls in pcretest, to keep
4189 Electric Fence happy when testing.
4190
4191
4192
4193 Version 5.0 13-Sep-04
4194 ---------------------
4195
4196 1. Internal change: literal characters are no longer packed up into items
4197 containing multiple characters in a single byte-string. Each character
4198 is now matched using a separate opcode. However, there may be more than one
4199 byte in the character in UTF-8 mode.
4200
4201 2. The pcre_callout_block structure has two new fields: pattern_position and
4202 next_item_length. These contain the offset in the pattern to the next match
4203 item, and its length, respectively.
4204
4205 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4206 insertion of callouts before each pattern item. Added the /C option to
4207 pcretest to make use of this.
4208
4209 4. On the advice of a Windows user, the lines
4210
4211 #if defined(_WIN32) || defined(WIN32)
4212 _setmode( _fileno( stdout ), 0x8000 );
4213 #endif /* defined(_WIN32) || defined(WIN32) */
4214
4215 have been added to the source of pcretest. This apparently does useful
4216 magic in relation to line terminators.
4217
4218 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4219 for the benefit of those environments where the "b" makes a difference.
4220
4221 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4222 to know about it. I have put a hack into configure.in that adds in code
4223 to set GCC=yes if CC=icc. This seems to end up at a point in the
4224 generated configure script that is early enough to affect the setting of
4225 compiler options, which is what is needed, but I have no means of testing
4226 whether it really works. (The user who reported this had patched the
4227 generated configure script, which of course I cannot do.)
4228
4229 LATER: After change 22 below (new libtool files), the configure script
4230 seems to know about icc (and also ecc). Therefore, I have commented out
4231 this hack in configure.in.
4232
4233 7. Added support for pkg-config (2 patches were sent in).
4234
4235 8. Negated POSIX character classes that used a combination of internal tables
4236 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4237 [[:^ascii]]. Typically, they would match almost any characters. The other
4238 POSIX classes were not broken in this way.
4239
4240 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4241 to find the match, as PCRE was deluded into thinking that the match had to
4242 start at the start point or following a newline. The same bug applied to
4243 patterns with negative forward assertions or any backward assertions
4244 preceding ".*" at the start, unless the pattern required a fixed first
4245 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4246
4247 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4248 starting at the last subject character, bytes beyond the end of the subject
4249 string were read.
4250
4251 11. Renamed the variable "class" as "classbits" to make life easier for C++
4252 users. (Previously there was a macro definition, but it apparently wasn't
4253 enough.)
4254
4255 12. Added the new field "tables" to the extra data so that tables can be passed
4256 in at exec time, or the internal tables can be re-selected. This allows
4257 a compiled regex to be saved and re-used at a later time by a different
4258 program that might have everything at different addresses.
4259
4260 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4261 -R library as well as a -L library.
4262
4263 14. The debugging options of pcretest (-d on the command line or D on a
4264 pattern) showed incorrect output for anything following an extended class
4265 that contained multibyte characters and which was followed by a quantifier.
4266
4267 15. Added optional support for general category Unicode character properties
4268 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4269 support. It adds about 90K to the size of the library. The meanings of the
4270 inbuilt class escapes such as \d and \s have NOT been changed.
4271
4272 16. Updated pcredemo.c to include calls to free() to release the memory for the
4273 compiled pattern.
4274
4275 17. The generated file chartables.c was being created in the source directory
4276 instead of in the building directory. This caused the build to fail if the
4277 source directory was different from the building directory, and was
4278 read-only.
4279
4280 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4281 file. No doubt somebody will tell me if they don't make sense... Also added
4282 Dan Mooney's comments about building on OpenVMS.
4283
4284 19. Added support for partial matching via the PCRE_PARTIAL option for
4285 pcre_exec() and the \P data escape in pcretest.
4286
4287 20. Extended pcretest with 3 new pattern features:
4288
4289 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4290 write the compiled pattern to the file whose name is "rest-of-line".
4291 This is a straight binary dump of the data, with the saved pointer to
4292 the character tables forced to be NULL. The study data, if any, is
4293 written too. After writing, pcretest reads a new pattern.
4294
4295 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4296 compiled pattern from the given file. There must not be any
4297 occurrences of "<" in the file name (pretty unlikely); if there are,
4298 pcretest will instead treat the initial "<" as a pattern delimiter.
4299 After reading in the pattern, pcretest goes on to read data lines as
4300 usual.
4301
4302 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4303 and 16-bit fields in a compiled pattern, to simulate a pattern that
4304 was compiled on a host of opposite endianness.
4305
4306 21. The pcre-exec() function can now cope with patterns that were compiled on
4307 hosts of opposite endianness, with this restriction:
4308
4309 As for any compiled expression that is saved and used later, the tables
4310 pointer field cannot be preserved; the extra_data field in the arguments
4311 to pcre_exec() should be used to pass in a tables address if a value
4312 other than the default internal tables were used at compile time.
4313
4314 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4315 now diagnosed as an error. Previously, most of the time, a negative number
4316 would have been treated as zero, but if in addition "ovector" was passed as
4317 NULL, a crash could occur.
4318
4319 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4320 new versions from the libtool 1.5 distribution (the last one is a copy of
4321 a file called libtool.m4). This seems to have fixed the need to patch
4322 "configure" to support Darwin 1.3 (which I used to do). However, I still
4323 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4324 workstation).
4325
4326 24. Changed the PCRE licence to be the more standard "BSD" licence.
4327
4328
4329 Version 4.5 01-Dec-03
4330 ---------------------
4331
4332 1. There has been some re-arrangement of the code for the match() function so
4333 that it can be compiled in a version that does not call itself recursively.
4334 Instead, it keeps those local variables that need separate instances for
4335 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4336 needs to "recurse". Keeping track of where control must go is done by means
4337 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4338 hide most of the details from the main code, and operates only if
4339 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4340 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4341 operating.
4342
4343 To make it easier for callers to provide specially tailored get/free
4344 functions for this usage, two new functions, pcre_stack_malloc, and
4345 pcre_stack_free, are used. They are always called in strict stacking order,
4346 and the size of block requested is always the same.
4347
4348 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4349 PCRE has been compiled to use the stack or the heap for recursion. The
4350 -C option of pcretest uses this to show which version is compiled.
4351
4352 A new data escape \S, is added to pcretest; it causes the amounts of store
4353 obtained and freed by both kinds of malloc/free at match time to be added
4354 to the output.
4355
4356 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4357 what's available on my current Linux desktop machine.
4358
4359 3. When matching a UTF-8 string, the test for a valid string at the start has
4360 been extended. If start_offset is not zero, PCRE now checks that it points
4361 to a byte that is the start of a UTF-8 character. If not, it returns
4362 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4363 this is necessary because there may be backward assertions in the pattern.
4364 When matching the same subject several times, it may save resources to use
4365 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4366
4367 4. The code for checking the validity of UTF-8 strings has been tightened so
4368 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4369 containing "overlong sequences".
4370
4371 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4372 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4373 should have been "&", but it just so happened that all the cases this let
4374 through by mistake were picked up later in the function.
4375
4376 6. I had used a variable called "isblank" - this is a C99 function, causing
4377 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4378
4379 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4380 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4381 so the version is shown; (c) stop "make test" echoing "./RunTest".
4382
4383 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4384
4385 9. The prototype for memmove() for systems that don't have it was using
4386 size_t, but the inclusion of the header that defines size_t was later. I've
4387 moved the #includes for the C headers earlier to avoid this.
4388
4389 10. Added some adjustments to the code to make it easier to compiler on certain
4390 special systems:
4391
4392 (a) Some "const" qualifiers were missing.
4393 (b) Added the macro EXPORT before all exported functions; by default this
4394 is defined to be empty.
4395 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4396 that it reads its output file name as an argument instead of writing
4397 to the standard output and assuming this can be redirected.
4398
4399 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4400 class containing characters with values greater than 255, PCRE compilation
4401 went into a loop.
4402
4403 12. A recursive reference to a subpattern that was within another subpattern
4404 that had a minimum quantifier of zero caused PCRE to crash. For example,
4405 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4406 recursion. If the recursively-called subpattern itself had a zero repeat,
4407 that was OK.
4408
4409 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4410 buffer into which it was copied (for escape processing) was still set at
4411 1024, so long lines caused crashes.
4412
4413 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4414 "internal error: code overflow...". This applied to any character class
4415 that was followed by a possessive quantifier.
4416
4417 15. Modified the Makefile to add libpcre.la as a prerequisite for
4418 libpcreposix.la because I was told this is needed for a parallel build to
4419 work.
4420
4421 16. If a pattern that contained .* following optional items at the start was
4422 studied, the wrong optimizing data was generated, leading to matching
4423 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4424 matching string must start with a or b or c. The correct conclusion for
4425 this pattern is that a match can start with any character.
4426
4427
4428 Version 4.4 13-Aug-03
4429 ---------------------
4430
4431 1. In UTF-8 mode, a character class containing characters with values between
4432 127 and 255 was not handled correctly if the compiled pattern was studied.
4433 In fixing this, I have also improved the studying algorithm for such
4434 classes (slightly).
4435
4436 2. Three internal functions had redundant arguments passed to them. Removal
4437 might give a very teeny performance improvement.
4438
4439 3. Documentation bug: the value of the capture_top field in a callout is *one
4440 more than* the number of the hightest numbered captured substring.
4441
4442 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4443 in incorrectly linking with a previously installed version. They now link
4444 explicitly with libpcre.la.
4445
4446 5. configure.in no longer needs to recognize Cygwin specially.
4447
4448 6. A problem in pcre.in for Windows platforms is fixed.
4449
4450 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4451 pcretest, it used to include the size of the study block as part of its
4452 output. Unfortunately, the structure contains a field that has a different
4453 size on different hardware architectures. This meant that the tests that
4454 showed this size failed. As the block is currently always of a fixed size,
4455 this information isn't actually particularly useful in pcretest output, so
4456 I have just removed it.
4457
4458 8. Three pre-processor statements accidentally did not start in column 1.
4459 Sadly, there are *still* compilers around that complain, even though
4460 standard C has not required this for well over a decade. Sigh.
4461
4462 9. In pcretest, the code for checking callouts passed small integers in the
4463 callout_data field, which is a void * field. However, some picky compilers
4464 complained about the casts involved for this on 64-bit systems. Now
4465 pcretest passes the address of the small integer instead, which should get
4466 rid of the warnings.
4467
4468 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4469 both compile and run time, and gives an error if an invalid UTF-8 sequence
4470 is found. There is a option for disabling this check in cases where the
4471 string is known to be correct and/or the maximum performance is wanted.
4472
4473 11. In response to a bug report, I changed one line in Makefile.in from
4474
4475 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4476 to
4477 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4478
4479 to look similar to other lines, but I have no way of telling whether this
4480 is the right thing to do, as I do not use Windows. No doubt I'll get told
4481 if it's wrong...
4482
4483
4484 Version 4.3 21-May-03
4485 ---------------------
4486
4487 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4488 Makefile.
4489
4490 2. Some refactoring to improve the quality of the code:
4491
4492 (i) The utf8_table... variables are now declared "const".
4493
4494 (ii) The code for \cx, which used the "case flipping" table to upper case
4495 lower case letters, now just substracts 32. This is ASCII-specific,
4496 but the whole concept of \cx is ASCII-specific, so it seems
4497 reasonable.
4498
4499 (iii) PCRE was using its character types table to recognize decimal and
4500 hexadecimal digits in the pattern. This is silly, because it handles
4501 only 0-9, a-f, and A-F, but the character types table is locale-
4502 specific, which means strange things might happen. A private
4503 table is now used for this - though it costs 256 bytes, a table is
4504 much faster than multiple explicit tests. Of course, the standard
4505 character types table is still used for matching digits in subject
4506 strings against \d.
4507
4508 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4509 ending in _t are). So I've renamed it as ESC_tee.
4510
4511 3. The first argument for regexec() in the POSIX wrapper should have been
4512 defined as "const".
4513
4514 4. Changed pcretest to use malloc() for its buffers so that they can be
4515 Electric Fenced for debugging.
4516
4517 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4518 to read one or more bytes before the start of the subject string. Often this
4519 had no effect on PCRE's behaviour, but in some circumstances it could
4520 provoke a segmentation fault.
4521
4522 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4523 to try to read one or more bytes before the start of the subject string.
4524
4525 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4526 UTF-8 support could misbehave in various ways if the subject string
4527 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4528 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4529 back over UTF-8 characters.)
4530
4531
4532 Version 4.2 14-Apr-03
4533 ---------------------
4534
4535 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4536
4537 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4538 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4539 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4540 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4541 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4542 and BUILD_EXEEXT
4543 Note: automatic setting of the BUILD variables is not yet working
4544 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4545 compile-time but not at link-time
4546 [LINK]: use for linking executables only
4547 make different versions for Windows and non-Windows
4548 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4549 libraries
4550 [LINK_FOR_BUILD]: new variable
4551 [OBJEXT]: use throughout
4552 [EXEEXT]: use throughout
4553 <winshared>: new target
4554 <wininstall>: new target
4555 <dftables.o>: use native compiler
4556 <dftables>: use native linker
4557 <install>: handle Windows platform correctly
4558 <clean>: ditto
4559 <check>: ditto
4560 copy DLL to top builddir before testing
4561
4562 As part of these changes, -no-undefined was removed again. This was reported
4563 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4564 in any case.
4565
4566 3. Some tidies to get rid of compiler warnings:
4567
4568 . In the match_data structure, match_limit was an unsigned long int, whereas
4569 match_call_count was an int. I've made them both unsigned long ints.
4570
4571 . In pcretest the fact that a const uschar * doesn't automatically cast to
4572 a void * provoked a warning.
4573
4574 . Turning on some more compiler warnings threw up some "shadow" variables
4575 and a few more missing casts.
4576
4577 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4578 option, a class that contained a single character with a value between 128
4579 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4580
4581 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4582 option, a class that contained several characters, but with at least one
4583 whose value was between 128 and 255 caused PCRE to crash.
4584
4585
4586 Version 4.1 12-Mar-03
4587 ---------------------
4588
4589 1. Compiling with gcc -pedantic found a couple of places where casts were
4590 needed, and a string in dftables.c that was longer than standard compilers are
4591 required to support.
4592
4593 2. Compiling with Sun's compiler found a few more places where the code could
4594 be tidied up in order to avoid warnings.
4595
4596 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4597 first of these names is deprecated in the latest Autoconf in favour of the name
4598 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4599 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4600 analogy I have changed it to CFLAGS_FOR_BUILD.
4601
4602 4. Added -no-undefined to the linking command in the Makefile, because this is
4603 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4604 linking step for the pcreposix library.
4605
4606 5. PCRE was failing to diagnose the case of two named groups with the same
4607 name.
4608
4609 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4610 literal character that is needed in the subject for a match, and scans along to
4611 ensure that it is present before embarking on the full matching process. This
4612 saves time in cases of nested unlimited repeats that are never going to match.
4613 Problem: the scan can take a lot of time if the subject is very long (e.g.
4614 megabytes), thus penalizing straightforward matches. It is now done only if the
4615 amount of subject to be scanned is less than 1000 bytes.
4616
4617 7. A lesser problem with the same optimization is that it was recording the
4618 first character of an anchored pattern as "needed", thus provoking a search
4619 right along the subject, even when the first match of the pattern was going to
4620 fail. The "needed" character is now not set for anchored patterns, unless it
4621 follows something in the pattern that is of non-fixed length. Thus, it still
4622 fulfils its original purpose of finding quick non-matches in cases of nested
4623 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4624
4625
4626 Version 4.0 17-Feb-03
4627 ---------------------
4628
4629 1. If a comment in an extended regex that started immediately after a meta-item
4630 extended to the end of string, PCRE compiled incorrect data. This could lead to
4631 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4632
4633 2. Moved to autoconf 2.53 and libtool 1.4.2.
4634
4635 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4636 the special perltest8 script is no longer needed - all the tests can be run
4637 from a single perltest script.
4638
4639 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4640 by \s. It has now been removed in PCRE. This means it isn't recognized as
4641 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4642 class [:space:] *does* include VT, thereby creating a mess.
4643
4644 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4645 space and tab.
4646
4647 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4648 its new features into the main test script, reducing the number of scripts.
4649
4650 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4651 were backward compatible, and made the (?i) apply to the whole pattern, as if
4652 /i were given. Now it behaves more logically, and applies the option setting
4653 only to what follows. PCRE has been changed to follow suit. However, if it
4654 finds options settings right at the start of the pattern, it extracts them into
4655 the global options, as before. Thus, they show up in the info data.
4656
4657 8. Added support for the \Q...\E escape sequence. Characters in between are
4658 treated as literals. This is slightly different from Perl in that $ and @ are
4659 also handled as literals inside the quotes. In Perl, they will cause variable
4660 interpolation. Note the following examples:
4661
4662 Pattern PCRE matches Perl matches
4663
4664 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4665 \Qabc\$xyz\E abc\$xyz abc\$xyz
4666 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4667
4668 For compatibility with Perl, \Q...\E sequences are recognized inside character
4669 classes as well as outside them.
4670
4671 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4672 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4673 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4674 signed/unsigned warnings.
4675
4676 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4677 option for pcretest, so I've replaced it by a simple function that does just
4678 that job.
4679
4680 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4681 "pcregrep -".
4682
4683 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4684 Java package. This provides some syntactic sugar for simple cases of what my
4685 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4686 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4687 item, you can use this simplified notation. Note that only makes sense with
4688 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4689 greediness, whatever the setting of the PCRE_UNGREEDY option.
4690
4691 13. A change of greediness default within a pattern was not taking effect at
4692 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4693 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4694 was abstracted outside.
4695
4696 14. PCRE now supports the \G assertion. It is true when the current matching
4697 position is at the start point of the match. This differs from \A when the
4698 starting offset is non-zero. Used with the /g option of pcretest (or similar
4699 code), it works in the same way as it does for Perl's /g option. If all
4700 alternatives of a regex begin with \G, the expression is anchored to the start
4701 match position, and the "anchored" flag is set in the compiled expression.
4702
4703 15. Some bugs concerning the handling of certain option changes within patterns
4704 have been fixed. These applied to options other than (?ims). For example,
4705 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4706 been the other way round. Some of this was related to change 7 above.
4707
4708 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4709 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4710 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4711 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4712
4713 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4714 mode. Unlike ".", it always matches newline, whatever the setting of
4715 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4716 assertions. Perl allows it, but it doesn't (in general) work because it can't
4717 calculate the length of the lookbehind. At least, that's the case for Perl
4718 5.8.0 - I've been told they are going to document that it doesn't work in
4719 future.
4720
4721 18. Added an error diagnosis for escapes that PCRE does not support: these are
4722 \L, \l, \N, \P, \p, \U, \u, and \X.
4723
4724 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4725 reading past the end of the pattern in cases such as /[abcd/.
4726
4727 20. PCRE was getting more memory than necessary for patterns with classes that
4728 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4729
4730 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4731 compiling PCRE for use with Virtual Pascal.
4732
4733 22. Small fix to the Makefile to make it work properly if the build is done
4734 outside the source tree.
4735
4736 23. Added a new extension: a condition to go with recursion. If a conditional
4737 subpattern starts with (?(R) the "true" branch is used if recursion has
4738 happened, whereas the "false" branch is used only at the top level.
4739
4740 24. When there was a very long string of literal characters (over 255 bytes
4741 without UTF support, over 250 bytes with UTF support), the computation of how
4742 much memory was required could be incorrect, leading to segfaults or other
4743 strange effects.
4744
4745 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4746 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4747 there was a subsequent back reference to those brackets. This meant that, for
4748 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4749 possible to check for precisely this case. All we can do is abandon the
4750 optimization if .* occurs inside capturing brackets when there are any back
4751 references whatsoever. (See below for a better fix that came later.)
4752
4753 26. The handling of the optimization for finding the first character of a
4754 non-anchored pattern, and for finding a character that is required later in the
4755 match were failing in some cases. This didn't break the matching; it just
4756 failed to optimize when it could. The way this is done has been re-implemented.
4757
4758 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4759
4760 28. Added a new feature that provides some of the functionality that Perl
4761 provides with (?{...}). The facility is termed a "callout". The way it is done
4762 in PCRE is for the caller to provide an optional function, by setting
4763 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4764 global variable. By default it is unset, which disables all calling out. To get
4765 the function called, the regex must include (?C) at appropriate points. This
4766 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4767 This provides a means of identifying different callout points. When PCRE
4768 reaches such a point in the regex, if pcre_callout has been set, the external
4769 function is called. It is provided with data in a structure called
4770 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4771 matching continues; if it returns a non-zero value, the match at the current
4772 point fails. However, backtracking will occur if possible. [This was changed
4773 later and other features added - see item 49 below.]
4774
4775 29. pcretest is upgraded to test the callout functionality. It provides a
4776 callout function that displays information. By default, it shows the start of
4777 the match and the current position in the text. There are some new data escapes
4778 to vary what happens:
4779
4780 \C+ in addition, show current contents of captured substrings
4781 \C- do not supply a callout function
4782 \C!n return 1 when callout number n is reached
4783 \C!n!m return 1 when callout number n is reached for the mth time
4784
4785 30. If pcregrep was called with the -l option and just a single file name, it
4786 output "<stdin>" if a match was found, instead of the file name.
4787
4788 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4789 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4790 pcre_exec(). This saves a malloc/free per call. The default value of
4791 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4792 when configuring.
4793
4794 32. The default maximum size of a compiled pattern is 64K. There have been a
4795 few cases of people hitting this limit. The code now uses macros to handle the
4796 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4797 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4798 configuring. Tests 2 and 5 work only with 2-byte links because they output
4799 debugging information about compiled patterns.
4800
4801 33. Internal code re-arrangements:
4802
4803 (a) Moved the debugging function for printing out a compiled regex into
4804 its own source file (printint.c) and used #include to pull it into
4805 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4806 separate copies.
4807
4808 (b) Defined the list of op-code names for debugging as a macro in
4809 internal.h so that it is next to the definition of the opcodes.
4810
4811 (c) Defined a table of op-code lengths for simpler skipping along compiled
4812 code. This is again a macro in internal.h so that it is next to the
4813 definition of the opcodes.
4814
4815 34. Added support for recursive calls to individual subpatterns, along the
4816 lines of Robin Houston's patch (but implemented somewhat differently).
4817
4818 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4819 allow it to read and process whole directories in Win32. This code was
4820 contributed by Lionel Fourquaux; it has not been tested by me.
4821
4822 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4823 used to name a group. Names consist of alphanumerics and underscores, and must
4824 be unique. Back references use the syntax (?P=name) and recursive calls use
4825 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4826 numbers. The function pcre_fullinfo() can be used after compilation to extract
4827 a name/number map. There are three relevant calls:
4828
4829 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4830 PCRE_INFO_NAMECOUNT yields the number of entries
4831 PCRE_INFO_NAMETABLE yields a pointer to the map.
4832
4833 The map is a vector of fixed-size entries. The size of each entry depends on
4834 the length of the longest name used. The first two bytes of each entry are the
4835 group number, most significant byte first. There follows the corresponding
4836 name, zero terminated. The names are in alphabetical order.
4837
4838 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4839 case instead of 255. Making it the same both with and without UTF-8 support
4840 means that the same test output works with both.
4841
4842 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4843 calling malloc() with a zero argument.
4844
4845 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4846 optimization. I've improved things by keeping a bitmap of backreferences with
4847 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4848 fact referenced, the optimization can be applied. It is unlikely that a
4849 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4850 the match to follow \n) will appear inside brackets with a number greater than
4851 31, but if it does, any back reference > 31 suppresses the optimization.
4852
4853 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4854 of disabling numbered capturing parentheses. Any opening parenthesis that is
4855 not followed by ? behaves as if it were followed by ?: but named parentheses
4856 can still be used for capturing (and they will acquire numbers in the usual
4857 way).
4858
4859 41. Redesigned the return codes from the match() function into yes/no/error so
4860 that errors can be passed back from deep inside the nested calls. A malloc
4861 failure while inside a recursive subpattern call now causes the
4862 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4863
4864 42. It is now possible to set a limit on the number of times the match()
4865 function is called in a call to pcre_exec(). This facility makes it possible to
4866 limit the amount of recursion and backtracking, though not in a directly
4867 obvious way, because the match() function is used in a number of different
4868 circumstances. The count starts from zero for each position in the subject
4869 string (for non-anchored patterns). The default limit is, for compatibility, a
4870 large number, namely 10 000 000. You can change this in two ways:
4871
4872 (a) When configuring PCRE before making, you can use --with-match-limit=n
4873 to set a default value for the compiled library.
4874
4875 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4876 a different value is set. See 45 below.
4877
4878 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4879
4880 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4881 of things that can be changed at compile time. The first argument specifies
4882 what is wanted and the second points to where the information is to be placed.
4883 The current list of available information is:
4884
4885 PCRE_CONFIG_UTF8
4886
4887 The output is an integer that is set to one if UTF-8 support is available;
4888 otherwise it is set to zero.
4889
4890 PCRE_CONFIG_NEWLINE
4891
4892 The output is an integer that it set to the value of the code that is used for
4893 newline. It is either LF (10) or CR (13).
4894
4895 PCRE_CONFIG_LINK_SIZE
4896
4897 The output is an integer that contains the number of bytes used for internal
4898 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4899
4900 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4901
4902 The output is an integer that contains the threshold above which the POSIX
4903 interface uses malloc() for output vectors. See item 31 above.
4904
4905 PCRE_CONFIG_MATCH_LIMIT
4906
4907 The output is an unsigned integer that contains the default limit of the number
4908 of match() calls in a pcre_exec() execution. See 42 above.
4909
4910 44. pcretest has been upgraded by the addition of the -C option. This causes it
4911 to extract all the available output from the new pcre_config() function, and to
4912 output it. The program then exits immediately.
4913
4914 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4915 order to support additional features. One way would have been to define
4916 pcre_exec2() (for example) with extra arguments, but this would not have been
4917 extensible, and would also have required all calls to the original function to
4918 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4919 is used for passing in "extra" data from pcre_study().
4920
4921 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4922 contains the following fields:
4923
4924 flags a bitmap indicating which of the following fields are set
4925 study_data opaque data from pcre_study()
4926 match_limit a way of specifying a limit on match() calls for a specific
4927 call to pcre_exec()
4928 callout_data data for callouts (see 49 below)
4929
4930 The flag bits are also defined in pcre.h, and are
4931
4932 PCRE_EXTRA_STUDY_DATA
4933 PCRE_EXTRA_MATCH_LIMIT
4934 PCRE_EXTRA_CALLOUT_DATA
4935
4936 The pcre_study() function now returns one of these new pcre_extra blocks, with
4937 the actual study data pointed to by the study_data field, and the
4938 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4939 before. That is, this change is entirely upwards-compatible and requires no
4940 change to existing code.
4941
4942 If you want to pass in additional data to pcre_exec(), you can either place it
4943 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4944 block.
4945
4946 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4947 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4948 times with different match limits, until it finds the minimum value needed for
4949 pcre_exec() to complete. The value is then output. This can be instructive; for
4950 most simple matches the number is quite small, but for pathological cases it
4951 gets very large very quickly.
4952
4953 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4954 returns the size of the data block pointed to by the study_data field in a
4955 pcre_extra block, that is, the value that was passed as the argument to
4956 pcre_malloc() when PCRE was getting memory in which to place the information
4957 created by pcre_study(). The fourth argument should point to a size_t variable.
4958 pcretest has been extended so that this information is shown after a successful
4959 pcre_study() call when information about the compiled regex is being displayed.
4960
4961 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4962 because what follows is always an absolute path. (Later: it turns out that this
4963 is more than cosmetic for MinGW, because it doesn't like empty path
4964 components.)
4965
4966 49. Some changes have been made to the callout feature (see 28 above):
4967
4968 (i) A callout function now has three choices for what it returns:
4969
4970 0 => success, carry on matching
4971 > 0 => failure at this point, but backtrack if possible
4972 < 0 => serious error, return this value from pcre_exec()
4973
4974 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4975 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4976 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4977 use by callout functions. It will never be used by PCRE itself.
4978
4979 (ii) The pcre_extra structure (see 45 above) has a void * field called
4980 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4981 pcre_callout_block structure has a field of the same name. The contents of
4982 the field passed in the pcre_extra structure are passed to the callout
4983 function in the corresponding field in the callout block. This makes it
4984 easier to use the same callout-containing regex from multiple threads. For
4985 testing, the pcretest program has a new data escape
4986
4987 \C*n pass the number n (may be negative) as callout_data
4988
4989 If the callout function in pcretest receives a non-zero value as
4990 callout_data, it returns that value.
4991
4992 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4993 there were some redundant $(CFLAGS) in commands that are now specified as
4994 $(LINK), which already includes $(CFLAGS).
4995
4996 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4997 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4998 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4999 one-byte characters throughout. Note that case-insensitive matching applies
5000 only to characters whose values are less than 256. PCRE doesn't support the
5001 notion of cases for higher-valued characters.
5002
5003 (i) A character class whose characters are all within 0-255 is handled as
5004 a bit map, and the map is inverted for negative classes. Previously, a
5005 character > 255 always failed to match such a class; however it should
5006 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5007
5008 (ii) A negated character class with a single character < 255 is coded as
5009 "not this character" (OP_NOT). This wasn't working properly when the test
5010 character was multibyte, either singly or repeated.
5011
5012 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5013 mode, for example: \x{100}{2,3}.
5014
5015 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5016 singly or repeated) now correctly test multibyte characters. However,
5017 PCRE doesn't recognize any characters with values greater than 255 as
5018 digits, spaces, or word characters. Such characters always match \D, \S,
5019 and \W, and never match \d, \s, or \w.
5020
5021 (v) Classes may now contain characters and character ranges with values
5022 greater than 255. For example: [ab\x{100}-\x{400}].
5023
5024 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5025 PCRE in UTF-8 mode.
5026
5027 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5028 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5029 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5030 value.)
5031
5032 53. The single man page has become too large. I have therefore split it up into
5033 a number of separate man pages. These also give rise to individual HTML pages;
5034 these are now put in a separate directory, and there is an index.html page that
5035 lists them all. Some hyperlinking between the pages has been installed.
5036
5037 54. Added convenience functions for handling named capturing parentheses.
5038
5039 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5040 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5041 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5042 are faulted.
5043
5044 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5045 calling configure. These values are used when compiling the dftables.c program
5046 which is run to generate the source of the default character tables. They
5047 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5048 you will need to set these values.
5049
5050 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5051
5052
5053 Version 3.9 02-Jan-02
5054 ---------------------
5055
5056 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5057
5058 2. If --disable-static was given, the building process failed when trying to
5059 build pcretest and pcregrep. (For some reason it was using libtool to compile
5060 them, which is not right, as they aren't part of the library.)
5061
5062
5063 Version 3.8 18-Dec-01
5064 ---------------------
5065
5066 1. The experimental UTF-8 code was completely screwed up. It was packing the
5067 bytes in the wrong order. How dumb can you get?
5068
5069
5070 Version 3.7 29-Oct-01
5071 ---------------------
5072
5073 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5074 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5075 this didn't happen under Solaris 8, where I normally test things.
5076
5077 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5078 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5079 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5080 this fix an hour or so after the initial 3.7 release.)
5081
5082
5083 Version 3.6 23-Oct-01
5084 ---------------------
5085
5086 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5087 offsets passed as NULL with zero offset count.
5088
5089 2. The config.guess and config.sub files had not been updated when I moved to
5090 the latest autoconf.
5091
5092
5093 Version 3.5 15-Aug-01
5094 ---------------------
5095
5096 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5097 had been forgotten.
5098
5099 2. By using declared but undefined structures, we can avoid using "void"
5100 definitions in pcre.h while keeping the internal definitions of the structures
5101 private.
5102
5103 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5104 user point of view, this means that both static and shared libraries are built
5105 by default, but this can be individually controlled. More of the work of
5106 handling this static/shared cases is now inside libtool instead of PCRE's make
5107 file.
5108
5109 4. The pcretest utility is now installed along with pcregrep because it is
5110 useful for users (to test regexs) and by doing this, it automatically gets
5111 relinked by libtool. The documentation has been turned into a man page, so
5112 there are now .1, .txt, and .html versions in /doc.
5113
5114 5. Upgrades to pcregrep:
5115 (i) Added long-form option names like gnu grep.
5116 (ii) Added --help to list all options with an explanatory phrase.
5117 (iii) Added -r, --recursive to recurse into sub-directories.
5118 (iv) Added -f, --file to read patterns from a file.
5119
5120 6. pcre_exec() was referring to its "code" argument before testing that
5121 argument for NULL (and giving an error if it was NULL).
5122
5123 7. Upgraded Makefile.in to allow for compiling in a different directory from
5124 the source directory.
5125
5126 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5127 options bits, the pointer it was passed was to an int instead of to an unsigned
5128 long int. This mattered only on 64-bit systems.
5129
5130 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5131 generated) instead of pcre.in, which it its source. Also made the same change
5132 in several of the .c files.
5133
5134 10. A new release of gcc defines printf() as a macro, which broke pcretest
5135 because it had an ifdef in the middle of a string argument for printf(). Fixed
5136 by using separate calls to printf().
5137
5138 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5139 script, to force use of CR or LF instead of \n in the source. On non-Unix
5140 systems, the value can be set in config.h.
5141
5142 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5143 absolute limit. Changed the text of the error message to make this clear, and
5144 likewise updated the man page.
5145
5146 13. The limit of 99 on the number of capturing subpatterns has been removed.
5147 The new limit is 65535, which I hope will not be a "real" limit.
5148
5149
5150 Version 3.4 22-Aug-00
5151 ---------------------
5152
5153 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5154
5155 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5156
5157
5158 Version 3.3 01-Aug-00
5159 ---------------------
5160
5161 1. If an octal character was given, but the value was greater than \377, it
5162 was not getting masked to the least significant bits, as documented. This could
5163 lead to crashes in some systems.
5164
5165 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5166 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5167
5168 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5169 These just pass their arguments on to (pcre_free)(), but they are provided
5170 because some uses of PCRE bind it to non-C systems that can call its functions,
5171 but cannot call free() or pcre_free() directly.
5172
5173 4. Add "make test" as a synonym for "make check". Corrected some comments in
5174 the Makefile.
5175
5176 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5177 Makefile.
5178
5179 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5180 command called pgrep for grepping around the active processes.
5181
5182 7. Added the beginnings of support for UTF-8 character strings.
5183
5184 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5185 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5186 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5187 out for the ar command.)
5188
5189
5190 Version 3.2 12-May-00
5191 ---------------------
5192
5193 This is purely a bug fixing release.
5194
5195 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5196 of ZA. This was just one example of several cases that could provoke this bug,
5197 which was introduced by change 9 of version 2.00. The code for breaking
5198 infinite loops after an iteration that matches an empty string was't working
5199 correctly.
5200
5201 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5202 when matched against abbab (for example). After matching an empty string, it
5203 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5204 caused it to match further down the string than it should.
5205
5206 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5207 was there because it doesn't seem to be needed, and it causes trouble on some
5208 systems, as it is not a Standard C header. It has been removed.
5209
5210 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5211 were reported on the Macintosh. The changes were from
5212
5213 while ((c = *(++ptr)) != 0 && c != '\n');
5214 to
5215 while ((c = *(++ptr)) != 0 && c != '\n') ;
5216
5217 Totally extraordinary, but if that's what it takes...
5218
5219 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5220 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5221 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5222 assumes the way PCRE uses memmove() (always moving upwards).
5223
5224 6. PCRE is being used in one environment where strchr() is not available. There
5225 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5226 faster code anyway.
5227
5228
5229 Version 3.1 09-Feb-00
5230 ---------------------
5231
5232 The only change in this release is the fixing of some bugs in Makefile.in for
5233 the "install" target:
5234
5235 (1) It was failing to install pcreposix.h.
5236
5237 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5238
5239
5240 Version 3.0 01-Feb-00
5241 ---------------------
5242
5243 1. Add support for the /+ modifier to perltest (to output $` like it does in
5244 pcretest).
5245
5246 2. Add support for the /g modifier to perltest.
5247
5248 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5249 matches null strings.
5250
5251 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5252 pattern. Perl treats empty patterns specially - it reuses the most recent
5253 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5254 effect.
5255
5256 5. The POSIX interface was broken in that it was just handing over the POSIX
5257 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5258 required a bigger vector, with some working space on the end. This means that
5259 the POSIX wrapper now has to get and free some memory, and copy the results.
5260
5261 6. Added some simple autoconf support, placing the test data and the
5262 documentation in separate directories, re-organizing some of the
5263 information files, and making it build pcre-config (a GNU standard). Also added
5264 libtool support for building PCRE as a shared library, which is now the
5265 default.
5266
5267 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5268 09 are not valid octal constants. Single digits will be used for minor values
5269 less than 10.
5270
5271 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5272 existing programs that set these in the POSIX interface can use PCRE without
5273 modification.
5274
5275 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5276 return all that pcre_info() returns, plus additional data. The pcre_info()
5277 function is retained for compatibility, but is considered to be obsolete.
5278
5279 10. Added experimental recursion feature (?R) to handle one common case that
5280 Perl 5.6 will be able to do with (?p{...}).
5281
5282 11. Added support for POSIX character classes like [:alpha:], which Perl is
5283 adopting.
5284
5285
5286 Version 2.08 31-Aug-99
5287 ----------------------
5288
5289 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5290 trying to match at the startoffset position, but instead was moving forward to
5291 the next newline as if a previous match had failed.
5292
5293 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5294 and could get into a loop if a null string was matched other than at the start
5295 of the subject.
5296
5297 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5298 be distinguished at compile time, and for completeness also added PCRE_DATE.
5299
5300 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5301 in GnuWin32 environments.
5302
5303
5304 Version 2.07 29-Jul-99
5305 ----------------------
5306
5307 1. The documentation is now supplied in plain text form and HTML as well as in
5308 the form of man page sources.
5309
5310 2. C++ compilers don't like assigning (void *) values to other pointer types.
5311 In particular this affects malloc(). Although there is no problem in Standard
5312 C, I've put in casts to keep C++ compilers happy.
5313
5314 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5315 should be (const char *).
5316
5317 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5318 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5319 However, I haven't made this a standard facility. The documentation doesn't
5320 mention it, and the Makefile doesn't support it.
5321
5322 5. The Makefile now contains an "install" target, with editable destinations at
5323 the top of the file. The pcretest program is not installed.
5324
5325 6. pgrep -V now gives the PCRE version number and date.
5326
5327 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5328 causing the entire string to be ignored, instead of just the last character.
5329
5330 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5331 non-matching string, it can take a very, very long time, even for strings of
5332 quite modest length, because of the nested recursion. PCRE now does better in
5333 some of these cases. It does this by remembering the last required literal
5334 character in the pattern, and pre-searching the subject to ensure it is present
5335 before running the real match. In other words, it applies a heuristic to detect
5336 some types of certain failure quickly, and in the above example, if presented
5337 with a string that has no trailing " it gives "no match" very quickly.
5338
5339 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5340 other alternatives are tried instead.
5341
5342
5343 Version 2.06 09-Jun-99
5344 ----------------------
5345
5346 1. Change pcretest's output for amount of store used to show just the code
5347 space, because the remainder (the data block) varies in size between 32-bit and
5348 64-bit systems.
5349
5350 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5351 start matching at. This allows lookbehinds to work when searching for multiple
5352 occurrences in a string.
5353
5354 3. Added additional options to pcretest for testing multiple occurrences:
5355
5356 /+ outputs the rest of the string that follows a match
5357 /g loops for multiple occurrences, using the new startoffset argument
5358 /G loops for multiple occurrences by passing an incremented pointer
5359
5360 4. PCRE wasn't doing the "first character" optimization for patterns starting
5361 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5362 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5363 the letter 'x'. On long subject strings, this gives a significant speed-up.
5364
5365
5366 Version 2.05 21-Apr-99
5367 ----------------------
5368
5369 1. Changed the type of magic_number from int to long int so that it works
5370 properly on 16-bit systems.
5371
5372 2. Fixed a bug which caused patterns starting with .* not to work correctly
5373 when the subject string contained newline characters. PCRE was assuming
5374 anchoring for such patterns in all cases, which is not correct because .* will
5375 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5376 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5377 must be retried after every newline in the subject.
5378
5379
5380 Version 2.04 18-Feb-99
5381 ----------------------
5382
5383 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5384 computation of the store needed to hold the pattern was incorrect (too large).
5385 If such patterns were nested a few deep, this could multiply and become a real
5386 problem.
5387
5388 2. Added /M option to pcretest to show the memory requirement of a specific
5389 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5390
5391 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5392 compiled in such a way that the backtracking after subsequent failure was
5393 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5394 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5395
5396
5397 Version 2.03 02-Feb-99
5398 ----------------------
5399
5400 1. Fixed typo and small mistake in man page.
5401
5402 2. Added 4th condition (GPL supersedes if conflict) and created separate
5403 LICENCE file containing the conditions.
5404
5405 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5406 Perl, that is the internal \ allows the delimiter to be included in the
5407 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5408 the final delimiter, add \ to the end of the pattern (to test the error).
5409
5410 4. Added the convenience functions for extracting substrings after a successful
5411 match. Updated pcretest to make it able to test these functions.
5412
5413
5414 Version 2.02 14-Jan-99
5415 ----------------------
5416
5417 1. Initialized the working variables associated with each extraction so that
5418 their saving and restoring doesn't refer to uninitialized store.
5419
5420 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5421 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5422 fix the problem.
5423
5424 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5425 calls, and wasn't printing the correct value for compiling calls. Increased the
5426 default value of LOOPREPEAT, and the number of significant figures in the
5427 times.
5428
5429 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5430
5431 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5432 a building problem on Windows NT with a FAT file system.
5433
5434
5435 Version 2.01 21-Oct-98
5436 ----------------------
5437
5438 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5439 to character tables built by pcre_maketables() in the current locale. If NULL
5440 is passed, the default tables are used.
5441
5442
5443 Version 2.00 24-Sep-98
5444 ----------------------
5445
5446 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5447 it any more.
5448
5449 2. Allow quantification of (?>) groups, and make it work correctly.
5450
5451 3. The first character computation wasn't working for (?>) groups.
5452
5453 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5454 end of the subject) and add 5.005's \z, which really does match only at the
5455 very end of the subject.
5456
5457 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5458
5459 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5460 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5461 localized options. All options to pcre_study() were also removed.
5462
5463 7. Add other new features from 5.005:
5464
5465 $(?<= positive lookbehind
5466 $(?<! negative lookbehind
5467 (?imsx-imsx) added the unsetting capability
5468 such a setting is global if at outer level; local otherwise
5469 (?imsx-imsx:) non-capturing groups with option setting
5470 (?(cond)re|re) conditional pattern matching
5471
5472 A backreference to itself in a repeated group matches the previous
5473 captured string.
5474
5475 8. General tidying up of studying (both automatic and via "study")
5476 consequential on the addition of new assertions.
5477
5478 9. As in 5.005, unlimited repeated groups that could match an empty substring
5479 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5480 runtime if any iteration does actually match an empty substring.
5481
5482 10. Include the RunTest script in the distribution.
5483
5484 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5485 discrepancies, some of which were old and were also with respect to 5.004. They
5486 have now been fixed.
5487
5488
5489 Version 1.09 28-Apr-98
5490 ----------------------
5491
5492 1. A negated single character class followed by a quantifier with a minimum
5493 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5494 program crashes, or just wrong answers. This did not apply to negated classes
5495 containing more than one character, or to minima other than one.
5496
5497
5498 Version 1.08 27-Mar-98
5499 ----------------------
5500
5501 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5502
5503 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5504 latter must appear before anything that relies on it in the pattern.
5505
5506
5507 Version 1.07 16-Feb-98
5508 ----------------------
5509
5510 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5511 repeat of a potentially empty string).
5512
5513
5514 Version 1.06 23-Jan-98
5515 ----------------------
5516
5517 1. Added Markus Oberhumer's little patches for C++.
5518
5519 2. Literal strings longer than 255 characters were broken.
5520
5521
5522 Version 1.05 23-Dec-97
5523 ----------------------
5524
5525 1. Negated character classes containing more than one character were failing if
5526 PCRE_CASELESS was set at run time.
5527
5528
5529 Version 1.04 19-Dec-97
5530 ----------------------
5531
5532 1. Corrected the man page, where some "const" qualifiers had been omitted.
5533
5534 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5535 input syntax.
5536
5537 3. Fixed memory leak which occurred when a regex with back references was
5538 matched with an offsets vector that wasn't big enough. The temporary memory
5539 that is used in this case wasn't being freed if the match failed.
5540
5541 4. Tidied pcretest to ensure it frees memory that it gets.
5542
5543 5. Temporary memory was being obtained in the case where the passed offsets
5544 vector was exactly big enough.
5545
5546 6. Corrected definition of offsetof() from change 5 below.
5547
5548 7. I had screwed up change 6 below and broken the rules for the use of
5549 setjmp(). Now fixed.
5550
5551
5552 Version 1.03 18-Dec-97
5553 ----------------------
5554
5555 1. A erroneous regex with a missing opening parenthesis was correctly
5556 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5557 on some systems.
5558
5559 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5560 it was reported that one broken compiler failed on the former because "code" is
5561 also an independent variable.
5562
5563 3. The erroneous regex a[]b caused an array overrun reference.
5564
5565 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5566 fail on data ending with that character. (It was going on too far, and checking
5567 the next character, typically a binary zero.) This was specific to the
5568 optimized code for single-character negative classes.
5569
5570 5. Added a contributed patch from the TIN world which does the following:
5571
5572 + Add an undef for memmove, in case the the system defines a macro for it.
5573
5574 + Add a definition of offsetof(), in case there isn't one. (I don't know
5575 the reason behind this - offsetof() is part of the ANSI standard - but
5576 it does no harm).
5577
5578 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5579 most of the places where whitespace preceded '#'. I have given up and
5580 allowed the remaining 2 cases to be at the margin.
5581
5582 + Rename some variables in pcre to eliminate shadowing. This seems very
5583 pedantic, but does no harm, of course.
5584
5585 6. Moved the call to setjmp() into its own function, to get rid of warnings
5586 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5587
5588 7. Constructs such as \d{8,} were compiling into the equivalent of
5589 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5590 outcome, but in this particular case used more store than had been allocated,
5591 which caused the bug to be discovered because it threw up an internal error.
5592
5593 8. The debugging code in both pcre and pcretest for outputting the compiled
5594 form of a regex was going wrong in the case of back references followed by
5595 curly-bracketed repeats.
5596
5597
5598 Version 1.02 12-Dec-97
5599 ----------------------
5600
5601 1. Typos in pcre.3 and comments in the source fixed.
5602
5603 2. Applied a contributed patch to get rid of places where it used to remove
5604 'const' from variables, and fixed some signed/unsigned and uninitialized
5605 variable warnings.
5606
5607 3. Added the "runtest" target to Makefile.
5608
5609 4. Set default compiler flag to -O2 rather than just -O.
5610
5611
5612 Version 1.01 19-Nov-97
5613 ----------------------
5614
5615 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5616 like /([ab]*)*/, that is, for classes with more than one character in them.
5617
5618 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5619 as /((?>a*))*/ (a PCRE_EXTRA facility).
5620
5621
5622 Version 1.00 18-Nov-97
5623 ----------------------
5624
5625 1. Added compile-time macros to support systems such as SunOS4 which don't have
5626 memmove() or strerror() but have other things that can be used instead.
5627
5628 2. Arranged that "make clean" removes the executables.
5629
5630
5631 Version 0.99 27-Oct-97
5632 ----------------------
5633
5634 1. Fixed bug in code for optimizing classes with only one character. It was
5635 initializing a 32-byte map regardless, which could cause it to run off the end
5636 of the memory it had got.
5637
5638 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5639
5640
5641 Version 0.98 22-Oct-97
5642 ----------------------
5643
5644 1. Fixed bug in code for handling temporary memory usage when there are more
5645 back references than supplied space in the ovector. This could cause segfaults.
5646
5647
5648 Version 0.97 21-Oct-97
5649 ----------------------
5650
5651 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5652
5653 2. Optimized negated single characters not to use a bit map.
5654
5655 3. Brought error texts together as macro definitions; clarified some of them;
5656 fixed one that was wrong - it said "range out of order" when it meant "invalid
5657 escape sequence".
5658
5659 4. Changed some char * arguments to const char *.
5660
5661 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5662
5663 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5664 pcretest.
5665
5666
5667 Version 0.96 16-Oct-97
5668 ----------------------
5669
5670 1. Added a simple "pgrep" utility to the distribution.
5671
5672 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5673 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5674 where "ddd" means "one or more decimal digits".
5675
5676 3. Fixed serious bug. If a pattern had a back reference, but the call to
5677 pcre_exec() didn't supply a large enough ovector to record the related
5678 identifying subpattern, the match always failed. PCRE now remembers the number
5679 of the largest back reference, and gets some temporary memory in which to save
5680 the offsets during matching if necessary, in order to ensure that
5681 backreferences always work.
5682
5683 4. Increased the compatibility with Perl in a number of ways:
5684
5685 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5686 to request this handling. The option can be set at compile or exec time.
5687
5688 (b) $ matches before a terminating newline by default; an option
5689 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5690 mode). The option can be set at compile or exec time.
5691
5692 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5693 the same as Perl's. If the decimal number it represents is less than 10
5694 or there aren't that many previous left capturing parentheses, an octal
5695 escape is read. Inside a character class, it's always an octal escape,
5696 even if it is a single digit.
5697
5698 (d) An escaped but undefined alphabetic character is taken as a literal,
5699 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5700 escapes.
5701
5702 (e) {0} is now permitted. (The previous item is removed from the compiled
5703 pattern).
5704
5705 5. Changed all the names of code files so that the basic parts are no longer
5706 than 10 characters, and abolished the teeny "globals.c" file.
5707
5708 6. Changed the handling of character classes; they are now done with a 32-byte
5709 bit map always.
5710
5711 7. Added the -d and /D options to pcretest to make it possible to look at the
5712 internals of compilation without having to recompile pcre.
5713
5714
5715 Version 0.95 23-Sep-97
5716 ----------------------
5717
5718 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5719 \x20 at the start of a run of normal characters. These were being treated as
5720 real characters, instead of the source characters being re-checked.
5721
5722
5723 Version 0.94 18-Sep-97
5724 ----------------------
5725
5726 1. The functions are now thread-safe, with the caveat that the global variables
5727 containing pointers to malloc() and free() or alternative functions are the
5728 same for all threads.
5729
5730 2. Get pcre_study() to generate a bitmap of initial characters for non-
5731 anchored patterns when this is possible, and use it if passed to pcre_exec().
5732
5733
5734 Version 0.93 15-Sep-97
5735 ----------------------
5736
5737 1. /(b)|(:+)/ was computing an incorrect first character.
5738
5739 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5740 but not actually doing anything yet.
5741
5742 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5743 as Perl does (e.g. [-az] or [az-]).
5744
5745 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5746 all possible positions.
5747
5748 5. Split up into different modules to avoid including unneeded functions in a
5749 compiled binary. However, compile and exec are still in one module. The "study"
5750 function is split off.
5751
5752 6. The character tables are now in a separate module whose source is generated
5753 by an auxiliary program - but can then be edited by hand if required. There are
5754 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5755 toupper() in the code.
5756
5757 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5758 make them global. Abolish the function for setting them, as the caller can now
5759 set them directly.
5760
5761
5762 Version 0.92 11-Sep-97
5763 ----------------------
5764
5765 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5766 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5767
5768 2. Caseless matching was not working in character classes if the characters in
5769 the pattern were in upper case.
5770
5771 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5772
5773 4. Make PCRE_ANCHORED public and accept as a compile option.
5774
5775 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5776 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5777 pass them.
5778
5779 6. Give an error if bad option bits passed at compile or run time.
5780
5781 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5782 pcretest to cause it to pass that flag.
5783
5784 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5785 options, and the first character, if set.
5786
5787 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5788
5789
5790 Version 0.91 10-Sep-97
5791 ----------------------
5792
5793 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5794 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5795
5796 2. PCRE was looping on encountering an indefinitely repeated back reference to
5797 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5798 Perl does - treats the match as successful.
5799
5800 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5