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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1559 - (show annotations)
Sat May 16 11:05:40 2015 UTC (4 years, 6 months ago) by ph10
File size: 278599 byte(s)
Error occurred while calculating annotation data.
Fix named forward reference to duplicate group number overflow bug.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.38 xx-xxx-xxxx
5 ------------------------
6
7 1. If a group that contained a recursive back reference also contained a
8 forward reference subroutine call followed by a non-forward-reference
9 subroutine call, for example /.((?2)(?R)\1)()/, pcre2_compile() failed to
10 compile correct code, leading to undefined behaviour or an internally
11 detected error. This bug was discovered by the LLVM fuzzer.
12
13 2. Quantification of certain items (e.g. atomic back references) could cause
14 incorrect code to be compiled when recursive forward references were
15 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
16 This bug was discovered by the LLVM fuzzer.
17
18 3. A repeated conditional group whose condition was a reference by name caused
19 a buffer overflow if there was more than one group with the given name.
20 This bug was discovered by the LLVM fuzzer.
21
22 4. A recursive back reference by name within a group that had the same name as
23 another group caused a buffer overflow. For example:
24 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
25
26 30. A forward reference by name to a group whose number is the same as the
27 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
28 a buffer overflow at compile time. This bug was discovered by the LLVM
29 fuzzer.
30
31
32 Version 8.37 28-April-2015
33 --------------------------
34
35 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
36 for those parentheses to be closed with whatever has been captured so far.
37 However, it was failing to mark any other groups between the hightest
38 capture so far and the currrent group as "unset". Thus, the ovector for
39 those groups contained whatever was previously there. An example is the
40 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
41
42 2. If an assertion condition was quantified with a minimum of zero (an odd
43 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
44
45 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
46 unrecognized modifier, a crash could occur.
47
48 4. An attempt to do global matching in pcretest with a zero-length ovector
49 caused a crash.
50
51 5. Fixed a memory leak during matching that could occur for a subpattern
52 subroutine call (recursive or otherwise) if the number of captured groups
53 that had to be saved was greater than ten.
54
55 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
56 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
57 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
58
59 7. A UTF pattern containing a "not" match of a non-ASCII character and a
60 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
61
62 8. When a pattern is compiled, it remembers the highest back reference so that
63 when matching, if the ovector is too small, extra memory can be obtained to
64 use instead. A conditional subpattern whose condition is a check on a
65 capture having happened, such as, for example in the pattern
66 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
67 setting the highest backreference number. This mattered only if pcre_exec()
68 was called with an ovector that was too small to hold the capture, and there
69 was no other kind of back reference (a situation which is probably quite
70 rare). The effect of the bug was that the condition was always treated as
71 FALSE when the capture could not be consulted, leading to a incorrect
72 behaviour by pcre_exec(). This bug has been fixed.
73
74 9. A reference to a duplicated named group (either a back reference or a test
75 for being set in a conditional) that occurred in a part of the pattern where
76 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
77 to be incorrectly calculated, leading to overwriting.
78
79 10. A mutually recursive set of back references such as (\2)(\1) caused a
80 segfault at study time (while trying to find the minimum matching length).
81 The infinite loop is now broken (with the minimum length unset, that is,
82 zero).
83
84 11. If an assertion that was used as a condition was quantified with a minimum
85 of zero, matching went wrong. In particular, if the whole group had
86 unlimited repetition and could match an empty string, a segfault was
87 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
88 assertions to be quantified, but not if they are being used as conditions,
89 so the above pattern is faulted by Perl. PCRE has now been changed so that
90 it also rejects such patterns.
91
92 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
93 failed to allow the zero-repeat case if pcre2_exec() was called with an
94 ovector too small to capture the group.
95
96 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
97 Red Hat Product Security:
98
99 (a) A crash if /K and /F were both set with the option to save the compiled
100 pattern.
101
102 (b) Another crash if the option to print captured substrings in a callout
103 was combined with setting a null ovector, for example \O\C+ as a subject
104 string.
105
106 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
107 forward reference repeated a large (but limited) number of times within a
108 repeated outer group that has a zero minimum quantifier, caused incorrect
109 code to be compiled, leading to the error "internal error:
110 previously-checked referenced subpattern not found" when an incorrect
111 memory address was read. This bug was reported as "heap overflow",
112 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
113 CVE-2015-2325.
114
115 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
116 call within a group that also contained a recursive back reference caused
117 incorrect code to be compiled. This bug was reported as "heap overflow",
118 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
119 number CVE-2015-2326.
120
121 24. Computing the size of the JIT read-only data in advance has been a source
122 of various issues, and new ones are still appear unfortunately. To fix
123 existing and future issues, size computation is eliminated from the code,
124 and replaced by on-demand memory allocation.
125
126 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
127 adjacent to the end of the range, and the range contained characters with
128 more than one other case, caused incorrect behaviour when compiled in UTF
129 mode. In that example, the range a-j was left out of the class.
130
131 26. Fix JIT compilation of conditional blocks, which assertion
132 is converted to (*FAIL). E.g: /(?(?!))/.
133
134 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
135 discovered by the LLVM fuzzer.
136
137 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
138 when this assertion was used as a condition, for example (?(?!)a|b). In
139 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
140 error about an unsupported item.
141
142 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
143 possessification code could take exponential time to complete. A recursion
144 depth limit of 1000 has been imposed to limit the resources used by this
145 optimization.
146
147 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
148 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
149 because \S ensures they are all in the class. The code for doing this was
150 interacting badly with the code for computing the amount of space needed to
151 compile the pattern, leading to a buffer overflow. This bug was discovered
152 by the LLVM fuzzer.
153
154 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
155 other kinds of group caused stack overflow at compile time. This bug was
156 discovered by the LLVM fuzzer.
157
158 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
159 between a subroutine call and its quantifier was incorrectly compiled,
160 leading to buffer overflow or other errors. This bug was discovered by the
161 LLVM fuzzer.
162
163 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
164 assertion after (?(. The code was failing to check the character after
165 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
166 was discovered by the LLVM fuzzer.
167
168 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
169 a fixed maximum following a group that contains a subroutine reference was
170 incorrectly compiled and could trigger buffer overflow. This bug was
171 discovered by the LLVM fuzzer.
172
173 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
174 caused a stack overflow instead of the diagnosis of a non-fixed length
175 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
176
177 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
178 (e.g. /(?<=\Ka)/) could make pcregrep loop.
179
180 37. There was a similar problem to 36 in pcretest for global matches.
181
182 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
183 and a subsequent item in the pattern caused a non-match, backtracking over
184 the repeated \X did not stop, but carried on past the start of the subject,
185 causing reference to random memory and/or a segfault. There were also some
186 other cases where backtracking after \C could crash. This set of bugs was
187 discovered by the LLVM fuzzer.
188
189 39. The function for finding the minimum length of a matching string could take
190 a very long time if mutual recursion was present many times in a pattern,
191 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
192 method has been implemented. This infelicity was discovered by the LLVM
193 fuzzer.
194
195 40. Static linking against the PCRE library using the pkg-config module was
196 failing on missing pthread symbols.
197
198
199 Version 8.36 26-September-2014
200 ------------------------------
201
202 1. Got rid of some compiler warnings in the C++ modules that were shown up by
203 -Wmissing-field-initializers and -Wunused-parameter.
204
205 2. The tests for quantifiers being too big (greater than 65535) were being
206 applied after reading the number, and stupidly assuming that integer
207 overflow would give a negative number. The tests are now applied as the
208 numbers are read.
209
210 3. Tidy code in pcre_exec.c where two branches that used to be different are
211 now the same.
212
213 4. The JIT compiler did not generate match limit checks for certain
214 bracketed expressions with quantifiers. This may lead to exponential
215 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
216 issue should be resolved now.
217
218 5. Fixed an issue, which occures when nested alternatives are optimized
219 with table jumps.
220
221 6. Inserted two casts and changed some ints to size_t in the light of some
222 reported 64-bit compiler warnings (Bugzilla 1477).
223
224 7. Fixed a bug concerned with zero-minimum possessive groups that could match
225 an empty string, which sometimes were behaving incorrectly in the
226 interpreter (though correctly in the JIT matcher). This pcretest input is
227 an example:
228
229 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
230 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
231
232 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
233 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
234 for an empty string was breaking the inner loop and carrying on at a lower
235 level, when possessive repeated groups should always return to a higher
236 level as they have no backtrack points in them. The empty string test now
237 occurs at the outer level.
238
239 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
240 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
241
242 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
243 doesn't).
244
245 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
246 to hold a pattern to be incorrectly computed (too small) when there were
247 named back references to duplicated names. This could cause "internal
248 error: code overflow" or "double free or corruption" or other memory
249 handling errors.
250
251 11. When named subpatterns had the same prefixes, back references could be
252 confused. For example, in this pattern:
253
254 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
255
256 the reference to 'Name' was incorrectly treated as a reference to a
257 duplicate name.
258
259 12. A pattern such as /^s?c/mi8 where the optional character has more than
260 one "other case" was incorrectly compiled such that it would only try to
261 match starting at "c".
262
263 13. When a pattern starting with \s was studied, VT was not included in the
264 list of possible starting characters; this should have been part of the
265 8.34/18 patch.
266
267 14. If a character class started [\Qx]... where x is any character, the class
268 was incorrectly terminated at the ].
269
270 15. If a pattern that started with a caseless match for a character with more
271 than one "other case" was studied, PCRE did not set up the starting code
272 unit bit map for the list of possible characters. Now it does. This is an
273 optimization improvement, not a bug fix.
274
275 16. The Unicode data tables have been updated to Unicode 7.0.0.
276
277 17. Fixed a number of memory leaks in pcregrep.
278
279 18. Avoid a compiler warning (from some compilers) for a function call with
280 a cast that removes "const" from an lvalue by using an intermediate
281 variable (to which the compiler does not object).
282
283 19. Incorrect code was compiled if a group that contained an internal recursive
284 back reference was optional (had quantifier with a minimum of zero). This
285 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
286 caused segmentation faults because of stack overflows at compile time.
287
288 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
289 group that is quantified with an indefinite repeat, caused a compile-time
290 loop which used up all the system stack and provoked a segmentation fault.
291 This was not the same bug as 19 above.
292
293 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
294 Patch by Mike Frysinger.
295
296
297 Version 8.35 04-April-2014
298 --------------------------
299
300 1. A new flag is set, when property checks are present in an XCLASS.
301 When this flag is not set, PCRE can perform certain optimizations
302 such as studying these XCLASS-es.
303
304 2. The auto-possessification of character sets were improved: a normal
305 and an extended character set can be compared now. Furthermore
306 the JIT compiler optimizes more character set checks.
307
308 3. Got rid of some compiler warnings for potentially uninitialized variables
309 that show up only when compiled with -O2.
310
311 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
312 of a match later then the end of the match. The pcretest program was not
313 handling the case sensibly - it was outputting from the start to the next
314 binary zero. It now reports this situation in a message, and outputs the
315 text from the end to the start.
316
317 5. Fast forward search is improved in JIT. Instead of the first three
318 characters, any three characters with fixed position can be searched.
319 Search order: first, last, middle.
320
321 6. Improve character range checks in JIT. Characters are read by an inprecise
322 function now, which returns with an unknown value if the character code is
323 above a certain threshold (e.g: 256). The only limitation is that the value
324 must be bigger than the threshold as well. This function is useful when
325 the characters above the threshold are handled in the same way.
326
327 7. The macros whose names start with RAWUCHAR are placeholders for a future
328 mode in which only the bottom 21 bits of 32-bit data items are used. To
329 make this more memorable for those maintaining the code, the names have
330 been changed to start with UCHAR21, and an extensive comment has been added
331 to their definition.
332
333 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
334 to the export list in Makefile.am (they were accidentally omitted from the
335 8.34 tarball).
336
337 9. The informational output from pcretest used the phrase "starting byte set"
338 which is inappropriate for the 16-bit and 32-bit libraries. As the output
339 for "first char" and "need char" really means "non-UTF-char", I've changed
340 "byte" to "char", and slightly reworded the output. The documentation about
341 these values has also been (I hope) clarified.
342
343 10. Another JIT related optimization: use table jumps for selecting the correct
344 backtracking path, when more than four alternatives are present inside a
345 bracket.
346
347 11. Empty match is not possible, when the minimum length is greater than zero,
348 and there is no \K in the pattern. JIT should avoid empty match checks in
349 such cases.
350
351 12. In a caseless character class with UCP support, when a character with more
352 than one alternative case was not the first character of a range, not all
353 the alternative cases were added to the class. For example, s and \x{17f}
354 are both alternative cases for S: the class [RST] was handled correctly,
355 but [R-T] was not.
356
357 13. The configure.ac file always checked for pthread support when JIT was
358 enabled. This is not used in Windows, so I have put this test inside a
359 check for the presence of windows.h (which was already tested for).
360
361 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
362 The algorithm provides a way to skip certain starting offsets, and usually
363 faster than linear prefix searches.
364
365 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
366 as for 'fr_FR' and 'french'. For some reason, however, it then used the
367 Windows-specific input and output files, which have 'french' screwed in.
368 So this could never have worked. One of the problems with locales is that
369 they aren't always the same. I have now updated RunTest so that it checks
370 the output of the locale test (test 3) against three different output
371 files, and it allows the test to pass if any one of them matches. With luck
372 this should make the test pass on some versions of Solaris where it was
373 failing. Because of the uncertainty, the script did not used to stop if
374 test 3 failed; it now does. If further versions of a French locale ever
375 come to light, they can now easily be added.
376
377 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
378 there was a message during ./configure, but it did not stop. This now
379 provokes an error. The invalid example in README has been corrected.
380 If a value less than the minimum is given, the minimum value has always
381 been used, but now a warning is given.
382
383 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
384 was a bug in the test system, which is now fixed. Also, the list of various
385 configurations that are tested for each release did not have one with both
386 16/32 bits and --enable-bar-anycrlf. It now does.
387
388 18. pcretest was missing "-C bsr" for displaying the \R default setting.
389
390 19. Little endian PowerPC systems are supported now by the JIT compiler.
391
392 20. The fast forward newline mechanism could enter to an infinite loop on
393 certain invalid UTF-8 input. Although we don't support these cases
394 this issue can be fixed by a performance optimization.
395
396 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
397 not take account if existing stack usage. There is now a new global
398 variable called pcre_stack_guard that can be set to point to an external
399 function to check stack availability. It is called at the start of
400 processing every parenthesized group.
401
402 22. A typo in the code meant that in ungreedy mode the max/min qualifier
403 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
404 not match "ab".
405
406 23. When UTF was disabled, the JIT program reported some incorrect compile
407 errors. These messages are silenced now.
408
409 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
410 compiler.
411
412 25. Change all the temporary files used in RunGrepTest to be different to those
413 used by RunTest so that the tests can be run simultaneously, for example by
414 "make -j check".
415
416
417 Version 8.34 15-December-2013
418 -----------------------------
419
420 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
421 executable memory. Patch inspired by Carsten Klein.
422
423 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
424 this macro is never tested and has no effect, because the work to support
425 coverage involves only compiling and linking options and special targets in
426 the Makefile. The comment in config.h implied that defining the macro would
427 enable coverage support, which is totally false. There was also support for
428 setting this macro in the CMake files (my fault, I just copied it from
429 configure). SUPPORT_GCOV has now been removed.
430
431 3. Make a small performance improvement in strlen16() and strlen32() in
432 pcretest.
433
434 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
435 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
436 these cases). There was also one in pcretest.c.
437
438 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
439
440 6. In UTF mode, the code for checking whether a group could match an empty
441 string (which is used for indefinitely repeated groups to allow for
442 breaking an infinite loop) was broken when the group contained a repeated
443 negated single-character class with a character that occupied more than one
444 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
445 UTF-8 mode). The effect was undefined: the group might or might not be
446 deemed as matching an empty string, or the program might have crashed.
447
448 7. The code for checking whether a group could match an empty string was not
449 recognizing that \h, \H, \v, \V, and \R must match a character.
450
451 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
452 an empty string. If it can, pcretest shows this in its information output.
453
454 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
455 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
456 matched by pcre_exec() without using JIT:
457
458 (a) If the rest of the pattern did not match after a maximal run of
459 grapheme clusters, the code for backing up to try with fewer of them
460 did not always back up over a full grapheme when characters that do not
461 have the modifier quality were involved, e.g. Hangul syllables.
462
463 (b) If the match point in a subject started with modifier character, and
464 there was no match, the code could incorrectly back up beyond the match
465 point, and potentially beyond the first character in the subject,
466 leading to a segfault or an incorrect match result.
467
468 10. A conditional group with an assertion condition could lead to PCRE
469 recording an incorrect first data item for a match if no other first data
470 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
471 first data item, and therefore matched "ca" after "c" instead of at the
472 start.
473
474 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
475 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
476
477 12. The source of pcregrep now includes z/OS-specific code so that it can be
478 compiled for z/OS as part of the special z/OS distribution.
479
480 13. Added the -T and -TM options to pcretest.
481
482 14. The code in pcre_compile.c for creating the table of named capturing groups
483 has been refactored. Instead of creating the table dynamically during the
484 actual compiling pass, the information is remembered during the pre-compile
485 pass (on the stack unless there are more than 20 named groups, in which
486 case malloc() is used) and the whole table is created before the actual
487 compile happens. This has simplified the code (it is now nearly 150 lines
488 shorter) and prepared the way for better handling of references to groups
489 with duplicate names.
490
491 15. A back reference to a named subpattern when there is more than one of the
492 same name now checks them in the order in which they appear in the pattern.
493 The first one that is set is used for the reference. Previously only the
494 first one was inspected. This change makes PCRE more compatible with Perl.
495
496 16. Unicode character properties were updated from Unicode 6.3.0.
497
498 17. The compile-time code for auto-possessification has been refactored, based
499 on a patch by Zoltan Herczeg. It now happens after instead of during
500 compilation. The code is cleaner, and more cases are handled. The option
501 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
502 options in pcretest are provided to set it. It can also be set by
503 (*NO_AUTO_POSSESS) at the start of a pattern.
504
505 18. The character VT has been added to the default ("C" locale) set of
506 characters that match \s and are generally treated as white space,
507 following this same change in Perl 5.18. There is now no difference between
508 "Perl space" and "POSIX space". Whether VT is treated as white space in
509 other locales depends on the locale.
510
511 19. The code for checking named groups as conditions, either for being set or
512 for being recursed, has been refactored (this is related to 14 and 15
513 above). Processing unduplicated named groups should now be as fast at
514 numerical groups, and processing duplicated groups should be faster than
515 before.
516
517 20. Two patches to the CMake build system, by Alexander Barkov:
518
519 (1) Replace the "source" command by "." in CMakeLists.txt because
520 "source" is a bash-ism.
521
522 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
523 without these the CMake build does not work on Solaris.
524
525 21. Perl has changed its handling of \8 and \9. If there is no previously
526 encountered capturing group of those numbers, they are treated as the
527 literal characters 8 and 9 instead of a binary zero followed by the
528 literals. PCRE now does the same.
529
530 22. Following Perl, added \o{} to specify codepoints in octal, making it
531 possible to specify values greater than 0777 and also making them
532 unambiguous.
533
534 23. Perl now gives an error for missing closing braces after \x{... instead of
535 treating the string as literal. PCRE now does the same.
536
537 24. RunTest used to grumble if an inappropriate test was selected explicitly,
538 but just skip it when running all tests. This make it awkward to run ranges
539 of tests when one of them was inappropriate. Now it just skips any
540 inappropriate tests, as it always did when running all tests.
541
542 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
543 character types such as \d or \w, too many callouts were inserted, and the
544 data that they returned was rubbish.
545
546 26. In UCP mode, \s was not matching two of the characters that Perl matches,
547 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
548 were matched by \h. The code has now been refactored so that the lists of
549 the horizontal and vertical whitespace characters used for \h and \v (which
550 are defined only in one place) are now also used for \s.
551
552 27. Add JIT support for the 64 bit TileGX architecture.
553 Patch by Jiong Wang (Tilera Corporation).
554
555 28. Possessive quantifiers for classes (both explicit and automatically
556 generated) now use special opcodes instead of wrapping in ONCE brackets.
557
558 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
559 (because it's meaningless), this was not happening when PCRE_CASELESS was
560 set. Not wrong, but inefficient.
561
562 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
563 properties for \w, \d, etc) is present in a test regex. Otherwise if the
564 test contains no characters greater than 255, Perl doesn't realise it
565 should be using Unicode semantics.
566
567 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
568 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
569 does in Unicode mode.
570
571 32. Added the "forbid" facility to pcretest so that putting tests into the
572 wrong test files can sometimes be quickly detected.
573
574 33. There is now a limit (default 250) on the depth of nesting of parentheses.
575 This limit is imposed to control the amount of system stack used at compile
576 time. It can be changed at build time by --with-parens-nest-limit=xxx or
577 the equivalent in CMake.
578
579 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
580 errors. Perl warns for these when in warning mode, but PCRE has no facility
581 for giving warnings.
582
583 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
584 However, this was not working for (?!) because it is optimized to (*FAIL),
585 for which PCRE does not allow quantifiers. The optimization is now disabled
586 when a quantifier follows (?!). I can't see any use for this, but it makes
587 things uniform.
588
589 36. Perl no longer allows group names to start with digits, so I have made this
590 change also in PCRE. It simplifies the code a bit.
591
592 37. In extended mode, Perl ignores spaces before a + that indicates a
593 possessive quantifier. PCRE allowed a space before the quantifier, but not
594 before the possessive +. It now does.
595
596 38. The use of \K (reset reported match start) within a repeated possessive
597 group such as (a\Kb)*+ was not working.
598
599 40. Document that the same character tables must be used at compile time and
600 run time, and that the facility to pass tables to pcre_exec() and
601 pcre_dfa_exec() is for use only with saved/restored patterns.
602
603 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
604 features for Builds with MSVC:
605
606 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
607 builds). The .rc files can be used to set FileDescription and many other
608 attributes.
609
610 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
611 This allows higher-level build scripts which want .pdb files to avoid
612 hard-coding the exact files needed."
613
614 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
615 mean "start of word" and "end of word", respectively, as a transition aid.
616
617 43. A minimizing repeat of a class containing codepoints greater than 255 in
618 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
619 compiled to use the heap for recursion.
620
621 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
622 is configured.
623
624
625 Version 8.33 28-May-2013
626 ------------------------
627
628 1. Added 'U' to some constants that are compared to unsigned integers, to
629 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
630 variables that are added to signed variables, to ensure the result is
631 signed and can be negated.
632
633 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
634 CMake config files.
635
636 3. Revise the creation of config.h.generic so that all boolean macros are
637 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
638 overriding via -D on the command line possible.
639
640 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
641 to unsigned int is reported to make a quite noticeable speed difference in
642 a specific Windows environment. Testing on Linux did also appear to show
643 some benefit (and it is clearly not harmful). Also fixed the definition of
644 Xop which should be unsigned.
645
646 5. Related to (4), changing the definition of the intermediate variable cc
647 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
648 improvements.
649
650 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
651 unnecessary spaces.
652
653 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
654 and later.
655
656 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
657
658 9. Optimizing fast_forward_start_bits in JIT.
659
660 10. Adding support for callouts in JIT, and fixing some issues revealed
661 during this work. Namely:
662
663 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
664
665 (b) Minimum length was not checked before the matching is started.
666
667 11. The value of capture_last that is passed to callouts was incorrect in some
668 cases when there was a capture on one path that was subsequently abandoned
669 after a backtrack. Also, the capture_last value is now reset after a
670 recursion, since all captures are also reset in this case.
671
672 12. The interpreter no longer returns the "too many substrings" error in the
673 case when an overflowing capture is in a branch that is subsequently
674 abandoned after a backtrack.
675
676 13. In the pathological case when an offset vector of size 2 is used, pcretest
677 now prints out the matched string after a yield of 0 or 1.
678
679 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
680 Only supported by the JIT compiler at the moment.
681
682 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
683
684 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
685 offset of the starting point of the matching process, provided the offsets
686 vector is large enough.
687
688 17. The \A escape now records a lookbehind value of 1, though its execution
689 does not actually inspect the previous character. This is to ensure that,
690 in partial multi-segment matching, at least one character from the old
691 segment is retained when a new segment is processed. Otherwise, if there
692 are no lookbehinds in the pattern, \A might match incorrectly at the start
693 of a new segment.
694
695 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
696
697 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
698 gives some modest performance improvement in 8-bit mode.
699
700 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
701 be expressed in certain programming languages using Universal Character
702 Names.
703
704 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
705 which points out that "non characters" are not "characters that may not
706 appear in Unicode strings" but rather "characters that are reserved for
707 internal use and have only local meaning".
708
709 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
710 there was a conditional group that depended on an assertion, if the
711 assertion was false, the callout that immediately followed the alternation
712 in the condition was skipped when pcre_exec() was used for matching.
713
714 23. Allow an explicit callout to be inserted before an assertion that is the
715 condition for a conditional group, for compatibility with automatic
716 callouts, which always insert a callout at this point.
717
718 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
719 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
720
721 25. (*PRUNE) is now supported by the JIT compiler.
722
723 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
724
725 27. Fix the case where there are two or more SKIPs with arguments that may be
726 ignored.
727
728 28. (*SKIP) is now supported by the JIT compiler.
729
730 29. (*THEN) is now supported by the JIT compiler.
731
732 30. Update RunTest with additional test selector options.
733
734 31. The way PCRE handles backtracking verbs has been changed in two ways.
735
736 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
737 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
738 backtracking. In some cases this makes it more Perl-compatible, but Perl's
739 rather obscure rules do not always do the same thing.
740
741 (2) Previously, backtracking verbs were confined within assertions. This is
742 no longer the case for positive assertions, except for (*ACCEPT). Again,
743 this sometimes improves Perl compatibility, and sometimes does not.
744
745 32. A number of tests that were in test 2 because Perl did things differently
746 have been moved to test 1, because either Perl or PCRE has changed, and
747 these tests are now compatible.
748
749 32. Backtracking control verbs are now handled in the same way in JIT and
750 interpreter.
751
752 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
753 contained a forward subroutine reference caused a compile error.
754
755 34. Auto-detect and optimize limited repetitions in JIT.
756
757 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
758 blocking (*UTF) etc.
759
760 36. In the interpreter, maximizing pattern repetitions for characters and
761 character types now use tail recursion, which reduces stack usage.
762
763 37. The value of the max lookbehind was not correctly preserved if a compiled
764 and saved regex was reloaded on a host of different endianness.
765
766 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
767 of the compiled pattern block, expand the flags field from 16 to 32 bits
768 because it was almost full.
769
770 39. Try madvise first before posix_madvise.
771
772 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
773 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
774 original change locked out all matches of empty strings. This has been
775 changed so that one match of an empty string per line is recognized.
776 Subsequent searches on the same line (for colouring or for --only-matching,
777 for example) do not recognize empty strings.
778
779 41. Applied a user patch to fix a number of spelling mistakes in comments.
780
781 42. Data lines longer than 65536 caused pcretest to crash.
782
783 43. Clarified the data type for length and startoffset arguments for pcre_exec
784 and pcre_dfa_exec in the function-specific man pages, where they were
785 explicitly stated to be in bytes, never having been updated. I also added
786 some clarification to the pcreapi man page.
787
788 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
789 a segmentation fault.
790
791
792 Version 8.32 30-November-2012
793 -----------------------------
794
795 1. Improved JIT compiler optimizations for first character search and single
796 character iterators.
797
798 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
799 Patch by Daniel Richard G.
800
801 3. Single character iterator optimizations in the JIT compiler.
802
803 4. Improved JIT compiler optimizations for character ranges.
804
805 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
806 Reported by Giuseppe D'Angelo.
807
808 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
809 of a line, was being set incorrectly in cases where .* appeared inside
810 atomic brackets at the start of a pattern, or where there was a subsequent
811 *PRUNE or *SKIP.
812
813 7. Improved instruction cache flush for POWER/PowerPC.
814 Patch by Daniel Richard G.
815
816 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
817 grep:
818
819 (a) There is now no limit to the number of patterns to be matched.
820
821 (b) An error is given if a pattern is too long.
822
823 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
824 are now supported.
825
826 (d) --exclude-from and --include-from (multiple use) have been added.
827
828 (e) Exclusions and inclusions now apply to all files and directories, not
829 just to those obtained from scanning a directory recursively.
830
831 (f) Multiple uses of -f and --file-list are now supported.
832
833 (g) In a Windows environment, the default for -d has been changed from
834 "read" (the GNU grep default) to "skip", because otherwise the presence
835 of a directory in the file list provokes an error.
836
837 (h) The documentation has been revised and clarified in places.
838
839 9. Improve the matching speed of capturing brackets.
840
841 10. Changed the meaning of \X so that it now matches a Unicode extended
842 grapheme cluster.
843
844 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
845 out POSIX threads when JIT support is configured.
846
847 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
848
849 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
850 structure could go wrong in environments where size_t is not the same size
851 as int.
852
853 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
854 set.
855
856 15. The EBCDIC support had decayed; later updates to the code had included
857 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
858 general tidy up of EBCDIC-related issues, and the documentation was also
859 not quite right. There is now a test that can be run on ASCII systems to
860 check some of the EBCDIC-related things (but is it not a full test).
861
862 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
863 in a small tidy to the code.
864
865 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
866
867 18. If the --only-matching (-o) option in pcregrep is specified multiple
868 times, each one causes appropriate output. For example, -o1 -o2 outputs the
869 substrings matched by the 1st and 2nd capturing parentheses. A separating
870 string can be specified by --om-separator (default empty).
871
872 19. Improving the first n character searches.
873
874 20. Turn case lists for horizontal and vertical white space into macros so that
875 they are defined only once.
876
877 21. This set of changes together give more compatible Unicode case-folding
878 behaviour for characters that have more than one other case when UCP
879 support is available.
880
881 (a) The Unicode property table now has offsets into a new table of sets of
882 three or more characters that are case-equivalent. The MultiStage2.py
883 script that generates these tables (the pcre_ucd.c file) now scans
884 CaseFolding.txt instead of UnicodeData.txt for character case
885 information.
886
887 (b) The code for adding characters or ranges of characters to a character
888 class has been abstracted into a generalized function that also handles
889 case-independence. In UTF-mode with UCP support, this uses the new data
890 to handle characters with more than one other case.
891
892 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
893 whose other case is greater than 256 are now correctly matched
894 caselessly. Previously, the high codepoint matched the low one, but not
895 vice versa.
896
897 (d) The processing of \h, \H, \v, and \ in character classes now makes use
898 of the new class addition function, using character lists defined as
899 macros alongside the case definitions of 20 above.
900
901 (e) Caseless back references now work with characters that have more than
902 one other case.
903
904 (f) General caseless matching of characters with more than one other case
905 is supported.
906
907 22. Unicode character properties were updated from Unicode 6.2.0
908
909 23. Improved CMake support under Windows. Patch by Daniel Richard G.
910
911 24. Add support for 32-bit character strings, and UTF-32
912
913 25. Major JIT compiler update (code refactoring and bugfixing).
914 Experimental Sparc 32 support is added.
915
916 26. Applied a modified version of Daniel Richard G's patch to create
917 pcre.h.generic and config.h.generic by "make" instead of in the
918 PrepareRelease script.
919
920 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
921 pcre_compile.c when checking for a zero character.
922
923 28. Introducing a native interface for JIT. Through this interface, the compiled
924 machine code can be directly executed. The purpose of this interface is to
925 provide fast pattern matching, so several sanity checks are not performed.
926 However, feature tests are still performed. The new interface provides
927 1.4x speedup compared to the old one.
928
929 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
930 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
931 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
932
933 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
934 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
935 the "old" RFC 2279). Instead, it ended up passing a negative length to
936 pcre_exec().
937
938 31. Add support for GCC's visibility feature to hide internal functions.
939
940 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
941 "unknown -C option" after outputting 0 or 1.
942
943 33. There is now support for generating a code coverage report for the test
944 suite in environments where gcc is the compiler and lcov is installed. This
945 is mainly for the benefit of the developers.
946
947 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
948 unaddressable using valgrind annotations, allowing valgrind to detect
949 invalid memory accesses. This is mainly for the benefit of the developers.
950
951 25. (*UTF) can now be used to start a pattern in any of the three libraries.
952
953 26. Give configure error if --enable-cpp but no C++ compiler found.
954
955
956 Version 8.31 06-July-2012
957 -------------------------
958
959 1. Fixing a wrong JIT test case and some compiler warnings.
960
961 2. Removed a bashism from the RunTest script.
962
963 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
964 to unsigned type, result still unsigned" that was given by an MS compiler
965 on encountering the code "-sizeof(xxx)".
966
967 4. Partial matching support is added to the JIT compiler.
968
969 5. Fixed several bugs concerned with partial matching of items that consist
970 of more than one character:
971
972 (a) /^(..)\1/ did not partially match "aba" because checking references was
973 done on an "all or nothing" basis. This also applied to repeated
974 references.
975
976 (b) \R did not give a hard partial match if \r was found at the end of the
977 subject.
978
979 (c) \X did not give a hard partial match after matching one or more
980 characters at the end of the subject.
981
982 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
983 a partial match for the string "\r".
984
985 (e) When newline was set to CRLF, the metacharacter "." did not recognize
986 a partial match for a CR character at the end of the subject string.
987
988 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
989 running pcretest, the text "(JIT)" added to the output whenever JIT is
990 actually used to run the match.
991
992 7. Individual JIT compile options can be set in pcretest by following -s+[+]
993 or /S+[+] with a digit between 1 and 7.
994
995 8. OP_NOT now supports any UTF character not just single-byte ones.
996
997 9. (*MARK) control verb is now supported by the JIT compiler.
998
999 10. The command "./RunTest list" lists the available tests without actually
1000 running any of them. (Because I keep forgetting what they all are.)
1001
1002 11. Add PCRE_INFO_MAXLOOKBEHIND.
1003
1004 12. Applied a (slightly modified) user-supplied patch that improves performance
1005 when the heap is used for recursion (compiled with --disable-stack-for-
1006 recursion). Instead of malloc and free for each heap frame each time a
1007 logical recursion happens, frames are retained on a chain and re-used where
1008 possible. This sometimes gives as much as 30% improvement.
1009
1010 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1011 call.
1012
1013 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1014
1015 15. It is now possible to link pcretest with libedit as an alternative to
1016 libreadline.
1017
1018 16. (*COMMIT) control verb is now supported by the JIT compiler.
1019
1020 17. The Unicode data tables have been updated to Unicode 6.1.0.
1021
1022 18. Added --file-list option to pcregrep.
1023
1024 19. Added binary file support to pcregrep, including the -a, --binary-files,
1025 -I, and --text options.
1026
1027 20. The madvise function is renamed for posix_madvise for QNX compatibility
1028 reasons. Fixed by Giuseppe D'Angelo.
1029
1030 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1031
1032 22. Changed the option for creating symbolic links for 16-bit man pages from
1033 -s to -sf so that re-installing does not cause issues.
1034
1035 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1036
1037 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1038 matches in certain environments (the workspace was not being correctly
1039 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1040 some of the workspace data at the beginning of a restart.
1041
1042 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1043 was not doing so when it should - probably a typo introduced by SVN 528
1044 (change 8.10/14).
1045
1046 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1047 \w+ when the character tables indicated that \x{c4} was a word character.
1048 There were several related cases, all because the tests for doing a table
1049 lookup were testing for characters less than 127 instead of 255.
1050
1051 27. If a pattern contains capturing parentheses that are not used in a match,
1052 their slots in the ovector are set to -1. For those that are higher than
1053 any matched groups, this happens at the end of processing. In the case when
1054 there were back references that the ovector was too small to contain
1055 (causing temporary malloc'd memory to be used during matching), and the
1056 highest capturing number was not used, memory off the end of the ovector
1057 was incorrectly being set to -1. (It was using the size of the temporary
1058 memory instead of the true size.)
1059
1060 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1061 ovector size, it uses memory at the end of the block that it has got.
1062
1063 29. Check for an overlong MARK name and give an error at compile time. The
1064 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1065
1066 30. JIT compiler update.
1067
1068 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1069 Rill for the patch.
1070
1071 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1072
1073 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1074
1075 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1076 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1077 was enabled.
1078
1079 35. Improve JIT code generation for greedy plus quantifier.
1080
1081 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1082 1 to "aa" instead of to an empty string. The bug affected repeated groups
1083 that could potentially match an empty string.
1084
1085 37. Optimizing single character iterators in JIT.
1086
1087 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1088 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1089 codepoints that are too big for the mode are faulted, and in a UTF mode,
1090 disallowed codepoints are also faulted.
1091
1092 39. If PCRE was compiled with UTF support, in three places in the DFA
1093 matcher there was code that should only have been obeyed in UTF mode, but
1094 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1095 processing when bytes with values greater than 127 were present. In 16-bit
1096 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1097 both cases the values are those that cannot be the first data item in a UTF
1098 character. The three items that might have provoked this were recursions,
1099 possessively repeated groups, and atomic groups.
1100
1101 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1102 and pcregrep, because some OS require shared objects to be explicitly
1103 passed to ld, causing the link step to fail if they are not.
1104
1105 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1106 mode, patterns that started with \h* or \R* might be incorrectly matched.
1107
1108
1109 Version 8.30 04-February-2012
1110 -----------------------------
1111
1112 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1113 name is defined in ctype.h.
1114
1115 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1116 only in quite long subpatterns.
1117
1118 3. Removed the function pcre_info(), which has been obsolete and deprecated
1119 since it was replaced by pcre_fullinfo() in February 2000.
1120
1121 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1122 match a (*MARK), and the match failed at the start of the subject, a
1123 reference to memory before the start of the subject could occur. This bug
1124 was introduced by fix 17 of release 8.21.
1125
1126 5. A reference to an unset group with zero minimum repetition was giving
1127 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1128 /(another)?(\1?)test/ matched against "hello world test". This bug was
1129 introduced in release 8.13.
1130
1131 6. Add support for 16-bit character strings (a large amount of work involving
1132 many changes and refactorings).
1133
1134 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1135 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1136 from a file.
1137
1138 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1139 rounding is not applied in this particular case).
1140
1141 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1142 if they appear, or are escaped, in patterns.
1143
1144 10. Get rid of a number of -Wunused-but-set-variable warnings.
1145
1146 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1147 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1148 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1149 also returns the mark "x". This bug applied to capturing parentheses,
1150 non-capturing parentheses, and atomic parentheses. It also applied to some
1151 assertions.
1152
1153 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1154 information out of configure.ac instead of relying on pcre.h.generic, which
1155 is not stored in the repository.
1156
1157 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1158 -lreadline.
1159
1160 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1161
1162 15. Applied Graycode's patch to put the top-level frame on the stack rather
1163 than the heap when not using the stack for recursion. This gives a
1164 performance improvement in many cases when recursion is not deep.
1165
1166 16. Experimental code added to "pcretest -C" to output the stack frame size.
1167
1168
1169 Version 8.21 12-Dec-2011
1170 ------------------------
1171
1172 1. Updating the JIT compiler.
1173
1174 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1175 are added as well.
1176
1177 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1178 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1179 calling _pcre_jit_exec. Some extra comments are added.
1180
1181 4. (*MARK) settings inside atomic groups that do not contain any capturing
1182 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1183 was introduced by change 18 for 8.20.
1184
1185 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1186 ECMA-262 standard.
1187
1188 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1189 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1190 This bug was probably introduced by change 9 of 8.13.
1191
1192 7. While fixing 6 above, I noticed that a number of other items were being
1193 incorrectly rejected as "not fixed length". This arose partly because newer
1194 opcodes had not been added to the fixed-length checking code. I have (a)
1195 corrected the bug and added tests for these items, and (b) arranged for an
1196 error to occur if an unknown opcode is encountered while checking for fixed
1197 length instead of just assuming "not fixed length". The items that were
1198 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1199 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1200 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1201
1202 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1203 being incorrectly compiled and would have given unpredicatble results.
1204
1205 9. A possessively repeated subpattern with minimum repeat count greater than
1206 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1207 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1208 the first (A) could occur when it should not.
1209
1210 10. Add a cast and remove a redundant test from the code.
1211
1212 11. JIT should use pcre_malloc/pcre_free for allocation.
1213
1214 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1215 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1216 is anything other than /usr, -L is still shown).
1217
1218 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1219
1220 14. Perl does not support \N without a following name in a [] class; PCRE now
1221 also gives an error.
1222
1223 15. If a forward reference was repeated with an upper limit of around 2000,
1224 it caused the error "internal error: overran compiling workspace". The
1225 maximum number of forward references (including repeats) was limited by the
1226 internal workspace, and dependent on the LINK_SIZE. The code has been
1227 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1228 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1229 of around 200,000 forward references. While doing this, I also speeded up
1230 the filling in of repeated forward references.
1231
1232 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1233 incorrectly expecting the subject to contain another "a" after the start.
1234
1235 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1236 in the match, the SKIP should be ignored. This was not happening; instead
1237 the SKIP was being treated as NOMATCH. For patterns such as
1238 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1239 tested.
1240
1241 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1242 now much more compatible with Perl, in particular in cases where the result
1243 is a non-match for a non-anchored pattern. For example, if
1244 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1245 "m", where previously it did not return a name. A side effect of this
1246 change is that for partial matches, the last encountered mark name is
1247 returned, as for non matches. A number of tests that were previously not
1248 Perl-compatible have been moved into the Perl-compatible test files. The
1249 refactoring has had the pleasing side effect of removing one argument from
1250 the match() function, thus reducing its stack requirements.
1251
1252 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1253 subsequent uses of /S (without +) incorrectly behaved like /S+.
1254
1255 21. Retrieve executable code size support for the JIT compiler and fixing
1256 some warnings.
1257
1258 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1259 not work when the shorter character appeared right at the end of the
1260 subject string.
1261
1262 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1263 systems.
1264
1265 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1266 output it when the /M option is used in pcretest.
1267
1268 25. The CheckMan script was not being included in the distribution. Also, added
1269 an explicit "perl" to run Perl scripts from the PrepareRelease script
1270 because this is reportedly needed in Windows.
1271
1272 26. If study data was being save in a file and studying had not found a set of
1273 "starts with" bytes for the pattern, the data written to the file (though
1274 never used) was taken from uninitialized memory and so caused valgrind to
1275 complain.
1276
1277 27. Updated RunTest.bat as provided by Sheri Pierce.
1278
1279 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1280
1281 29. Computation of memory usage for the table of capturing group names was
1282 giving an unnecessarily large value.
1283
1284
1285 Version 8.20 21-Oct-2011
1286 ------------------------
1287
1288 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1289 a POSIX class. After further experiments with Perl, which convinced me that
1290 Perl has bugs and confusions, a closing square bracket is no longer allowed
1291 in a POSIX name. This bug also affected patterns with classes that started
1292 with full stops.
1293
1294 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1295 captured substring, but while checking the failing first alternative,
1296 substring 1 is temporarily captured. If the output vector supplied to
1297 pcre_exec() was not big enough for this capture, the yield of the function
1298 was still zero ("insufficient space for captured substrings"). This cannot
1299 be totally fixed without adding another stack variable, which seems a lot
1300 of expense for a edge case. However, I have improved the situation in cases
1301 such as /(a)(b)x|abc/ matched against "abc", where the return code
1302 indicates that fewer than the maximum number of slots in the ovector have
1303 been set.
1304
1305 3. Related to (2) above: when there are more back references in a pattern than
1306 slots in the output vector, pcre_exec() uses temporary memory during
1307 matching, and copies in the captures as far as possible afterwards. It was
1308 using the entire output vector, but this conflicts with the specification
1309 that only 2/3 is used for passing back captured substrings. Now it uses
1310 only the first 2/3, for compatibility. This is, of course, another edge
1311 case.
1312
1313 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1314 main code base, and can be used by building with --enable-jit. When this is
1315 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1316 runtime --no-jit option is given.
1317
1318 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1319 ovector, the return from the function was zero, implying that there were
1320 other matches that did not fit. The correct "exactly full" value is now
1321 returned.
1322
1323 6. If a subpattern that was called recursively or as a subroutine contained
1324 (*PRUNE) or any other control that caused it to give a non-standard return,
1325 invalid errors such as "Error -26 (nested recursion at the same subject
1326 position)" or even infinite loops could occur.
1327
1328 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1329 computing the minimum length on reaching *ACCEPT, and so ended up with the
1330 wrong value of 1 rather than 0. Further investigation indicates that
1331 computing a minimum subject length in the presence of *ACCEPT is difficult
1332 (think back references, subroutine calls), and so I have changed the code
1333 so that no minimum is registered for a pattern that contains *ACCEPT.
1334
1335 8. If (*THEN) was present in the first (true) branch of a conditional group,
1336 it was not handled as intended. [But see 16 below.]
1337
1338 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1339 Sheri Pierce.
1340
1341 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1342 the first byte in a match must be "a".
1343
1344 11. Change 17 for 8.13 increased the recursion depth for patterns like
1345 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1346 pattern contains any instances of (*THEN). If it does not, the old
1347 optimizations are restored. It would be nice to do this on a per-group
1348 basis, but at the moment that is not feasible.
1349
1350 12. In some environments, the output of pcretest -C is CRLF terminated. This
1351 broke RunTest's code that checks for the link size. A single white space
1352 character after the value is now allowed for.
1353
1354 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1355 For "fr", it uses the Windows-specific input and output files.
1356
1357 14. If (*THEN) appeared in a group that was called recursively or as a
1358 subroutine, it did not work as intended. [But see next item.]
1359
1360 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1361 pattern fragments (but not containing any | characters). If A and B are
1362 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1363 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1364 D. In other words, Perl considers parentheses that do not contain any |
1365 characters to be part of a surrounding alternative, whereas PCRE was
1366 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1367 differently. PCRE now behaves in the same way as Perl, except in the case
1368 of subroutine/recursion calls such as (?1) which have in any case always
1369 been different (but PCRE had them first :-).
1370
1371 16. Related to 15 above: Perl does not treat the | in a conditional group as
1372 creating alternatives. Such a group is treated in the same way as an
1373 ordinary group without any | characters when processing (*THEN). PCRE has
1374 been changed to match Perl's behaviour.
1375
1376 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1377 RunGrepTest script failed.
1378
1379 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1380 inevitable for groups that contain captures, but it can lead to a lot of
1381 stack use in large patterns. The old behaviour has been restored for atomic
1382 groups that do not contain any capturing parentheses.
1383
1384 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1385 suppress the check for a minimum subject length at run time. (If it was
1386 given to pcre_exec() or pcre_dfa_exec() it did work.)
1387
1388 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1389 fail to work when decoding hex characters in data strings in EBCDIC
1390 environments.
1391
1392 21. It appears that in at least one Mac OS environment, the isxdigit() function
1393 is implemented as a macro that evaluates to its argument more than once,
1394 contravening the C 90 Standard (I haven't checked a later standard). There
1395 was an instance in pcretest which caused it to go wrong when processing
1396 \x{...} escapes in subject strings. The has been rewritten to avoid using
1397 things like p++ in the argument of isxdigit().
1398
1399
1400 Version 8.13 16-Aug-2011
1401 ------------------------
1402
1403 1. The Unicode data tables have been updated to Unicode 6.0.0.
1404
1405 2. Two minor typos in pcre_internal.h have been fixed.
1406
1407 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1408 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1409 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1410
1411 4. There were a number of related bugs in the code for matching backrefences
1412 caselessly in UTF-8 mode when codes for the characters concerned were
1413 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1414 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1415 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1416 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1417 2-byte code at the end of the subject (it thought there wasn't enough data
1418 left).
1419
1420 5. Comprehensive information about what went wrong is now returned by
1421 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1422 as the output vector has at least 2 elements. The offset of the start of
1423 the failing character and a reason code are placed in the vector.
1424
1425 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1426 now returned is for the first byte of the failing character, instead of the
1427 last byte inspected. This is an incompatible change, but I hope it is small
1428 enough not to be a problem. It makes the returned offset consistent with
1429 pcre_exec() and pcre_dfa_exec().
1430
1431 7. pcretest now gives a text phrase as well as the error number when
1432 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1433 failure, the offset and reason code are output.
1434
1435 8. When \R was used with a maximizing quantifier it failed to skip backwards
1436 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1437 back over a single character (\n). This seems wrong (because it treated the
1438 two characters as a single entity when going forwards), conflicts with the
1439 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1440 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1441 has been changed.
1442
1443 9. Some internal refactoring has changed the processing so that the handling
1444 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1445 time (the PCRE_DOTALL option was changed this way some time ago: version
1446 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1447 which was always a bit of a fudge. It also means that there is one less
1448 argument for the match() function, which reduces its stack requirements
1449 slightly. This change also fixes an incompatibility with Perl: the pattern
1450 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1451
1452 10. More internal refactoring has drastically reduced the number of recursive
1453 calls to match() for possessively repeated groups such as (abc)++ when
1454 using pcre_exec().
1455
1456 11. While implementing 10, a number of bugs in the handling of groups were
1457 discovered and fixed:
1458
1459 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1460 (a|)*(?1) gave a compile-time internal error.
1461 ((a|)+)+ did not notice that the outer group could match an empty string.
1462 (^a|^)+ was not marked as anchored.
1463 (.*a|.*)+ was not marked as matching at start or after a newline.
1464
1465 12. Yet more internal refactoring has removed another argument from the match()
1466 function. Special calls to this function are now indicated by setting a
1467 value in a variable in the "match data" data block.
1468
1469 13. Be more explicit in pcre_study() instead of relying on "default" for
1470 opcodes that mean there is no starting character; this means that when new
1471 ones are added and accidentally left out of pcre_study(), testing should
1472 pick them up.
1473
1474 14. The -s option of pcretest has been documented for ages as being an old
1475 synonym of -m (show memory usage). I have changed it to mean "force study
1476 for every regex", that is, assume /S for every regex. This is similar to -i
1477 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1478 using it. It makes it easier to run collections of tests with and without
1479 study enabled, and thereby test pcre_study() more easily. All the standard
1480 tests are now run with and without -s (but some patterns can be marked as
1481 "never study" - see 20 below).
1482
1483 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1484 restoration of the capturing data to the outer values was not happening
1485 correctly.
1486
1487 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1488 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1489 pattern had matched an empty string, and so incorrectly returned a no
1490 match.
1491
1492 17. There was optimizing code for the last branch of non-capturing parentheses,
1493 and also for the obeyed branch of a conditional subexpression, which used
1494 tail recursion to cut down on stack usage. Unfortunately, now that there is
1495 the possibility of (*THEN) occurring in these branches, tail recursion is
1496 no longer possible because the return has to be checked for (*THEN). These
1497 two optimizations have therefore been removed. [But see 8.20/11 above.]
1498
1499 18. If a pattern containing \R was studied, it was assumed that \R always
1500 matched two bytes, thus causing the minimum subject length to be
1501 incorrectly computed because \R can also match just one byte.
1502
1503 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1504 was incorrectly computed.
1505
1506 20. If /S is present twice on a test pattern in pcretest input, it now
1507 *disables* studying, thereby overriding the use of -s on the command line
1508 (see 14 above). This is necessary for one or two tests to keep the output
1509 identical in both cases.
1510
1511 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1512 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1513
1514 22. When an atomic group that contained a capturing parenthesis was
1515 successfully matched, but the branch in which it appeared failed, the
1516 capturing was not being forgotten if a higher numbered group was later
1517 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1518 group 1 to "a", when in fact it should be unset. This applied to multi-
1519 branched capturing and non-capturing groups, repeated or not, and also to
1520 positive assertions (capturing in negative assertions does not happen
1521 in PCRE) and also to nested atomic groups.
1522
1523 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1524 subject after a captured substring, to make it easier to tell which of a
1525 number of identical substrings has been captured.
1526
1527 24. The way atomic groups are processed by pcre_exec() has been changed so that
1528 if they are repeated, backtracking one repetition now resets captured
1529 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1530 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1531 "aaa". Previously, it would have been "a". As part of this code
1532 refactoring, the way recursive calls are handled has also been changed.
1533
1534 25. If an assertion condition captured any substrings, they were not passed
1535 back unless some other capturing happened later. For example, if
1536 (?(?=(a))a) was matched against "a", no capturing was returned.
1537
1538 26. When studying a pattern that contained subroutine calls or assertions,
1539 the code for finding the minimum length of a possible match was handling
1540 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1541 group 1 called group 2 while simultaneously a separate group 2 called group
1542 1). A stack overflow occurred in this case. I have fixed this by limiting
1543 the recursion depth to 10.
1544
1545 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1546 Fortmann. This supports explicit test numbers on the command line, and has
1547 argument validation and error reporting.
1548
1549 28. An instance of \X with an unlimited repeat could fail if at any point the
1550 first character it looked at was a mark character.
1551
1552 29. Some minor code refactoring concerning Unicode properties and scripts
1553 should reduce the stack requirement of match() slightly.
1554
1555 30. Added the '=' option to pcretest to check the setting of unused capturing
1556 slots at the end of the pattern, which are documented as being -1, but are
1557 not included in the return count.
1558
1559 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1560 compiled something random. Now it gives a compile-time error (as does
1561 Perl).
1562
1563 32. A *MARK encountered during the processing of a positive assertion is now
1564 recorded and passed back (compatible with Perl).
1565
1566 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1567 had alternative anchored branches, the search for a second match in a line
1568 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1569 matched the line "0102" twice. The same bug affected patterns that started
1570 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1571 twice.
1572
1573 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1574 does, and because of capturing effects, quantifying parenthesized
1575 assertions may at times be useful. Quantifiers are now allowed for
1576 parenthesized assertions.
1577
1578 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1579
1580 36. \g was being checked for fancy things in a character class, when it should
1581 just be a literal "g".
1582
1583 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1584 appearance of a nested POSIX class supersedes an apparent external class.
1585 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1586 unescaped square brackets may also appear as part of class names. For
1587 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1588 more like Perl. (But see 8.20/1 above.)
1589
1590 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1591 was because it thought it was \N{name}, which is not supported).
1592
1593 39. Add minix to OS list not supporting the -S option in pcretest.
1594
1595 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1596 cannot analyze patterns in sufficient detail to catch mutual recursions
1597 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1598 subgroup is called recursively as a subpattern for a second time at the
1599 same position in the subject string. In previous releases this might have
1600 been caught by the recursion limit, or it might have run out of stack.
1601
1602 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1603 happen only once. PCRE was, however incorrectly giving a compile time error
1604 "recursive call could loop indefinitely" because it cannot analyze the
1605 pattern in sufficient detail. The compile time test no longer happens when
1606 PCRE is compiling a conditional subpattern, but actual runaway loops are
1607 now caught at runtime (see 40 above).
1608
1609 42. It seems that Perl allows any characters other than a closing parenthesis
1610 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1611 has been changed to be the same.
1612
1613 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1614 as not to get warnings when autogen.sh is called. Also changed
1615 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1616
1617 44. To help people who use pcregrep to scan files containing exceedingly long
1618 lines, the following changes have been made:
1619
1620 (a) The default value of the buffer size parameter has been increased from
1621 8K to 20K. (The actual buffer used is three times this size.)
1622
1623 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1624 PCRE is built.
1625
1626 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1627 to be set at run time.
1628
1629 (d) Numerical values in pcregrep options can be followed by K or M, for
1630 example --buffer-size=50K.
1631
1632 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1633 given and the return code is set to 2.
1634
1635 45. Add a pointer to the latest mark to the callout data block.
1636
1637 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1638 partial match of an empty string instead of no match. This was specific to
1639 the use of ".".
1640
1641 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1642 complete match instead of a partial match. This bug was dependent on both
1643 the PCRE_UTF8 and PCRE_DOTALL options being set.
1644
1645 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1646 starting byte set, because \b was not being ignored.
1647
1648
1649 Version 8.12 15-Jan-2011
1650 ------------------------
1651
1652 1. Fixed some typos in the markup of the man pages, and wrote a script that
1653 checks for such things as part of the documentation building process.
1654
1655 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1656 --match-limit and --recursion-limit options (added for 8.11). In
1657 particular, this made one of the standard tests fail. (The integer value
1658 went into the wrong half of a long int.)
1659
1660 3. If the --colour option was given to pcregrep with -v (invert match), it
1661 did strange things, either producing crazy output, or crashing. It should,
1662 of course, ignore a request for colour when reporting lines that do not
1663 match.
1664
1665 4. Another pcregrep bug caused similar problems if --colour was specified with
1666 -M (multiline) and the pattern match finished with a line ending.
1667
1668 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1669 matched in multiline mode, the following line was shown as part of the
1670 match. This seems wrong, so I have changed it.
1671
1672 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1673 the check for further matches in the same line (so they could be coloured)
1674 to overrun the end of the current line. If another match was found, it was
1675 incorrectly shown (and then shown again when found in the next line).
1676
1677 7. If pcregrep was compiled under Windows, there was a reference to the
1678 function pcregrep_exit() before it was defined. I am assuming this was
1679 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1680 reported by a user. I've moved the definition above the reference.
1681
1682
1683 Version 8.11 10-Dec-2010
1684 ------------------------
1685
1686 1. (*THEN) was not working properly if there were untried alternatives prior
1687 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1688 backtracked to try for "b" instead of moving to the next alternative branch
1689 at the same level (in this case, to look for "c"). The Perl documentation
1690 is clear that when (*THEN) is backtracked onto, it goes to the "next
1691 alternative in the innermost enclosing group".
1692
1693 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1694 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1695 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1696 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1697 (*THEN).
1698
1699 3. If \s appeared in a character class, it removed the VT character from
1700 the class, even if it had been included by some previous item, for example
1701 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1702 of \s, but is part of the POSIX "space" class.)
1703
1704 4. A partial match never returns an empty string (because you can always
1705 match an empty string at the end of the subject); however the checking for
1706 an empty string was starting at the "start of match" point. This has been
1707 changed to the "earliest inspected character" point, because the returned
1708 data for a partial match starts at this character. This means that, for
1709 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1710 (previously it gave "no match").
1711
1712 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1713 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1714 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1715 has an implication that the given string is incomplete (because a partial
1716 match is preferred over a full match). For this reason, these items now
1717 give a partial match in this situation. [Aside: previously, the one case
1718 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1719 match rather than a full match, which was wrong by the old rules, but is
1720 now correct.]
1721
1722 6. There was a bug in the handling of #-introduced comments, recognized when
1723 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1724 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1725 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1726 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1727 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1728 places in pcre_compile().
1729
1730 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1731 comments when looking ahead for named forward references to subpatterns,
1732 the only newline sequence it recognized was NL. It now handles newlines
1733 according to the set newline convention.
1734
1735 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1736 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1737 cater for a lack of strerror(). These oversights have been fixed.
1738
1739 9. Added --match-limit and --recursion-limit to pcregrep.
1740
1741 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1742
1743 11. When the -o option was used, pcregrep was setting a return code of 1, even
1744 when matches were found, and --line-buffered was not being honoured.
1745
1746 12. Added an optional parentheses number to the -o and --only-matching options
1747 of pcregrep.
1748
1749 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1750 can match an empty string. The code to do it in pcretest and pcredemo
1751 needed fixing:
1752
1753 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1754 only one byte after an empty string match just before CRLF (this case
1755 just got forgotten; "any" and "anycrlf" were OK).
1756
1757 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1758 mode when an empty string match preceded an ASCII character followed by
1759 a non-ASCII character. (The code for advancing by one character rather
1760 than one byte was nonsense.)
1761
1762 (c) The pcredemo.c sample program did not have any code at all to handle
1763 the cases when CRLF is a valid newline sequence.
1764
1765 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1766 as a starting offset was within the subject string. There is now a new
1767 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1768 negative or greater than the length of the string. In order to test this,
1769 pcretest is extended to allow the setting of negative starting offsets.
1770
1771 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1772 starting offset points to the beginning of a UTF-8 character was
1773 unnecessarily clumsy. I tidied it up.
1774
1775 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1776 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1777
1778 17. Nobody had reported that the --include_dir option, which was added in
1779 release 7.7 should have been called --include-dir (hyphen, not underscore)
1780 for compatibility with GNU grep. I have changed it to --include-dir, but
1781 left --include_dir as an undocumented synonym, and the same for
1782 --exclude-dir, though that is not available in GNU grep, at least as of
1783 release 2.5.4.
1784
1785 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1786 characters from a string of bytes) have been redefined so as not to use
1787 loops, in order to improve performance in some environments. At the same
1788 time, I abstracted some of the common code into auxiliary macros to save
1789 repetition (this should not affect the compiled code).
1790
1791 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1792 compile-time error is now given if \c is not followed by an ASCII
1793 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1794 different, and any byte value is allowed.)
1795
1796 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1797 START_OPTIMIZE option, which is now allowed at compile time - but just
1798 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1799 to pcregrep and other applications that have no direct access to PCRE
1800 options. The new /Y option in pcretest sets this option when calling
1801 pcre_compile().
1802
1803 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1804 back references. Groups containing recursive back references were forced to
1805 be atomic by that change, but in the case of named groups, the amount of
1806 memory required was incorrectly computed, leading to "Failed: internal
1807 error: code overflow". This has been fixed.
1808
1809 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1810 pcretest.c, to avoid build problems in some Borland environments.
1811
1812
1813 Version 8.10 25-Jun-2010
1814 ------------------------
1815
1816 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1817 THEN.
1818
1819 2. (*ACCEPT) was not working when inside an atomic group.
1820
1821 3. Inside a character class, \B is treated as a literal by default, but
1822 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1823 causes the error). The code is unchanged, but I tidied the documentation.
1824
1825 4. Inside a character class, PCRE always treated \R and \X as literals,
1826 whereas Perl faults them if its -w option is set. I have changed PCRE so
1827 that it faults them when PCRE_EXTRA is set.
1828
1829 5. Added support for \N, which always matches any character other than
1830 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1831
1832 6. When compiling pcregrep with newer versions of gcc which may have
1833 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1834 declared with attribute warn_unused_result" were given. Just casting the
1835 result to (void) does not stop the warnings; a more elaborate fudge is
1836 needed. I've used a macro to implement this.
1837
1838 7. Minor change to pcretest.c to avoid a compiler warning.
1839
1840 8. Added four artifical Unicode properties to help with an option to make
1841 \s etc use properties (see next item). The new properties are: Xan
1842 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1843
1844 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1845 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1846 this option. Modified pcretest to add /W to test this facility. Added
1847 REG_UCP to make it available via the POSIX interface.
1848
1849 10. Added --line-buffered to pcregrep.
1850
1851 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1852 studied, and the match started with a letter with a code point greater than
1853 127 whose first byte was different to the first byte of the other case of
1854 the letter, the other case of this starting letter was not recognized
1855 (#976).
1856
1857 12. If a pattern that was studied started with a repeated Unicode property
1858 test, for example, \p{Nd}+, there was the theoretical possibility of
1859 setting up an incorrect bitmap of starting bytes, but fortunately it could
1860 not have actually happened in practice until change 8 above was made (it
1861 added property types that matched character-matching opcodes).
1862
1863 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1864 possible starting bytes for non-anchored patterns.
1865
1866 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1867 \R, and also a number of cases that involve Unicode properties, both
1868 explicit and implicit when PCRE_UCP is set.
1869
1870 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1871 input, it could crash or give wrong results if characters with values
1872 greater than 0xc0 were present in the subject string. (Detail: it assumed
1873 UTF-8 input when processing these items.)
1874
1875 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1876 size_t is 64-bit (#991).
1877
1878 17. Added a check for running out of memory when PCRE is compiled with
1879 --disable-stack-for-recursion (#990).
1880
1881 18. If the last data line in a file for pcretest does not have a newline on
1882 the end, a newline was missing in the output.
1883
1884 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1885 less than 128) in its various bitmaps. However, there is a facility for
1886 generating tables according to the current locale when PCRE is compiled. It
1887 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1888 characters, are recognized by isspace() and therefore were getting set in
1889 these tables, and indeed these tables seem to approximate to ISO 8859. This
1890 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1891 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1892 which of course cannot start UTF-8 characters. I have changed the code so
1893 that only real ASCII characters (less than 128) and the correct starting
1894 bytes for UTF-8 encodings are set for characters greater than 127 when in
1895 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1896 altogether.)
1897
1898 20. Added the /T option to pcretest so as to be able to run tests with non-
1899 standard character tables, thus making it possible to include the tests
1900 used for 19 above in the standard set of tests.
1901
1902 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1903 reference to a subpattern the other side of a comment that contains an
1904 opening parenthesis caused either an internal compiling error, or a
1905 reference to the wrong subpattern.
1906
1907
1908 Version 8.02 19-Mar-2010
1909 ------------------------
1910
1911 1. The Unicode data tables have been updated to Unicode 5.2.0.
1912
1913 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1914 configured.
1915
1916 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1917 original author of that file, following a query about its status.
1918
1919 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1920 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1921
1922 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1923 quantifier applied to a forward-referencing subroutine call, could compile
1924 incorrect code or give the error "internal error: previously-checked
1925 referenced subpattern not found".
1926
1927 6. Both MS Visual Studio and Symbian OS have problems with initializing
1928 variables to point to external functions. For these systems, therefore,
1929 pcre_malloc etc. are now initialized to local functions that call the
1930 relevant global functions.
1931
1932 7. There were two entries missing in the vectors called coptable and poptable
1933 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1934 I've fixed the data, and added a kludgy way of testing at compile time that
1935 the lengths are correct (equal to the number of opcodes).
1936
1937 8. Following on from 7, I added a similar kludge to check the length of the
1938 eint vector in pcreposix.c.
1939
1940 9. Error texts for pcre_compile() are held as one long string to avoid too
1941 much relocation at load time. To find a text, the string is searched,
1942 counting zeros. There was no check for running off the end of the string,
1943 which could happen if a new error number was added without updating the
1944 string.
1945
1946 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1947
1948 11. \K was not working if it appeared in an atomic group or in a group that
1949 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1950 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1951 the assertion is positive, but not if it is negative.
1952
1953 12. Change 11 fortuitously reduced the size of the stack frame used in the
1954 "match()" function of pcre_exec.c by one pointer. Forthcoming
1955 implementation of support for (*MARK) will need an extra pointer on the
1956 stack; I have reserved it now, so that the stack frame size does not
1957 decrease.
1958
1959 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1960 item in branch that calls a recursion is a subroutine call - as in the
1961 second branch in the above example - was incorrectly given the compile-
1962 time error "recursive call could loop indefinitely" because pcre_compile()
1963 was not correctly checking the subroutine for matching a non-empty string.
1964
1965 14. The checks for overrunning compiling workspace could trigger after an
1966 overrun had occurred. This is a "should never occur" error, but it can be
1967 triggered by pathological patterns such as hundreds of nested parentheses.
1968 The checks now trigger 100 bytes before the end of the workspace.
1969
1970 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1971
1972
1973 Version 8.01 19-Jan-2010
1974 ------------------------
1975
1976 1. If a pattern contained a conditional subpattern with only one branch (in
1977 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1978 computed the wrong minimum data length (which is of course zero for such
1979 subpatterns). This could cause incorrect "no match" results.
1980
1981 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1982 the pattern is reset in the first branch, pcre_compile() failed with
1983 "internal error: code overflow at offset...". This happened only when
1984 the reset was to the original external option setting. (An optimization
1985 abstracts leading options settings into an external setting, which was the
1986 cause of this.)
1987
1988 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1989 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1990 assertion pattern did not match (meaning that the assertion was true), it
1991 was incorrectly treated as false if the SKIP had been reached during the
1992 matching. This also applied to assertions used as conditions.
1993
1994 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1995 assertion subpattern, including such a pattern used as a condition,
1996 unpredictable results occurred, instead of the error return
1997 PCRE_ERROR_DFA_UITEM.
1998
1999 5. The C++ GlobalReplace function was not working like Perl for the special
2000 situation when an empty string is matched. It now does the fancy magic
2001 stuff that is necessary.
2002
2003 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2004 removed. (These were left over from very, very early versions of PCRE.)
2005
2006 7. Some cosmetic changes to the code to make life easier when compiling it
2007 as part of something else:
2008
2009 (a) Change DEBUG to PCRE_DEBUG.
2010
2011 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2012 called "current" as "current_branch", to prevent a collision with the
2013 Linux macro when compiled as a kernel module.
2014
2015 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2016 prevent a collision with the Linux macro when compiled as a kernel
2017 module.
2018
2019 8. In pcre_compile() there are some checks for integer overflows that used to
2020 cast potentially large values to (double). This has been changed to that
2021 when building, a check for int64_t is made, and if it is found, it is used
2022 instead, thus avoiding the use of floating point arithmetic. (There is no
2023 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2024 double.
2025
2026 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2027 2005 (difference between two addresses compared to an unsigned value).
2028
2029 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2030 custom one, because of the following reported problem in Windows:
2031
2032 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2033 under Win32.
2034 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2035 therefore missing the function definition.
2036 - The compiler thus generates a "C" signature for the test function.
2037 - The linker fails to find the "C" function.
2038 - PCRE fails to configure if asked to do so against libbz2.
2039
2040 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2041 messages were output:
2042
2043 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2044 rerunning libtoolize, to keep the correct libtool macros in-tree.
2045 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2046
2047 I have done both of these things.
2048
2049 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2050 most of the time, it *can* run out if it is given a pattern that contains a
2051 runaway infinite recursion. I updated the discussion in the pcrestack man
2052 page.
2053
2054 13. Now that we have gone to the x.xx style of version numbers, the minor
2055 version may start with zero. Using 08 or 09 is a bad idea because users
2056 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2057 interpreted as invalid octal numbers. I've updated the previous comment in
2058 configure.ac, and also added a check that gives an error if 08 or 09 are
2059 used.
2060
2061 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2062 causing partial matching to fail when the end of the subject matched \W
2063 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2064
2065 15. There were some discrepancies between the declarations in pcre_internal.h
2066 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2067 their definitions. The declarations used "const uschar *" and the
2068 definitions used USPTR. Even though USPTR is normally defined as "const
2069 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2070 reported that: "This difference in casting confuses some C++ compilers, for
2071 example, SunCC recognizes above declarations as different functions and
2072 generates broken code for hbpcre." I have changed the declarations to use
2073 USPTR.
2074
2075 16. GNU libtool is named differently on some systems. The autogen.sh script now
2076 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2077 (FreeBSD).
2078
2079 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2080 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2081 comment: "Figure out how to create a longlong from a string: strtoll and
2082 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2083 instance, but it only takes 2 args instead of 3!"
2084
2085 18. A subtle bug concerned with back references has been fixed by a change of
2086 specification, with a corresponding code fix. A pattern such as
2087 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2088 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2089 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2090 same bug. Such groups have to be quantified to be useful, or contained
2091 inside another quantified group. (If there's no repetition, the reference
2092 can never match.) The problem arises because, having left the group and
2093 moved on to the rest of the pattern, a later failure that backtracks into
2094 the group uses the captured value from the final iteration of the group
2095 rather than the correct earlier one. I have fixed this in PCRE by forcing
2096 any group that contains a reference to itself to be an atomic group; that
2097 is, there cannot be any backtracking into it once it has completed. This is
2098 similar to recursive and subroutine calls.
2099
2100
2101 Version 8.00 19-Oct-09
2102 ----------------------
2103
2104 1. The table for translating pcre_compile() error codes into POSIX error codes
2105 was out-of-date, and there was no check on the pcre_compile() error code
2106 being within the table. This could lead to an OK return being given in
2107 error.
2108
2109 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2110 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2111 in a Windows environment.
2112
2113 3. The pcregrep --count option prints the count for each file even when it is
2114 zero, as does GNU grep. However, pcregrep was also printing all files when
2115 --files-with-matches was added. Now, when both options are given, it prints
2116 counts only for those files that have at least one match. (GNU grep just
2117 prints the file name in this circumstance, but including the count seems
2118 more useful - otherwise, why use --count?) Also ensured that the
2119 combination -clh just lists non-zero counts, with no names.
2120
2121 4. The long form of the pcregrep -F option was incorrectly implemented as
2122 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2123 but it seems right to fix it, and I didn't think it was worth preserving
2124 the old behaviour.
2125
2126 5. The command line items --regex=pattern and --regexp=pattern were not
2127 recognized by pcregrep, which required --regex pattern or --regexp pattern
2128 (with a space rather than an '='). The man page documented the '=' forms,
2129 which are compatible with GNU grep; these now work.
2130
2131 6. No libpcreposix.pc file was created for pkg-config; there was just
2132 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2133
2134 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2135 when UCP support is not needed, by modifying the Python script that
2136 generates it from Unicode data files. This should not matter if the module
2137 is correctly used as a library, but I received one complaint about 50K of
2138 unwanted data. My guess is that the person linked everything into his
2139 program rather than using a library. Anyway, it does no harm.
2140
2141 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2142 was a minimum greater than 1 for a wide character in a possessive
2143 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2144 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2145 character. Chaos in the form of incorrect output or a compiling loop could
2146 result.
2147
2148 9. The restrictions on what a pattern can contain when partial matching is
2149 requested for pcre_exec() have been removed. All patterns can now be
2150 partially matched by this function. In addition, if there are at least two
2151 slots in the offset vector, the offset of the earliest inspected character
2152 for the match and the offset of the end of the subject are set in them when
2153 PCRE_ERROR_PARTIAL is returned.
2154
2155 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2156 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2157 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2158 and may be more useful for multi-segment matching.
2159
2160 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2161 used to be given if ever the end of the subject was reached; now it is
2162 given only if matching could not proceed because another character was
2163 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2164 string "Z", which now yields "no match" instead of "partial match". In the
2165 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2166 final character ended with (*FAIL).
2167
2168 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2169 if the pattern had a "must contain" character that was already found in the
2170 earlier partial match, unless partial matching was again requested. For
2171 example, with the pattern /dog.(body)?/, the "must contain" character is
2172 "g". If the first part-match was for the string "dog", restarting with
2173 "sbody" failed. This bug has been fixed.
2174
2175 13. The string returned by pcre_dfa_exec() after a partial match has been
2176 changed so that it starts at the first inspected character rather than the
2177 first character of the match. This makes a difference only if the pattern
2178 starts with a lookbehind assertion or \b or \B (\K is not supported by
2179 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2180 matching functions compatible, and I think it's the right thing to do.
2181
2182 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2183 so that the demonstration program is easily available in environments where
2184 PCRE has not been installed from source.
2185
2186 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2187 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2188 library.
2189
2190 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2191 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2192 is not the first non-POSIX option to be added. Clearly some people find
2193 these options useful.
2194
2195 17. If a caller to the POSIX matching function regexec() passes a non-zero
2196 value for nmatch with a NULL value for pmatch, the value of
2197 nmatch is forced to zero.
2198
2199 18. RunGrepTest did not have a test for the availability of the -u option of
2200 the diff command, as RunTest does. It now checks in the same way as
2201 RunTest, and also checks for the -b option.
2202
2203 19. If an odd number of negated classes containing just a single character
2204 interposed, within parentheses, between a forward reference to a named
2205 subpattern and the definition of the subpattern, compilation crashed with
2206 an internal error, complaining that it could not find the referenced
2207 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2208 [The bug was that it was starting one character too far in when skipping
2209 over the character class, thus treating the ] as data rather than
2210 terminating the class. This meant it could skip too much.]
2211
2212 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2213 /g option in pcretest when the pattern contains \K, which makes it possible
2214 to have an empty string match not at the start, even when the pattern is
2215 anchored. Updated pcretest and pcredemo to use this option.
2216
2217 21. If the maximum number of capturing subpatterns in a recursion was greater
2218 than the maximum at the outer level, the higher number was returned, but
2219 with unset values at the outer level. The correct (outer level) value is
2220 now given.
2221
2222 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2223 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2224 make it do so. The string so far is captured, making this feature
2225 compatible with Perl.
2226
2227 23. The tests have been re-organized, adding tests 11 and 12, to make it
2228 possible to check the Perl 5.10 features against Perl 5.10.
2229
2230 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2231 pattern matches a fixed length string. PCRE did not allow this; now it
2232 does. Neither allows recursion.
2233
2234 25. I finally figured out how to implement a request to provide the minimum
2235 length of subject string that was needed in order to match a given pattern.
2236 (It was back references and recursion that I had previously got hung up
2237 on.) This code has now been added to pcre_study(); it finds a lower bound
2238 to the length of subject needed. It is not necessarily the greatest lower
2239 bound, but using it to avoid searching strings that are too short does give
2240 some useful speed-ups. The value is available to calling programs via
2241 pcre_fullinfo().
2242
2243 26. While implementing 25, I discovered to my embarrassment that pcretest had
2244 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2245 study optimizations had never been tested with that matching function.
2246 Oops. What is worse, even when it was passed study data, there was a bug in
2247 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2248 were also very few tests of studied patterns with pcre_dfa_exec().
2249
2250 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2251 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2252 on the other side of the coin, they are no longer allowed to have different
2253 names, because these cannot be distinguished in PCRE, and this has caused
2254 confusion. (This is a difference from Perl.)
2255
2256 28. When duplicate subpattern names are present (necessarily with different
2257 numbers, as required by 27 above), and a test is made by name in a
2258 conditional pattern, either for a subpattern having been matched, or for
2259 recursion in such a pattern, all the associated numbered subpatterns are
2260 tested, and the overall condition is true if the condition is true for any
2261 one of them. This is the way Perl works, and is also more like the way
2262 testing by number works.
2263
2264
2265 Version 7.9 11-Apr-09
2266 ---------------------
2267
2268 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2269 (pcretest), all targets were linked against these libraries. This included
2270 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2271 libraries. This caused unwanted dependencies to be created. This problem
2272 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2273 pcretest is linked with readline.
2274
2275 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2276 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2277 moved outside it again, because FALSE and TRUE are already defined in AIX,
2278 but BOOL is not.
2279
2280 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2281 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2282
2283 4. The pcregrep documentation said spaces were inserted as well as colons (or
2284 hyphens) following file names and line numbers when outputting matching
2285 lines. This is not true; no spaces are inserted. I have also clarified the
2286 wording for the --colour (or --color) option.
2287
2288 5. In pcregrep, when --colour was used with -o, the list of matching strings
2289 was not coloured; this is different to GNU grep, so I have changed it to be
2290 the same.
2291
2292 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2293 each matching line was coloured. Now it goes on to look for further matches
2294 of any of the test patterns, which is the same behaviour as GNU grep.
2295
2296 7. A pattern that could match an empty string could cause pcregrep to loop; it
2297 doesn't make sense to accept an empty string match in pcregrep, so I have
2298 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2299 seems to be how GNU grep behaves. [But see later change 40 for release
2300 8.33.]
2301
2302 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2303 start or after a newline", because the conditional assertion was not being
2304 correctly handled. The rule now is that both the assertion and what follows
2305 in the first alternative must satisfy the test.
2306
2307 9. If auto-callout was enabled in a pattern with a conditional group whose
2308 condition was an assertion, PCRE could crash during matching, both with
2309 pcre_exec() and pcre_dfa_exec().
2310
2311 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2312 used for matching.
2313
2314 11. Unicode property support in character classes was not working for
2315 characters (bytes) greater than 127 when not in UTF-8 mode.
2316
2317 12. Added the -M command line option to pcretest.
2318
2319 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2320
2321 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2322
2323 16. Added comments and documentation about mis-use of no_arg in the C++
2324 wrapper.
2325
2326 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2327 from Martin Jerabek that uses macro names for all relevant character and
2328 string constants.
2329
2330 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2331 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2332 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2333 these, but not everybody uses configure.
2334
2335 19. A conditional group that had only one branch was not being correctly
2336 recognized as an item that could match an empty string. This meant that an
2337 enclosing group might also not be so recognized, causing infinite looping
2338 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2339 with the subject "ab", where knowledge that the repeated group can match
2340 nothing is needed in order to break the loop.
2341
2342 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2343 exec(), but without supplying a callout function, matching went wrong.
2344
2345 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2346 leak if the size of the offset vector was greater than 30. When the vector
2347 is smaller, the saved offsets during recursion go onto a local stack
2348 vector, but for larger vectors malloc() is used. It was failing to free
2349 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2350 error, in fact).
2351
2352 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2353 heapframe that is used only when UTF-8 support is enabled. This caused no
2354 problem, but was untidy.
2355
2356 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2357 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2358 included within another project.
2359
2360 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2361 slightly modified by me:
2362
2363 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2364 not building pcregrep.
2365
2366 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2367 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2368
2369 25. Forward references, both numeric and by name, in patterns that made use of
2370 duplicate group numbers, could behave incorrectly or give incorrect errors,
2371 because when scanning forward to find the reference group, PCRE was not
2372 taking into account the duplicate group numbers. A pattern such as
2373 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2374
2375 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2376 the feature of a custom pointer more persuasive (as requested by a user).
2377
2378 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2379 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2380 pre-defined.
2381
2382 28. Added support for (*UTF8) at the start of a pattern.
2383
2384 29. Arrange for flags added by the "release type" setting in CMake to be shown
2385 in the configuration summary.
2386
2387
2388 Version 7.8 05-Sep-08
2389 ---------------------
2390
2391 1. Replaced UCP searching code with optimized version as implemented for Ad
2392 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2393 stage table and inline lookup instead of a function, giving speed ups of 2
2394 to 5 times on some simple patterns that I tested. Permission was given to
2395 distribute the MultiStage2.py script that generates the tables (it's not in
2396 the tarball, but is in the Subversion repository).
2397
2398 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2399 scripts.
2400
2401 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2402 a group with a zero qualifier. The result of the study could be incorrect,
2403 or the function might crash, depending on the pattern.
2404
2405 4. Caseless matching was not working for non-ASCII characters in back
2406 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2407 It now works when Unicode Property Support is available.
2408
2409 5. In pcretest, an escape such as \x{de} in the data was always generating
2410 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2411 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2412 truncation.
2413
2414 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2415
2416 7. Added two (int) casts to pcregrep when printing the difference of two
2417 pointers, in case they are 64-bit values.
2418
2419 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2420 test 2 if it fails.
2421
2422 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2423 and a #define of that name to empty if it is not externally set. This is to
2424 allow users of MSVC to set it if necessary.
2425
2426 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2427 the convenience functions in the pcre_get.c source file.
2428
2429 11. An option change at the start of a pattern that had top-level alternatives
2430 could cause overwriting and/or a crash. This command provoked a crash in
2431 some environments:
2432
2433 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2434
2435 This potential security problem was recorded as CVE-2008-2371.
2436
2437 12. For a pattern where the match had to start at the beginning or immediately
2438 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2439 pcre_dfa_exec() could read past the end of the passed subject if there was
2440 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2441 pcretest so that it places the subject at the end of its malloc-ed buffer.
2442
2443 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2444 exec() might read past the end of the data buffer in UTF-8 mode.
2445
2446 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2447 the data contained the byte 0x85 as part of a UTF-8 character within its
2448 first line. This applied both to normal and DFA matching.
2449
2450 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2451 /^[^d]*?$/8 failed to match "abc".
2452
2453 16. Added a missing copyright notice to pcrecpp_internal.h.
2454
2455 17. Make it more clear in the documentation that values returned from
2456 pcre_exec() in ovector are byte offsets, not character counts.
2457
2458 18. Tidied a few places to stop certain compilers from issuing warnings.
2459
2460 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2461 supplied by Stefan Weber. I made a further small update for 7.8 because
2462 there is a change of source arrangements: the pcre_searchfuncs.c module is
2463 replaced by pcre_ucd.c.
2464
2465
2466 Version 7.7 07-May-08
2467 ---------------------
2468
2469 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2470 a string to a long long, pretend we don't even have a long long." This is
2471 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2472
2473 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2474 pre-7.6 versions, which defined a global no_arg variable instead of putting
2475 it in the RE class. (See also #8 below.)
2476
2477 3. Remove a line of dead code, identified by coverity and reported by Nuno
2478 Lopes.
2479
2480 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2481
2482 (1) The include/exclude patterns were being applied to the whole pathnames
2483 of files, instead of just to the final components.
2484
2485 (2) If there was more than one level of directory, the subdirectories were
2486 skipped unless they satisfied the include/exclude conditions. This is
2487 inconsistent with GNU grep (and could even be seen as contrary to the
2488 pcregrep specification - which I improved to make it absolutely clear).
2489 The action now is always to scan all levels of directory, and just
2490 apply the include/exclude patterns to regular files.
2491
2492 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2493 --exclude_dir in the tests to avoid scanning .svn directories.
2494
2495 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2496 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2497 doesn't support NULs in patterns.
2498
2499 7. Added some missing "const"s to declarations of static tables in
2500 pcre_compile.c and pcre_dfa_exec.c.
2501
2502 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2503 caused by fix #2 above. (Subsequently also a second patch to fix the
2504 first patch. And a third patch - this was a messy problem.)
2505
2506 9. Applied Craig's patch to remove the use of push_back().
2507
2508 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2509 matching function regexec().
2510
2511 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2512 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2513 references. PCRE supports relative numbers with this syntax (I don't think
2514 Oniguruma does).
2515
2516 12. Previously, a group with a zero repeat such as (...){0} was completely
2517 omitted from the compiled regex. However, this means that if the group
2518 was called as a subroutine from elsewhere in the pattern, things went wrong
2519 (an internal error was given). Such groups are now left in the compiled
2520 pattern, with a new opcode that causes them to be skipped at execution
2521 time.
2522
2523 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2524 to the way PCRE behaves:
2525
2526 (a) A lone ] character is dis-allowed (Perl treats it as data).
2527
2528 (b) A back reference to an unmatched subpattern matches an empty string
2529 (Perl fails the current match path).
2530
2531 (c) A data ] in a character class must be notated as \] because if the
2532 first data character in a class is ], it defines an empty class. (In
2533 Perl it is not possible to have an empty class.) The empty class []
2534 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2535 The negative empty class [^] matches any one character, independently
2536 of the DOTALL setting.
2537
2538 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2539 non-existent subpattern following a character class starting with ']' and
2540 containing () gave an internal compiling error instead of "reference to
2541 non-existent subpattern". Fortunately, when the pattern did exist, the
2542 compiled code was correct. (When scanning forwards to check for the
2543 existence of the subpattern, it was treating the data ']' as terminating
2544 the class, so got the count wrong. When actually compiling, the reference
2545 was subsequently set up correctly.)
2546
2547 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2548 it was being rejected as not supported by pcre_dfa_exec(), even though
2549 other assertions are supported. I have made pcre_dfa_exec() support
2550 (*FAIL).
2551
2552 16. The implementation of 13c above involved the invention of a new opcode,
2553 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2554 cannot be changed at match time, I realized I could make a small
2555 improvement to matching performance by compiling OP_ALLANY instead of
2556 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2557 on the OP_ANY path.
2558
2559 17. Compiling pcretest on Windows with readline support failed without the
2560 following two fixes: (1) Make the unistd.h include conditional on
2561 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2562
2563 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2564 ncurses library to be included for pcretest when ReadLine support is
2565 requested, but also to allow for it to be overridden. This patch came from
2566 Daniel Bergström.
2567
2568 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2569 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2570 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2571 spotting this.
2572
2573
2574 Version 7.6 28-Jan-08
2575 ---------------------
2576
2577 1. A character class containing a very large number of characters with
2578 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2579 overflow.
2580
2581 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2582 HAVE_LONG_LONG is not defined.
2583
2584 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2585 bring them up to date and include new features. This patch includes:
2586
2587 - Fixed PH's badly added libz and libbz2 support.
2588 - Fixed a problem with static linking.
2589 - Added pcredemo. [But later removed - see 7 below.]
2590 - Fixed dftables problem and added an option.
2591 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2592 HAVE_LONG_LONG.
2593 - Added readline support for pcretest.
2594 - Added an listing of the option settings after cmake has run.
2595
2596 4. A user submitted a patch to Makefile that makes it easy to create
2597 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2598 Makefile.am that cause it to include this special target, without
2599 affecting anything else. Note that the same mingw target plus all
2600 the other distribution libraries and programs are now supported
2601 when configuring with CMake (see 6 below) instead of with
2602 Configure/Make.
2603
2604 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2605 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2606 exported in the Windows port". It has not yet been confirmed that the patch
2607 solves the problem, but it does no harm.
2608
2609 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2610 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2611 with CMake, and also correct the comment about stack recursion.
2612
2613 7. Remove the automatic building of pcredemo from the ./configure system and
2614 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2615 of a program that users should build themselves after PCRE is installed, so
2616 building it automatically is not really right. What is more, it gave
2617 trouble in some build environments.
2618
2619 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2620
2621
2622 Version 7.5 10-Jan-08
2623 ---------------------
2624
2625 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2626 values in parens when parsing an RE using the C++ wrapper."
2627
2628 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2629 Characters greater than 255 were excluded from the class instead of being
2630 included.
2631
2632 3. The same bug as (2) above applied to negated POSIX classes such as
2633 [:^space:].
2634
2635 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2636 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2637 I have changed it.
2638
2639 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2640 first named subpattern) and a construct such as (?&a) would reference the
2641 first named subpattern whose name started with "a" (in other words, the
2642 length check was missing). Both these problems are fixed. "Subpattern name
2643 expected" is now given for (?&) (a zero-length name), and this patch also
2644 makes it give the same error for \k'' (previously it complained that that
2645 was a reference to a non-existent subpattern).
2646
2647 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2648 this is right because (?- can be followed by option settings as well as by
2649 digits. I have, however, made the messages clearer.
2650
2651 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2652 than the number used in the conditional) now cause a compile-time error.
2653 This is actually not compatible with Perl, which accepts such patterns, but
2654 treats the conditional as always being FALSE (as PCRE used to), but it
2655 seems to me that giving a diagnostic is better.
2656
2657 8. Change "alphameric" to the more common word "alphanumeric" in comments
2658 and messages.
2659
2660 9. Fix two occurrences of "backslash" in comments that should have been
2661 "backspace".
2662
2663 10. Remove two redundant lines of code that can never be obeyed (their function
2664 was moved elsewhere).
2665
2666 11. The program that makes PCRE's Unicode character property table had a bug
2667 which caused it to generate incorrect table entries for sequences of
2668 characters that have the same character type, but are in different scripts.
2669 It amalgamated them into a single range, with the script of the first of
2670 them. In other words, some characters were in the wrong script. There were
2671 thirteen such cases, affecting characters in the following ranges:
2672
2673 U+002b0 - U+002c1
2674 U+0060c - U+0060d
2675 U+0061e - U+00612
2676 U+0064b - U+0065e
2677 U+0074d - U+0076d
2678 U+01800 - U+01805
2679 U+01d00 - U+01d77
2680 U+01d9b - U+01dbf
2681 U+0200b - U+0200f
2682 U+030fc - U+030fe
2683 U+03260 - U+0327f
2684 U+0fb46 - U+0fbb1
2685 U+10450 - U+1049d
2686
2687 12. The -o option (show only the matching part of a line) for pcregrep was not
2688 compatible with GNU grep in that, if there was more than one match in a
2689 line, it showed only the first of them. It now behaves in the same way as
2690 GNU grep.
2691
2692 13. If the -o and -v options were combined for pcregrep, it printed a blank
2693 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2694 does the same. The return code can be used to tell if there were any
2695 non-matching lines.
2696
2697 14. Added --file-offsets and --line-offsets to pcregrep.
2698
2699 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2700 infinitely looping recursion. The bug was that positive lookaheads were not
2701 being skipped when checking for a possible empty match (negative lookaheads
2702 and both kinds of lookbehind were skipped).
2703
2704 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2705 inclusion of <windows.h> to before rather than after the definition of
2706 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2707
2708 17. Specifying a possessive quantifier with a specific limit for a Unicode
2709 character property caused pcre_compile() to compile bad code, which led at
2710 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2711 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2712 caused the error; without that there was no problem.
2713
2714 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2715
2716 19. Added --enable-pcretest-libreadline.
2717
2718 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2719 RE::GlobalReplace(). As a result, the number of replacements returned was
2720 double what it should be. I removed one of the increments, but Craig sent a
2721 later patch that removed the other one (the right fix) and added unit tests
2722 that check the return values (which was not done before).
2723
2724 21. Several CMake things:
2725
2726 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2727 the names libpcre and libpcreposix, not just pcre and pcreposix.
2728
2729 (2) The above change means that pcretest and pcregrep are now correctly
2730 linked with the newly-built libraries, not previously installed ones.
2731
2732 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2733
2734 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2735 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2736 UTF-8 newline character). The key issue is that the pattern starts .*;
2737 this means that the match must be either at the beginning, or after a
2738 newline. The bug was in the code for advancing after a failed match and
2739 checking that the new position followed a newline. It was not taking
2740 account of UTF-8 characters correctly.
2741
2742 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2743 character classes. PCRE was not treating the sequence [:...:] as a
2744 character class unless the ... were all letters. Perl, however, seems to
2745 allow any characters between [: and :], though of course it rejects as
2746 unknown any "names" that contain non-letters, because all the known class
2747 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2748 for example, whereas PCRE did not - it did not recognize a POSIX character
2749 class. This seemed a bit dangerous, so the code has been changed to be
2750 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2751 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2752 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2753 Perl does, and where it didn't before.
2754
2755 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2756 Windows environments %n is disabled by default.
2757
2758
2759 Version 7.4 21-Sep-07
2760 ---------------------
2761
2762 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2763 means that a class such as [\s] counted as "explicit reference to CR or
2764 LF". That isn't really right - the whole point of the change was to try to
2765 help when there was an actual mention of one of the two characters. So now
2766 the change happens only if \r or \n (or a literal CR or LF) character is
2767 encountered.
2768
2769 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2770 of both had grown to the point where there were only 3 bits left.
2771 Fortunately, there was spare space in the data structure, and so I have
2772 moved the internal flags into a new 16-bit field to free up more option
2773 bits.
2774
2775 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2776 but did not set the internal JCHANGED flag - either of these is enough to
2777 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2778 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2779 start sets both bits.
2780
2781 4. Added options (at build time, compile time, exec time) to change \R from
2782 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2783
2784 5. doc/pcresyntax.html was missing from the distribution.
2785
2786 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2787 compatibility, even though it is no longer used.
2788
2789 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2790 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2791 windows.h file is present (where different names are used). [This was
2792 reversed later after testing - see 16 below.]
2793
2794 8. Changed all #include <config.h> to #include "config.h". There were also
2795 some further <pcre.h> cases that I changed to "pcre.h".
2796
2797 9. When pcregrep was used with the --colour option, it missed the line ending
2798 sequence off the lines that it output.
2799
2800 10. It was pointed out to me that arrays of string pointers cause lots of
2801 relocations when a shared library is dynamically loaded. A technique of
2802 using a single long string with a table of offsets can drastically reduce
2803 these. I have refactored PCRE in four places to do this. The result is
2804 dramatic:
2805
2806 Originally: 290
2807 After changing UCP table: 187
2808 After changing error message table: 43
2809 After changing table of "verbs" 36
2810 After changing table of Posix names 22
2811
2812 Thanks to the folks working on Gregex for glib for this insight.
2813
2814 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2815 unicode-properties was also set.
2816
2817 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2818
2819 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2820 checked only for CRLF.
2821
2822 14. Added casts to pcretest.c to avoid compiler warnings.
2823
2824 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2825
2826 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2827 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2828 entirely. This removes changes made in 7 above.
2829
2830 17. The CMake files have been updated, and there is now more information about
2831 building with CMake in the NON-UNIX-USE document.
2832
2833
2834 Version 7.3 28-Aug-07
2835 ---------------------
2836
2837 1. In the rejigging of the build system that eventually resulted in 7.1, the
2838 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2839 brackets there is not right, since it causes compilers to look for an
2840 installed pcre.h, not the version that is in the source that is being
2841 compiled (which of course may be different). I have changed it back to:
2842
2843 #include "pcre.h"
2844
2845 I have a vague recollection that the change was concerned with compiling in
2846 different directories, but in the new build system, that is taken care of
2847 by the VPATH setting the Makefile.
2848
2849 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2850 when the subject happened to end in the byte 0x85 (e.g. if the last
2851 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2852 characters but of course it shouldn't be taken as a newline when it is part
2853 of another character. The bug was that, for an unlimited repeat of . in
2854 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2855 characters when looking for a newline.
2856
2857 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2858
2859 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2860 in debug output.
2861
2862 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2863 long printing in the pcrecpp unittest when running under MinGW.
2864
2865 6. ESC_K was left out of the EBCDIC table.
2866
2867 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2868 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2869 limit also applies to "virtual nesting" when a pattern is recursive, and in
2870 this case 1000 isn't so big. I have been able to remove this limit at the
2871 expense of backing off one optimization in certain circumstances. Normally,
2872 when pcre_exec() would call its internal match() function recursively and
2873 immediately return the result unconditionally, it uses a "tail recursion"
2874 feature to save stack. However, when a subpattern that can match an empty
2875 string has an unlimited repetition quantifier, it no longer makes this
2876 optimization. That gives it a stack frame in which to save the data for
2877 checking that an empty string has been matched. Previously this was taken
2878 from the 1000-entry workspace that had been reserved. So now there is no
2879 explicit limit, but more stack is used.
2880
2881 8. Applied Daniel's patches to solve problems with the import/export magic
2882 syntax that is required for Windows, and which was going wrong for the
2883 pcreposix and pcrecpp parts of the library. These were overlooked when this
2884 problem was solved for the main library.
2885
2886 9. There were some crude static tests to avoid integer overflow when computing
2887 the size of patterns that contain repeated groups with explicit upper
2888 limits. As the maximum quantifier is 65535, the maximum group length was
2889 set at 30,000 so that the product of these two numbers did not overflow a
2890 32-bit integer. However, it turns out that people want to use groups that
2891 are longer than 30,000 bytes (though not repeat them that many times).
2892 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2893 made it possible to implement the integer overflow checks in a much more
2894 dynamic way, which I have now done. The artificial limitation on group
2895 length has been removed - we now have only the limit on the total length of
2896 the compiled pattern, which depends on the LINK_SIZE setting.
2897
2898 10. Fixed a bug in the documentation for get/copy named substring when
2899 duplicate names are permitted. If none of the named substrings are set, the
2900 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2901 empty string.
2902
2903 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2904 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2905 because the ] is interpreted as the first data character and the
2906 terminating ] is not found. PCRE has been made compatible with Perl in this
2907 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2908 cause memory overwriting.
2909
2910 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2911 string has been matched (to stop an infinite loop). It was not recognizing
2912 a conditional subpattern that could match an empty string if that
2913 subpattern was within another subpattern. For example, it looped when
2914 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2915 condition was not nested. This bug has been fixed.
2916
2917 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2918 past the start of the subject in the presence of bytes with the top bit
2919 set, for example "\x8aBCD".
2920
2921 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2922 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2923
2924 14. Optimized (?!) to (*FAIL).
2925
2926 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2927 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2928 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2929 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2930 does: it's just the validity check that is more restrictive.
2931
2932 16. Inserted checks for integer overflows during escape sequence (backslash)
2933 processing, and also fixed erroneous offset values for syntax errors during
2934 backslash processing.
2935
2936 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2937 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2938
2939 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2940 caused an overrun.
2941
2942 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2943 something other than just ASCII characters) inside a group that had an
2944 unlimited repeat caused a loop at compile time (while checking to see
2945 whether the group could match an empty string).
2946
2947 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2948 [\P{Any}] did so. (Error in the code for printing property names.)
2949
2950 21. An orphan \E inside a character class could cause a crash.
2951
2952 22. A repeated capturing bracket such as (A)? could cause a wild memory
2953 reference during compilation.
2954
2955 23. There are several functions in pcre_compile() that scan along a compiled
2956 expression for various reasons (e.g. to see if it's fixed length for look
2957 behind). There were bugs in these functions when a repeated \p or \P was
2958 present in the pattern. These operators have additional parameters compared
2959 with \d, etc, and these were not being taken into account when moving along
2960 the compiled data. Specifically:
2961
2962 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2963 length.
2964
2965 (b) An item such as \pL+ within a repeated group could cause crashes or
2966 loops.
2967
2968 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2969 "reference to non-existent subpattern" error.
2970
2971 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2972
2973 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2974 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2975
2976 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2977
2978 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2979 character were causing crashes (broken optimization).
2980
2981 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2982 \p or \P) caused a compile-time loop.
2983
2984 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2985 break. For example, the unstudied pattern [\r\n]A does not match the string
2986 "\r\nA" because change 7.0/46 below moves the current point on by two
2987 characters after failing to match at the start. However, the pattern \nA
2988 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2989 the same is true. There doesn't seem any very clean way out of this, but
2990 what I have chosen to do makes the common cases work: PCRE now takes note
2991 of whether there can be an explicit match for \r or \n anywhere in the
2992 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2993 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2994 pattern has explicit CR or LF references.
2995
2996 29. Added (*CR) etc for changing newline setting at start of pattern.
2997
2998
2999 Version 7.2 19-Jun-07
3000 ---------------------
3001
3002 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3003 which is apparently normally available under Windows.
3004
3005 2. Re-jig the pcregrep tests with different newline settings in an attempt
3006 to make them independent of the local environment's newline setting.
3007
3008 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3009
3010 4. Some of the "internals" tests were previously cut out when the link size
3011 was not 2, because the output contained actual offsets. The recent new
3012 "Z" feature of pcretest means that these can be cut out, making the tests
3013 usable with all link sizes.
3014
3015 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3016 stack recursion. This gives a massive performance boost under BSD, but just
3017 a small improvement under Linux. However, it saves one field in the frame
3018 in all cases.
3019
3020 6. Added more features from the forthcoming Perl 5.10:
3021
3022 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3023 recursion call. It refers to the nth most recently opened parentheses.
3024
3025 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3026 to be opened parentheses.
3027
3028 (c) Conditions that refer to capturing parentheses can be specified
3029 relatively, for example, (?(-2)... or (?(+3)...
3030
3031 (d) \K resets the start of the current match so that everything before
3032 is not part of it.
3033
3034 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3035
3036 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3037 reference syntax.
3038
3039 (g) (?| introduces a group in which the numbering of parentheses in each
3040 alternative starts with the same number.
3041
3042 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3043
3044 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3045 PCRE_INFO_JCHANGED.
3046
3047 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3048 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3049 for detecting groups that can match an empty string.
3050
3051 9. A pattern with a very large number of alternatives (more than several
3052 hundred) was running out of internal workspace during the pre-compile
3053 phase, where pcre_compile() figures out how much memory will be needed. A
3054 bit of new cunning has reduced the workspace needed for groups with
3055 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3056 workspace instead of running out of the 4096 that are available.
3057
3058 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3059
3060 11. Applied patch from Google to remove an optimization that didn't quite work.
3061 The report of the bug said:
3062
3063 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3064 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3065 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3066
3067 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3068 it matched the wrong number of bytes.
3069
3070
3071 Version 7.1 24-Apr-07
3072 ---------------------
3073
3074 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3075 that is more "standard", making use of automake and other Autotools. There
3076 is some re-arrangement of the files and adjustment of comments consequent
3077 on this.
3078
3079 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3080 for recursive directory scanning broke on some systems because the files
3081 are not scanned in any specific order and on different systems the order
3082 was different. A call to "sort" has been inserted into RunGrepTest for the
3083 approprate test as a short-term fix. In the longer term there may be an
3084 alternative.
3085
3086 3. I had an email from Eric Raymond about problems translating some of PCRE's
3087 man pages to HTML (despite the fact that I distribute HTML pages, some
3088 people do their own conversions for various reasons). The problems
3089 concerned the use of low-level troff macros .br and .in. I have therefore
3090 removed all such uses from the man pages (some were redundant, some could
3091 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3092 HTML has been updated to handle .nf/.fi and to complain if it encounters
3093 .br or .in.
3094
3095 4. Updated comments in configure.ac that get placed in config.h.in and also
3096 arranged for config.h to be included in the distribution, with the name
3097 config.h.generic, for the benefit of those who have to compile without
3098 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3099
3100 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3101 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3102 makevp.bat for the current PCRE, using the additional files
3103 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3104
3105 6. A Windows user reported a minor discrepancy with test 2, which turned out
3106 to be caused by a trailing space on an input line that had got lost in his
3107 copy. The trailing space was an accident, so I've just removed it.
3108
3109 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3110 that is needed.
3111
3112 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3113 as "const" (a) because they are and (b) because it helps the PHP
3114 maintainers who have recently made a script to detect big data structures
3115 in the php code that should be moved to the .rodata section. I remembered
3116 to update Builducptable as well, so it won't revert if ucptable.h is ever
3117 re-created.
3118
3119 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3120 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3121 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3122 support is not required. This saves 1.5-2K of code, which is important in
3123 some applications.
3124
3125 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3126 so as not to refer to the tables, even though these functions will never be
3127 called when UTF-8 support is disabled. Otherwise there are problems with a
3128 shared library.
3129
3130 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3131
3132 (a) It was defining its arguments as char * instead of void *.
3133
3134 (b) It was assuming that all moves were upwards in memory; this was true
3135 a long time ago when I wrote it, but is no longer the case.
3136
3137 The emulated memove() is provided for those environments that have neither
3138 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3139 is clearly not the case, as these two bugs were recently reported.
3140
3141 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3142 and Detrail to create the HTML documentation, the .txt form of the man
3143 pages, and it removes trailing spaces from listed files. It also creates
3144 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3145 case, it wraps all the #defines with #ifndefs. This script should be run
3146 before "make dist".
3147
3148 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3149 with Unicode property support.
3150
3151 (a) For a maximizing quantifier, if the two different cases of the
3152 character were of different lengths in their UTF-8 codings (there are
3153 some cases like this - I found 11), and the matching function had to
3154 back up over a mixture of the two cases, it incorrectly assumed they
3155 were both the same length.
3156
3157 (b) When PCRE was configured to use the heap rather than the stack for
3158 recursion during matching, it was not correctly preserving the data for
3159 the other case of a UTF-8 character when checking ahead for a match
3160 while processing a minimizing repeat. If the check also involved
3161 matching a wide character, but failed, corruption could cause an
3162 erroneous result when trying to check for a repeat of the original
3163 character.
3164
3165 13. Some tidying changes to the testing mechanism:
3166
3167 (a) The RunTest script now detects the internal link size and whether there
3168 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3169 values substituted by "configure". (The RunGrepTest script already did
3170 this for UTF-8.) The configure.ac script no longer substitutes the
3171 relevant variables.
3172
3173 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3174 with length and offset values. This means that the output is different
3175 for different internal link sizes. Test 2 is skipped for link sizes
3176 other than 2 because of this, bypassing the problem. Unfortunately,
3177 there was also a test in test 3 (the locale tests) that used /B and
3178 failed for link sizes other than 2. Rather than cut the whole test out,
3179 I have added a new /Z option to pcretest that replaces the length and
3180 offset values with spaces. This is now used to make test 3 independent
3181 of link size. (Test 2 will be tidied up later.)
3182
3183 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3184 segmentation fault instead of returning the appropriate error message.
3185
3186 15. In multiline mode when the newline sequence was set to "any", the pattern
3187 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3188 This doesn't seem right; it now treats the CRLF combination as the line
3189 ending, and so does not match in that case. It's only a pattern such as ^$
3190 that would hit this one: something like ^ABC$ would have failed after \r
3191 and then tried again after \r\n.
3192
3193 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3194 in an attempt to make files that differ only in their line terminators
3195 compare equal. This works on Linux.
3196
3197 17. Under certain error circumstances pcregrep might try to free random memory
3198 as it exited. This is now fixed, thanks to valgrind.
3199
3200 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3201 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3202 was because its rules for how to advance for /g after matching an empty
3203 string at the end of a line did not allow for this case. They now check for
3204 it specially.
3205
3206 20. pcretest is supposed to handle patterns and data of any length, by
3207 extending its buffers when necessary. It was getting this wrong when the
3208 buffer for a data line had to be extended.
3209
3210 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3211 CRLF as a newline sequence.
3212
3213 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3214 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3215 I have nevertheless tidied it up.
3216
3217 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3218
3219 24. Added a man page for pcre-config.
3220
3221
3222 Version 7.0 19-Dec-06
3223 ---------------------
3224
3225 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3226 moving to gcc 4.1.1.
3227
3228 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3229 sys/time.h, which is documented as needed for this function. It doesn't
3230 seem to matter on Linux, but it showed up on some releases of OS X.
3231
3232 3. It seems that there are systems where bytes whose values are greater than
3233 127 match isprint() in the "C" locale. The "C" locale should be the
3234 default when a C program starts up. In most systems, only ASCII printing
3235 characters match isprint(). This difference caused the output from pcretest
3236 to vary, making some of the tests fail. I have changed pcretest so that:
3237
3238 (a) When it is outputting text in the compiled version of a pattern, bytes
3239 other than 32-126 are always shown as hex escapes.
3240
3241 (b) When it is outputting text that is a matched part of a subject string,
3242 it does the same, unless a different locale has been set for the match
3243 (using the /L modifier). In this case, it uses isprint() to decide.
3244
3245 4. Fixed a major bug that caused incorrect computation of the amount of memory
3246 required for a compiled pattern when options that changed within the
3247 pattern affected the logic of the preliminary scan that determines the
3248 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3249 that the computed length was too small. The symptoms of this bug were
3250 either the PCRE error "internal error: code overflow" from pcre_compile(),
3251 or a glibc crash with a message such as "pcretest: free(): invalid next
3252 size (fast)". Examples of patterns that provoked this bug (shown in
3253 pcretest format) are:
3254
3255 /(?-x: )/x
3256 /(?x)(?-x: \s*#\s*)/
3257 /((?i)[\x{c0}])/8
3258 /(?i:[\x{c0}])/8
3259
3260 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3261 is now done differently.
3262
3263 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3264 wrapper classes; (b) implement a new function in the C++ scanner that is
3265 more efficient than the old way of doing things because it avoids levels of
3266 recursion in the regex matching; (c) add a paragraph to the documentation
3267 for the FullMatch() function.
3268
3269 6. The escape sequence \n was being treated as whatever was defined as
3270 "newline". Not only was this contrary to the documentation, which states
3271 that \n is character 10 (hex 0A), but it also went horribly wrong when
3272 "newline" was defined as CRLF. This has been fixed.
3273
3274 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3275 was being set to -1 for the "end of line" case (supposedly a value that no
3276 character can have). Though this value is never used (the check for end of
3277 line is "zero bytes in current character"), it caused compiler complaints.
3278 I've changed it to 0xffffffff.
3279
3280 8. In pcre_version.c, the version string was being built by a sequence of
3281 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3282 string (as it is for production releases) called a macro with an empty
3283 argument. The C standard says the result of this is undefined. The gcc
3284 compiler treats it as an empty string (which was what was wanted) but it is
3285 reported that Visual C gives an error. The source has been hacked around to
3286 avoid this problem.
3287
3288 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3289 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3290 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3291 of them did).
3292
3293 10. Originally, pcretest opened its input and output without "b"; then I was
3294 told that "b" was needed in some environments, so it was added for release
3295 5.0 to both the input and output. (It makes no difference on Unix-like
3296 systems.) Later I was told that it is wrong for the input on Windows. I've
3297 now abstracted the modes into two macros, to make it easier to fiddle with
3298 them, and removed "b" from the input mode under Windows.
3299
3300 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3301
3302 12. Added -help and --help to pcretest as an official way of being reminded
3303 of the options.
3304
3305 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3306 and pcrecpp.cc because they annoy compilers at high warning levels.
3307
3308 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3309
3310 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3311 scripts are not C programs :-) and which was not noticed because it works
3312 on Linux.
3313
3314 16. pcretest is supposed to handle any length of pattern and data line (as one
3315 line or as a continued sequence of lines) by extending its input buffer if
3316 necessary. This feature was broken for very long pattern lines, leading to
3317 a string of junk being passed to pcre_compile() if the pattern was longer
3318 than about 50K.
3319
3320 17. I have done a major re-factoring of the way pcre_compile() computes the
3321 amount of memory needed for a compiled pattern. Previously, there was code
3322 that made a preliminary scan of the pattern in order to do this. That was
3323 OK when PCRE was new, but as the facilities have expanded, it has become
3324 harder and harder to keep it in step with the real compile phase, and there
3325 have been a number of bugs (see for example, 4 above). I have now found a
3326 cunning way of running the real compile function in a "fake" mode that
3327 enables it to compute how much memory it would need, while actually only
3328 ever using a few hundred bytes of working memory and without too many
3329 tests of the mode. This should make future maintenance and development
3330 easier. A side effect of this work is that the limit of 200 on the nesting
3331 depth of parentheses has been removed (though this was never a serious
3332 limitation, I suspect). However, there is a downside: pcre_compile() now
3333 runs more slowly than before (30% or more, depending on the pattern). I
3334 hope this isn't a big issue. There is no effect on runtime performance.
3335
3336 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3337 newline (only possible for the last line of a file) and it was a
3338 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3339
3340 19. Added additional timing features to pcretest. (1) The -tm option now times
3341 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3342 separate command line item, by a number that specifies the number of
3343 repeats to use when timing. The default is 50000; this gives better
3344 precision, but takes uncomfortably long for very large patterns.
3345
3346 20. Extended pcre_study() to be more clever in cases where a branch of a
3347 subpattern has no definite first character. For example, (a*|b*)[cd] would
3348 previously give no result from pcre_study(). Now it recognizes that the
3349 first character must be a, b, c, or d.
3350
3351 21. There was an incorrect error "recursive call could loop indefinitely" if
3352 a subpattern (or the entire pattern) that was being tested for matching an
3353 empty string contained only one non-empty item after a nested subpattern.
3354 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3355 incorrectly, because the \d was being skipped in the check.
3356
3357 22. The pcretest program now has a new pattern option /B and a command line
3358 option -b, which is equivalent to adding /B to every pattern. This causes
3359 it to show the compiled bytecode, without the additional information that
3360 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3361 is the same as /B/I).
3362
3363 23. A new optimization is now able automatically to treat some sequences such
3364 as a*b as a*+b. More specifically, if something simple (such as a character
3365 or a simple class like \d) has an unlimited quantifier, and is followed by
3366 something that cannot possibly match the quantified thing, the quantifier
3367 is automatically "possessified".
3368
3369 24. A recursive reference to a subpattern whose number was greater than 39
3370 went wrong under certain circumstances in UTF-8 mode. This bug could also
3371 have affected the operation of pcre_study().
3372
3373 25. Realized that a little bit of performance could be had by replacing
3374 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3375
3376 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3377
3378 27. Possessive quantifiers such as a++ were previously implemented by turning
3379 them into atomic groups such as ($>a+). Now they have their own opcodes,
3380 which improves performance. This includes the automatically created ones
3381 from 23 above.
3382
3383 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3384 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3385 the first matched character to be a colon. This applied both to named and
3386 numbered groups.
3387
3388 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3389
3390 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3391 building PCRE on Windows easier, so I have included it in the distribution.
3392
3393 31. There is now a check in pcretest against a ridiculously large number being
3394 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3395 loop, the loop is abandoned.
3396
3397 32. Forward references to subpatterns in conditions such as (?(2)...) where
3398 subpattern 2 is defined later cause pcre_compile() to search forwards in
3399 the pattern for the relevant set of parentheses. This search went wrong
3400 when there were unescaped parentheses in a character class, parentheses
3401 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3402
3403 33. "Subroutine" calls and backreferences were previously restricted to
3404 referencing subpatterns earlier in the regex. This restriction has now
3405 been removed.
3406
3407 34. Added a number of extra features that are going to be in Perl 5.10. On the
3408 whole, these are just syntactic alternatives for features that PCRE had
3409 previously implemented using the Python syntax or my own invention. The
3410 other formats are all retained for compatibility.
3411
3412 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3413 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3414 also .NET compatible.
3415
3416 (b) A recursion or subroutine call to a named group can now be defined as
3417 (?&name) as well as (?P>name).
3418
3419 (c) A backreference to a named group can now be defined as \k<name> or
3420 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3421 5.10, are also .NET compatible.
3422
3423 (d) A conditional reference to a named group can now use the syntax
3424 (?(<name>) or (?('name') as well as (?(name).
3425
3426 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3427 groups (named and numbered) that are never evaluated inline, but can be
3428 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3429 is always false. There may be only one alternative in such a group.
3430
3431 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3432 as the simple (?(R). The condition is true only if the most recent
3433 recursion is that of the given number or name. It does not search out
3434 through the entire recursion stack.
3435
3436 (g) The escape \gN or \g{N} has been added, where N is a positive or
3437 negative number, specifying an absolute or relative reference.
3438
3439 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3440 some "unreachable code" warnings.
3441
3442 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3443 things, this adds five new scripts.
3444
3445 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3446 There were also incompatibilities regarding the handling of \Q..\E inside
3447 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3448 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3449
3450 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3451 matches an empty string, and forcibly breaks the loop. There were bugs in
3452 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3453 against aaaa the result was just "a" rather than "aaaa", for example. Two
3454 separate and independent bugs (that affected different cases) have been
3455 fixed.
3456
3457 39. Refactored the code to abolish the use of different opcodes for small
3458 capturing bracket numbers. This is a tidy that I avoided doing when I
3459 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3460 The new approach is not only tidier, it makes it possible to reduce the
3461 memory needed to fix the previous bug (38).
3462
3463 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3464 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3465 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3466 mode.
3467
3468 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3469 report.
3470
3471 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3472 copy construction and assignment in the C++ wrapper.
3473
3474 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3475 couple of bugs that could have given wrong results in the "--newline=crlf"
3476 case.
3477
3478 44. Added a number of casts and did some reorganization of signed/unsigned int
3479 variables following suggestions from Dair Grant. Also renamed the variable
3480 "this" as "item" because it is a C++ keyword.
3481
3482 45. Arranged for dftables to add
3483
3484 #include "pcre_internal.h"
3485
3486 to pcre_chartables.c because without it, gcc 4.x may remove the array
3487 definition from the final binary if PCRE is built into a static library and
3488 dead code stripping is activated.
3489
3490 46. For an unanchored pattern, if a match attempt fails at the start of a
3491 newline sequence, and the newline setting is CRLF or ANY, and the next two
3492 characters are CRLF, advance by two characters instead of one.
3493
3494
3495 Version 6.7 04-Jul-06
3496 ---------------------
3497
3498 1. In order to handle tests when input lines are enormously long, pcretest has
3499 been re-factored so that it automatically extends its buffers when
3500 necessary. The code is crude, but this _is_ just a test program. The
3501 default size has been increased from 32K to 50K.
3502
3503 2. The code in pcre_study() was using the value of the re argument before
3504 testing it for NULL. (Of course, in any sensible call of the function, it
3505 won't be NULL.)
3506
3507 3. The memmove() emulation function in pcre_internal.h, which is used on
3508 systems that lack both memmove() and bcopy() - that is, hardly ever -
3509 was missing a "static" storage class specifier.
3510
3511 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3512 containing an extended class (one that cannot be represented by a bitmap
3513 because it contains high-valued characters or Unicode property items, e.g.
3514 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3515 pattern, but PCRE should not loop if you do not (it no longer does).
3516 [Detail: two cases were found: (a) a repeated subpattern containing an
3517 extended class; (b) a recursive reference to a subpattern that followed a
3518 previous extended class. It wasn't skipping over the extended class
3519 correctly when UTF-8 mode was not set.]
3520
3521 5. A negated single-character class was not being recognized as fixed-length
3522 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3523 compile error "lookbehind assertion is not fixed length".
3524
3525 6. The RunPerlTest auxiliary script was showing an unexpected difference
3526 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3527 write a Perl script that can interpret lines of an input file either as
3528 byte characters or as UTF-8, which is what "perltest" was being required to
3529 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3530 can't do is switch easily at run time between having the "use utf8;" pragma
3531 or not. In the end, I fudged it by using the RunPerlTest script to insert
3532 "use utf8;" explicitly for the UTF-8 tests.
3533
3534 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3535 the end of the subject string, contrary to the documentation and to what
3536 Perl does. This was true of both matching functions. Now it matches only at
3537 the start of the subject and immediately after *internal* newlines.
3538
3539 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3540 a pointer to an int instead of a pointer to an unsigned long int. This
3541 caused problems on 64-bit systems.
3542
3543 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3544 instance of the 'standard' template library not being so standard".
3545
3546 10. There was no check on the number of named subpatterns nor the maximum
3547 length of a subpattern name. The product of these values is used to compute
3548 the size of the memory block for a compiled pattern. By supplying a very
3549 long subpattern name and a large number of named subpatterns, the size
3550 computation could be caused to overflow. This is now prevented by limiting
3551 the length of names to 32 characters, and the number of named subpatterns
3552 to 10,000.
3553
3554 11. Subpatterns that are repeated with specific counts have to be replicated in
3555 the compiled pattern. The size of memory for this was computed from the
3556 length of the subpattern and the repeat count. The latter is limited to
3557 65535, but there was no limit on the former, meaning that integer overflow
3558 could in principle occur. The compiled length of a repeated subpattern is
3559 now limited to 30,000 bytes in order to prevent this.
3560
3561 12. Added the optional facility to have named substrings with the same name.
3562
3563 13. Added the ability to use a named substring as a condition, using the
3564 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3565 are numbers (not recommended). Forward references are permitted.
3566
3567 14. Added forward references in named backreferences (if you see what I mean).
3568
3569 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3570 pattern could run off the end of the subject. For example, the pattern
3571 "(?s)(.{1,5})"8 did this with the subject "ab".
3572
3573 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3574 PCRE_CASELESS was set when matching characters that were quantified with ?
3575 or *.
3576
3577 17. A character class other than a single negated character that had a minimum
3578 but no maximum quantifier - for example [ab]{6,} - was not handled
3579 correctly by pce_dfa_exec(). It would match only one character.
3580
3581 18. A valid (though odd) pattern that looked like a POSIX character
3582 class but used an invalid character after [ (for example [[,abc,]]) caused
3583 pcre_compile() to give the error "Failed: internal error: code overflow" or
3584 in some cases to crash with a glibc free() error. This could even happen if
3585 the pattern terminated after [[ but there just happened to be a sequence of
3586 letters, a binary zero, and a closing ] in the memory that followed.
3587
3588 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3589 over the years. Originally (before any Unicode support), just the bottom 8
3590 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3591 output from "man perlunicode" includes this:
3592
3593 The regular expression compiler produces polymorphic opcodes. That
3594 is, the pattern adapts to the data and automatically switches to
3595 the Unicode character scheme when presented with Unicode data--or
3596 instead uses a traditional byte scheme when presented with byte
3597 data.
3598
3599 Sadly, a wide octal escape does not cause a switch, and in a string with
3600 no other multibyte characters, these octal escapes are treated as before.
3601 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3602 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3603 Unicode string.
3604
3605 I have not perpetrated such confusion in PCRE. Up till now, it took just
3606 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3607 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3608 translate to the appropriate multibyte character.
3609
3610 29. Applied some refactoring to reduce the number of warnings from Microsoft
3611 and Borland compilers. This has included removing the fudge introduced
3612 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3613 a warning about an unused variable.
3614
3615 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3616 characters since release 4.0, because Perl (from release 5.004) does not.
3617 [Or at least, is documented not to: some releases seem to be in conflict
3618 with the documentation.] However, when a pattern was studied with
3619 pcre_study() and all its branches started with \s, PCRE still included VT
3620 as a possible starting character. Of course, this did no harm; it just
3621 caused an unnecessary match attempt.
3622
3623 22. Removed a now-redundant internal flag bit that recorded the fact that case
3624 dependency changed within the pattern. This was once needed for "required
3625 byte" processing, but is no longer used. This recovers a now-scarce options
3626 bit. Also moved the least significant internal flag bit to the most-
3627 significant bit of the word, which was not previously used (hangover from
3628 the days when it was an int rather than a uint) to free up another bit for
3629 the future.
3630
3631 23. Added support for CRLF line endings as well as CR and LF. As well as the
3632 default being selectable at build time, it can now be changed at runtime
3633 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3634 specify that it is scanning data with non-default line endings.
3635
3636 24. Changed the definition of CXXLINK to make it agree with the definition of
3637 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3638
3639 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3640 recursions. This makes a big different to stack usage for some patterns.
3641
3642 26. If a subpattern containing a named recursion or subroutine reference such
3643 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3644 the space required for the compiled pattern went wrong and gave too small a
3645 value. Depending on the environment, this could lead to "Failed: internal
3646 error: code overflow at offset 49" or "glibc detected double free or
3647 corruption" errors.
3648
3649 27. Applied patches from Google (a) to support the new newline modes and (b) to
3650 advance over multibyte UTF-8 characters in GlobalReplace.
3651
3652 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3653 difference for some implementation of PCRE in some Windows version.
3654
3655 29. Added some extra testing facilities to pcretest:
3656
3657 \q<number> in a data line sets the "match limit" value
3658 \Q<number> in a data line sets the "match recursion limt" value
3659 -S <number> sets the stack size, where <number> is in megabytes
3660
3661 The -S option isn't available for Windows.
3662
3663
3664 Version 6.6 06-Feb-06
3665 ---------------------
3666
3667 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3668 in pcreposix.h. I have copied the definition from pcre.h.
3669
3670 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3671 because pcre.h is no longer a built file.
3672
3673 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3674 not normally included in the compiled code.
3675
3676
3677 Version 6.5 01-Feb-06
3678 ---------------------
3679
3680 1. When using the partial match feature with pcre_dfa_exec(), it was not
3681 anchoring the second and subsequent partial matches at the new starting
3682 point. This could lead to incorrect results. For example, with the pattern
3683 /1234/, partially matching against "123" and then "a4" gave a match.
3684
3685 2. Changes to pcregrep:
3686
3687 (a) All non-match returns from pcre_exec() were being treated as failures
3688 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3689 error message is output. Some extra information is given for the
3690 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3691 probably the only errors that are likely to be caused by users (by
3692 specifying a regex that has nested indefinite repeats, for instance).
3693 If there are more than 20 of these errors, pcregrep is abandoned.
3694
3695 (b) A binary zero was treated as data while matching, but terminated the
3696 output line if it was written out. This has been fixed: binary zeroes
3697 are now no different to any other data bytes.
3698
3699 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3700 used to set a locale for matching. The --locale=xxxx long option has
3701 been added (no short equivalent) to specify a locale explicitly on the
3702 pcregrep command, overriding the environment variables.
3703
3704 (d) When -B was used with -n, some line numbers in the output were one less
3705 than they should have been.
3706
3707 (e) Added the -o (--only-matching) option.
3708
3709 (f) If -A or -C was used with -c (count only), some lines of context were
3710 accidentally printed for the final match.
3711
3712 (g) Added the -H (--with-filename) option.
3713
3714 (h) The combination of options -rh failed to suppress file names for files
3715 that were found from directory arguments.
3716
3717 (i) Added the -D (--devices) and -d (--directories) options.
3718
3719 (j) Added the -F (--fixed-strings) option.
3720
3721 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3722
3723 (l) Added the --colo(u)r option.
3724
3725 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3726 is not present by default.
3727
3728 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3729 items such as (?R) or (?1), when the recursion could match a number of
3730 alternatives. If it matched one of the alternatives, but subsequently,
3731 outside the recursion, there was a failure, the code tried to back up into
3732 the recursion. However, because of the way PCRE is implemented, this is not
3733 possible, and the result was an incorrect result from the match.
3734
3735 In order to prevent this happening, the specification of recursion has
3736 been changed so that all such subpatterns are automatically treated as
3737 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3738
3739 4. I had overlooked the fact that, in some locales, there are characters for
3740 which isalpha() is true but neither isupper() nor islower() are true. In
3741 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3742 and ordfeminine) are like this. This affected the treatment of \w and \W
3743 when they appeared in character classes, but not when they appeared outside
3744 a character class. The bit map for "word" characters is now created
3745 separately from the results of isalnum() instead of just taking it from the
3746 upper, lower, and digit maps. (Plus the underscore character, of course.)
3747
3748 5. The above bug also affected the handling of POSIX character classes such as
3749 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3750 permanent tables. Instead, the bit maps for such a class were previously
3751 created as the appropriate unions of the upper, lower, and digit bitmaps.
3752 Now they are created by subtraction from the [[:word:]] class, which has
3753 its own bitmap.
3754
3755 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3756 It is created by subtracting the vertical space characters (\x09, \x0a,
3757 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3758 subtraction was done in the overall bitmap for a character class, meaning
3759 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3760 be recognized. This bug has been fixed.
3761
3762 7. Patches from the folks at Google:
3763
3764 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3765 real life, but is still worth protecting against".
3766
3767 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3768 regular expressions".
3769
3770 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3771 have it.
3772
3773 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3774 "configure" and the latter not, in order to fix a problem somebody had
3775 with compiling the Arg class on HP-UX.
3776
3777 (e) Improve the error-handling of the C++ wrapper a little bit.
3778
3779 (f) New tests for checking recursion limiting.
3780
3781 8. The pcre_memmove() function, which is used only if the environment does not
3782 have a standard memmove() function (and is therefore rarely compiled),
3783 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3784 returning a result (though PCRE never actually uses the result).
3785
3786 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3787 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3788 returned instead of calling malloc() with an overflowing number that would
3789 most likely cause subsequent chaos.
3790
3791 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3792
3793 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3794 with this option is matched, the nmatch and pmatch options of regexec() are
3795 ignored.
3796
3797 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3798 provided in case anyone wants to the the POSIX interface with UTF-8
3799 strings.
3800
3801 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3802 C++ linking (needed for some HP-UX environments).
3803
3804 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3805 (unused parameter) and in the pcre_printint() function (omitted "default"
3806 switch label when the default is to do nothing).
3807
3808 15. Added some code to make it possible, when PCRE is compiled as a C++
3809 library, to replace subject pointers for pcre_exec() with a smart pointer
3810 class, thus making it possible to process discontinuous strings.
3811
3812 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3813 much the same function. They were added by different people who were trying
3814 to make PCRE easy to compile on non-Unix systems. It has been suggested
3815 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3816 for compiling on Windows systems. I have therefore replaced it with
3817 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3818 defaults to "extern" for C or "extern C" for C++, which works fine on
3819 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3820 SCOPE with something explicit in config.h. In addition:
3821
3822 (a) pcreposix.h still had just "extern" instead of either of these macros;
3823 I have replaced it with PCRE_DATA_SCOPE.
3824
3825 (b) Functions such as _pcre_xclass(), which are internal to the library,
3826 but external in the C sense, all had PCRE_EXPORT in their definitions.
3827 This is apparently wrong for the Windows case, so I have removed it.
3828 (It makes no difference on Unix-like systems.)
3829
3830 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3831 of recursive calls to match(). This is different to MATCH_LIMIT because
3832 that limits the total number of calls to match(), not all of which increase
3833 the depth of recursion. Limiting the recursion depth limits the amount of
3834 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3835 when PCRE is compiled, and changed at run time. A patch from Google adds
3836 this functionality to the C++ interface.
3837
3838 18. Changes to the handling of Unicode character properties:
3839
3840 (a) Updated the table to Unicode 4.1.0.
3841
3842 (b) Recognize characters that are not in the table as "Cn" (undefined).
3843
3844 (c) I revised the way the table is implemented to a much improved format
3845 which includes recognition of ranges. It now supports the ranges that
3846 are defined in UnicodeData.txt, and it also amalgamates other
3847 characters into ranges. This has reduced the number of entries in the
3848 table from around 16,000 to around 3,000, thus reducing its size
3849 considerably. I realized I did not need to use a tree structure after
3850 all - a binary chop search is just as efficient. Having reduced the
3851 number of entries, I extended their size from 6 bytes to 8 bytes to
3852 allow for more data.
3853
3854 (d) Added support for Unicode script names via properties such as \p{Han}.
3855
3856 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3857 matching that character.
3858
3859 20. When matching a repeated Unicode property with a minimum greater than zero,
3860 (for example \pL{2,}), PCRE could look past the end of the subject if it
3861 reached it while seeking the minimum number of characters. This could
3862 happen only if some of the characters were more than one byte long, because
3863 there is a check for at least the minimum number of bytes.
3864
3865 21. Refactored the implementation of \p and \P so as to be more general, to
3866 allow for more different types of property in future. This has changed the
3867 compiled form incompatibly. Anybody with saved compiled patterns that use
3868 \p or \P will have to recompile them.
3869
3870 22. Added "Any" and "L&" to the supported property types.
3871
3872 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3873 but give a compile time error if the value is greater than 0xff.
3874
3875 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3876 accidentally not being installed or uninstalled.
3877
3878 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3879 made were to insert the current release number. This seemed silly, because
3880 it made things harder for people building PCRE on systems that don't run
3881 "configure". I have turned pcre.h into a distributed file, no longer built
3882 by "configure", with the version identification directly included. There is
3883 no longer a pcre.h.in file.
3884
3885 However, this change necessitated a change to the pcre-config script as
3886 well. It is built from pcre-config.in, and one of the substitutions was the
3887 release number. I have updated configure.ac so that ./configure now finds
3888 the release number by grepping pcre.h.
3889
3890 26. Added the ability to run the tests under valgrind.
3891
3892
3893 Version 6.4 05-Sep-05
3894 ---------------------
3895
3896 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3897 "--" to be printed when multiple files were scanned, even when none of the
3898 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3899 consider it to be a bug, and have restored the previous behaviour.
3900
3901 2. A couple of code tidies to get rid of compiler warnings.
3902
3903 3. The pcretest program used to cheat by referring to symbols in the library
3904 whose names begin with _pcre_. These are internal symbols that are not
3905 really supposed to be visible externally, and in some environments it is
3906 possible to suppress them. The cheating is now confined to including
3907 certain files from the library's source, which is a bit cleaner.
3908
3909 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3910 file's purpose clearer.
3911
3912 5. Reorganized pcre_ucp_findchar().
3913
3914
3915 Version 6.3 15-Aug-05
3916 ---------------------
3917
3918 1. The file libpcre.pc.in did not have general read permission in the tarball.
3919
3920 2. There were some problems when building without C++ support:
3921
3922 (a) If C++ support was not built, "make install" and "make test" still
3923 tried to test it.
3924
3925 (b) There were problems when the value of CXX was explicitly set. Some
3926 changes have been made to try to fix these, and ...
3927
3928 (c) --disable-cpp can now be used to explicitly disable C++ support.
3929
3930 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3931 backslash in a target when C++ was disabled. This confuses some
3932 versions of "make", apparently. Using an intermediate variable solves
3933 this. (Same for CPP_LOBJ.)
3934
3935 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3936 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3937 necessary on certain architectures.
3938
3939 4. Added a setting of -export-symbols-regex to the link command to remove
3940 those symbols that are exported in the C sense, but actually are local
3941 within the library, and not documented. Their names all begin with
3942 "_pcre_". This is not a perfect job, because (a) we have to except some
3943 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3944 available (and never for static libraries). I have made a note to try to
3945 find a way round (a) in the future.
3946
3947
3948 Version 6.2 01-Aug-05
3949 ---------------------
3950
3951 1. There was no test for integer overflow of quantifier values. A construction
3952 such as {1111111111111111} would give undefined results. What is worse, if
3953 a minimum quantifier for a parenthesized subpattern overflowed and became
3954 negative, the calculation of the memory size went wrong. This could have
3955 led to memory overwriting.
3956
3957 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3958
3959 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3960 operating environments where this matters.
3961
3962 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3963 PCRE options from within the C++ wrapper.
3964
3965 5. Named capturing subpatterns were not being correctly counted when a pattern
3966 was compiled. This caused two problems: (a) If there were more than 100
3967 such subpatterns, the calculation of the memory needed for the whole
3968 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3969 back references of the form \12, where the number was greater than 9, were
3970 not recognized as back references, even though there were sufficient
3971 previous subpatterns.
3972
3973 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3974 versions of gcc, e.g. 2.95.4.
3975
3976
3977 Version 6.1 21-Jun-05
3978 ---------------------
3979
3980 1. There was one reference to the variable "posix" in pcretest.c that was not
3981 surrounded by "#if !defined NOPOSIX".
3982
3983 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3984 the cross-check on the old pcre_info() function, for the benefit of the
3985 cut-down version of PCRE that is currently imported into Exim.
3986
3987 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3988 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3989 patterns that start (?i) but I don't think that matters. The use of (?i) is
3990 just an example; this all applies to the other options as well.
3991
3992 4. Since libtool seems to echo the compile commands it is issuing, the output
3993 from "make" can be reduced a bit by putting "@" in front of each libtool
3994 compile command.
3995
3996 5. Patch from the folks at Google for configure.in to be a bit more thorough
3997 in checking for a suitable C++ installation before trying to compile the
3998 C++ stuff. This should fix a reported problem when a compiler was present,
3999 but no suitable headers.
4000
4001 6. The man pages all had just "PCRE" as their title. I have changed them to
4002 be the relevant file name. I have also arranged that these names are
4003 retained in the file doc/pcre.txt, which is a concatenation in text format
4004 of all the man pages except the little individual ones for each function.
4005
4006 7. The NON-UNIX-USE file had not been updated for the different set of source
4007 files that come with release 6. I also added a few comments about the C++
4008 wrapper.
4009
4010
4011 Version 6.0 07-Jun-05
4012 ---------------------
4013
4014 1. Some minor internal re-organization to help with my DFA experiments.
4015
4016 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4017 didn't matter for the library itself when fully configured, but did matter
4018 when compiling without UCP support, or within Exim, where the ucp files are
4019 not imported.
4020
4021 3. Refactoring of the library code to split up the various functions into
4022 different source modules. The addition of the new DFA matching code (see
4023 below) to a single monolithic source would have made it really too
4024 unwieldy, quite apart from causing all the code to be include in a
4025 statically linked application, when only some functions are used. This is
4026 relevant even without the DFA addition now that patterns can be compiled in
4027 one application and matched in another.
4028
4029 The downside of splitting up is that there have to be some external
4030 functions and data tables that are used internally in different modules of
4031 the library but which are not part of the API. These have all had their
4032 names changed to start with "_pcre_" so that they are unlikely to clash
4033 with other external names.
4034
4035 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4036 a different (DFA) algorithm. Although it is slower than the original
4037 function, it does have some advantages for certain types of matching
4038 problem.
4039
4040 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4041 including restarting after a partial match.
4042
4043 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4044 defined when compiling for Windows was sent to me. I have put it into the
4045 code, though I have no means of testing or verifying it.
4046
4047 7. Added the pcre_refcount() auxiliary function.
4048
4049 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4050 match before or at the first newline in the subject string. In pcretest,
4051 the /f option on a pattern can be used to set this.
4052
4053 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4054 would behave wrongly. This has been present in PCRE since release 4.0.
4055
4056 10. A number of changes to the pcregrep command:
4057
4058 (a) Refactored how -x works; insert ^(...)$ instead of setting
4059 PCRE_ANCHORED and checking the length, in preparation for adding
4060 something similar for -w.
4061
4062 (b) Added the -w (match as a word) option.
4063
4064 (c) Refactored the way lines are read and buffered so as to have more
4065 than one at a time available.
4066
4067 (d) Implemented a pcregrep test script.
4068
4069 (e) Added the -M (multiline match) option. This allows patterns to match
4070 over several lines of the subject. The buffering ensures that at least
4071 8K, or the rest of the document (whichever is the shorter) is available
4072 for matching (and similarly the previous 8K for lookbehind assertions).
4073
4074 (f) Changed the --help output so that it now says
4075
4076 -w, --word-regex(p)
4077
4078 instead of two lines, one with "regex" and the other with "regexp"
4079 because that confused at least one person since the short forms are the
4080 same. (This required a bit of code, as the output is generated
4081 automatically from a table. It wasn't just a text change.)
4082
4083 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4084 option but starts with a hyphen. Could be a pattern or a path name
4085 starting with a hyphen, for instance.
4086
4087 (h) "-" can be given as a file name to represent stdin.
4088
4089 (i) When file names are being printed, "(standard input)" is used for
4090 the standard input, for compatibility with GNU grep. Previously
4091 "<stdin>" was used.
4092
4093 (j) The option --label=xxx can be used to supply a name to be used for
4094 stdin when file names are being printed. There is no short form.
4095
4096 (k) Re-factored the options decoding logic because we are going to add
4097 two more options that take data. Such options can now be given in four
4098 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4099
4100 (l) Added the -A, -B, and -C options for requesting that lines of context
4101 around matches be printed.
4102
4103 (m) Added the -L option to print the names of files that do not contain
4104 any matching lines, that is, the complement of -l.
4105
4106 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4107 continue to scan other files.
4108
4109 (o) The -s option was incorrectly implemented. For compatibility with other
4110 greps, it now suppresses the error message for a non-existent or non-
4111 accessible file (but not the return code). There is a new option called
4112 -q that suppresses the output of matching lines, which was what -s was
4113 previously doing.
4114
4115 (p) Added --include and --exclude options to specify files for inclusion
4116 and exclusion when recursing.
4117
4118 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4119 Hopefully, it now does.
4120
4121 12. Missing cast in pcre_study().
4122
4123 13. Added an "uninstall" target to the makefile.
4124
4125 14. Replaced "extern" in the function prototypes in Makefile.in with
4126 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4127 world, but is set differently for Windows.
4128
4129 15. Added a second compiling function called pcre_compile2(). The only
4130 difference is that it has an extra argument, which is a pointer to an
4131 integer error code. When there is a compile-time failure, this is set
4132 non-zero, in addition to the error test pointer being set to point to an
4133 error message. The new argument may be NULL if no error number is required
4134 (but then you may as well call pcre_compile(), which is now just a
4135 wrapper). This facility is provided because some applications need a
4136 numeric error indication, but it has also enabled me to tidy up the way
4137 compile-time errors are handled in the POSIX wrapper.
4138
4139 16. Added VPATH=.libs to the makefile; this should help when building with one
4140 prefix path and installing with another. (Or so I'm told by someone who
4141 knows more about this stuff than I do.)
4142
4143 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4144 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4145 match everything, including newlines. This is not POSIX-compatible, but
4146 somebody wanted the feature. From pcretest it can be activated by using
4147 both the P and the s flags.
4148
4149 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4150
4151 19. libpcre.pc was being incorrectly installed as executable.
4152
4153 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4154 it now also looks for '\r' so that it will work unmodified on Windows.
4155
4156 21. Added Google's contributed C++ wrapper to the distribution.
4157
4158 22. Added some untidy missing memory free() calls in pcretest, to keep
4159 Electric Fence happy when testing.
4160
4161
4162
4163 Version 5.0 13-Sep-04
4164 ---------------------
4165
4166 1. Internal change: literal characters are no longer packed up into items
4167 containing multiple characters in a single byte-string. Each character
4168 is now matched using a separate opcode. However, there may be more than one
4169 byte in the character in UTF-8 mode.
4170
4171 2. The pcre_callout_block structure has two new fields: pattern_position and
4172 next_item_length. These contain the offset in the pattern to the next match
4173 item, and its length, respectively.
4174
4175 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4176 insertion of callouts before each pattern item. Added the /C option to
4177 pcretest to make use of this.
4178
4179 4. On the advice of a Windows user, the lines
4180
4181 #if defined(_WIN32) || defined(WIN32)
4182 _setmode( _fileno( stdout ), 0x8000 );
4183 #endif /* defined(_WIN32) || defined(WIN32) */
4184
4185 have been added to the source of pcretest. This apparently does useful
4186 magic in relation to line terminators.
4187
4188 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4189 for the benefit of those environments where the "b" makes a difference.
4190
4191 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4192 to know about it. I have put a hack into configure.in that adds in code
4193 to set GCC=yes if CC=icc. This seems to end up at a point in the
4194 generated configure script that is early enough to affect the setting of
4195 compiler options, which is what is needed, but I have no means of testing
4196 whether it really works. (The user who reported this had patched the
4197 generated configure script, which of course I cannot do.)
4198
4199 LATER: After change 22 below (new libtool files), the configure script
4200 seems to know about icc (and also ecc). Therefore, I have commented out
4201 this hack in configure.in.
4202
4203 7. Added support for pkg-config (2 patches were sent in).
4204
4205 8. Negated POSIX character classes that used a combination of internal tables
4206 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4207 [[:^ascii]]. Typically, they would match almost any characters. The other
4208 POSIX classes were not broken in this way.
4209
4210 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4211 to find the match, as PCRE was deluded into thinking that the match had to
4212 start at the start point or following a newline. The same bug applied to
4213 patterns with negative forward assertions or any backward assertions
4214 preceding ".*" at the start, unless the pattern required a fixed first
4215 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4216
4217 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4218 starting at the last subject character, bytes beyond the end of the subject
4219 string were read.
4220
4221 11. Renamed the variable "class" as "classbits" to make life easier for C++
4222 users. (Previously there was a macro definition, but it apparently wasn't
4223 enough.)
4224
4225 12. Added the new field "tables" to the extra data so that tables can be passed
4226 in at exec time, or the internal tables can be re-selected. This allows
4227 a compiled regex to be saved and re-used at a later time by a different
4228 program that might have everything at different addresses.
4229
4230 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4231 -R library as well as a -L library.
4232
4233 14. The debugging options of pcretest (-d on the command line or D on a
4234 pattern) showed incorrect output for anything following an extended class
4235 that contained multibyte characters and which was followed by a quantifier.
4236
4237 15. Added optional support for general category Unicode character properties
4238 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4239 support. It adds about 90K to the size of the library. The meanings of the
4240 inbuilt class escapes such as \d and \s have NOT been changed.
4241
4242 16. Updated pcredemo.c to include calls to free() to release the memory for the
4243 compiled pattern.
4244
4245 17. The generated file chartables.c was being created in the source directory
4246 instead of in the building directory. This caused the build to fail if the
4247 source directory was different from the building directory, and was
4248 read-only.
4249
4250 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4251 file. No doubt somebody will tell me if they don't make sense... Also added
4252 Dan Mooney's comments about building on OpenVMS.
4253
4254 19. Added support for partial matching via the PCRE_PARTIAL option for
4255 pcre_exec() and the \P data escape in pcretest.
4256
4257 20. Extended pcretest with 3 new pattern features:
4258
4259 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4260 write the compiled pattern to the file whose name is "rest-of-line".
4261 This is a straight binary dump of the data, with the saved pointer to
4262 the character tables forced to be NULL. The study data, if any, is
4263 written too. After writing, pcretest reads a new pattern.
4264
4265 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4266 compiled pattern from the given file. There must not be any
4267 occurrences of "<" in the file name (pretty unlikely); if there are,
4268 pcretest will instead treat the initial "<" as a pattern delimiter.
4269 After reading in the pattern, pcretest goes on to read data lines as
4270 usual.
4271
4272 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4273 and 16-bit fields in a compiled pattern, to simulate a pattern that
4274 was compiled on a host of opposite endianness.
4275
4276 21. The pcre-exec() function can now cope with patterns that were compiled on
4277 hosts of opposite endianness, with this restriction:
4278
4279 As for any compiled expression that is saved and used later, the tables
4280 pointer field cannot be preserved; the extra_data field in the arguments
4281 to pcre_exec() should be used to pass in a tables address if a value
4282 other than the default internal tables were used at compile time.
4283
4284 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4285 now diagnosed as an error. Previously, most of the time, a negative number
4286 would have been treated as zero, but if in addition "ovector" was passed as
4287 NULL, a crash could occur.
4288
4289 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4290 new versions from the libtool 1.5 distribution (the last one is a copy of
4291 a file called libtool.m4). This seems to have fixed the need to patch
4292 "configure" to support Darwin 1.3 (which I used to do). However, I still
4293 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4294 workstation).
4295
4296 24. Changed the PCRE licence to be the more standard "BSD" licence.
4297
4298
4299 Version 4.5 01-Dec-03
4300 ---------------------
4301
4302 1. There has been some re-arrangement of the code for the match() function so
4303 that it can be compiled in a version that does not call itself recursively.
4304 Instead, it keeps those local variables that need separate instances for
4305 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4306 needs to "recurse". Keeping track of where control must go is done by means
4307 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4308 hide most of the details from the main code, and operates only if
4309 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4310 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4311 operating.
4312
4313 To make it easier for callers to provide specially tailored get/free
4314 functions for this usage, two new functions, pcre_stack_malloc, and
4315 pcre_stack_free, are used. They are always called in strict stacking order,
4316 and the size of block requested is always the same.
4317
4318 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4319 PCRE has been compiled to use the stack or the heap for recursion. The
4320 -C option of pcretest uses this to show which version is compiled.
4321
4322 A new data escape \S, is added to pcretest; it causes the amounts of store
4323 obtained and freed by both kinds of malloc/free at match time to be added
4324 to the output.
4325
4326 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4327 what's available on my current Linux desktop machine.
4328
4329 3. When matching a UTF-8 string, the test for a valid string at the start has
4330 been extended. If start_offset is not zero, PCRE now checks that it points
4331 to a byte that is the start of a UTF-8 character. If not, it returns
4332 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4333 this is necessary because there may be backward assertions in the pattern.
4334 When matching the same subject several times, it may save resources to use
4335 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4336
4337 4. The code for checking the validity of UTF-8 strings has been tightened so
4338 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4339 containing "overlong sequences".
4340
4341 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4342 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4343 should have been "&", but it just so happened that all the cases this let
4344 through by mistake were picked up later in the function.
4345
4346 6. I had used a variable called "isblank" - this is a C99 function, causing
4347 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4348
4349 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4350 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4351 so the version is shown; (c) stop "make test" echoing "./RunTest".
4352
4353 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4354
4355 9. The prototype for memmove() for systems that don't have it was using
4356 size_t, but the inclusion of the header that defines size_t was later. I've
4357 moved the #includes for the C headers earlier to avoid this.
4358
4359 10. Added some adjustments to the code to make it easier to compiler on certain
4360 special systems:
4361
4362 (a) Some "const" qualifiers were missing.
4363 (b) Added the macro EXPORT before all exported functions; by default this
4364 is defined to be empty.
4365 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4366 that it reads its output file name as an argument instead of writing
4367 to the standard output and assuming this can be redirected.
4368
4369 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4370 class containing characters with values greater than 255, PCRE compilation
4371 went into a loop.
4372
4373 12. A recursive reference to a subpattern that was within another subpattern
4374 that had a minimum quantifier of zero caused PCRE to crash. For example,
4375 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4376 recursion. If the recursively-called subpattern itself had a zero repeat,
4377 that was OK.
4378
4379 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4380 buffer into which it was copied (for escape processing) was still set at
4381 1024, so long lines caused crashes.
4382
4383 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4384 "internal error: code overflow...". This applied to any character class
4385 that was followed by a possessive quantifier.
4386
4387 15. Modified the Makefile to add libpcre.la as a prerequisite for
4388 libpcreposix.la because I was told this is needed for a parallel build to
4389 work.
4390
4391 16. If a pattern that contained .* following optional items at the start was
4392 studied, the wrong optimizing data was generated, leading to matching
4393 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4394 matching string must start with a or b or c. The correct conclusion for
4395 this pattern is that a match can start with any character.
4396
4397
4398 Version 4.4 13-Aug-03
4399 ---------------------
4400
4401 1. In UTF-8 mode, a character class containing characters with values between
4402 127 and 255 was not handled correctly if the compiled pattern was studied.
4403 In fixing this, I have also improved the studying algorithm for such
4404 classes (slightly).
4405
4406 2. Three internal functions had redundant arguments passed to them. Removal
4407 might give a very teeny performance improvement.
4408
4409 3. Documentation bug: the value of the capture_top field in a callout is *one
4410 more than* the number of the hightest numbered captured substring.
4411
4412 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4413 in incorrectly linking with a previously installed version. They now link
4414 explicitly with libpcre.la.
4415
4416 5. configure.in no longer needs to recognize Cygwin specially.
4417
4418 6. A problem in pcre.in for Windows platforms is fixed.
4419
4420 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4421 pcretest, it used to include the size of the study block as part of its
4422 output. Unfortunately, the structure contains a field that has a different
4423 size on different hardware architectures. This meant that the tests that
4424 showed this size failed. As the block is currently always of a fixed size,
4425 this information isn't actually particularly useful in pcretest output, so
4426 I have just removed it.
4427
4428 8. Three pre-processor statements accidentally did not start in column 1.
4429 Sadly, there are *still* compilers around that complain, even though
4430 standard C has not required this for well over a decade. Sigh.
4431
4432 9. In pcretest, the code for checking callouts passed small integers in the
4433 callout_data field, which is a void * field. However, some picky compilers
4434 complained about the casts involved for this on 64-bit systems. Now
4435 pcretest passes the address of the small integer instead, which should get
4436 rid of the warnings.
4437
4438 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4439 both compile and run time, and gives an error if an invalid UTF-8 sequence
4440 is found. There is a option for disabling this check in cases where the
4441 string is known to be correct and/or the maximum performance is wanted.
4442
4443 11. In response to a bug report, I changed one line in Makefile.in from
4444
4445 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4446 to
4447 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4448
4449 to look similar to other lines, but I have no way of telling whether this
4450 is the right thing to do, as I do not use Windows. No doubt I'll get told
4451 if it's wrong...
4452
4453
4454 Version 4.3 21-May-03
4455 ---------------------
4456
4457 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4458 Makefile.
4459
4460 2. Some refactoring to improve the quality of the code:
4461
4462 (i) The utf8_table... variables are now declared "const".
4463
4464 (ii) The code for \cx, which used the "case flipping" table to upper case
4465 lower case letters, now just substracts 32. This is ASCII-specific,
4466 but the whole concept of \cx is ASCII-specific, so it seems
4467 reasonable.
4468
4469 (iii) PCRE was using its character types table to recognize decimal and
4470 hexadecimal digits in the pattern. This is silly, because it handles
4471 only 0-9, a-f, and A-F, but the character types table is locale-
4472 specific, which means strange things might happen. A private
4473 table is now used for this - though it costs 256 bytes, a table is
4474 much faster than multiple explicit tests. Of course, the standard
4475 character types table is still used for matching digits in subject
4476 strings against \d.
4477
4478 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4479 ending in _t are). So I've renamed it as ESC_tee.
4480
4481 3. The first argument for regexec() in the POSIX wrapper should have been
4482 defined as "const".
4483
4484 4. Changed pcretest to use malloc() for its buffers so that they can be
4485 Electric Fenced for debugging.
4486
4487 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4488 to read one or more bytes before the start of the subject string. Often this
4489 had no effect on PCRE's behaviour, but in some circumstances it could
4490 provoke a segmentation fault.
4491
4492 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4493 to try to read one or more bytes before the start of the subject string.
4494
4495 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4496 UTF-8 support could misbehave in various ways if the subject string
4497 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4498 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4499 back over UTF-8 characters.)
4500
4501
4502 Version 4.2 14-Apr-03
4503 ---------------------
4504
4505 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4506
4507 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4508 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4509 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4510 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4511 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4512 and BUILD_EXEEXT
4513 Note: automatic setting of the BUILD variables is not yet working
4514 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4515 compile-time but not at link-time
4516 [LINK]: use for linking executables only
4517 make different versions for Windows and non-Windows
4518 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4519 libraries
4520 [LINK_FOR_BUILD]: new variable
4521 [OBJEXT]: use throughout
4522 [EXEEXT]: use throughout
4523 <winshared>: new target
4524 <wininstall>: new target
4525 <dftables.o>: use native compiler
4526 <dftables>: use native linker
4527 <install>: handle Windows platform correctly
4528 <clean>: ditto
4529 <check>: ditto
4530 copy DLL to top builddir before testing
4531
4532 As part of these changes, -no-undefined was removed again. This was reported
4533 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4534 in any case.
4535
4536 3. Some tidies to get rid of compiler warnings:
4537
4538 . In the match_data structure, match_limit was an unsigned long int, whereas
4539 match_call_count was an int. I've made them both unsigned long ints.
4540
4541 . In pcretest the fact that a const uschar * doesn't automatically cast to
4542 a void * provoked a warning.
4543
4544 . Turning on some more compiler warnings threw up some "shadow" variables
4545 and a few more missing casts.
4546
4547 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4548 option, a class that contained a single character with a value between 128
4549 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4550
4551 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4552 option, a class that contained several characters, but with at least one
4553 whose value was between 128 and 255 caused PCRE to crash.
4554
4555
4556 Version 4.1 12-Mar-03
4557 ---------------------
4558
4559 1. Compiling with gcc -pedantic found a couple of places where casts were
4560 needed, and a string in dftables.c that was longer than standard compilers are
4561 required to support.
4562
4563 2. Compiling with Sun's compiler found a few more places where the code could
4564 be tidied up in order to avoid warnings.
4565
4566 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4567 first of these names is deprecated in the latest Autoconf in favour of the name
4568 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4569 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4570 analogy I have changed it to CFLAGS_FOR_BUILD.
4571
4572 4. Added -no-undefined to the linking command in the Makefile, because this is
4573 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4574 linking step for the pcreposix library.
4575
4576 5. PCRE was failing to diagnose the case of two named groups with the same
4577 name.
4578
4579 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4580 literal character that is needed in the subject for a match, and scans along to
4581 ensure that it is present before embarking on the full matching process. This
4582 saves time in cases of nested unlimited repeats that are never going to match.
4583 Problem: the scan can take a lot of time if the subject is very long (e.g.
4584 megabytes), thus penalizing straightforward matches. It is now done only if the
4585 amount of subject to be scanned is less than 1000 bytes.
4586
4587 7. A lesser problem with the same optimization is that it was recording the
4588 first character of an anchored pattern as "needed", thus provoking a search
4589 right along the subject, even when the first match of the pattern was going to
4590 fail. The "needed" character is now not set for anchored patterns, unless it
4591 follows something in the pattern that is of non-fixed length. Thus, it still
4592 fulfils its original purpose of finding quick non-matches in cases of nested
4593 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4594
4595
4596 Version 4.0 17-Feb-03
4597 ---------------------
4598
4599 1. If a comment in an extended regex that started immediately after a meta-item
4600 extended to the end of string, PCRE compiled incorrect data. This could lead to
4601 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4602
4603 2. Moved to autoconf 2.53 and libtool 1.4.2.
4604
4605 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4606 the special perltest8 script is no longer needed - all the tests can be run
4607 from a single perltest script.
4608
4609 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4610 by \s. It has now been removed in PCRE. This means it isn't recognized as
4611 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4612 class [:space:] *does* include VT, thereby creating a mess.
4613
4614 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4615 space and tab.
4616
4617 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4618 its new features into the main test script, reducing the number of scripts.
4619
4620 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4621 were backward compatible, and made the (?i) apply to the whole pattern, as if
4622 /i were given. Now it behaves more logically, and applies the option setting
4623 only to what follows. PCRE has been changed to follow suit. However, if it
4624 finds options settings right at the start of the pattern, it extracts them into
4625 the global options, as before. Thus, they show up in the info data.
4626
4627 8. Added support for the \Q...\E escape sequence. Characters in between are
4628 treated as literals. This is slightly different from Perl in that $ and @ are
4629 also handled as literals inside the quotes. In Perl, they will cause variable
4630 interpolation. Note the following examples:
4631
4632 Pattern PCRE matches Perl matches
4633
4634 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4635 \Qabc\$xyz\E abc\$xyz abc\$xyz
4636 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4637
4638 For compatibility with Perl, \Q...\E sequences are recognized inside character
4639 classes as well as outside them.
4640
4641 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4642 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4643 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4644 signed/unsigned warnings.
4645
4646 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4647 option for pcretest, so I've replaced it by a simple function that does just
4648 that job.
4649
4650 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4651 "pcregrep -".
4652
4653 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4654 Java package. This provides some syntactic sugar for simple cases of what my
4655 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4656 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4657 item, you can use this simplified notation. Note that only makes sense with
4658 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4659 greediness, whatever the setting of the PCRE_UNGREEDY option.
4660
4661 13. A change of greediness default within a pattern was not taking effect at
4662 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4663 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4664 was abstracted outside.
4665
4666 14. PCRE now supports the \G assertion. It is true when the current matching
4667 position is at the start point of the match. This differs from \A when the
4668 starting offset is non-zero. Used with the /g option of pcretest (or similar
4669 code), it works in the same way as it does for Perl's /g option. If all
4670 alternatives of a regex begin with \G, the expression is anchored to the start
4671 match position, and the "anchored" flag is set in the compiled expression.
4672
4673 15. Some bugs concerning the handling of certain option changes within patterns
4674 have been fixed. These applied to options other than (?ims). For example,
4675 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4676 been the other way round. Some of this was related to change 7 above.
4677
4678 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4679 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4680 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4681 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4682
4683 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4684 mode. Unlike ".", it always matches newline, whatever the setting of
4685 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4686 assertions. Perl allows it, but it doesn't (in general) work because it can't
4687 calculate the length of the lookbehind. At least, that's the case for Perl
4688 5.8.0 - I've been told they are going to document that it doesn't work in
4689 future.
4690
4691 18. Added an error diagnosis for escapes that PCRE does not support: these are
4692 \L, \l, \N, \P, \p, \U, \u, and \X.
4693
4694 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4695 reading past the end of the pattern in cases such as /[abcd/.
4696
4697 20. PCRE was getting more memory than necessary for patterns with classes that
4698 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4699
4700 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4701 compiling PCRE for use with Virtual Pascal.
4702
4703 22. Small fix to the Makefile to make it work properly if the build is done
4704 outside the source tree.
4705
4706 23. Added a new extension: a condition to go with recursion. If a conditional
4707 subpattern starts with (?(R) the "true" branch is used if recursion has
4708 happened, whereas the "false" branch is used only at the top level.
4709
4710 24. When there was a very long string of literal characters (over 255 bytes
4711 without UTF support, over 250 bytes with UTF support), the computation of how
4712 much memory was required could be incorrect, leading to segfaults or other
4713 strange effects.
4714
4715 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4716 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4717 there was a subsequent back reference to those brackets. This meant that, for
4718 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4719 possible to check for precisely this case. All we can do is abandon the
4720 optimization if .* occurs inside capturing brackets when there are any back
4721 references whatsoever. (See below for a better fix that came later.)
4722
4723 26. The handling of the optimization for finding the first character of a
4724 non-anchored pattern, and for finding a character that is required later in the
4725 match were failing in some cases. This didn't break the matching; it just
4726 failed to optimize when it could. The way this is done has been re-implemented.
4727
4728 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4729
4730 28. Added a new feature that provides some of the functionality that Perl
4731 provides with (?{...}). The facility is termed a "callout". The way it is done
4732 in PCRE is for the caller to provide an optional function, by setting
4733 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4734 global variable. By default it is unset, which disables all calling out. To get
4735 the function called, the regex must include (?C) at appropriate points. This
4736 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4737 This provides a means of identifying different callout points. When PCRE
4738 reaches such a point in the regex, if pcre_callout has been set, the external
4739 function is called. It is provided with data in a structure called
4740 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4741 matching continues; if it returns a non-zero value, the match at the current
4742 point fails. However, backtracking will occur if possible. [This was changed
4743 later and other features added - see item 49 below.]
4744
4745 29. pcretest is upgraded to test the callout functionality. It provides a
4746 callout function that displays information. By default, it shows the start of
4747 the match and the current position in the text. There are some new data escapes
4748 to vary what happens:
4749
4750 \C+ in addition, show current contents of captured substrings
4751 \C- do not supply a callout function
4752 \C!n return 1 when callout number n is reached
4753 \C!n!m return 1 when callout number n is reached for the mth time
4754
4755 30. If pcregrep was called with the -l option and just a single file name, it
4756 output "<stdin>" if a match was found, instead of the file name.
4757
4758 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4759 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4760 pcre_exec(). This saves a malloc/free per call. The default value of
4761 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4762 when configuring.
4763
4764 32. The default maximum size of a compiled pattern is 64K. There have been a
4765 few cases of people hitting this limit. The code now uses macros to handle the
4766 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4767 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4768 configuring. Tests 2 and 5 work only with 2-byte links because they output
4769 debugging information about compiled patterns.
4770
4771 33. Internal code re-arrangements:
4772
4773 (a) Moved the debugging function for printing out a compiled regex into
4774 its own source file (printint.c) and used #include to pull it into
4775 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4776 separate copies.
4777
4778 (b) Defined the list of op-code names for debugging as a macro in
4779 internal.h so that it is next to the definition of the opcodes.
4780
4781 (c) Defined a table of op-code lengths for simpler skipping along compiled
4782 code. This is again a macro in internal.h so that it is next to the
4783 definition of the opcodes.
4784
4785 34. Added support for recursive calls to individual subpatterns, along the
4786 lines of Robin Houston's patch (but implemented somewhat differently).
4787
4788 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4789 allow it to read and process whole directories in Win32. This code was
4790 contributed by Lionel Fourquaux; it has not been tested by me.
4791
4792 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4793 used to name a group. Names consist of alphanumerics and underscores, and must
4794 be unique. Back references use the syntax (?P=name) and recursive calls use
4795 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4796 numbers. The function pcre_fullinfo() can be used after compilation to extract
4797 a name/number map. There are three relevant calls:
4798
4799 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4800 PCRE_INFO_NAMECOUNT yields the number of entries
4801 PCRE_INFO_NAMETABLE yields a pointer to the map.
4802
4803 The map is a vector of fixed-size entries. The size of each entry depends on
4804 the length of the longest name used. The first two bytes of each entry are the
4805 group number, most significant byte first. There follows the corresponding
4806 name, zero terminated. The names are in alphabetical order.
4807
4808 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4809 case instead of 255. Making it the same both with and without UTF-8 support
4810 means that the same test output works with both.
4811
4812 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4813 calling malloc() with a zero argument.
4814
4815 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4816 optimization. I've improved things by keeping a bitmap of backreferences with
4817 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4818 fact referenced, the optimization can be applied. It is unlikely that a
4819 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4820 the match to follow \n) will appear inside brackets with a number greater than
4821 31, but if it does, any back reference > 31 suppresses the optimization.
4822
4823 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4824 of disabling numbered capturing parentheses. Any opening parenthesis that is
4825 not followed by ? behaves as if it were followed by ?: but named parentheses
4826 can still be used for capturing (and they will acquire numbers in the usual
4827 way).
4828
4829 41. Redesigned the return codes from the match() function into yes/no/error so
4830 that errors can be passed back from deep inside the nested calls. A malloc
4831 failure while inside a recursive subpattern call now causes the
4832 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4833
4834 42. It is now possible to set a limit on the number of times the match()
4835 function is called in a call to pcre_exec(). This facility makes it possible to
4836 limit the amount of recursion and backtracking, though not in a directly
4837 obvious way, because the match() function is used in a number of different
4838 circumstances. The count starts from zero for each position in the subject
4839 string (for non-anchored patterns). The default limit is, for compatibility, a
4840 large number, namely 10 000 000. You can change this in two ways:
4841
4842 (a) When configuring PCRE before making, you can use --with-match-limit=n
4843 to set a default value for the compiled library.
4844
4845 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4846 a different value is set. See 45 below.
4847
4848 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4849
4850 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4851 of things that can be changed at compile time. The first argument specifies
4852 what is wanted and the second points to where the information is to be placed.
4853 The current list of available information is:
4854
4855 PCRE_CONFIG_UTF8
4856
4857 The output is an integer that is set to one if UTF-8 support is available;
4858 otherwise it is set to zero.
4859
4860 PCRE_CONFIG_NEWLINE
4861
4862 The output is an integer that it set to the value of the code that is used for
4863 newline. It is either LF (10) or CR (13).
4864
4865 PCRE_CONFIG_LINK_SIZE
4866
4867 The output is an integer that contains the number of bytes used for internal
4868 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4869
4870 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4871
4872 The output is an integer that contains the threshold above which the POSIX
4873 interface uses malloc() for output vectors. See item 31 above.
4874
4875 PCRE_CONFIG_MATCH_LIMIT
4876
4877 The output is an unsigned integer that contains the default limit of the number
4878 of match() calls in a pcre_exec() execution. See 42 above.
4879
4880 44. pcretest has been upgraded by the addition of the -C option. This causes it
4881 to extract all the available output from the new pcre_config() function, and to
4882 output it. The program then exits immediately.
4883
4884 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4885 order to support additional features. One way would have been to define
4886 pcre_exec2() (for example) with extra arguments, but this would not have been
4887 extensible, and would also have required all calls to the original function to
4888 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4889 is used for passing in "extra" data from pcre_study().
4890
4891 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4892 contains the following fields:
4893
4894 flags a bitmap indicating which of the following fields are set
4895 study_data opaque data from pcre_study()
4896 match_limit a way of specifying a limit on match() calls for a specific
4897 call to pcre_exec()
4898 callout_data data for callouts (see 49 below)
4899
4900 The flag bits are also defined in pcre.h, and are
4901
4902 PCRE_EXTRA_STUDY_DATA
4903 PCRE_EXTRA_MATCH_LIMIT
4904 PCRE_EXTRA_CALLOUT_DATA
4905
4906 The pcre_study() function now returns one of these new pcre_extra blocks, with
4907 the actual study data pointed to by the study_data field, and the
4908 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4909 before. That is, this change is entirely upwards-compatible and requires no
4910 change to existing code.
4911
4912 If you want to pass in additional data to pcre_exec(), you can either place it
4913 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4914 block.
4915
4916 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4917 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4918 times with different match limits, until it finds the minimum value needed for
4919 pcre_exec() to complete. The value is then output. This can be instructive; for
4920 most simple matches the number is quite small, but for pathological cases it
4921 gets very large very quickly.
4922
4923 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4924 returns the size of the data block pointed to by the study_data field in a
4925 pcre_extra block, that is, the value that was passed as the argument to
4926 pcre_malloc() when PCRE was getting memory in which to place the information
4927 created by pcre_study(). The fourth argument should point to a size_t variable.
4928 pcretest has been extended so that this information is shown after a successful
4929 pcre_study() call when information about the compiled regex is being displayed.
4930
4931 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4932 because what follows is always an absolute path. (Later: it turns out that this
4933 is more than cosmetic for MinGW, because it doesn't like empty path
4934 components.)
4935
4936 49. Some changes have been made to the callout feature (see 28 above):
4937
4938 (i) A callout function now has three choices for what it returns:
4939
4940 0 => success, carry on matching
4941 > 0 => failure at this point, but backtrack if possible
4942 < 0 => serious error, return this value from pcre_exec()
4943
4944 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4945 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4946 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4947 use by callout functions. It will never be used by PCRE itself.
4948
4949 (ii) The pcre_extra structure (see 45 above) has a void * field called
4950 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4951 pcre_callout_block structure has a field of the same name. The contents of
4952 the field passed in the pcre_extra structure are passed to the callout
4953 function in the corresponding field in the callout block. This makes it
4954 easier to use the same callout-containing regex from multiple threads. For
4955 testing, the pcretest program has a new data escape
4956
4957 \C*n pass the number n (may be negative) as callout_data
4958
4959 If the callout function in pcretest receives a non-zero value as
4960 callout_data, it returns that value.
4961
4962 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4963 there were some redundant $(CFLAGS) in commands that are now specified as
4964 $(LINK), which already includes $(CFLAGS).
4965
4966 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4967 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4968 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4969 one-byte characters throughout. Note that case-insensitive matching applies
4970 only to characters whose values are less than 256. PCRE doesn't support the
4971 notion of cases for higher-valued characters.
4972
4973 (i) A character class whose characters are all within 0-255 is handled as
4974 a bit map, and the map is inverted for negative classes. Previously, a
4975 character > 255 always failed to match such a class; however it should
4976 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4977
4978 (ii) A negated character class with a single character < 255 is coded as
4979 "not this character" (OP_NOT). This wasn't working properly when the test
4980 character was multibyte, either singly or repeated.
4981
4982 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4983 mode, for example: \x{100}{2,3}.
4984
4985 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4986 singly or repeated) now correctly test multibyte characters. However,
4987 PCRE doesn't recognize any characters with values greater than 255 as
4988 digits, spaces, or word characters. Such characters always match \D, \S,
4989 and \W, and never match \d, \s, or \w.
4990
4991 (v) Classes may now contain characters and character ranges with values
4992 greater than 255. For example: [ab\x{100}-\x{400}].
4993
4994 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4995 PCRE in UTF-8 mode.
4996
4997 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4998 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4999 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5000 value.)
5001
5002 53. The single man page has become too large. I have therefore split it up into
5003 a number of separate man pages. These also give rise to individual HTML pages;
5004 these are now put in a separate directory, and there is an index.html page that
5005 lists them all. Some hyperlinking between the pages has been installed.
5006
5007 54. Added convenience functions for handling named capturing parentheses.
5008
5009 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5010 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5011 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5012 are faulted.
5013
5014 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5015 calling configure. These values are used when compiling the dftables.c program
5016 which is run to generate the source of the default character tables. They
5017 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5018 you will need to set these values.
5019
5020 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5021
5022
5023 Version 3.9 02-Jan-02
5024 ---------------------
5025
5026 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5027
5028 2. If --disable-static was given, the building process failed when trying to
5029 build pcretest and pcregrep. (For some reason it was using libtool to compile
5030 them, which is not right, as they aren't part of the library.)
5031
5032
5033 Version 3.8 18-Dec-01
5034 ---------------------
5035
5036 1. The experimental UTF-8 code was completely screwed up. It was packing the
5037 bytes in the wrong order. How dumb can you get?
5038
5039
5040 Version 3.7 29-Oct-01
5041 ---------------------
5042
5043 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5044 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5045 this didn't happen under Solaris 8, where I normally test things.
5046
5047 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5048 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5049 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5050 this fix an hour or so after the initial 3.7 release.)
5051
5052
5053 Version 3.6 23-Oct-01
5054 ---------------------
5055
5056 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5057 offsets passed as NULL with zero offset count.
5058
5059 2. The config.guess and config.sub files had not been updated when I moved to
5060 the latest autoconf.
5061
5062
5063 Version 3.5 15-Aug-01
5064 ---------------------
5065
5066 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5067 had been forgotten.
5068
5069 2. By using declared but undefined structures, we can avoid using "void"
5070 definitions in pcre.h while keeping the internal definitions of the structures
5071 private.
5072
5073 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5074 user point of view, this means that both static and shared libraries are built
5075 by default, but this can be individually controlled. More of the work of
5076 handling this static/shared cases is now inside libtool instead of PCRE's make
5077 file.
5078
5079 4. The pcretest utility is now installed along with pcregrep because it is
5080 useful for users (to test regexs) and by doing this, it automatically gets
5081 relinked by libtool. The documentation has been turned into a man page, so
5082 there are now .1, .txt, and .html versions in /doc.
5083
5084 5. Upgrades to pcregrep:
5085 (i) Added long-form option names like gnu grep.
5086 (ii) Added --help to list all options with an explanatory phrase.
5087 (iii) Added -r, --recursive to recurse into sub-directories.
5088 (iv) Added -f, --file to read patterns from a file.
5089
5090 6. pcre_exec() was referring to its "code" argument before testing that
5091 argument for NULL (and giving an error if it was NULL).
5092
5093 7. Upgraded Makefile.in to allow for compiling in a different directory from
5094 the source directory.
5095
5096 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5097 options bits, the pointer it was passed was to an int instead of to an unsigned
5098 long int. This mattered only on 64-bit systems.
5099
5100 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5101 generated) instead of pcre.in, which it its source. Also made the same change
5102 in several of the .c files.
5103
5104 10. A new release of gcc defines printf() as a macro, which broke pcretest
5105 because it had an ifdef in the middle of a string argument for printf(). Fixed
5106 by using separate calls to printf().
5107
5108 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5109 script, to force use of CR or LF instead of \n in the source. On non-Unix
5110 systems, the value can be set in config.h.
5111
5112 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5113 absolute limit. Changed the text of the error message to make this clear, and
5114 likewise updated the man page.
5115
5116 13. The limit of 99 on the number of capturing subpatterns has been removed.
5117 The new limit is 65535, which I hope will not be a "real" limit.
5118
5119
5120 Version 3.4 22-Aug-00
5121 ---------------------
5122
5123 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5124
5125 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5126
5127
5128 Version 3.3 01-Aug-00
5129 ---------------------
5130
5131 1. If an octal character was given, but the value was greater than \377, it
5132 was not getting masked to the least significant bits, as documented. This could
5133 lead to crashes in some systems.
5134
5135 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5136 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5137
5138 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5139 These just pass their arguments on to (pcre_free)(), but they are provided
5140 because some uses of PCRE bind it to non-C systems that can call its functions,
5141 but cannot call free() or pcre_free() directly.
5142
5143 4. Add "make test" as a synonym for "make check". Corrected some comments in
5144 the Makefile.
5145
5146 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5147 Makefile.
5148
5149 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5150 command called pgrep for grepping around the active processes.
5151
5152 7. Added the beginnings of support for UTF-8 character strings.
5153
5154 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5155 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5156 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5157 out for the ar command.)
5158
5159
5160 Version 3.2 12-May-00
5161 ---------------------
5162
5163 This is purely a bug fixing release.
5164
5165 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5166 of ZA. This was just one example of several cases that could provoke this bug,
5167 which was introduced by change 9 of version 2.00. The code for breaking
5168 infinite loops after an iteration that matches an empty string was't working
5169 correctly.
5170
5171 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5172 when matched against abbab (for example). After matching an empty string, it
5173 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5174 caused it to match further down the string than it should.
5175
5176 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5177 was there because it doesn't seem to be needed, and it causes trouble on some
5178 systems, as it is not a Standard C header. It has been removed.
5179
5180 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5181 were reported on the Macintosh. The changes were from
5182
5183 while ((c = *(++ptr)) != 0 && c != '\n');
5184 to
5185 while ((c = *(++ptr)) != 0 && c != '\n') ;
5186
5187 Totally extraordinary, but if that's what it takes...
5188
5189 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5190 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5191 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5192 assumes the way PCRE uses memmove() (always moving upwards).
5193
5194 6. PCRE is being used in one environment where strchr() is not available. There
5195 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5196 faster code anyway.
5197
5198
5199 Version 3.1 09-Feb-00
5200 ---------------------
5201
5202 The only change in this release is the fixing of some bugs in Makefile.in for
5203 the "install" target:
5204
5205 (1) It was failing to install pcreposix.h.
5206
5207 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5208
5209
5210 Version 3.0 01-Feb-00
5211 ---------------------
5212
5213 1. Add support for the /+ modifier to perltest (to output $` like it does in
5214 pcretest).
5215
5216 2. Add support for the /g modifier to perltest.
5217
5218 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5219 matches null strings.
5220
5221 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5222 pattern. Perl treats empty patterns specially - it reuses the most recent
5223 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5224 effect.
5225
5226 5. The POSIX interface was broken in that it was just handing over the POSIX
5227 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5228 required a bigger vector, with some working space on the end. This means that
5229 the POSIX wrapper now has to get and free some memory, and copy the results.
5230
5231 6. Added some simple autoconf support, placing the test data and the
5232 documentation in separate directories, re-organizing some of the
5233 information files, and making it build pcre-config (a GNU standard). Also added
5234 libtool support for building PCRE as a shared library, which is now the
5235 default.
5236
5237 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5238 09 are not valid octal constants. Single digits will be used for minor values
5239 less than 10.
5240
5241 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5242 existing programs that set these in the POSIX interface can use PCRE without
5243 modification.
5244
5245 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5246 return all that pcre_info() returns, plus additional data. The pcre_info()
5247 function is retained for compatibility, but is considered to be obsolete.
5248
5249 10. Added experimental recursion feature (?R) to handle one common case that
5250 Perl 5.6 will be able to do with (?p{...}).
5251
5252 11. Added support for POSIX character classes like [:alpha:], which Perl is
5253 adopting.
5254
5255
5256 Version 2.08 31-Aug-99
5257 ----------------------
5258
5259 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5260 trying to match at the startoffset position, but instead was moving forward to
5261 the next newline as if a previous match had failed.
5262
5263 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5264 and could get into a loop if a null string was matched other than at the start
5265 of the subject.
5266
5267 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5268 be distinguished at compile time, and for completeness also added PCRE_DATE.
5269
5270 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5271 in GnuWin32 environments.
5272
5273
5274 Version 2.07 29-Jul-99
5275 ----------------------
5276
5277 1. The documentation is now supplied in plain text form and HTML as well as in
5278 the form of man page sources.
5279
5280 2. C++ compilers don't like assigning (void *) values to other pointer types.
5281 In particular this affects malloc(). Although there is no problem in Standard
5282 C, I've put in casts to keep C++ compilers happy.
5283
5284 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5285 should be (const char *).
5286
5287 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5288 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5289 However, I haven't made this a standard facility. The documentation doesn't
5290 mention it, and the Makefile doesn't support it.
5291
5292 5. The Makefile now contains an "install" target, with editable destinations at
5293 the top of the file. The pcretest program is not installed.
5294
5295 6. pgrep -V now gives the PCRE version number and date.
5296
5297 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5298 causing the entire string to be ignored, instead of just the last character.
5299
5300 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5301 non-matching string, it can take a very, very long time, even for strings of
5302 quite modest length, because of the nested recursion. PCRE now does better in
5303 some of these cases. It does this by remembering the last required literal
5304 character in the pattern, and pre-searching the subject to ensure it is present
5305 before running the real match. In other words, it applies a heuristic to detect
5306 some types of certain failure quickly, and in the above example, if presented
5307 with a string that has no trailing " it gives "no match" very quickly.
5308
5309 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5310 other alternatives are tried instead.
5311
5312
5313 Version 2.06 09-Jun-99
5314 ----------------------
5315
5316 1. Change pcretest's output for amount of store used to show just the code
5317 space, because the remainder (the data block) varies in size between 32-bit and
5318 64-bit systems.
5319
5320 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5321 start matching at. This allows lookbehinds to work when searching for multiple
5322 occurrences in a string.
5323
5324 3. Added additional options to pcretest for testing multiple occurrences:
5325
5326 /+ outputs the rest of the string that follows a match
5327 /g loops for multiple occurrences, using the new startoffset argument
5328 /G loops for multiple occurrences by passing an incremented pointer
5329
5330 4. PCRE wasn't doing the "first character" optimization for patterns starting
5331 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5332 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5333 the letter 'x'. On long subject strings, this gives a significant speed-up.
5334
5335
5336 Version 2.05 21-Apr-99
5337 ----------------------
5338
5339 1. Changed the type of magic_number from int to long int so that it works
5340 properly on 16-bit systems.
5341
5342 2. Fixed a bug which caused patterns starting with .* not to work correctly
5343 when the subject string contained newline characters. PCRE was assuming
5344 anchoring for such patterns in all cases, which is not correct because .* will
5345 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5346 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5347 must be retried after every newline in the subject.
5348
5349
5350 Version 2.04 18-Feb-99
5351 ----------------------
5352
5353 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5354 computation of the store needed to hold the pattern was incorrect (too large).
5355 If such patterns were nested a few deep, this could multiply and become a real
5356 problem.
5357
5358 2. Added /M option to pcretest to show the memory requirement of a specific
5359 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5360
5361 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5362 compiled in such a way that the backtracking after subsequent failure was
5363 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5364 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5365
5366
5367 Version 2.03 02-Feb-99
5368 ----------------------
5369
5370 1. Fixed typo and small mistake in man page.
5371
5372 2. Added 4th condition (GPL supersedes if conflict) and created separate
5373 LICENCE file containing the conditions.
5374
5375 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5376 Perl, that is the internal \ allows the delimiter to be included in the
5377 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5378 the final delimiter, add \ to the end of the pattern (to test the error).
5379
5380 4. Added the convenience functions for extracting substrings after a successful
5381 match. Updated pcretest to make it able to test these functions.
5382
5383
5384 Version 2.02 14-Jan-99
5385 ----------------------
5386
5387 1. Initialized the working variables associated with each extraction so that
5388 their saving and restoring doesn't refer to uninitialized store.
5389
5390 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5391 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5392 fix the problem.
5393
5394 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5395 calls, and wasn't printing the correct value for compiling calls. Increased the
5396 default value of LOOPREPEAT, and the number of significant figures in the
5397 times.
5398
5399 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5400
5401 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5402 a building problem on Windows NT with a FAT file system.
5403
5404
5405 Version 2.01 21-Oct-98
5406 ----------------------
5407
5408 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5409 to character tables built by pcre_maketables() in the current locale. If NULL
5410 is passed, the default tables are used.
5411
5412
5413 Version 2.00 24-Sep-98
5414 ----------------------
5415
5416 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5417 it any more.
5418
5419 2. Allow quantification of (?>) groups, and make it work correctly.
5420
5421 3. The first character computation wasn't working for (?>) groups.
5422
5423 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5424 end of the subject) and add 5.005's \z, which really does match only at the
5425 very end of the subject.
5426
5427 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5428
5429 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5430 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5431 localized options. All options to pcre_study() were also removed.
5432
5433 7. Add other new features from 5.005:
5434
5435 $(?<= positive lookbehind
5436 $(?<! negative lookbehind
5437 (?imsx-imsx) added the unsetting capability
5438 such a setting is global if at outer level; local otherwise
5439 (?imsx-imsx:) non-capturing groups with option setting
5440 (?(cond)re|re) conditional pattern matching
5441
5442 A backreference to itself in a repeated group matches the previous
5443 captured string.
5444
5445 8. General tidying up of studying (both automatic and via "study")
5446 consequential on the addition of new assertions.
5447
5448 9. As in 5.005, unlimited repeated groups that could match an empty substring
5449 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5450 runtime if any iteration does actually match an empty substring.
5451
5452 10. Include the RunTest script in the distribution.
5453
5454 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5455 discrepancies, some of which were old and were also with respect to 5.004. They
5456 have now been fixed.
5457
5458
5459 Version 1.09 28-Apr-98
5460 ----------------------
5461
5462 1. A negated single character class followed by a quantifier with a minimum
5463 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5464 program crashes, or just wrong answers. This did not apply to negated classes
5465 containing more than one character, or to minima other than one.
5466
5467
5468 Version 1.08 27-Mar-98
5469 ----------------------
5470
5471 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5472
5473 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5474 latter must appear before anything that relies on it in the pattern.
5475
5476
5477 Version 1.07 16-Feb-98
5478 ----------------------
5479
5480 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5481 repeat of a potentially empty string).
5482
5483
5484 Version 1.06 23-Jan-98
5485 ----------------------
5486
5487 1. Added Markus Oberhumer's little patches for C++.
5488
5489 2. Literal strings longer than 255 characters were broken.
5490
5491
5492 Version 1.05 23-Dec-97
5493 ----------------------
5494
5495 1. Negated character classes containing more than one character were failing if
5496 PCRE_CASELESS was set at run time.
5497
5498
5499 Version 1.04 19-Dec-97
5500 ----------------------
5501
5502 1. Corrected the man page, where some "const" qualifiers had been omitted.
5503
5504 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5505 input syntax.
5506
5507 3. Fixed memory leak which occurred when a regex with back references was
5508 matched with an offsets vector that wasn't big enough. The temporary memory
5509 that is used in this case wasn't being freed if the match failed.
5510
5511 4. Tidied pcretest to ensure it frees memory that it gets.
5512
5513 5. Temporary memory was being obtained in the case where the passed offsets
5514 vector was exactly big enough.
5515
5516 6. Corrected definition of offsetof() from change 5 below.
5517
5518 7. I had screwed up change 6 below and broken the rules for the use of
5519 setjmp(). Now fixed.
5520
5521
5522 Version 1.03 18-Dec-97
5523 ----------------------
5524
5525 1. A erroneous regex with a missing opening parenthesis was correctly
5526 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5527 on some systems.
5528
5529 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5530 it was reported that one broken compiler failed on the former because "code" is
5531 also an independent variable.
5532
5533 3. The erroneous regex a[]b caused an array overrun reference.
5534
5535 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5536 fail on data ending with that character. (It was going on too far, and checking
5537 the next character, typically a binary zero.) This was specific to the
5538 optimized code for single-character negative classes.
5539
5540 5. Added a contributed patch from the TIN world which does the following:
5541
5542 + Add an undef for memmove, in case the the system defines a macro for it.
5543
5544 + Add a definition of offsetof(), in case there isn't one. (I don't know
5545 the reason behind this - offsetof() is part of the ANSI standard - but
5546 it does no harm).
5547
5548 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5549 most of the places where whitespace preceded '#'. I have given up and
5550 allowed the remaining 2 cases to be at the margin.
5551
5552 + Rename some variables in pcre to eliminate shadowing. This seems very
5553 pedantic, but does no harm, of course.
5554
5555 6. Moved the call to setjmp() into its own function, to get rid of warnings
5556 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5557
5558 7. Constructs such as \d{8,} were compiling into the equivalent of
5559 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5560 outcome, but in this particular case used more store than had been allocated,
5561 which caused the bug to be discovered because it threw up an internal error.
5562
5563 8. The debugging code in both pcre and pcretest for outputting the compiled
5564 form of a regex was going wrong in the case of back references followed by
5565 curly-bracketed repeats.
5566
5567
5568 Version 1.02 12-Dec-97
5569 ----------------------
5570
5571 1. Typos in pcre.3 and comments in the source fixed.
5572
5573 2. Applied a contributed patch to get rid of places where it used to remove
5574 'const' from variables, and fixed some signed/unsigned and uninitialized
5575 variable warnings.
5576
5577 3. Added the "runtest" target to Makefile.
5578
5579 4. Set default compiler flag to -O2 rather than just -O.
5580
5581
5582 Version 1.01 19-Nov-97
5583 ----------------------
5584
5585 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5586 like /([ab]*)*/, that is, for classes with more than one character in them.
5587
5588 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5589 as /((?>a*))*/ (a PCRE_EXTRA facility).
5590
5591
5592 Version 1.00 18-Nov-97
5593 ----------------------
5594
5595 1. Added compile-time macros to support systems such as SunOS4 which don't have
5596 memmove() or strerror() but have other things that can be used instead.
5597
5598 2. Arranged that "make clean" removes the executables.
5599
5600
5601 Version 0.99 27-Oct-97
5602 ----------------------
5603
5604 1. Fixed bug in code for optimizing classes with only one character. It was
5605 initializing a 32-byte map regardless, which could cause it to run off the end
5606 of the memory it had got.
5607
5608 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5609
5610
5611 Version 0.98 22-Oct-97
5612 ----------------------
5613
5614 1. Fixed bug in code for handling temporary memory usage when there are more
5615 back references than supplied space in the ovector. This could cause segfaults.
5616
5617
5618 Version 0.97 21-Oct-97
5619 ----------------------
5620
5621 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5622
5623 2. Optimized negated single characters not to use a bit map.
5624
5625 3. Brought error texts together as macro definitions; clarified some of them;
5626 fixed one that was wrong - it said "range out of order" when it meant "invalid
5627 escape sequence".
5628
5629 4. Changed some char * arguments to const char *.
5630
5631 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5632
5633 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5634 pcretest.
5635
5636
5637 Version 0.96 16-Oct-97
5638 ----------------------
5639
5640 1. Added a simple "pgrep" utility to the distribution.
5641
5642 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5643 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5644 where "ddd" means "one or more decimal digits".
5645
5646 3. Fixed serious bug. If a pattern had a back reference, but the call to
5647 pcre_exec() didn't supply a large enough ovector to record the related
5648 identifying subpattern, the match always failed. PCRE now remembers the number
5649 of the largest back reference, and gets some temporary memory in which to save
5650 the offsets during matching if necessary, in order to ensure that
5651 backreferences always work.
5652
5653 4. Increased the compatibility with Perl in a number of ways:
5654
5655 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5656 to request this handling. The option can be set at compile or exec time.
5657
5658 (b) $ matches before a terminating newline by default; an option
5659 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5660 mode). The option can be set at compile or exec time.
5661
5662 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5663 the same as Perl's. If the decimal number it represents is less than 10
5664 or there aren't that many previous left capturing parentheses, an octal
5665 escape is read. Inside a character class, it's always an octal escape,
5666 even if it is a single digit.
5667
5668 (d) An escaped but undefined alphabetic character is taken as a literal,
5669 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5670 escapes.
5671
5672 (e) {0} is now permitted. (The previous item is removed from the compiled
5673 pattern).
5674
5675 5. Changed all the names of code files so that the basic parts are no longer
5676 than 10 characters, and abolished the teeny "globals.c" file.
5677
5678 6. Changed the handling of character classes; they are now done with a 32-byte
5679 bit map always.
5680
5681 7. Added the -d and /D options to pcretest to make it possible to look at the
5682 internals of compilation without having to recompile pcre.
5683
5684
5685 Version 0.95 23-Sep-97
5686 ----------------------
5687
5688 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5689 \x20 at the start of a run of normal characters. These were being treated as
5690 real characters, instead of the source characters being re-checked.
5691
5692
5693 Version 0.94 18-Sep-97
5694 ----------------------
5695
5696 1. The functions are now thread-safe, with the caveat that the global variables
5697 containing pointers to malloc() and free() or alternative functions are the
5698 same for all threads.
5699
5700 2. Get pcre_study() to generate a bitmap of initial characters for non-
5701 anchored patterns when this is possible, and use it if passed to pcre_exec().
5702
5703
5704 Version 0.93 15-Sep-97
5705 ----------------------
5706
5707 1. /(b)|(:+)/ was computing an incorrect first character.
5708
5709 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5710 but not actually doing anything yet.
5711
5712 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5713 as Perl does (e.g. [-az] or [az-]).
5714
5715 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5716 all possible positions.
5717
5718 5. Split up into different modules to avoid including unneeded functions in a
5719 compiled binary. However, compile and exec are still in one module. The "study"
5720 function is split off.
5721
5722 6. The character tables are now in a separate module whose source is generated
5723 by an auxiliary program - but can then be edited by hand if required. There are
5724 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5725 toupper() in the code.
5726
5727 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5728 make them global. Abolish the function for setting them, as the caller can now
5729 set them directly.
5730
5731
5732 Version 0.92 11-Sep-97
5733 ----------------------
5734
5735 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5736 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5737
5738 2. Caseless matching was not working in character classes if the characters in
5739 the pattern were in upper case.
5740
5741 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5742
5743 4. Make PCRE_ANCHORED public and accept as a compile option.
5744
5745 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5746 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5747 pass them.
5748
5749 6. Give an error if bad option bits passed at compile or run time.
5750
5751 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5752 pcretest to cause it to pass that flag.
5753
5754 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5755 options, and the first character, if set.
5756
5757 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5758
5759
5760 Version 0.91 10-Sep-97
5761 ----------------------
5762
5763 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5764 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5765
5766 2. PCRE was looping on encountering an indefinitely repeated back reference to
5767 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5768 Perl does - treats the match as successful.
5769
5770 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5