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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1560 - (show annotations)
Tue May 19 16:02:06 2015 UTC (8 days, 22 hours ago) by ph10
File size: 278759 byte(s)
Fix buffer overflow for lookbehind within mutually recursive subroutines.
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 5. 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 6. A lookbehind assertion within a set of mutually recursive subpatterns could
32 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
33
34
35 Version 8.37 28-April-2015
36 --------------------------
37
38 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
39 for those parentheses to be closed with whatever has been captured so far.
40 However, it was failing to mark any other groups between the hightest
41 capture so far and the currrent group as "unset". Thus, the ovector for
42 those groups contained whatever was previously there. An example is the
43 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
44
45 2. If an assertion condition was quantified with a minimum of zero (an odd
46 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
47
48 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
49 unrecognized modifier, a crash could occur.
50
51 4. An attempt to do global matching in pcretest with a zero-length ovector
52 caused a crash.
53
54 5. Fixed a memory leak during matching that could occur for a subpattern
55 subroutine call (recursive or otherwise) if the number of captured groups
56 that had to be saved was greater than ten.
57
58 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
59 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
60 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
61
62 7. A UTF pattern containing a "not" match of a non-ASCII character and a
63 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
64
65 8. When a pattern is compiled, it remembers the highest back reference so that
66 when matching, if the ovector is too small, extra memory can be obtained to
67 use instead. A conditional subpattern whose condition is a check on a
68 capture having happened, such as, for example in the pattern
69 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
70 setting the highest backreference number. This mattered only if pcre_exec()
71 was called with an ovector that was too small to hold the capture, and there
72 was no other kind of back reference (a situation which is probably quite
73 rare). The effect of the bug was that the condition was always treated as
74 FALSE when the capture could not be consulted, leading to a incorrect
75 behaviour by pcre_exec(). This bug has been fixed.
76
77 9. A reference to a duplicated named group (either a back reference or a test
78 for being set in a conditional) that occurred in a part of the pattern where
79 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
80 to be incorrectly calculated, leading to overwriting.
81
82 10. A mutually recursive set of back references such as (\2)(\1) caused a
83 segfault at study time (while trying to find the minimum matching length).
84 The infinite loop is now broken (with the minimum length unset, that is,
85 zero).
86
87 11. If an assertion that was used as a condition was quantified with a minimum
88 of zero, matching went wrong. In particular, if the whole group had
89 unlimited repetition and could match an empty string, a segfault was
90 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
91 assertions to be quantified, but not if they are being used as conditions,
92 so the above pattern is faulted by Perl. PCRE has now been changed so that
93 it also rejects such patterns.
94
95 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
96 failed to allow the zero-repeat case if pcre2_exec() was called with an
97 ovector too small to capture the group.
98
99 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
100 Red Hat Product Security:
101
102 (a) A crash if /K and /F were both set with the option to save the compiled
103 pattern.
104
105 (b) Another crash if the option to print captured substrings in a callout
106 was combined with setting a null ovector, for example \O\C+ as a subject
107 string.
108
109 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
110 forward reference repeated a large (but limited) number of times within a
111 repeated outer group that has a zero minimum quantifier, caused incorrect
112 code to be compiled, leading to the error "internal error:
113 previously-checked referenced subpattern not found" when an incorrect
114 memory address was read. This bug was reported as "heap overflow",
115 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
116 CVE-2015-2325.
117
118 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
119 call within a group that also contained a recursive back reference caused
120 incorrect code to be compiled. This bug was reported as "heap overflow",
121 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
122 number CVE-2015-2326.
123
124 24. Computing the size of the JIT read-only data in advance has been a source
125 of various issues, and new ones are still appear unfortunately. To fix
126 existing and future issues, size computation is eliminated from the code,
127 and replaced by on-demand memory allocation.
128
129 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
130 adjacent to the end of the range, and the range contained characters with
131 more than one other case, caused incorrect behaviour when compiled in UTF
132 mode. In that example, the range a-j was left out of the class.
133
134 26. Fix JIT compilation of conditional blocks, which assertion
135 is converted to (*FAIL). E.g: /(?(?!))/.
136
137 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
138 discovered by the LLVM fuzzer.
139
140 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
141 when this assertion was used as a condition, for example (?(?!)a|b). In
142 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
143 error about an unsupported item.
144
145 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
146 possessification code could take exponential time to complete. A recursion
147 depth limit of 1000 has been imposed to limit the resources used by this
148 optimization.
149
150 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
151 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
152 because \S ensures they are all in the class. The code for doing this was
153 interacting badly with the code for computing the amount of space needed to
154 compile the pattern, leading to a buffer overflow. This bug was discovered
155 by the LLVM fuzzer.
156
157 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
158 other kinds of group caused stack overflow at compile time. This bug was
159 discovered by the LLVM fuzzer.
160
161 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
162 between a subroutine call and its quantifier was incorrectly compiled,
163 leading to buffer overflow or other errors. This bug was discovered by the
164 LLVM fuzzer.
165
166 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
167 assertion after (?(. The code was failing to check the character after
168 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
169 was discovered by the LLVM fuzzer.
170
171 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
172 a fixed maximum following a group that contains a subroutine reference was
173 incorrectly compiled and could trigger buffer overflow. This bug was
174 discovered by the LLVM fuzzer.
175
176 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
177 caused a stack overflow instead of the diagnosis of a non-fixed length
178 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
179
180 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
181 (e.g. /(?<=\Ka)/) could make pcregrep loop.
182
183 37. There was a similar problem to 36 in pcretest for global matches.
184
185 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
186 and a subsequent item in the pattern caused a non-match, backtracking over
187 the repeated \X did not stop, but carried on past the start of the subject,
188 causing reference to random memory and/or a segfault. There were also some
189 other cases where backtracking after \C could crash. This set of bugs was
190 discovered by the LLVM fuzzer.
191
192 39. The function for finding the minimum length of a matching string could take
193 a very long time if mutual recursion was present many times in a pattern,
194 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
195 method has been implemented. This infelicity was discovered by the LLVM
196 fuzzer.
197
198 40. Static linking against the PCRE library using the pkg-config module was
199 failing on missing pthread symbols.
200
201
202 Version 8.36 26-September-2014
203 ------------------------------
204
205 1. Got rid of some compiler warnings in the C++ modules that were shown up by
206 -Wmissing-field-initializers and -Wunused-parameter.
207
208 2. The tests for quantifiers being too big (greater than 65535) were being
209 applied after reading the number, and stupidly assuming that integer
210 overflow would give a negative number. The tests are now applied as the
211 numbers are read.
212
213 3. Tidy code in pcre_exec.c where two branches that used to be different are
214 now the same.
215
216 4. The JIT compiler did not generate match limit checks for certain
217 bracketed expressions with quantifiers. This may lead to exponential
218 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
219 issue should be resolved now.
220
221 5. Fixed an issue, which occures when nested alternatives are optimized
222 with table jumps.
223
224 6. Inserted two casts and changed some ints to size_t in the light of some
225 reported 64-bit compiler warnings (Bugzilla 1477).
226
227 7. Fixed a bug concerned with zero-minimum possessive groups that could match
228 an empty string, which sometimes were behaving incorrectly in the
229 interpreter (though correctly in the JIT matcher). This pcretest input is
230 an example:
231
232 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
233 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
234
235 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
236 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
237 for an empty string was breaking the inner loop and carrying on at a lower
238 level, when possessive repeated groups should always return to a higher
239 level as they have no backtrack points in them. The empty string test now
240 occurs at the outer level.
241
242 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
243 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
244
245 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
246 doesn't).
247
248 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
249 to hold a pattern to be incorrectly computed (too small) when there were
250 named back references to duplicated names. This could cause "internal
251 error: code overflow" or "double free or corruption" or other memory
252 handling errors.
253
254 11. When named subpatterns had the same prefixes, back references could be
255 confused. For example, in this pattern:
256
257 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
258
259 the reference to 'Name' was incorrectly treated as a reference to a
260 duplicate name.
261
262 12. A pattern such as /^s?c/mi8 where the optional character has more than
263 one "other case" was incorrectly compiled such that it would only try to
264 match starting at "c".
265
266 13. When a pattern starting with \s was studied, VT was not included in the
267 list of possible starting characters; this should have been part of the
268 8.34/18 patch.
269
270 14. If a character class started [\Qx]... where x is any character, the class
271 was incorrectly terminated at the ].
272
273 15. If a pattern that started with a caseless match for a character with more
274 than one "other case" was studied, PCRE did not set up the starting code
275 unit bit map for the list of possible characters. Now it does. This is an
276 optimization improvement, not a bug fix.
277
278 16. The Unicode data tables have been updated to Unicode 7.0.0.
279
280 17. Fixed a number of memory leaks in pcregrep.
281
282 18. Avoid a compiler warning (from some compilers) for a function call with
283 a cast that removes "const" from an lvalue by using an intermediate
284 variable (to which the compiler does not object).
285
286 19. Incorrect code was compiled if a group that contained an internal recursive
287 back reference was optional (had quantifier with a minimum of zero). This
288 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
289 caused segmentation faults because of stack overflows at compile time.
290
291 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
292 group that is quantified with an indefinite repeat, caused a compile-time
293 loop which used up all the system stack and provoked a segmentation fault.
294 This was not the same bug as 19 above.
295
296 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
297 Patch by Mike Frysinger.
298
299
300 Version 8.35 04-April-2014
301 --------------------------
302
303 1. A new flag is set, when property checks are present in an XCLASS.
304 When this flag is not set, PCRE can perform certain optimizations
305 such as studying these XCLASS-es.
306
307 2. The auto-possessification of character sets were improved: a normal
308 and an extended character set can be compared now. Furthermore
309 the JIT compiler optimizes more character set checks.
310
311 3. Got rid of some compiler warnings for potentially uninitialized variables
312 that show up only when compiled with -O2.
313
314 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
315 of a match later then the end of the match. The pcretest program was not
316 handling the case sensibly - it was outputting from the start to the next
317 binary zero. It now reports this situation in a message, and outputs the
318 text from the end to the start.
319
320 5. Fast forward search is improved in JIT. Instead of the first three
321 characters, any three characters with fixed position can be searched.
322 Search order: first, last, middle.
323
324 6. Improve character range checks in JIT. Characters are read by an inprecise
325 function now, which returns with an unknown value if the character code is
326 above a certain threshold (e.g: 256). The only limitation is that the value
327 must be bigger than the threshold as well. This function is useful when
328 the characters above the threshold are handled in the same way.
329
330 7. The macros whose names start with RAWUCHAR are placeholders for a future
331 mode in which only the bottom 21 bits of 32-bit data items are used. To
332 make this more memorable for those maintaining the code, the names have
333 been changed to start with UCHAR21, and an extensive comment has been added
334 to their definition.
335
336 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
337 to the export list in Makefile.am (they were accidentally omitted from the
338 8.34 tarball).
339
340 9. The informational output from pcretest used the phrase "starting byte set"
341 which is inappropriate for the 16-bit and 32-bit libraries. As the output
342 for "first char" and "need char" really means "non-UTF-char", I've changed
343 "byte" to "char", and slightly reworded the output. The documentation about
344 these values has also been (I hope) clarified.
345
346 10. Another JIT related optimization: use table jumps for selecting the correct
347 backtracking path, when more than four alternatives are present inside a
348 bracket.
349
350 11. Empty match is not possible, when the minimum length is greater than zero,
351 and there is no \K in the pattern. JIT should avoid empty match checks in
352 such cases.
353
354 12. In a caseless character class with UCP support, when a character with more
355 than one alternative case was not the first character of a range, not all
356 the alternative cases were added to the class. For example, s and \x{17f}
357 are both alternative cases for S: the class [RST] was handled correctly,
358 but [R-T] was not.
359
360 13. The configure.ac file always checked for pthread support when JIT was
361 enabled. This is not used in Windows, so I have put this test inside a
362 check for the presence of windows.h (which was already tested for).
363
364 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
365 The algorithm provides a way to skip certain starting offsets, and usually
366 faster than linear prefix searches.
367
368 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
369 as for 'fr_FR' and 'french'. For some reason, however, it then used the
370 Windows-specific input and output files, which have 'french' screwed in.
371 So this could never have worked. One of the problems with locales is that
372 they aren't always the same. I have now updated RunTest so that it checks
373 the output of the locale test (test 3) against three different output
374 files, and it allows the test to pass if any one of them matches. With luck
375 this should make the test pass on some versions of Solaris where it was
376 failing. Because of the uncertainty, the script did not used to stop if
377 test 3 failed; it now does. If further versions of a French locale ever
378 come to light, they can now easily be added.
379
380 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
381 there was a message during ./configure, but it did not stop. This now
382 provokes an error. The invalid example in README has been corrected.
383 If a value less than the minimum is given, the minimum value has always
384 been used, but now a warning is given.
385
386 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
387 was a bug in the test system, which is now fixed. Also, the list of various
388 configurations that are tested for each release did not have one with both
389 16/32 bits and --enable-bar-anycrlf. It now does.
390
391 18. pcretest was missing "-C bsr" for displaying the \R default setting.
392
393 19. Little endian PowerPC systems are supported now by the JIT compiler.
394
395 20. The fast forward newline mechanism could enter to an infinite loop on
396 certain invalid UTF-8 input. Although we don't support these cases
397 this issue can be fixed by a performance optimization.
398
399 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
400 not take account if existing stack usage. There is now a new global
401 variable called pcre_stack_guard that can be set to point to an external
402 function to check stack availability. It is called at the start of
403 processing every parenthesized group.
404
405 22. A typo in the code meant that in ungreedy mode the max/min qualifier
406 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
407 not match "ab".
408
409 23. When UTF was disabled, the JIT program reported some incorrect compile
410 errors. These messages are silenced now.
411
412 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
413 compiler.
414
415 25. Change all the temporary files used in RunGrepTest to be different to those
416 used by RunTest so that the tests can be run simultaneously, for example by
417 "make -j check".
418
419
420 Version 8.34 15-December-2013
421 -----------------------------
422
423 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
424 executable memory. Patch inspired by Carsten Klein.
425
426 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
427 this macro is never tested and has no effect, because the work to support
428 coverage involves only compiling and linking options and special targets in
429 the Makefile. The comment in config.h implied that defining the macro would
430 enable coverage support, which is totally false. There was also support for
431 setting this macro in the CMake files (my fault, I just copied it from
432 configure). SUPPORT_GCOV has now been removed.
433
434 3. Make a small performance improvement in strlen16() and strlen32() in
435 pcretest.
436
437 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
438 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
439 these cases). There was also one in pcretest.c.
440
441 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
442
443 6. In UTF mode, the code for checking whether a group could match an empty
444 string (which is used for indefinitely repeated groups to allow for
445 breaking an infinite loop) was broken when the group contained a repeated
446 negated single-character class with a character that occupied more than one
447 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
448 UTF-8 mode). The effect was undefined: the group might or might not be
449 deemed as matching an empty string, or the program might have crashed.
450
451 7. The code for checking whether a group could match an empty string was not
452 recognizing that \h, \H, \v, \V, and \R must match a character.
453
454 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
455 an empty string. If it can, pcretest shows this in its information output.
456
457 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
458 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
459 matched by pcre_exec() without using JIT:
460
461 (a) If the rest of the pattern did not match after a maximal run of
462 grapheme clusters, the code for backing up to try with fewer of them
463 did not always back up over a full grapheme when characters that do not
464 have the modifier quality were involved, e.g. Hangul syllables.
465
466 (b) If the match point in a subject started with modifier character, and
467 there was no match, the code could incorrectly back up beyond the match
468 point, and potentially beyond the first character in the subject,
469 leading to a segfault or an incorrect match result.
470
471 10. A conditional group with an assertion condition could lead to PCRE
472 recording an incorrect first data item for a match if no other first data
473 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
474 first data item, and therefore matched "ca" after "c" instead of at the
475 start.
476
477 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
478 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
479
480 12. The source of pcregrep now includes z/OS-specific code so that it can be
481 compiled for z/OS as part of the special z/OS distribution.
482
483 13. Added the -T and -TM options to pcretest.
484
485 14. The code in pcre_compile.c for creating the table of named capturing groups
486 has been refactored. Instead of creating the table dynamically during the
487 actual compiling pass, the information is remembered during the pre-compile
488 pass (on the stack unless there are more than 20 named groups, in which
489 case malloc() is used) and the whole table is created before the actual
490 compile happens. This has simplified the code (it is now nearly 150 lines
491 shorter) and prepared the way for better handling of references to groups
492 with duplicate names.
493
494 15. A back reference to a named subpattern when there is more than one of the
495 same name now checks them in the order in which they appear in the pattern.
496 The first one that is set is used for the reference. Previously only the
497 first one was inspected. This change makes PCRE more compatible with Perl.
498
499 16. Unicode character properties were updated from Unicode 6.3.0.
500
501 17. The compile-time code for auto-possessification has been refactored, based
502 on a patch by Zoltan Herczeg. It now happens after instead of during
503 compilation. The code is cleaner, and more cases are handled. The option
504 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
505 options in pcretest are provided to set it. It can also be set by
506 (*NO_AUTO_POSSESS) at the start of a pattern.
507
508 18. The character VT has been added to the default ("C" locale) set of
509 characters that match \s and are generally treated as white space,
510 following this same change in Perl 5.18. There is now no difference between
511 "Perl space" and "POSIX space". Whether VT is treated as white space in
512 other locales depends on the locale.
513
514 19. The code for checking named groups as conditions, either for being set or
515 for being recursed, has been refactored (this is related to 14 and 15
516 above). Processing unduplicated named groups should now be as fast at
517 numerical groups, and processing duplicated groups should be faster than
518 before.
519
520 20. Two patches to the CMake build system, by Alexander Barkov:
521
522 (1) Replace the "source" command by "." in CMakeLists.txt because
523 "source" is a bash-ism.
524
525 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
526 without these the CMake build does not work on Solaris.
527
528 21. Perl has changed its handling of \8 and \9. If there is no previously
529 encountered capturing group of those numbers, they are treated as the
530 literal characters 8 and 9 instead of a binary zero followed by the
531 literals. PCRE now does the same.
532
533 22. Following Perl, added \o{} to specify codepoints in octal, making it
534 possible to specify values greater than 0777 and also making them
535 unambiguous.
536
537 23. Perl now gives an error for missing closing braces after \x{... instead of
538 treating the string as literal. PCRE now does the same.
539
540 24. RunTest used to grumble if an inappropriate test was selected explicitly,
541 but just skip it when running all tests. This make it awkward to run ranges
542 of tests when one of them was inappropriate. Now it just skips any
543 inappropriate tests, as it always did when running all tests.
544
545 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
546 character types such as \d or \w, too many callouts were inserted, and the
547 data that they returned was rubbish.
548
549 26. In UCP mode, \s was not matching two of the characters that Perl matches,
550 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
551 were matched by \h. The code has now been refactored so that the lists of
552 the horizontal and vertical whitespace characters used for \h and \v (which
553 are defined only in one place) are now also used for \s.
554
555 27. Add JIT support for the 64 bit TileGX architecture.
556 Patch by Jiong Wang (Tilera Corporation).
557
558 28. Possessive quantifiers for classes (both explicit and automatically
559 generated) now use special opcodes instead of wrapping in ONCE brackets.
560
561 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
562 (because it's meaningless), this was not happening when PCRE_CASELESS was
563 set. Not wrong, but inefficient.
564
565 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
566 properties for \w, \d, etc) is present in a test regex. Otherwise if the
567 test contains no characters greater than 255, Perl doesn't realise it
568 should be using Unicode semantics.
569
570 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
571 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
572 does in Unicode mode.
573
574 32. Added the "forbid" facility to pcretest so that putting tests into the
575 wrong test files can sometimes be quickly detected.
576
577 33. There is now a limit (default 250) on the depth of nesting of parentheses.
578 This limit is imposed to control the amount of system stack used at compile
579 time. It can be changed at build time by --with-parens-nest-limit=xxx or
580 the equivalent in CMake.
581
582 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
583 errors. Perl warns for these when in warning mode, but PCRE has no facility
584 for giving warnings.
585
586 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
587 However, this was not working for (?!) because it is optimized to (*FAIL),
588 for which PCRE does not allow quantifiers. The optimization is now disabled
589 when a quantifier follows (?!). I can't see any use for this, but it makes
590 things uniform.
591
592 36. Perl no longer allows group names to start with digits, so I have made this
593 change also in PCRE. It simplifies the code a bit.
594
595 37. In extended mode, Perl ignores spaces before a + that indicates a
596 possessive quantifier. PCRE allowed a space before the quantifier, but not
597 before the possessive +. It now does.
598
599 38. The use of \K (reset reported match start) within a repeated possessive
600 group such as (a\Kb)*+ was not working.
601
602 40. Document that the same character tables must be used at compile time and
603 run time, and that the facility to pass tables to pcre_exec() and
604 pcre_dfa_exec() is for use only with saved/restored patterns.
605
606 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
607 features for Builds with MSVC:
608
609 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
610 builds). The .rc files can be used to set FileDescription and many other
611 attributes.
612
613 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
614 This allows higher-level build scripts which want .pdb files to avoid
615 hard-coding the exact files needed."
616
617 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
618 mean "start of word" and "end of word", respectively, as a transition aid.
619
620 43. A minimizing repeat of a class containing codepoints greater than 255 in
621 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
622 compiled to use the heap for recursion.
623
624 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
625 is configured.
626
627
628 Version 8.33 28-May-2013
629 ------------------------
630
631 1. Added 'U' to some constants that are compared to unsigned integers, to
632 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
633 variables that are added to signed variables, to ensure the result is
634 signed and can be negated.
635
636 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
637 CMake config files.
638
639 3. Revise the creation of config.h.generic so that all boolean macros are
640 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
641 overriding via -D on the command line possible.
642
643 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
644 to unsigned int is reported to make a quite noticeable speed difference in
645 a specific Windows environment. Testing on Linux did also appear to show
646 some benefit (and it is clearly not harmful). Also fixed the definition of
647 Xop which should be unsigned.
648
649 5. Related to (4), changing the definition of the intermediate variable cc
650 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
651 improvements.
652
653 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
654 unnecessary spaces.
655
656 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
657 and later.
658
659 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
660
661 9. Optimizing fast_forward_start_bits in JIT.
662
663 10. Adding support for callouts in JIT, and fixing some issues revealed
664 during this work. Namely:
665
666 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
667
668 (b) Minimum length was not checked before the matching is started.
669
670 11. The value of capture_last that is passed to callouts was incorrect in some
671 cases when there was a capture on one path that was subsequently abandoned
672 after a backtrack. Also, the capture_last value is now reset after a
673 recursion, since all captures are also reset in this case.
674
675 12. The interpreter no longer returns the "too many substrings" error in the
676 case when an overflowing capture is in a branch that is subsequently
677 abandoned after a backtrack.
678
679 13. In the pathological case when an offset vector of size 2 is used, pcretest
680 now prints out the matched string after a yield of 0 or 1.
681
682 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
683 Only supported by the JIT compiler at the moment.
684
685 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
686
687 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
688 offset of the starting point of the matching process, provided the offsets
689 vector is large enough.
690
691 17. The \A escape now records a lookbehind value of 1, though its execution
692 does not actually inspect the previous character. This is to ensure that,
693 in partial multi-segment matching, at least one character from the old
694 segment is retained when a new segment is processed. Otherwise, if there
695 are no lookbehinds in the pattern, \A might match incorrectly at the start
696 of a new segment.
697
698 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
699
700 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
701 gives some modest performance improvement in 8-bit mode.
702
703 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
704 be expressed in certain programming languages using Universal Character
705 Names.
706
707 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
708 which points out that "non characters" are not "characters that may not
709 appear in Unicode strings" but rather "characters that are reserved for
710 internal use and have only local meaning".
711
712 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
713 there was a conditional group that depended on an assertion, if the
714 assertion was false, the callout that immediately followed the alternation
715 in the condition was skipped when pcre_exec() was used for matching.
716
717 23. Allow an explicit callout to be inserted before an assertion that is the
718 condition for a conditional group, for compatibility with automatic
719 callouts, which always insert a callout at this point.
720
721 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
722 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
723
724 25. (*PRUNE) is now supported by the JIT compiler.
725
726 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
727
728 27. Fix the case where there are two or more SKIPs with arguments that may be
729 ignored.
730
731 28. (*SKIP) is now supported by the JIT compiler.
732
733 29. (*THEN) is now supported by the JIT compiler.
734
735 30. Update RunTest with additional test selector options.
736
737 31. The way PCRE handles backtracking verbs has been changed in two ways.
738
739 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
740 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
741 backtracking. In some cases this makes it more Perl-compatible, but Perl's
742 rather obscure rules do not always do the same thing.
743
744 (2) Previously, backtracking verbs were confined within assertions. This is
745 no longer the case for positive assertions, except for (*ACCEPT). Again,
746 this sometimes improves Perl compatibility, and sometimes does not.
747
748 32. A number of tests that were in test 2 because Perl did things differently
749 have been moved to test 1, because either Perl or PCRE has changed, and
750 these tests are now compatible.
751
752 32. Backtracking control verbs are now handled in the same way in JIT and
753 interpreter.
754
755 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
756 contained a forward subroutine reference caused a compile error.
757
758 34. Auto-detect and optimize limited repetitions in JIT.
759
760 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
761 blocking (*UTF) etc.
762
763 36. In the interpreter, maximizing pattern repetitions for characters and
764 character types now use tail recursion, which reduces stack usage.
765
766 37. The value of the max lookbehind was not correctly preserved if a compiled
767 and saved regex was reloaded on a host of different endianness.
768
769 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
770 of the compiled pattern block, expand the flags field from 16 to 32 bits
771 because it was almost full.
772
773 39. Try madvise first before posix_madvise.
774
775 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
776 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
777 original change locked out all matches of empty strings. This has been
778 changed so that one match of an empty string per line is recognized.
779 Subsequent searches on the same line (for colouring or for --only-matching,
780 for example) do not recognize empty strings.
781
782 41. Applied a user patch to fix a number of spelling mistakes in comments.
783
784 42. Data lines longer than 65536 caused pcretest to crash.
785
786 43. Clarified the data type for length and startoffset arguments for pcre_exec
787 and pcre_dfa_exec in the function-specific man pages, where they were
788 explicitly stated to be in bytes, never having been updated. I also added
789 some clarification to the pcreapi man page.
790
791 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
792 a segmentation fault.
793
794
795 Version 8.32 30-November-2012
796 -----------------------------
797
798 1. Improved JIT compiler optimizations for first character search and single
799 character iterators.
800
801 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
802 Patch by Daniel Richard G.
803
804 3. Single character iterator optimizations in the JIT compiler.
805
806 4. Improved JIT compiler optimizations for character ranges.
807
808 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
809 Reported by Giuseppe D'Angelo.
810
811 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
812 of a line, was being set incorrectly in cases where .* appeared inside
813 atomic brackets at the start of a pattern, or where there was a subsequent
814 *PRUNE or *SKIP.
815
816 7. Improved instruction cache flush for POWER/PowerPC.
817 Patch by Daniel Richard G.
818
819 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
820 grep:
821
822 (a) There is now no limit to the number of patterns to be matched.
823
824 (b) An error is given if a pattern is too long.
825
826 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
827 are now supported.
828
829 (d) --exclude-from and --include-from (multiple use) have been added.
830
831 (e) Exclusions and inclusions now apply to all files and directories, not
832 just to those obtained from scanning a directory recursively.
833
834 (f) Multiple uses of -f and --file-list are now supported.
835
836 (g) In a Windows environment, the default for -d has been changed from
837 "read" (the GNU grep default) to "skip", because otherwise the presence
838 of a directory in the file list provokes an error.
839
840 (h) The documentation has been revised and clarified in places.
841
842 9. Improve the matching speed of capturing brackets.
843
844 10. Changed the meaning of \X so that it now matches a Unicode extended
845 grapheme cluster.
846
847 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
848 out POSIX threads when JIT support is configured.
849
850 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
851
852 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
853 structure could go wrong in environments where size_t is not the same size
854 as int.
855
856 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
857 set.
858
859 15. The EBCDIC support had decayed; later updates to the code had included
860 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
861 general tidy up of EBCDIC-related issues, and the documentation was also
862 not quite right. There is now a test that can be run on ASCII systems to
863 check some of the EBCDIC-related things (but is it not a full test).
864
865 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
866 in a small tidy to the code.
867
868 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
869
870 18. If the --only-matching (-o) option in pcregrep is specified multiple
871 times, each one causes appropriate output. For example, -o1 -o2 outputs the
872 substrings matched by the 1st and 2nd capturing parentheses. A separating
873 string can be specified by --om-separator (default empty).
874
875 19. Improving the first n character searches.
876
877 20. Turn case lists for horizontal and vertical white space into macros so that
878 they are defined only once.
879
880 21. This set of changes together give more compatible Unicode case-folding
881 behaviour for characters that have more than one other case when UCP
882 support is available.
883
884 (a) The Unicode property table now has offsets into a new table of sets of
885 three or more characters that are case-equivalent. The MultiStage2.py
886 script that generates these tables (the pcre_ucd.c file) now scans
887 CaseFolding.txt instead of UnicodeData.txt for character case
888 information.
889
890 (b) The code for adding characters or ranges of characters to a character
891 class has been abstracted into a generalized function that also handles
892 case-independence. In UTF-mode with UCP support, this uses the new data
893 to handle characters with more than one other case.
894
895 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
896 whose other case is greater than 256 are now correctly matched
897 caselessly. Previously, the high codepoint matched the low one, but not
898 vice versa.
899
900 (d) The processing of \h, \H, \v, and \ in character classes now makes use
901 of the new class addition function, using character lists defined as
902 macros alongside the case definitions of 20 above.
903
904 (e) Caseless back references now work with characters that have more than
905 one other case.
906
907 (f) General caseless matching of characters with more than one other case
908 is supported.
909
910 22. Unicode character properties were updated from Unicode 6.2.0
911
912 23. Improved CMake support under Windows. Patch by Daniel Richard G.
913
914 24. Add support for 32-bit character strings, and UTF-32
915
916 25. Major JIT compiler update (code refactoring and bugfixing).
917 Experimental Sparc 32 support is added.
918
919 26. Applied a modified version of Daniel Richard G's patch to create
920 pcre.h.generic and config.h.generic by "make" instead of in the
921 PrepareRelease script.
922
923 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
924 pcre_compile.c when checking for a zero character.
925
926 28. Introducing a native interface for JIT. Through this interface, the compiled
927 machine code can be directly executed. The purpose of this interface is to
928 provide fast pattern matching, so several sanity checks are not performed.
929 However, feature tests are still performed. The new interface provides
930 1.4x speedup compared to the old one.
931
932 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
933 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
934 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
935
936 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
937 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
938 the "old" RFC 2279). Instead, it ended up passing a negative length to
939 pcre_exec().
940
941 31. Add support for GCC's visibility feature to hide internal functions.
942
943 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
944 "unknown -C option" after outputting 0 or 1.
945
946 33. There is now support for generating a code coverage report for the test
947 suite in environments where gcc is the compiler and lcov is installed. This
948 is mainly for the benefit of the developers.
949
950 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
951 unaddressable using valgrind annotations, allowing valgrind to detect
952 invalid memory accesses. This is mainly for the benefit of the developers.
953
954 25. (*UTF) can now be used to start a pattern in any of the three libraries.
955
956 26. Give configure error if --enable-cpp but no C++ compiler found.
957
958
959 Version 8.31 06-July-2012
960 -------------------------
961
962 1. Fixing a wrong JIT test case and some compiler warnings.
963
964 2. Removed a bashism from the RunTest script.
965
966 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
967 to unsigned type, result still unsigned" that was given by an MS compiler
968 on encountering the code "-sizeof(xxx)".
969
970 4. Partial matching support is added to the JIT compiler.
971
972 5. Fixed several bugs concerned with partial matching of items that consist
973 of more than one character:
974
975 (a) /^(..)\1/ did not partially match "aba" because checking references was
976 done on an "all or nothing" basis. This also applied to repeated
977 references.
978
979 (b) \R did not give a hard partial match if \r was found at the end of the
980 subject.
981
982 (c) \X did not give a hard partial match after matching one or more
983 characters at the end of the subject.
984
985 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
986 a partial match for the string "\r".
987
988 (e) When newline was set to CRLF, the metacharacter "." did not recognize
989 a partial match for a CR character at the end of the subject string.
990
991 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
992 running pcretest, the text "(JIT)" added to the output whenever JIT is
993 actually used to run the match.
994
995 7. Individual JIT compile options can be set in pcretest by following -s+[+]
996 or /S+[+] with a digit between 1 and 7.
997
998 8. OP_NOT now supports any UTF character not just single-byte ones.
999
1000 9. (*MARK) control verb is now supported by the JIT compiler.
1001
1002 10. The command "./RunTest list" lists the available tests without actually
1003 running any of them. (Because I keep forgetting what they all are.)
1004
1005 11. Add PCRE_INFO_MAXLOOKBEHIND.
1006
1007 12. Applied a (slightly modified) user-supplied patch that improves performance
1008 when the heap is used for recursion (compiled with --disable-stack-for-
1009 recursion). Instead of malloc and free for each heap frame each time a
1010 logical recursion happens, frames are retained on a chain and re-used where
1011 possible. This sometimes gives as much as 30% improvement.
1012
1013 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1014 call.
1015
1016 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1017
1018 15. It is now possible to link pcretest with libedit as an alternative to
1019 libreadline.
1020
1021 16. (*COMMIT) control verb is now supported by the JIT compiler.
1022
1023 17. The Unicode data tables have been updated to Unicode 6.1.0.
1024
1025 18. Added --file-list option to pcregrep.
1026
1027 19. Added binary file support to pcregrep, including the -a, --binary-files,
1028 -I, and --text options.
1029
1030 20. The madvise function is renamed for posix_madvise for QNX compatibility
1031 reasons. Fixed by Giuseppe D'Angelo.
1032
1033 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1034
1035 22. Changed the option for creating symbolic links for 16-bit man pages from
1036 -s to -sf so that re-installing does not cause issues.
1037
1038 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1039
1040 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1041 matches in certain environments (the workspace was not being correctly
1042 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1043 some of the workspace data at the beginning of a restart.
1044
1045 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1046 was not doing so when it should - probably a typo introduced by SVN 528
1047 (change 8.10/14).
1048
1049 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1050 \w+ when the character tables indicated that \x{c4} was a word character.
1051 There were several related cases, all because the tests for doing a table
1052 lookup were testing for characters less than 127 instead of 255.
1053
1054 27. If a pattern contains capturing parentheses that are not used in a match,
1055 their slots in the ovector are set to -1. For those that are higher than
1056 any matched groups, this happens at the end of processing. In the case when
1057 there were back references that the ovector was too small to contain
1058 (causing temporary malloc'd memory to be used during matching), and the
1059 highest capturing number was not used, memory off the end of the ovector
1060 was incorrectly being set to -1. (It was using the size of the temporary
1061 memory instead of the true size.)
1062
1063 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1064 ovector size, it uses memory at the end of the block that it has got.
1065
1066 29. Check for an overlong MARK name and give an error at compile time. The
1067 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1068
1069 30. JIT compiler update.
1070
1071 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1072 Rill for the patch.
1073
1074 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1075
1076 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1077
1078 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1079 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1080 was enabled.
1081
1082 35. Improve JIT code generation for greedy plus quantifier.
1083
1084 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1085 1 to "aa" instead of to an empty string. The bug affected repeated groups
1086 that could potentially match an empty string.
1087
1088 37. Optimizing single character iterators in JIT.
1089
1090 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1091 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1092 codepoints that are too big for the mode are faulted, and in a UTF mode,
1093 disallowed codepoints are also faulted.
1094
1095 39. If PCRE was compiled with UTF support, in three places in the DFA
1096 matcher there was code that should only have been obeyed in UTF mode, but
1097 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1098 processing when bytes with values greater than 127 were present. In 16-bit
1099 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1100 both cases the values are those that cannot be the first data item in a UTF
1101 character. The three items that might have provoked this were recursions,
1102 possessively repeated groups, and atomic groups.
1103
1104 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1105 and pcregrep, because some OS require shared objects to be explicitly
1106 passed to ld, causing the link step to fail if they are not.
1107
1108 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1109 mode, patterns that started with \h* or \R* might be incorrectly matched.
1110
1111
1112 Version 8.30 04-February-2012
1113 -----------------------------
1114
1115 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1116 name is defined in ctype.h.
1117
1118 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1119 only in quite long subpatterns.
1120
1121 3. Removed the function pcre_info(), which has been obsolete and deprecated
1122 since it was replaced by pcre_fullinfo() in February 2000.
1123
1124 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1125 match a (*MARK), and the match failed at the start of the subject, a
1126 reference to memory before the start of the subject could occur. This bug
1127 was introduced by fix 17 of release 8.21.
1128
1129 5. A reference to an unset group with zero minimum repetition was giving
1130 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1131 /(another)?(\1?)test/ matched against "hello world test". This bug was
1132 introduced in release 8.13.
1133
1134 6. Add support for 16-bit character strings (a large amount of work involving
1135 many changes and refactorings).
1136
1137 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1138 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1139 from a file.
1140
1141 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1142 rounding is not applied in this particular case).
1143
1144 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1145 if they appear, or are escaped, in patterns.
1146
1147 10. Get rid of a number of -Wunused-but-set-variable warnings.
1148
1149 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1150 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1151 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1152 also returns the mark "x". This bug applied to capturing parentheses,
1153 non-capturing parentheses, and atomic parentheses. It also applied to some
1154 assertions.
1155
1156 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1157 information out of configure.ac instead of relying on pcre.h.generic, which
1158 is not stored in the repository.
1159
1160 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1161 -lreadline.
1162
1163 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1164
1165 15. Applied Graycode's patch to put the top-level frame on the stack rather
1166 than the heap when not using the stack for recursion. This gives a
1167 performance improvement in many cases when recursion is not deep.
1168
1169 16. Experimental code added to "pcretest -C" to output the stack frame size.
1170
1171
1172 Version 8.21 12-Dec-2011
1173 ------------------------
1174
1175 1. Updating the JIT compiler.
1176
1177 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1178 are added as well.
1179
1180 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1181 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1182 calling _pcre_jit_exec. Some extra comments are added.
1183
1184 4. (*MARK) settings inside atomic groups that do not contain any capturing
1185 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1186 was introduced by change 18 for 8.20.
1187
1188 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1189 ECMA-262 standard.
1190
1191 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1192 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1193 This bug was probably introduced by change 9 of 8.13.
1194
1195 7. While fixing 6 above, I noticed that a number of other items were being
1196 incorrectly rejected as "not fixed length". This arose partly because newer
1197 opcodes had not been added to the fixed-length checking code. I have (a)
1198 corrected the bug and added tests for these items, and (b) arranged for an
1199 error to occur if an unknown opcode is encountered while checking for fixed
1200 length instead of just assuming "not fixed length". The items that were
1201 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1202 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1203 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1204
1205 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1206 being incorrectly compiled and would have given unpredicatble results.
1207
1208 9. A possessively repeated subpattern with minimum repeat count greater than
1209 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1210 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1211 the first (A) could occur when it should not.
1212
1213 10. Add a cast and remove a redundant test from the code.
1214
1215 11. JIT should use pcre_malloc/pcre_free for allocation.
1216
1217 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1218 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1219 is anything other than /usr, -L is still shown).
1220
1221 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1222
1223 14. Perl does not support \N without a following name in a [] class; PCRE now
1224 also gives an error.
1225
1226 15. If a forward reference was repeated with an upper limit of around 2000,
1227 it caused the error "internal error: overran compiling workspace". The
1228 maximum number of forward references (including repeats) was limited by the
1229 internal workspace, and dependent on the LINK_SIZE. The code has been
1230 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1231 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1232 of around 200,000 forward references. While doing this, I also speeded up
1233 the filling in of repeated forward references.
1234
1235 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1236 incorrectly expecting the subject to contain another "a" after the start.
1237
1238 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1239 in the match, the SKIP should be ignored. This was not happening; instead
1240 the SKIP was being treated as NOMATCH. For patterns such as
1241 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1242 tested.
1243
1244 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1245 now much more compatible with Perl, in particular in cases where the result
1246 is a non-match for a non-anchored pattern. For example, if
1247 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1248 "m", where previously it did not return a name. A side effect of this
1249 change is that for partial matches, the last encountered mark name is
1250 returned, as for non matches. A number of tests that were previously not
1251 Perl-compatible have been moved into the Perl-compatible test files. The
1252 refactoring has had the pleasing side effect of removing one argument from
1253 the match() function, thus reducing its stack requirements.
1254
1255 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1256 subsequent uses of /S (without +) incorrectly behaved like /S+.
1257
1258 21. Retrieve executable code size support for the JIT compiler and fixing
1259 some warnings.
1260
1261 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1262 not work when the shorter character appeared right at the end of the
1263 subject string.
1264
1265 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1266 systems.
1267
1268 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1269 output it when the /M option is used in pcretest.
1270
1271 25. The CheckMan script was not being included in the distribution. Also, added
1272 an explicit "perl" to run Perl scripts from the PrepareRelease script
1273 because this is reportedly needed in Windows.
1274
1275 26. If study data was being save in a file and studying had not found a set of
1276 "starts with" bytes for the pattern, the data written to the file (though
1277 never used) was taken from uninitialized memory and so caused valgrind to
1278 complain.
1279
1280 27. Updated RunTest.bat as provided by Sheri Pierce.
1281
1282 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1283
1284 29. Computation of memory usage for the table of capturing group names was
1285 giving an unnecessarily large value.
1286
1287
1288 Version 8.20 21-Oct-2011
1289 ------------------------
1290
1291 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1292 a POSIX class. After further experiments with Perl, which convinced me that
1293 Perl has bugs and confusions, a closing square bracket is no longer allowed
1294 in a POSIX name. This bug also affected patterns with classes that started
1295 with full stops.
1296
1297 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1298 captured substring, but while checking the failing first alternative,
1299 substring 1 is temporarily captured. If the output vector supplied to
1300 pcre_exec() was not big enough for this capture, the yield of the function
1301 was still zero ("insufficient space for captured substrings"). This cannot
1302 be totally fixed without adding another stack variable, which seems a lot
1303 of expense for a edge case. However, I have improved the situation in cases
1304 such as /(a)(b)x|abc/ matched against "abc", where the return code
1305 indicates that fewer than the maximum number of slots in the ovector have
1306 been set.
1307
1308 3. Related to (2) above: when there are more back references in a pattern than
1309 slots in the output vector, pcre_exec() uses temporary memory during
1310 matching, and copies in the captures as far as possible afterwards. It was
1311 using the entire output vector, but this conflicts with the specification
1312 that only 2/3 is used for passing back captured substrings. Now it uses
1313 only the first 2/3, for compatibility. This is, of course, another edge
1314 case.
1315
1316 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1317 main code base, and can be used by building with --enable-jit. When this is
1318 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1319 runtime --no-jit option is given.
1320
1321 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1322 ovector, the return from the function was zero, implying that there were
1323 other matches that did not fit. The correct "exactly full" value is now
1324 returned.
1325
1326 6. If a subpattern that was called recursively or as a subroutine contained
1327 (*PRUNE) or any other control that caused it to give a non-standard return,
1328 invalid errors such as "Error -26 (nested recursion at the same subject
1329 position)" or even infinite loops could occur.
1330
1331 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1332 computing the minimum length on reaching *ACCEPT, and so ended up with the
1333 wrong value of 1 rather than 0. Further investigation indicates that
1334 computing a minimum subject length in the presence of *ACCEPT is difficult
1335 (think back references, subroutine calls), and so I have changed the code
1336 so that no minimum is registered for a pattern that contains *ACCEPT.
1337
1338 8. If (*THEN) was present in the first (true) branch of a conditional group,
1339 it was not handled as intended. [But see 16 below.]
1340
1341 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1342 Sheri Pierce.
1343
1344 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1345 the first byte in a match must be "a".
1346
1347 11. Change 17 for 8.13 increased the recursion depth for patterns like
1348 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1349 pattern contains any instances of (*THEN). If it does not, the old
1350 optimizations are restored. It would be nice to do this on a per-group
1351 basis, but at the moment that is not feasible.
1352
1353 12. In some environments, the output of pcretest -C is CRLF terminated. This
1354 broke RunTest's code that checks for the link size. A single white space
1355 character after the value is now allowed for.
1356
1357 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1358 For "fr", it uses the Windows-specific input and output files.
1359
1360 14. If (*THEN) appeared in a group that was called recursively or as a
1361 subroutine, it did not work as intended. [But see next item.]
1362
1363 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1364 pattern fragments (but not containing any | characters). If A and B are
1365 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1366 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1367 D. In other words, Perl considers parentheses that do not contain any |
1368 characters to be part of a surrounding alternative, whereas PCRE was
1369 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1370 differently. PCRE now behaves in the same way as Perl, except in the case
1371 of subroutine/recursion calls such as (?1) which have in any case always
1372 been different (but PCRE had them first :-).
1373
1374 16. Related to 15 above: Perl does not treat the | in a conditional group as
1375 creating alternatives. Such a group is treated in the same way as an
1376 ordinary group without any | characters when processing (*THEN). PCRE has
1377 been changed to match Perl's behaviour.
1378
1379 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1380 RunGrepTest script failed.
1381
1382 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1383 inevitable for groups that contain captures, but it can lead to a lot of
1384 stack use in large patterns. The old behaviour has been restored for atomic
1385 groups that do not contain any capturing parentheses.
1386
1387 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1388 suppress the check for a minimum subject length at run time. (If it was
1389 given to pcre_exec() or pcre_dfa_exec() it did work.)
1390
1391 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1392 fail to work when decoding hex characters in data strings in EBCDIC
1393 environments.
1394
1395 21. It appears that in at least one Mac OS environment, the isxdigit() function
1396 is implemented as a macro that evaluates to its argument more than once,
1397 contravening the C 90 Standard (I haven't checked a later standard). There
1398 was an instance in pcretest which caused it to go wrong when processing
1399 \x{...} escapes in subject strings. The has been rewritten to avoid using
1400 things like p++ in the argument of isxdigit().
1401
1402
1403 Version 8.13 16-Aug-2011
1404 ------------------------
1405
1406 1. The Unicode data tables have been updated to Unicode 6.0.0.
1407
1408 2. Two minor typos in pcre_internal.h have been fixed.
1409
1410 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1411 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1412 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1413
1414 4. There were a number of related bugs in the code for matching backrefences
1415 caselessly in UTF-8 mode when codes for the characters concerned were
1416 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1417 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1418 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1419 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1420 2-byte code at the end of the subject (it thought there wasn't enough data
1421 left).
1422
1423 5. Comprehensive information about what went wrong is now returned by
1424 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1425 as the output vector has at least 2 elements. The offset of the start of
1426 the failing character and a reason code are placed in the vector.
1427
1428 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1429 now returned is for the first byte of the failing character, instead of the
1430 last byte inspected. This is an incompatible change, but I hope it is small
1431 enough not to be a problem. It makes the returned offset consistent with
1432 pcre_exec() and pcre_dfa_exec().
1433
1434 7. pcretest now gives a text phrase as well as the error number when
1435 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1436 failure, the offset and reason code are output.
1437
1438 8. When \R was used with a maximizing quantifier it failed to skip backwards
1439 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1440 back over a single character (\n). This seems wrong (because it treated the
1441 two characters as a single entity when going forwards), conflicts with the
1442 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1443 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1444 has been changed.
1445
1446 9. Some internal refactoring has changed the processing so that the handling
1447 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1448 time (the PCRE_DOTALL option was changed this way some time ago: version
1449 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1450 which was always a bit of a fudge. It also means that there is one less
1451 argument for the match() function, which reduces its stack requirements
1452 slightly. This change also fixes an incompatibility with Perl: the pattern
1453 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1454
1455 10. More internal refactoring has drastically reduced the number of recursive
1456 calls to match() for possessively repeated groups such as (abc)++ when
1457 using pcre_exec().
1458
1459 11. While implementing 10, a number of bugs in the handling of groups were
1460 discovered and fixed:
1461
1462 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1463 (a|)*(?1) gave a compile-time internal error.
1464 ((a|)+)+ did not notice that the outer group could match an empty string.
1465 (^a|^)+ was not marked as anchored.
1466 (.*a|.*)+ was not marked as matching at start or after a newline.
1467
1468 12. Yet more internal refactoring has removed another argument from the match()
1469 function. Special calls to this function are now indicated by setting a
1470 value in a variable in the "match data" data block.
1471
1472 13. Be more explicit in pcre_study() instead of relying on "default" for
1473 opcodes that mean there is no starting character; this means that when new
1474 ones are added and accidentally left out of pcre_study(), testing should
1475 pick them up.
1476
1477 14. The -s option of pcretest has been documented for ages as being an old
1478 synonym of -m (show memory usage). I have changed it to mean "force study
1479 for every regex", that is, assume /S for every regex. This is similar to -i
1480 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1481 using it. It makes it easier to run collections of tests with and without
1482 study enabled, and thereby test pcre_study() more easily. All the standard
1483 tests are now run with and without -s (but some patterns can be marked as
1484 "never study" - see 20 below).
1485
1486 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1487 restoration of the capturing data to the outer values was not happening
1488 correctly.
1489
1490 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1491 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1492 pattern had matched an empty string, and so incorrectly returned a no
1493 match.
1494
1495 17. There was optimizing code for the last branch of non-capturing parentheses,
1496 and also for the obeyed branch of a conditional subexpression, which used
1497 tail recursion to cut down on stack usage. Unfortunately, now that there is
1498 the possibility of (*THEN) occurring in these branches, tail recursion is
1499 no longer possible because the return has to be checked for (*THEN). These
1500 two optimizations have therefore been removed. [But see 8.20/11 above.]
1501
1502 18. If a pattern containing \R was studied, it was assumed that \R always
1503 matched two bytes, thus causing the minimum subject length to be
1504 incorrectly computed because \R can also match just one byte.
1505
1506 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1507 was incorrectly computed.
1508
1509 20. If /S is present twice on a test pattern in pcretest input, it now
1510 *disables* studying, thereby overriding the use of -s on the command line
1511 (see 14 above). This is necessary for one or two tests to keep the output
1512 identical in both cases.
1513
1514 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1515 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1516
1517 22. When an atomic group that contained a capturing parenthesis was
1518 successfully matched, but the branch in which it appeared failed, the
1519 capturing was not being forgotten if a higher numbered group was later
1520 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1521 group 1 to "a", when in fact it should be unset. This applied to multi-
1522 branched capturing and non-capturing groups, repeated or not, and also to
1523 positive assertions (capturing in negative assertions does not happen
1524 in PCRE) and also to nested atomic groups.
1525
1526 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1527 subject after a captured substring, to make it easier to tell which of a
1528 number of identical substrings has been captured.
1529
1530 24. The way atomic groups are processed by pcre_exec() has been changed so that
1531 if they are repeated, backtracking one repetition now resets captured
1532 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1533 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1534 "aaa". Previously, it would have been "a". As part of this code
1535 refactoring, the way recursive calls are handled has also been changed.
1536
1537 25. If an assertion condition captured any substrings, they were not passed
1538 back unless some other capturing happened later. For example, if
1539 (?(?=(a))a) was matched against "a", no capturing was returned.
1540
1541 26. When studying a pattern that contained subroutine calls or assertions,
1542 the code for finding the minimum length of a possible match was handling
1543 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1544 group 1 called group 2 while simultaneously a separate group 2 called group
1545 1). A stack overflow occurred in this case. I have fixed this by limiting
1546 the recursion depth to 10.
1547
1548 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1549 Fortmann. This supports explicit test numbers on the command line, and has
1550 argument validation and error reporting.
1551
1552 28. An instance of \X with an unlimited repeat could fail if at any point the
1553 first character it looked at was a mark character.
1554
1555 29. Some minor code refactoring concerning Unicode properties and scripts
1556 should reduce the stack requirement of match() slightly.
1557
1558 30. Added the '=' option to pcretest to check the setting of unused capturing
1559 slots at the end of the pattern, which are documented as being -1, but are
1560 not included in the return count.
1561
1562 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1563 compiled something random. Now it gives a compile-time error (as does
1564 Perl).
1565
1566 32. A *MARK encountered during the processing of a positive assertion is now
1567 recorded and passed back (compatible with Perl).
1568
1569 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1570 had alternative anchored branches, the search for a second match in a line
1571 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1572 matched the line "0102" twice. The same bug affected patterns that started
1573 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1574 twice.
1575
1576 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1577 does, and because of capturing effects, quantifying parenthesized
1578 assertions may at times be useful. Quantifiers are now allowed for
1579 parenthesized assertions.
1580
1581 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1582
1583 36. \g was being checked for fancy things in a character class, when it should
1584 just be a literal "g".
1585
1586 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1587 appearance of a nested POSIX class supersedes an apparent external class.
1588 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1589 unescaped square brackets may also appear as part of class names. For
1590 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1591 more like Perl. (But see 8.20/1 above.)
1592
1593 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1594 was because it thought it was \N{name}, which is not supported).
1595
1596 39. Add minix to OS list not supporting the -S option in pcretest.
1597
1598 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1599 cannot analyze patterns in sufficient detail to catch mutual recursions
1600 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1601 subgroup is called recursively as a subpattern for a second time at the
1602 same position in the subject string. In previous releases this might have
1603 been caught by the recursion limit, or it might have run out of stack.
1604
1605 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1606 happen only once. PCRE was, however incorrectly giving a compile time error
1607 "recursive call could loop indefinitely" because it cannot analyze the
1608 pattern in sufficient detail. The compile time test no longer happens when
1609 PCRE is compiling a conditional subpattern, but actual runaway loops are
1610 now caught at runtime (see 40 above).
1611
1612 42. It seems that Perl allows any characters other than a closing parenthesis
1613 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1614 has been changed to be the same.
1615
1616 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1617 as not to get warnings when autogen.sh is called. Also changed
1618 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1619
1620 44. To help people who use pcregrep to scan files containing exceedingly long
1621 lines, the following changes have been made:
1622
1623 (a) The default value of the buffer size parameter has been increased from
1624 8K to 20K. (The actual buffer used is three times this size.)
1625
1626 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1627 PCRE is built.
1628
1629 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1630 to be set at run time.
1631
1632 (d) Numerical values in pcregrep options can be followed by K or M, for
1633 example --buffer-size=50K.
1634
1635 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1636 given and the return code is set to 2.
1637
1638 45. Add a pointer to the latest mark to the callout data block.
1639
1640 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1641 partial match of an empty string instead of no match. This was specific to
1642 the use of ".".
1643
1644 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1645 complete match instead of a partial match. This bug was dependent on both
1646 the PCRE_UTF8 and PCRE_DOTALL options being set.
1647
1648 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1649 starting byte set, because \b was not being ignored.
1650
1651
1652 Version 8.12 15-Jan-2011
1653 ------------------------
1654
1655 1. Fixed some typos in the markup of the man pages, and wrote a script that
1656 checks for such things as part of the documentation building process.
1657
1658 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1659 --match-limit and --recursion-limit options (added for 8.11). In
1660 particular, this made one of the standard tests fail. (The integer value
1661 went into the wrong half of a long int.)
1662
1663 3. If the --colour option was given to pcregrep with -v (invert match), it
1664 did strange things, either producing crazy output, or crashing. It should,
1665 of course, ignore a request for colour when reporting lines that do not
1666 match.
1667
1668 4. Another pcregrep bug caused similar problems if --colour was specified with
1669 -M (multiline) and the pattern match finished with a line ending.
1670
1671 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1672 matched in multiline mode, the following line was shown as part of the
1673 match. This seems wrong, so I have changed it.
1674
1675 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1676 the check for further matches in the same line (so they could be coloured)
1677 to overrun the end of the current line. If another match was found, it was
1678 incorrectly shown (and then shown again when found in the next line).
1679
1680 7. If pcregrep was compiled under Windows, there was a reference to the
1681 function pcregrep_exit() before it was defined. I am assuming this was
1682 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1683 reported by a user. I've moved the definition above the reference.
1684
1685
1686 Version 8.11 10-Dec-2010
1687 ------------------------
1688
1689 1. (*THEN) was not working properly if there were untried alternatives prior
1690 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1691 backtracked to try for "b" instead of moving to the next alternative branch
1692 at the same level (in this case, to look for "c"). The Perl documentation
1693 is clear that when (*THEN) is backtracked onto, it goes to the "next
1694 alternative in the innermost enclosing group".
1695
1696 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1697 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1698 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1699 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1700 (*THEN).
1701
1702 3. If \s appeared in a character class, it removed the VT character from
1703 the class, even if it had been included by some previous item, for example
1704 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1705 of \s, but is part of the POSIX "space" class.)
1706
1707 4. A partial match never returns an empty string (because you can always
1708 match an empty string at the end of the subject); however the checking for
1709 an empty string was starting at the "start of match" point. This has been
1710 changed to the "earliest inspected character" point, because the returned
1711 data for a partial match starts at this character. This means that, for
1712 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1713 (previously it gave "no match").
1714
1715 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1716 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1717 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1718 has an implication that the given string is incomplete (because a partial
1719 match is preferred over a full match). For this reason, these items now
1720 give a partial match in this situation. [Aside: previously, the one case
1721 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1722 match rather than a full match, which was wrong by the old rules, but is
1723 now correct.]
1724
1725 6. There was a bug in the handling of #-introduced comments, recognized when
1726 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1727 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1728 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1729 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1730 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1731 places in pcre_compile().
1732
1733 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1734 comments when looking ahead for named forward references to subpatterns,
1735 the only newline sequence it recognized was NL. It now handles newlines
1736 according to the set newline convention.
1737
1738 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1739 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1740 cater for a lack of strerror(). These oversights have been fixed.
1741
1742 9. Added --match-limit and --recursion-limit to pcregrep.
1743
1744 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1745
1746 11. When the -o option was used, pcregrep was setting a return code of 1, even
1747 when matches were found, and --line-buffered was not being honoured.
1748
1749 12. Added an optional parentheses number to the -o and --only-matching options
1750 of pcregrep.
1751
1752 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1753 can match an empty string. The code to do it in pcretest and pcredemo
1754 needed fixing:
1755
1756 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1757 only one byte after an empty string match just before CRLF (this case
1758 just got forgotten; "any" and "anycrlf" were OK).
1759
1760 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1761 mode when an empty string match preceded an ASCII character followed by
1762 a non-ASCII character. (The code for advancing by one character rather
1763 than one byte was nonsense.)
1764
1765 (c) The pcredemo.c sample program did not have any code at all to handle
1766 the cases when CRLF is a valid newline sequence.
1767
1768 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1769 as a starting offset was within the subject string. There is now a new
1770 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1771 negative or greater than the length of the string. In order to test this,
1772 pcretest is extended to allow the setting of negative starting offsets.
1773
1774 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1775 starting offset points to the beginning of a UTF-8 character was
1776 unnecessarily clumsy. I tidied it up.
1777
1778 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1779 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1780
1781 17. Nobody had reported that the --include_dir option, which was added in
1782 release 7.7 should have been called --include-dir (hyphen, not underscore)
1783 for compatibility with GNU grep. I have changed it to --include-dir, but
1784 left --include_dir as an undocumented synonym, and the same for
1785 --exclude-dir, though that is not available in GNU grep, at least as of
1786 release 2.5.4.
1787
1788 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1789 characters from a string of bytes) have been redefined so as not to use
1790 loops, in order to improve performance in some environments. At the same
1791 time, I abstracted some of the common code into auxiliary macros to save
1792 repetition (this should not affect the compiled code).
1793
1794 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1795 compile-time error is now given if \c is not followed by an ASCII
1796 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1797 different, and any byte value is allowed.)
1798
1799 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1800 START_OPTIMIZE option, which is now allowed at compile time - but just
1801 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1802 to pcregrep and other applications that have no direct access to PCRE
1803 options. The new /Y option in pcretest sets this option when calling
1804 pcre_compile().
1805
1806 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1807 back references. Groups containing recursive back references were forced to
1808 be atomic by that change, but in the case of named groups, the amount of
1809 memory required was incorrectly computed, leading to "Failed: internal
1810 error: code overflow". This has been fixed.
1811
1812 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1813 pcretest.c, to avoid build problems in some Borland environments.
1814
1815
1816 Version 8.10 25-Jun-2010
1817 ------------------------
1818
1819 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1820 THEN.
1821
1822 2. (*ACCEPT) was not working when inside an atomic group.
1823
1824 3. Inside a character class, \B is treated as a literal by default, but
1825 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1826 causes the error). The code is unchanged, but I tidied the documentation.
1827
1828 4. Inside a character class, PCRE always treated \R and \X as literals,
1829 whereas Perl faults them if its -w option is set. I have changed PCRE so
1830 that it faults them when PCRE_EXTRA is set.
1831
1832 5. Added support for \N, which always matches any character other than
1833 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1834
1835 6. When compiling pcregrep with newer versions of gcc which may have
1836 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1837 declared with attribute warn_unused_result" were given. Just casting the
1838 result to (void) does not stop the warnings; a more elaborate fudge is
1839 needed. I've used a macro to implement this.
1840
1841 7. Minor change to pcretest.c to avoid a compiler warning.
1842
1843 8. Added four artifical Unicode properties to help with an option to make
1844 \s etc use properties (see next item). The new properties are: Xan
1845 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1846
1847 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1848 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1849 this option. Modified pcretest to add /W to test this facility. Added
1850 REG_UCP to make it available via the POSIX interface.
1851
1852 10. Added --line-buffered to pcregrep.
1853
1854 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1855 studied, and the match started with a letter with a code point greater than
1856 127 whose first byte was different to the first byte of the other case of
1857 the letter, the other case of this starting letter was not recognized
1858 (#976).
1859
1860 12. If a pattern that was studied started with a repeated Unicode property
1861 test, for example, \p{Nd}+, there was the theoretical possibility of
1862 setting up an incorrect bitmap of starting bytes, but fortunately it could
1863 not have actually happened in practice until change 8 above was made (it
1864 added property types that matched character-matching opcodes).
1865
1866 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1867 possible starting bytes for non-anchored patterns.
1868
1869 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1870 \R, and also a number of cases that involve Unicode properties, both
1871 explicit and implicit when PCRE_UCP is set.
1872
1873 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1874 input, it could crash or give wrong results if characters with values
1875 greater than 0xc0 were present in the subject string. (Detail: it assumed
1876 UTF-8 input when processing these items.)
1877
1878 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1879 size_t is 64-bit (#991).
1880
1881 17. Added a check for running out of memory when PCRE is compiled with
1882 --disable-stack-for-recursion (#990).
1883
1884 18. If the last data line in a file for pcretest does not have a newline on
1885 the end, a newline was missing in the output.
1886
1887 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1888 less than 128) in its various bitmaps. However, there is a facility for
1889 generating tables according to the current locale when PCRE is compiled. It
1890 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1891 characters, are recognized by isspace() and therefore were getting set in
1892 these tables, and indeed these tables seem to approximate to ISO 8859. This
1893 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1894 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1895 which of course cannot start UTF-8 characters. I have changed the code so
1896 that only real ASCII characters (less than 128) and the correct starting
1897 bytes for UTF-8 encodings are set for characters greater than 127 when in
1898 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1899 altogether.)
1900
1901 20. Added the /T option to pcretest so as to be able to run tests with non-
1902 standard character tables, thus making it possible to include the tests
1903 used for 19 above in the standard set of tests.
1904
1905 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1906 reference to a subpattern the other side of a comment that contains an
1907 opening parenthesis caused either an internal compiling error, or a
1908 reference to the wrong subpattern.
1909
1910
1911 Version 8.02 19-Mar-2010
1912 ------------------------
1913
1914 1. The Unicode data tables have been updated to Unicode 5.2.0.
1915
1916 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1917 configured.
1918
1919 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1920 original author of that file, following a query about its status.
1921
1922 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1923 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1924
1925 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1926 quantifier applied to a forward-referencing subroutine call, could compile
1927 incorrect code or give the error "internal error: previously-checked
1928 referenced subpattern not found".
1929
1930 6. Both MS Visual Studio and Symbian OS have problems with initializing
1931 variables to point to external functions. For these systems, therefore,
1932 pcre_malloc etc. are now initialized to local functions that call the
1933 relevant global functions.
1934
1935 7. There were two entries missing in the vectors called coptable and poptable
1936 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1937 I've fixed the data, and added a kludgy way of testing at compile time that
1938 the lengths are correct (equal to the number of opcodes).
1939
1940 8. Following on from 7, I added a similar kludge to check the length of the
1941 eint vector in pcreposix.c.
1942
1943 9. Error texts for pcre_compile() are held as one long string to avoid too
1944 much relocation at load time. To find a text, the string is searched,
1945 counting zeros. There was no check for running off the end of the string,
1946 which could happen if a new error number was added without updating the
1947 string.
1948
1949 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1950
1951 11. \K was not working if it appeared in an atomic group or in a group that
1952 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1953 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1954 the assertion is positive, but not if it is negative.
1955
1956 12. Change 11 fortuitously reduced the size of the stack frame used in the
1957 "match()" function of pcre_exec.c by one pointer. Forthcoming
1958 implementation of support for (*MARK) will need an extra pointer on the
1959 stack; I have reserved it now, so that the stack frame size does not
1960 decrease.
1961
1962 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1963 item in branch that calls a recursion is a subroutine call - as in the
1964 second branch in the above example - was incorrectly given the compile-
1965 time error "recursive call could loop indefinitely" because pcre_compile()
1966 was not correctly checking the subroutine for matching a non-empty string.
1967
1968 14. The checks for overrunning compiling workspace could trigger after an
1969 overrun had occurred. This is a "should never occur" error, but it can be
1970 triggered by pathological patterns such as hundreds of nested parentheses.
1971 The checks now trigger 100 bytes before the end of the workspace.
1972
1973 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1974
1975
1976 Version 8.01 19-Jan-2010
1977 ------------------------
1978
1979 1. If a pattern contained a conditional subpattern with only one branch (in
1980 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1981 computed the wrong minimum data length (which is of course zero for such
1982 subpatterns). This could cause incorrect "no match" results.
1983
1984 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1985 the pattern is reset in the first branch, pcre_compile() failed with
1986 "internal error: code overflow at offset...". This happened only when
1987 the reset was to the original external option setting. (An optimization
1988 abstracts leading options settings into an external setting, which was the
1989 cause of this.)
1990
1991 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1992 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1993 assertion pattern did not match (meaning that the assertion was true), it
1994 was incorrectly treated as false if the SKIP had been reached during the
1995 matching. This also applied to assertions used as conditions.
1996
1997 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1998 assertion subpattern, including such a pattern used as a condition,
1999 unpredictable results occurred, instead of the error return
2000 PCRE_ERROR_DFA_UITEM.
2001
2002 5. The C++ GlobalReplace function was not working like Perl for the special
2003 situation when an empty string is matched. It now does the fancy magic
2004 stuff that is necessary.
2005
2006 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2007 removed. (These were left over from very, very early versions of PCRE.)
2008
2009 7. Some cosmetic changes to the code to make life easier when compiling it
2010 as part of something else:
2011
2012 (a) Change DEBUG to PCRE_DEBUG.
2013
2014 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2015 called "current" as "current_branch", to prevent a collision with the
2016 Linux macro when compiled as a kernel module.
2017
2018 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2019 prevent a collision with the Linux macro when compiled as a kernel
2020 module.
2021
2022 8. In pcre_compile() there are some checks for integer overflows that used to
2023 cast potentially large values to (double). This has been changed to that
2024 when building, a check for int64_t is made, and if it is found, it is used
2025 instead, thus avoiding the use of floating point arithmetic. (There is no
2026 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2027 double.
2028
2029 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2030 2005 (difference between two addresses compared to an unsigned value).
2031
2032 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2033 custom one, because of the following reported problem in Windows:
2034
2035 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2036 under Win32.
2037 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2038 therefore missing the function definition.
2039 - The compiler thus generates a "C" signature for the test function.
2040 - The linker fails to find the "C" function.
2041 - PCRE fails to configure if asked to do so against libbz2.
2042
2043 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2044 messages were output:
2045
2046 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2047 rerunning libtoolize, to keep the correct libtool macros in-tree.
2048 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2049
2050 I have done both of these things.
2051
2052 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2053 most of the time, it *can* run out if it is given a pattern that contains a
2054 runaway infinite recursion. I updated the discussion in the pcrestack man
2055 page.
2056
2057 13. Now that we have gone to the x.xx style of version numbers, the minor
2058 version may start with zero. Using 08 or 09 is a bad idea because users
2059 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2060 interpreted as invalid octal numbers. I've updated the previous comment in
2061 configure.ac, and also added a check that gives an error if 08 or 09 are
2062 used.
2063
2064 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2065 causing partial matching to fail when the end of the subject matched \W
2066 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2067
2068 15. There were some discrepancies between the declarations in pcre_internal.h
2069 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2070 their definitions. The declarations used "const uschar *" and the
2071 definitions used USPTR. Even though USPTR is normally defined as "const
2072 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2073 reported that: "This difference in casting confuses some C++ compilers, for
2074 example, SunCC recognizes above declarations as different functions and
2075 generates broken code for hbpcre." I have changed the declarations to use
2076 USPTR.
2077
2078 16. GNU libtool is named differently on some systems. The autogen.sh script now
2079 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2080 (FreeBSD).
2081
2082 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2083 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2084 comment: "Figure out how to create a longlong from a string: strtoll and
2085 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2086 instance, but it only takes 2 args instead of 3!"
2087
2088 18. A subtle bug concerned with back references has been fixed by a change of
2089 specification, with a corresponding code fix. A pattern such as
2090 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2091 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2092 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2093 same bug. Such groups have to be quantified to be useful, or contained
2094 inside another quantified group. (If there's no repetition, the reference
2095 can never match.) The problem arises because, having left the group and
2096 moved on to the rest of the pattern, a later failure that backtracks into
2097 the group uses the captured value from the final iteration of the group
2098 rather than the correct earlier one. I have fixed this in PCRE by forcing
2099 any group that contains a reference to itself to be an atomic group; that
2100 is, there cannot be any backtracking into it once it has completed. This is
2101 similar to recursive and subroutine calls.
2102
2103
2104 Version 8.00 19-Oct-09
2105 ----------------------
2106
2107 1. The table for translating pcre_compile() error codes into POSIX error codes
2108 was out-of-date, and there was no check on the pcre_compile() error code
2109 being within the table. This could lead to an OK return being given in
2110 error.
2111
2112 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2113 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2114 in a Windows environment.
2115
2116 3. The pcregrep --count option prints the count for each file even when it is
2117 zero, as does GNU grep. However, pcregrep was also printing all files when
2118 --files-with-matches was added. Now, when both options are given, it prints
2119 counts only for those files that have at least one match. (GNU grep just
2120 prints the file name in this circumstance, but including the count seems
2121 more useful - otherwise, why use --count?) Also ensured that the
2122 combination -clh just lists non-zero counts, with no names.
2123
2124 4. The long form of the pcregrep -F option was incorrectly implemented as
2125 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2126 but it seems right to fix it, and I didn't think it was worth preserving
2127 the old behaviour.
2128
2129 5. The command line items --regex=pattern and --regexp=pattern were not
2130 recognized by pcregrep, which required --regex pattern or --regexp pattern
2131 (with a space rather than an '='). The man page documented the '=' forms,
2132 which are compatible with GNU grep; these now work.
2133
2134 6. No libpcreposix.pc file was created for pkg-config; there was just
2135 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2136
2137 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2138 when UCP support is not needed, by modifying the Python script that
2139 generates it from Unicode data files. This should not matter if the module
2140 is correctly used as a library, but I received one complaint about 50K of
2141 unwanted data. My guess is that the person linked everything into his
2142 program rather than using a library. Anyway, it does no harm.
2143
2144 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2145 was a minimum greater than 1 for a wide character in a possessive
2146 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2147 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2148 character. Chaos in the form of incorrect output or a compiling loop could
2149 result.
2150
2151 9. The restrictions on what a pattern can contain when partial matching is
2152 requested for pcre_exec() have been removed. All patterns can now be
2153 partially matched by this function. In addition, if there are at least two
2154 slots in the offset vector, the offset of the earliest inspected character
2155 for the match and the offset of the end of the subject are set in them when
2156 PCRE_ERROR_PARTIAL is returned.
2157
2158 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2159 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2160 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2161 and may be more useful for multi-segment matching.
2162
2163 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2164 used to be given if ever the end of the subject was reached; now it is
2165 given only if matching could not proceed because another character was
2166 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2167 string "Z", which now yields "no match" instead of "partial match". In the
2168 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2169 final character ended with (*FAIL).
2170
2171 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2172 if the pattern had a "must contain" character that was already found in the
2173 earlier partial match, unless partial matching was again requested. For
2174 example, with the pattern /dog.(body)?/, the "must contain" character is
2175 "g". If the first part-match was for the string "dog", restarting with
2176 "sbody" failed. This bug has been fixed.
2177
2178 13. The string returned by pcre_dfa_exec() after a partial match has been
2179 changed so that it starts at the first inspected character rather than the
2180 first character of the match. This makes a difference only if the pattern
2181 starts with a lookbehind assertion or \b or \B (\K is not supported by
2182 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2183 matching functions compatible, and I think it's the right thing to do.
2184
2185 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2186 so that the demonstration program is easily available in environments where
2187 PCRE has not been installed from source.
2188
2189 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2190 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2191 library.
2192
2193 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2194 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2195 is not the first non-POSIX option to be added. Clearly some people find
2196 these options useful.
2197
2198 17. If a caller to the POSIX matching function regexec() passes a non-zero
2199 value for nmatch with a NULL value for pmatch, the value of
2200 nmatch is forced to zero.
2201
2202 18. RunGrepTest did not have a test for the availability of the -u option of
2203 the diff command, as RunTest does. It now checks in the same way as
2204 RunTest, and also checks for the -b option.
2205
2206 19. If an odd number of negated classes containing just a single character
2207 interposed, within parentheses, between a forward reference to a named
2208 subpattern and the definition of the subpattern, compilation crashed with
2209 an internal error, complaining that it could not find the referenced
2210 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2211 [The bug was that it was starting one character too far in when skipping
2212 over the character class, thus treating the ] as data rather than
2213 terminating the class. This meant it could skip too much.]
2214
2215 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2216 /g option in pcretest when the pattern contains \K, which makes it possible
2217 to have an empty string match not at the start, even when the pattern is
2218 anchored. Updated pcretest and pcredemo to use this option.
2219
2220 21. If the maximum number of capturing subpatterns in a recursion was greater
2221 than the maximum at the outer level, the higher number was returned, but
2222 with unset values at the outer level. The correct (outer level) value is
2223 now given.
2224
2225 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2226 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2227 make it do so. The string so far is captured, making this feature
2228 compatible with Perl.
2229
2230 23. The tests have been re-organized, adding tests 11 and 12, to make it
2231 possible to check the Perl 5.10 features against Perl 5.10.
2232
2233 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2234 pattern matches a fixed length string. PCRE did not allow this; now it
2235 does. Neither allows recursion.
2236
2237 25. I finally figured out how to implement a request to provide the minimum
2238 length of subject string that was needed in order to match a given pattern.
2239 (It was back references and recursion that I had previously got hung up
2240 on.) This code has now been added to pcre_study(); it finds a lower bound
2241 to the length of subject needed. It is not necessarily the greatest lower
2242 bound, but using it to avoid searching strings that are too short does give
2243 some useful speed-ups. The value is available to calling programs via
2244 pcre_fullinfo().
2245
2246 26. While implementing 25, I discovered to my embarrassment that pcretest had
2247 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2248 study optimizations had never been tested with that matching function.
2249 Oops. What is worse, even when it was passed study data, there was a bug in
2250 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2251 were also very few tests of studied patterns with pcre_dfa_exec().
2252
2253 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2254 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2255 on the other side of the coin, they are no longer allowed to have different
2256 names, because these cannot be distinguished in PCRE, and this has caused
2257 confusion. (This is a difference from Perl.)
2258
2259 28. When duplicate subpattern names are present (necessarily with different
2260 numbers, as required by 27 above), and a test is made by name in a
2261 conditional pattern, either for a subpattern having been matched, or for
2262 recursion in such a pattern, all the associated numbered subpatterns are
2263 tested, and the overall condition is true if the condition is true for any
2264 one of them. This is the way Perl works, and is also more like the way
2265 testing by number works.
2266
2267
2268 Version 7.9 11-Apr-09
2269 ---------------------
2270
2271 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2272 (pcretest), all targets were linked against these libraries. This included
2273 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2274 libraries. This caused unwanted dependencies to be created. This problem
2275 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2276 pcretest is linked with readline.
2277
2278 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2279 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2280 moved outside it again, because FALSE and TRUE are already defined in AIX,
2281 but BOOL is not.
2282
2283 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2284 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2285
2286 4. The pcregrep documentation said spaces were inserted as well as colons (or
2287 hyphens) following file names and line numbers when outputting matching
2288 lines. This is not true; no spaces are inserted. I have also clarified the
2289 wording for the --colour (or --color) option.
2290
2291 5. In pcregrep, when --colour was used with -o, the list of matching strings
2292 was not coloured; this is different to GNU grep, so I have changed it to be
2293 the same.
2294
2295 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2296 each matching line was coloured. Now it goes on to look for further matches
2297 of any of the test patterns, which is the same behaviour as GNU grep.
2298
2299 7. A pattern that could match an empty string could cause pcregrep to loop; it
2300 doesn't make sense to accept an empty string match in pcregrep, so I have
2301 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2302 seems to be how GNU grep behaves. [But see later change 40 for release
2303 8.33.]
2304
2305 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2306 start or after a newline", because the conditional assertion was not being
2307 correctly handled. The rule now is that both the assertion and what follows
2308 in the first alternative must satisfy the test.
2309
2310 9. If auto-callout was enabled in a pattern with a conditional group whose
2311 condition was an assertion, PCRE could crash during matching, both with
2312 pcre_exec() and pcre_dfa_exec().
2313
2314 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2315 used for matching.
2316
2317 11. Unicode property support in character classes was not working for
2318 characters (bytes) greater than 127 when not in UTF-8 mode.
2319
2320 12. Added the -M command line option to pcretest.
2321
2322 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2323
2324 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2325
2326 16. Added comments and documentation about mis-use of no_arg in the C++
2327 wrapper.
2328
2329 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2330 from Martin Jerabek that uses macro names for all relevant character and
2331 string constants.
2332
2333 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2334 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2335 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2336 these, but not everybody uses configure.
2337
2338 19. A conditional group that had only one branch was not being correctly
2339 recognized as an item that could match an empty string. This meant that an
2340 enclosing group might also not be so recognized, causing infinite looping
2341 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2342 with the subject "ab", where knowledge that the repeated group can match
2343 nothing is needed in order to break the loop.
2344
2345 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2346 exec(), but without supplying a callout function, matching went wrong.
2347
2348 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2349 leak if the size of the offset vector was greater than 30. When the vector
2350 is smaller, the saved offsets during recursion go onto a local stack
2351 vector, but for larger vectors malloc() is used. It was failing to free
2352 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2353 error, in fact).
2354
2355 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2356 heapframe that is used only when UTF-8 support is enabled. This caused no
2357 problem, but was untidy.
2358
2359 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2360 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2361 included within another project.
2362
2363 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2364 slightly modified by me:
2365
2366 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2367 not building pcregrep.
2368
2369 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2370 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2371
2372 25. Forward references, both numeric and by name, in patterns that made use of
2373 duplicate group numbers, could behave incorrectly or give incorrect errors,
2374 because when scanning forward to find the reference group, PCRE was not
2375 taking into account the duplicate group numbers. A pattern such as
2376 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2377
2378 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2379 the feature of a custom pointer more persuasive (as requested by a user).
2380
2381 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2382 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2383 pre-defined.
2384
2385 28. Added support for (*UTF8) at the start of a pattern.
2386
2387 29. Arrange for flags added by the "release type" setting in CMake to be shown
2388 in the configuration summary.
2389
2390
2391 Version 7.8 05-Sep-08
2392 ---------------------
2393
2394 1. Replaced UCP searching code with optimized version as implemented for Ad
2395 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2396 stage table and inline lookup instead of a function, giving speed ups of 2
2397 to 5 times on some simple patterns that I tested. Permission was given to
2398 distribute the MultiStage2.py script that generates the tables (it's not in
2399 the tarball, but is in the Subversion repository).
2400
2401 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2402 scripts.
2403
2404 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2405 a group with a zero qualifier. The result of the study could be incorrect,
2406 or the function might crash, depending on the pattern.
2407
2408 4. Caseless matching was not working for non-ASCII characters in back
2409 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2410 It now works when Unicode Property Support is available.
2411
2412 5. In pcretest, an escape such as \x{de} in the data was always generating
2413 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2414 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2415 truncation.
2416
2417 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2418
2419 7. Added two (int) casts to pcregrep when printing the difference of two
2420 pointers, in case they are 64-bit values.
2421
2422 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2423 test 2 if it fails.
2424
2425 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2426 and a #define of that name to empty if it is not externally set. This is to
2427 allow users of MSVC to set it if necessary.
2428
2429 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2430 the convenience functions in the pcre_get.c source file.
2431
2432 11. An option change at the start of a pattern that had top-level alternatives
2433 could cause overwriting and/or a crash. This command provoked a crash in
2434 some environments:
2435
2436 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2437
2438 This potential security problem was recorded as CVE-2008-2371.
2439
2440 12. For a pattern where the match had to start at the beginning or immediately
2441 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2442 pcre_dfa_exec() could read past the end of the passed subject if there was
2443 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2444 pcretest so that it places the subject at the end of its malloc-ed buffer.
2445
2446 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2447 exec() might read past the end of the data buffer in UTF-8 mode.
2448
2449 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2450 the data contained the byte 0x85 as part of a UTF-8 character within its
2451 first line. This applied both to normal and DFA matching.
2452
2453 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2454 /^[^d]*?$/8 failed to match "abc".
2455
2456 16. Added a missing copyright notice to pcrecpp_internal.h.
2457
2458 17. Make it more clear in the documentation that values returned from
2459 pcre_exec() in ovector are byte offsets, not character counts.
2460
2461 18. Tidied a few places to stop certain compilers from issuing warnings.
2462
2463 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2464 supplied by Stefan Weber. I made a further small update for 7.8 because
2465 there is a change of source arrangements: the pcre_searchfuncs.c module is
2466 replaced by pcre_ucd.c.
2467
2468
2469 Version 7.7 07-May-08
2470 ---------------------
2471
2472 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2473 a string to a long long, pretend we don't even have a long long." This is
2474 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2475
2476 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2477 pre-7.6 versions, which defined a global no_arg variable instead of putting
2478 it in the RE class. (See also #8 below.)
2479
2480 3. Remove a line of dead code, identified by coverity and reported by Nuno
2481 Lopes.
2482
2483 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2484
2485 (1) The include/exclude patterns were being applied to the whole pathnames
2486 of files, instead of just to the final components.
2487
2488 (2) If there was more than one level of directory, the subdirectories were
2489 skipped unless they satisfied the include/exclude conditions. This is
2490 inconsistent with GNU grep (and could even be seen as contrary to the
2491 pcregrep specification - which I improved to make it absolutely clear).
2492 The action now is always to scan all levels of directory, and just
2493 apply the include/exclude patterns to regular files.
2494
2495 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2496 --exclude_dir in the tests to avoid scanning .svn directories.
2497
2498 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2499 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2500 doesn't support NULs in patterns.
2501
2502 7. Added some missing "const"s to declarations of static tables in
2503 pcre_compile.c and pcre_dfa_exec.c.
2504
2505 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2506 caused by fix #2 above. (Subsequently also a second patch to fix the
2507 first patch. And a third patch - this was a messy problem.)
2508
2509 9. Applied Craig's patch to remove the use of push_back().
2510
2511 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2512 matching function regexec().
2513
2514 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2515 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2516 references. PCRE supports relative numbers with this syntax (I don't think
2517 Oniguruma does).
2518
2519 12. Previously, a group with a zero repeat such as (...){0} was completely
2520 omitted from the compiled regex. However, this means that if the group
2521 was called as a subroutine from elsewhere in the pattern, things went wrong
2522 (an internal error was given). Such groups are now left in the compiled
2523 pattern, with a new opcode that causes them to be skipped at execution
2524 time.
2525
2526 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2527 to the way PCRE behaves:
2528
2529 (a) A lone ] character is dis-allowed (Perl treats it as data).
2530
2531 (b) A back reference to an unmatched subpattern matches an empty string
2532 (Perl fails the current match path).
2533
2534 (c) A data ] in a character class must be notated as \] because if the
2535 first data character in a class is ], it defines an empty class. (In
2536 Perl it is not possible to have an empty class.) The empty class []
2537 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2538 The negative empty class [^] matches any one character, independently
2539 of the DOTALL setting.
2540
2541 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2542 non-existent subpattern following a character class starting with ']' and
2543 containing () gave an internal compiling error instead of "reference to
2544 non-existent subpattern". Fortunately, when the pattern did exist, the
2545 compiled code was correct. (When scanning forwards to check for the
2546 existence of the subpattern, it was treating the data ']' as terminating
2547 the class, so got the count wrong. When actually compiling, the reference
2548 was subsequently set up correctly.)
2549
2550 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2551 it was being rejected as not supported by pcre_dfa_exec(), even though
2552 other assertions are supported. I have made pcre_dfa_exec() support
2553 (*FAIL).
2554
2555 16. The implementation of 13c above involved the invention of a new opcode,
2556 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2557 cannot be changed at match time, I realized I could make a small
2558 improvement to matching performance by compiling OP_ALLANY instead of
2559 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2560 on the OP_ANY path.
2561
2562 17. Compiling pcretest on Windows with readline support failed without the
2563 following two fixes: (1) Make the unistd.h include conditional on
2564 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2565
2566 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2567 ncurses library to be included for pcretest when ReadLine support is
2568 requested, but also to allow for it to be overridden. This patch came from
2569 Daniel Bergström.
2570
2571 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2572 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2573 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2574 spotting this.
2575
2576
2577 Version 7.6 28-Jan-08
2578 ---------------------
2579
2580 1. A character class containing a very large number of characters with
2581 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2582 overflow.
2583
2584 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2585 HAVE_LONG_LONG is not defined.
2586
2587 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2588 bring them up to date and include new features. This patch includes:
2589
2590 - Fixed PH's badly added libz and libbz2 support.
2591 - Fixed a problem with static linking.
2592 - Added pcredemo. [But later removed - see 7 below.]
2593 - Fixed dftables problem and added an option.
2594 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2595 HAVE_LONG_LONG.
2596 - Added readline support for pcretest.
2597 - Added an listing of the option settings after cmake has run.
2598
2599 4. A user submitted a patch to Makefile that makes it easy to create
2600 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2601 Makefile.am that cause it to include this special target, without
2602 affecting anything else. Note that the same mingw target plus all
2603 the other distribution libraries and programs are now supported
2604 when configuring with CMake (see 6 below) instead of with
2605 Configure/Make.
2606
2607 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2608 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2609 exported in the Windows port". It has not yet been confirmed that the patch
2610 solves the problem, but it does no harm.
2611
2612 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2613 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2614 with CMake, and also correct the comment about stack recursion.
2615
2616 7. Remove the automatic building of pcredemo from the ./configure system and
2617 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2618 of a program that users should build themselves after PCRE is installed, so
2619 building it automatically is not really right. What is more, it gave
2620 trouble in some build environments.
2621
2622 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2623
2624
2625 Version 7.5 10-Jan-08
2626 ---------------------
2627
2628 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2629 values in parens when parsing an RE using the C++ wrapper."
2630
2631 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2632 Characters greater than 255 were excluded from the class instead of being
2633 included.
2634
2635 3. The same bug as (2) above applied to negated POSIX classes such as
2636 [:^space:].
2637
2638 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2639 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2640 I have changed it.
2641
2642 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2643 first named subpattern) and a construct such as (?&a) would reference the
2644 first named subpattern whose name started with "a" (in other words, the
2645 length check was missing). Both these problems are fixed. "Subpattern name
2646 expected" is now given for (?&) (a zero-length name), and this patch also
2647 makes it give the same error for \k'' (previously it complained that that
2648 was a reference to a non-existent subpattern).
2649
2650 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2651 this is right because (?- can be followed by option settings as well as by
2652 digits. I have, however, made the messages clearer.
2653
2654 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2655 than the number used in the conditional) now cause a compile-time error.
2656 This is actually not compatible with Perl, which accepts such patterns, but
2657 treats the conditional as always being FALSE (as PCRE used to), but it
2658 seems to me that giving a diagnostic is better.
2659
2660 8. Change "alphameric" to the more common word "alphanumeric" in comments
2661 and messages.
2662
2663 9. Fix two occurrences of "backslash" in comments that should have been
2664 "backspace".
2665
2666 10. Remove two redundant lines of code that can never be obeyed (their function
2667 was moved elsewhere).
2668
2669 11. The program that makes PCRE's Unicode character property table had a bug
2670 which caused it to generate incorrect table entries for sequences of
2671 characters that have the same character type, but are in different scripts.
2672 It amalgamated them into a single range, with the script of the first of
2673 them. In other words, some characters were in the wrong script. There were
2674 thirteen such cases, affecting characters in the following ranges:
2675
2676 U+002b0 - U+002c1
2677 U+0060c - U+0060d
2678 U+0061e - U+00612
2679 U+0064b - U+0065e
2680 U+0074d - U+0076d
2681 U+01800 - U+01805
2682 U+01d00 - U+01d77
2683 U+01d9b - U+01dbf
2684 U+0200b - U+0200f
2685 U+030fc - U+030fe
2686 U+03260 - U+0327f
2687 U+0fb46 - U+0fbb1
2688 U+10450 - U+1049d
2689
2690 12. The -o option (show only the matching part of a line) for pcregrep was not
2691 compatible with GNU grep in that, if there was more than one match in a
2692 line, it showed only the first of them. It now behaves in the same way as
2693 GNU grep.
2694
2695 13. If the -o and -v options were combined for pcregrep, it printed a blank
2696 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2697 does the same. The return code can be used to tell if there were any
2698 non-matching lines.
2699
2700 14. Added --file-offsets and --line-offsets to pcregrep.
2701
2702 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2703 infinitely looping recursion. The bug was that positive lookaheads were not
2704 being skipped when checking for a possible empty match (negative lookaheads
2705 and both kinds of lookbehind were skipped).
2706
2707 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2708 inclusion of <windows.h> to before rather than after the definition of
2709 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2710
2711 17. Specifying a possessive quantifier with a specific limit for a Unicode
2712 character property caused pcre_compile() to compile bad code, which led at
2713 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2714 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2715 caused the error; without that there was no problem.
2716
2717 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2718
2719 19. Added --enable-pcretest-libreadline.
2720
2721 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2722 RE::GlobalReplace(). As a result, the number of replacements returned was
2723 double what it should be. I removed one of the increments, but Craig sent a
2724 later patch that removed the other one (the right fix) and added unit tests
2725 that check the return values (which was not done before).
2726
2727 21. Several CMake things:
2728
2729 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2730 the names libpcre and libpcreposix, not just pcre and pcreposix.
2731
2732 (2) The above change means that pcretest and pcregrep are now correctly
2733 linked with the newly-built libraries, not previously installed ones.
2734
2735 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2736
2737 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2738 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2739 UTF-8 newline character). The key issue is that the pattern starts .*;
2740 this means that the match must be either at the beginning, or after a
2741 newline. The bug was in the code for advancing after a failed match and
2742 checking that the new position followed a newline. It was not taking
2743 account of UTF-8 characters correctly.
2744
2745 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2746 character classes. PCRE was not treating the sequence [:...:] as a
2747 character class unless the ... were all letters. Perl, however, seems to
2748 allow any characters between [: and :], though of course it rejects as
2749 unknown any "names" that contain non-letters, because all the known class
2750 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2751 for example, whereas PCRE did not - it did not recognize a POSIX character
2752 class. This seemed a bit dangerous, so the code has been changed to be
2753 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2754 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2755 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2756 Perl does, and where it didn't before.
2757
2758 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2759 Windows environments %n is disabled by default.
2760
2761
2762 Version 7.4 21-Sep-07
2763 ---------------------
2764
2765 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2766 means that a class such as [\s] counted as "explicit reference to CR or
2767 LF". That isn't really right - the whole point of the change was to try to
2768 help when there was an actual mention of one of the two characters. So now
2769 the change happens only if \r or \n (or a literal CR or LF) character is
2770 encountered.
2771
2772 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2773 of both had grown to the point where there were only 3 bits left.
2774 Fortunately, there was spare space in the data structure, and so I have
2775 moved the internal flags into a new 16-bit field to free up more option
2776 bits.
2777
2778 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2779 but did not set the internal JCHANGED flag - either of these is enough to
2780 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2781 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2782 start sets both bits.
2783
2784 4. Added options (at build time, compile time, exec time) to change \R from
2785 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2786
2787 5. doc/pcresyntax.html was missing from the distribution.
2788
2789 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2790 compatibility, even though it is no longer used.
2791
2792 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2793 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2794 windows.h file is present (where different names are used). [This was
2795 reversed later after testing - see 16 below.]
2796
2797 8. Changed all #include <config.h> to #include "config.h". There were also
2798 some further <pcre.h> cases that I changed to "pcre.h".
2799
2800 9. When pcregrep was used with the --colour option, it missed the line ending
2801 sequence off the lines that it output.
2802
2803 10. It was pointed out to me that arrays of string pointers cause lots of
2804 relocations when a shared library is dynamically loaded. A technique of
2805 using a single long string with a table of offsets can drastically reduce
2806 these. I have refactored PCRE in four places to do this. The result is
2807 dramatic:
2808
2809 Originally: 290
2810 After changing UCP table: 187
2811 After changing error message table: 43
2812 After changing table of "verbs" 36
2813 After changing table of Posix names 22
2814
2815 Thanks to the folks working on Gregex for glib for this insight.
2816
2817 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2818 unicode-properties was also set.
2819
2820 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2821
2822 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2823 checked only for CRLF.
2824
2825 14. Added casts to pcretest.c to avoid compiler warnings.
2826
2827 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2828
2829 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2830 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2831 entirely. This removes changes made in 7 above.
2832
2833 17. The CMake files have been updated, and there is now more information about
2834 building with CMake in the NON-UNIX-USE document.
2835
2836
2837 Version 7.3 28-Aug-07
2838 ---------------------
2839
2840 1. In the rejigging of the build system that eventually resulted in 7.1, the
2841 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2842 brackets there is not right, since it causes compilers to look for an
2843 installed pcre.h, not the version that is in the source that is being
2844 compiled (which of course may be different). I have changed it back to:
2845
2846 #include "pcre.h"
2847
2848 I have a vague recollection that the change was concerned with compiling in
2849 different directories, but in the new build system, that is taken care of
2850 by the VPATH setting the Makefile.
2851
2852 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2853 when the subject happened to end in the byte 0x85 (e.g. if the last
2854 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2855 characters but of course it shouldn't be taken as a newline when it is part
2856 of another character. The bug was that, for an unlimited repeat of . in
2857 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2858 characters when looking for a newline.
2859
2860 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2861
2862 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2863 in debug output.
2864
2865 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2866 long printing in the pcrecpp unittest when running under MinGW.
2867
2868 6. ESC_K was left out of the EBCDIC table.
2869
2870 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2871 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2872 limit also applies to "virtual nesting" when a pattern is recursive, and in
2873 this case 1000 isn't so big. I have been able to remove this limit at the
2874 expense of backing off one optimization in certain circumstances. Normally,
2875 when pcre_exec() would call its internal match() function recursively and
2876 immediately return the result unconditionally, it uses a "tail recursion"
2877 feature to save stack. However, when a subpattern that can match an empty
2878 string has an unlimited repetition quantifier, it no longer makes this
2879 optimization. That gives it a stack frame in which to save the data for
2880 checking that an empty string has been matched. Previously this was taken
2881 from the 1000-entry workspace that had been reserved. So now there is no
2882 explicit limit, but more stack is used.
2883
2884 8. Applied Daniel's patches to solve problems with the import/export magic
2885 syntax that is required for Windows, and which was going wrong for the
2886 pcreposix and pcrecpp parts of the library. These were overlooked when this
2887 problem was solved for the main library.
2888
2889 9. There were some crude static tests to avoid integer overflow when computing
2890 the size of patterns that contain repeated groups with explicit upper
2891 limits. As the maximum quantifier is 65535, the maximum group length was
2892 set at 30,000 so that the product of these two numbers did not overflow a
2893 32-bit integer. However, it turns out that people want to use groups that
2894 are longer than 30,000 bytes (though not repeat them that many times).
2895 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2896 made it possible to implement the integer overflow checks in a much more
2897 dynamic way, which I have now done. The artificial limitation on group
2898 length has been removed - we now have only the limit on the total length of
2899 the compiled pattern, which depends on the LINK_SIZE setting.
2900
2901 10. Fixed a bug in the documentation for get/copy named substring when
2902 duplicate names are permitted. If none of the named substrings are set, the
2903 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2904 empty string.
2905
2906 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2907 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2908 because the ] is interpreted as the first data character and the
2909 terminating ] is not found. PCRE has been made compatible with Perl in this
2910 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2911 cause memory overwriting.
2912
2913 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2914 string has been matched (to stop an infinite loop). It was not recognizing
2915 a conditional subpattern that could match an empty string if that
2916 subpattern was within another subpattern. For example, it looped when
2917 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2918 condition was not nested. This bug has been fixed.
2919
2920 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2921 past the start of the subject in the presence of bytes with the top bit
2922 set, for example "\x8aBCD".
2923
2924 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2925 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2926
2927 14. Optimized (?!) to (*FAIL).
2928
2929 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2930 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2931 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2932 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2933 does: it's just the validity check that is more restrictive.
2934
2935 16. Inserted checks for integer overflows during escape sequence (backslash)
2936 processing, and also fixed erroneous offset values for syntax errors during
2937 backslash processing.
2938
2939 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2940 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2941
2942 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2943 caused an overrun.
2944
2945 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2946 something other than just ASCII characters) inside a group that had an
2947 unlimited repeat caused a loop at compile time (while checking to see
2948 whether the group could match an empty string).
2949
2950 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2951 [\P{Any}] did so. (Error in the code for printing property names.)
2952
2953 21. An orphan \E inside a character class could cause a crash.
2954
2955 22. A repeated capturing bracket such as (A)? could cause a wild memory
2956 reference during compilation.
2957
2958 23. There are several functions in pcre_compile() that scan along a compiled
2959 expression for various reasons (e.g. to see if it's fixed length for look
2960 behind). There were bugs in these functions when a repeated \p or \P was
2961 present in the pattern. These operators have additional parameters compared
2962 with \d, etc, and these were not being taken into account when moving along
2963 the compiled data. Specifically:
2964
2965 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2966 length.
2967
2968 (b) An item such as \pL+ within a repeated group could cause crashes or
2969 loops.
2970
2971 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2972 "reference to non-existent subpattern" error.
2973
2974 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2975
2976 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2977 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2978
2979 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2980
2981 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2982 character were causing crashes (broken optimization).
2983
2984 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2985 \p or \P) caused a compile-time loop.
2986
2987 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2988 break. For example, the unstudied pattern [\r\n]A does not match the string
2989 "\r\nA" because change 7.0/46 below moves the current point on by two
2990 characters after failing to match at the start. However, the pattern \nA
2991 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2992 the same is true. There doesn't seem any very clean way out of this, but
2993 what I have chosen to do makes the common cases work: PCRE now takes note
2994 of whether there can be an explicit match for \r or \n anywhere in the
2995 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2996 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2997 pattern has explicit CR or LF references.
2998
2999 29. Added (*CR) etc for changing newline setting at start of pattern.
3000
3001
3002 Version 7.2 19-Jun-07
3003 ---------------------
3004
3005 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3006 which is apparently normally available under Windows.
3007
3008 2. Re-jig the pcregrep tests with different newline settings in an attempt
3009 to make them independent of the local environment's newline setting.
3010
3011 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3012
3013 4. Some of the "internals" tests were previously cut out when the link size
3014 was not 2, because the output contained actual offsets. The recent new
3015 "Z" feature of pcretest means that these can be cut out, making the tests
3016 usable with all link sizes.
3017
3018 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3019 stack recursion. This gives a massive performance boost under BSD, but just
3020 a small improvement under Linux. However, it saves one field in the frame
3021 in all cases.
3022
3023 6. Added more features from the forthcoming Perl 5.10:
3024
3025 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3026 recursion call. It refers to the nth most recently opened parentheses.
3027
3028 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3029 to be opened parentheses.
3030
3031 (c) Conditions that refer to capturing parentheses can be specified
3032 relatively, for example, (?(-2)... or (?(+3)...
3033
3034 (d) \K resets the start of the current match so that everything before
3035 is not part of it.
3036
3037 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3038
3039 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3040 reference syntax.
3041
3042 (g) (?| introduces a group in which the numbering of parentheses in each
3043 alternative starts with the same number.
3044
3045 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3046
3047 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3048 PCRE_INFO_JCHANGED.
3049
3050 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3051 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3052 for detecting groups that can match an empty string.
3053
3054 9. A pattern with a very large number of alternatives (more than several
3055 hundred) was running out of internal workspace during the pre-compile
3056 phase, where pcre_compile() figures out how much memory will be needed. A
3057 bit of new cunning has reduced the workspace needed for groups with
3058 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3059 workspace instead of running out of the 4096 that are available.
3060
3061 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3062
3063 11. Applied patch from Google to remove an optimization that didn't quite work.
3064 The report of the bug said:
3065
3066 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3067 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3068 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3069
3070 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3071 it matched the wrong number of bytes.
3072
3073
3074 Version 7.1 24-Apr-07
3075 ---------------------
3076
3077 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3078 that is more "standard", making use of automake and other Autotools. There
3079 is some re-arrangement of the files and adjustment of comments consequent
3080 on this.
3081
3082 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3083 for recursive directory scanning broke on some systems because the files
3084 are not scanned in any specific order and on different systems the order
3085 was different. A call to "sort" has been inserted into RunGrepTest for the
3086 approprate test as a short-term fix. In the longer term there may be an
3087 alternative.
3088
3089 3. I had an email from Eric Raymond about problems translating some of PCRE's
3090 man pages to HTML (despite the fact that I distribute HTML pages, some
3091 people do their own conversions for various reasons). The problems
3092 concerned the use of low-level troff macros .br and .in. I have therefore
3093 removed all such uses from the man pages (some were redundant, some could
3094 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3095 HTML has been updated to handle .nf/.fi and to complain if it encounters
3096 .br or .in.
3097
3098 4. Updated comments in configure.ac that get placed in config.h.in and also
3099 arranged for config.h to be included in the distribution, with the name
3100 config.h.generic, for the benefit of those who have to compile without
3101 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3102
3103 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3104 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3105 makevp.bat for the current PCRE, using the additional files
3106 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3107
3108 6. A Windows user reported a minor discrepancy with test 2, which turned out
3109 to be caused by a trailing space on an input line that had got lost in his
3110 copy. The trailing space was an accident, so I've just removed it.
3111
3112 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3113 that is needed.
3114
3115 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3116 as "const" (a) because they are and (b) because it helps the PHP
3117 maintainers who have recently made a script to detect big data structures
3118 in the php code that should be moved to the .rodata section. I remembered
3119 to update Builducptable as well, so it won't revert if ucptable.h is ever
3120 re-created.
3121
3122 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3123 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3124 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3125 support is not required. This saves 1.5-2K of code, which is important in
3126 some applications.
3127
3128 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3129 so as not to refer to the tables, even though these functions will never be
3130 called when UTF-8 support is disabled. Otherwise there are problems with a
3131 shared library.
3132
3133 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3134
3135 (a) It was defining its arguments as char * instead of void *.
3136
3137 (b) It was assuming that all moves were upwards in memory; this was true
3138 a long time ago when I wrote it, but is no longer the case.
3139
3140 The emulated memove() is provided for those environments that have neither
3141 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3142 is clearly not the case, as these two bugs were recently reported.
3143
3144 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3145 and Detrail to create the HTML documentation, the .txt form of the man
3146 pages, and it removes trailing spaces from listed files. It also creates
3147 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3148 case, it wraps all the #defines with #ifndefs. This script should be run
3149 before "make dist".
3150
3151 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3152 with Unicode property support.
3153
3154 (a) For a maximizing quantifier, if the two different cases of the
3155 character were of different lengths in their UTF-8 codings (there are
3156 some cases like this - I found 11), and the matching function had to
3157 back up over a mixture of the two cases, it incorrectly assumed they
3158 were both the same length.
3159
3160 (b) When PCRE was configured to use the heap rather than the stack for
3161 recursion during matching, it was not correctly preserving the data for
3162 the other case of a UTF-8 character when checking ahead for a match
3163 while processing a minimizing repeat. If the check also involved
3164 matching a wide character, but failed, corruption could cause an
3165 erroneous result when trying to check for a repeat of the original
3166 character.
3167
3168 13. Some tidying changes to the testing mechanism:
3169
3170 (a) The RunTest script now detects the internal link size and whether there
3171 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3172 values substituted by "configure". (The RunGrepTest script already did
3173 this for UTF-8.) The configure.ac script no longer substitutes the
3174 relevant variables.
3175
3176 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3177 with length and offset values. This means that the output is different
3178 for different internal link sizes. Test 2 is skipped for link sizes
3179 other than 2 because of this, bypassing the problem. Unfortunately,
3180 there was also a test in test 3 (the locale tests) that used /B and
3181 failed for link sizes other than 2. Rather than cut the whole test out,
3182 I have added a new /Z option to pcretest that replaces the length and
3183 offset values with spaces. This is now used to make test 3 independent
3184 of link size. (Test 2 will be tidied up later.)
3185
3186 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3187 segmentation fault instead of returning the appropriate error message.
3188
3189 15. In multiline mode when the newline sequence was set to "any", the pattern
3190 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3191 This doesn't seem right; it now treats the CRLF combination as the line
3192 ending, and so does not match in that case. It's only a pattern such as ^$
3193 that would hit this one: something like ^ABC$ would have failed after \r
3194 and then tried again after \r\n.
3195
3196 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3197 in an attempt to make files that differ only in their line terminators
3198 compare equal. This works on Linux.
3199
3200 17. Under certain error circumstances pcregrep might try to free random memory
3201 as it exited. This is now fixed, thanks to valgrind.
3202
3203 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3204 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3205 was because its rules for how to advance for /g after matching an empty
3206 string at the end of a line did not allow for this case. They now check for
3207 it specially.
3208
3209 20. pcretest is supposed to handle patterns and data of any length, by
3210 extending its buffers when necessary. It was getting this wrong when the
3211 buffer for a data line had to be extended.
3212
3213 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3214 CRLF as a newline sequence.
3215
3216 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3217 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3218 I have nevertheless tidied it up.
3219
3220 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3221
3222 24. Added a man page for pcre-config.
3223
3224
3225 Version 7.0 19-Dec-06
3226 ---------------------
3227
3228 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3229 moving to gcc 4.1.1.
3230
3231 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3232 sys/time.h, which is documented as needed for this function. It doesn't
3233 seem to matter on Linux, but it showed up on some releases of OS X.
3234
3235 3. It seems that there are systems where bytes whose values are greater than
3236 127 match isprint() in the "C" locale. The "C" locale should be the
3237 default when a C program starts up. In most systems, only ASCII printing
3238 characters match isprint(). This difference caused the output from pcretest
3239 to vary, making some of the tests fail. I have changed pcretest so that:
3240
3241 (a) When it is outputting text in the compiled version of a pattern, bytes
3242 other than 32-126 are always shown as hex escapes.
3243
3244 (b) When it is outputting text that is a matched part of a subject string,
3245 it does the same, unless a different locale has been set for the match
3246 (using the /L modifier). In this case, it uses isprint() to decide.
3247
3248 4. Fixed a major bug that caused incorrect computation of the amount of memory
3249 required for a compiled pattern when options that changed within the
3250 pattern affected the logic of the preliminary scan that determines the
3251 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3252 that the computed length was too small. The symptoms of this bug were
3253 either the PCRE error "internal error: code overflow" from pcre_compile(),
3254 or a glibc crash with a message such as "pcretest: free(): invalid next
3255 size (fast)". Examples of patterns that provoked this bug (shown in
3256 pcretest format) are:
3257
3258 /(?-x: )/x
3259 /(?x)(?-x: \s*#\s*)/
3260 /((?i)[\x{c0}])/8
3261 /(?i:[\x{c0}])/8
3262
3263 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3264 is now done differently.
3265
3266 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3267 wrapper classes; (b) implement a new function in the C++ scanner that is
3268 more efficient than the old way of doing things because it avoids levels of
3269 recursion in the regex matching; (c) add a paragraph to the documentation
3270 for the FullMatch() function.
3271
3272 6. The escape sequence \n was being treated as whatever was defined as
3273 "newline". Not only was this contrary to the documentation, which states
3274 that \n is character 10 (hex 0A), but it also went horribly wrong when
3275 "newline" was defined as CRLF. This has been fixed.
3276
3277 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3278 was being set to -1 for the "end of line" case (supposedly a value that no
3279 character can have). Though this value is never used (the check for end of
3280 line is "zero bytes in current character"), it caused compiler complaints.
3281 I've changed it to 0xffffffff.
3282
3283 8. In pcre_version.c, the version string was being built by a sequence of
3284 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3285 string (as it is for production releases) called a macro with an empty
3286 argument. The C standard says the result of this is undefined. The gcc
3287 compiler treats it as an empty string (which was what was wanted) but it is
3288 reported that Visual C gives an error. The source has been hacked around to
3289 avoid this problem.
3290
3291 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3292 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3293 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3294 of them did).
3295
3296 10. Originally, pcretest opened its input and output without "b"; then I was
3297 told that "b" was needed in some environments, so it was added for release
3298 5.0 to both the input and output. (It makes no difference on Unix-like
3299 systems.) Later I was told that it is wrong for the input on Windows. I've
3300 now abstracted the modes into two macros, to make it easier to fiddle with
3301 them, and removed "b" from the input mode under Windows.
3302
3303 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3304
3305 12. Added -help and --help to pcretest as an official way of being reminded
3306 of the options.
3307
3308 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3309 and pcrecpp.cc because they annoy compilers at high warning levels.
3310
3311 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3312
3313 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3314 scripts are not C programs :-) and which was not noticed because it works
3315 on Linux.
3316
3317 16. pcretest is supposed to handle any length of pattern and data line (as one
3318 line or as a continued sequence of lines) by extending its input buffer if
3319 necessary. This feature was broken for very long pattern lines, leading to
3320 a string of junk being passed to pcre_compile() if the pattern was longer
3321 than about 50K.
3322
3323 17. I have done a major re-factoring of the way pcre_compile() computes the
3324 amount of memory needed for a compiled pattern. Previously, there was code
3325 that made a preliminary scan of the pattern in order to do this. That was
3326 OK when PCRE was new, but as the facilities have expanded, it has become
3327 harder and harder to keep it in step with the real compile phase, and there
3328 have been a number of bugs (see for example, 4 above). I have now found a
3329 cunning way of running the real compile function in a "fake" mode that
3330 enables it to compute how much memory it would need, while actually only
3331 ever using a few hundred bytes of working memory and without too many
3332 tests of the mode. This should make future maintenance and development
3333 easier. A side effect of this work is that the limit of 200 on the nesting
3334 depth of parentheses has been removed (though this was never a serious
3335 limitation, I suspect). However, there is a downside: pcre_compile() now
3336 runs more slowly than before (30% or more, depending on the pattern). I
3337 hope this isn't a big issue. There is no effect on runtime performance.
3338
3339 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3340 newline (only possible for the last line of a file) and it was a
3341 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3342
3343 19. Added additional timing features to pcretest. (1) The -tm option now times
3344 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3345 separate command line item, by a number that specifies the number of
3346 repeats to use when timing. The default is 50000; this gives better
3347 precision, but takes uncomfortably long for very large patterns.
3348
3349 20. Extended pcre_study() to be more clever in cases where a branch of a
3350 subpattern has no definite first character. For example, (a*|b*)[cd] would
3351 previously give no result from pcre_study(). Now it recognizes that the
3352 first character must be a, b, c, or d.
3353
3354 21. There was an incorrect error "recursive call could loop indefinitely" if
3355 a subpattern (or the entire pattern) that was being tested for matching an
3356 empty string contained only one non-empty item after a nested subpattern.
3357 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3358 incorrectly, because the \d was being skipped in the check.
3359
3360 22. The pcretest program now has a new pattern option /B and a command line
3361 option -b, which is equivalent to adding /B to every pattern. This causes
3362 it to show the compiled bytecode, without the additional information that
3363 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3364 is the same as /B/I).
3365
3366 23. A new optimization is now able automatically to treat some sequences such
3367 as a*b as a*+b. More specifically, if something simple (such as a character
3368 or a simple class like \d) has an unlimited quantifier, and is followed by
3369 something that cannot possibly match the quantified thing, the quantifier
3370 is automatically "possessified".
3371
3372 24. A recursive reference to a subpattern whose number was greater than 39
3373 went wrong under certain circumstances in UTF-8 mode. This bug could also
3374 have affected the operation of pcre_study().
3375
3376 25. Realized that a little bit of performance could be had by replacing
3377 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3378
3379 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3380
3381 27. Possessive quantifiers such as a++ were previously implemented by turning
3382 them into atomic groups such as ($>a+). Now they have their own opcodes,
3383 which improves performance. This includes the automatically created ones
3384 from 23 above.
3385
3386 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3387 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3388 the first matched character to be a colon. This applied both to named and
3389 numbered groups.
3390
3391 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3392
3393 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3394 building PCRE on Windows easier, so I have included it in the distribution.
3395
3396 31. There is now a check in pcretest against a ridiculously large number being
3397 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3398 loop, the loop is abandoned.
3399
3400 32. Forward references to subpatterns in conditions such as (?(2)...) where
3401 subpattern 2 is defined later cause pcre_compile() to search forwards in
3402 the pattern for the relevant set of parentheses. This search went wrong
3403 when there were unescaped parentheses in a character class, parentheses
3404 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3405
3406 33. "Subroutine" calls and backreferences were previously restricted to
3407 referencing subpatterns earlier in the regex. This restriction has now
3408 been removed.
3409
3410 34. Added a number of extra features that are going to be in Perl 5.10. On the
3411 whole, these are just syntactic alternatives for features that PCRE had
3412 previously implemented using the Python syntax or my own invention. The
3413 other formats are all retained for compatibility.
3414
3415 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3416 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3417 also .NET compatible.
3418
3419 (b) A recursion or subroutine call to a named group can now be defined as
3420 (?&name) as well as (?P>name).
3421
3422 (c) A backreference to a named group can now be defined as \k<name> or
3423 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3424 5.10, are also .NET compatible.
3425
3426 (d) A conditional reference to a named group can now use the syntax
3427 (?(<name>) or (?('name') as well as (?(name).
3428
3429 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3430 groups (named and numbered) that are never evaluated inline, but can be
3431 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3432 is always false. There may be only one alternative in such a group.
3433
3434 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3435 as the simple (?(R). The condition is true only if the most recent
3436 recursion is that of the given number or name. It does not search out
3437 through the entire recursion stack.
3438
3439 (g) The escape \gN or \g{N} has been added, where N is a positive or
3440 negative number, specifying an absolute or relative reference.
3441
3442 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3443 some "unreachable code" warnings.
3444
3445 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3446 things, this adds five new scripts.
3447
3448 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3449 There were also incompatibilities regarding the handling of \Q..\E inside
3450 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3451 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3452
3453 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3454 matches an empty string, and forcibly breaks the loop. There were bugs in
3455 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3456 against aaaa the result was just "a" rather than "aaaa", for example. Two
3457 separate and independent bugs (that affected different cases) have been
3458 fixed.
3459
3460 39. Refactored the code to abolish the use of different opcodes for small
3461 capturing bracket numbers. This is a tidy that I avoided doing when I
3462 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3463 The new approach is not only tidier, it makes it possible to reduce the
3464 memory needed to fix the previous bug (38).
3465
3466 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3467 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3468 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3469 mode.
3470
3471 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3472 report.
3473
3474 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3475 copy construction and assignment in the C++ wrapper.
3476
3477 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3478 couple of bugs that could have given wrong results in the "--newline=crlf"
3479 case.
3480
3481 44. Added a number of casts and did some reorganization of signed/unsigned int
3482 variables following suggestions from Dair Grant. Also renamed the variable
3483 "this" as "item" because it is a C++ keyword.
3484
3485 45. Arranged for dftables to add
3486
3487 #include "pcre_internal.h"
3488
3489 to pcre_chartables.c because without it, gcc 4.x may remove the array
3490 definition from the final binary if PCRE is built into a static library and
3491 dead code stripping is activated.
3492
3493 46. For an unanchored pattern, if a match attempt fails at the start of a
3494 newline sequence, and the newline setting is CRLF or ANY, and the next two
3495 characters are CRLF, advance by two characters instead of one.
3496
3497
3498 Version 6.7 04-Jul-06
3499 ---------------------
3500
3501 1. In order to handle tests when input lines are enormously long, pcretest has
3502 been re-factored so that it automatically extends its buffers when
3503 necessary. The code is crude, but this _is_ just a test program. The
3504 default size has been increased from 32K to 50K.
3505
3506 2. The code in pcre_study() was using the value of the re argument before
3507 testing it for NULL. (Of course, in any sensible call of the function, it
3508 won't be NULL.)
3509
3510 3. The memmove() emulation function in pcre_internal.h, which is used on
3511 systems that lack both memmove() and bcopy() - that is, hardly ever -
3512 was missing a "static" storage class specifier.
3513
3514 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3515 containing an extended class (one that cannot be represented by a bitmap
3516 because it contains high-valued characters or Unicode property items, e.g.
3517 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3518 pattern, but PCRE should not loop if you do not (it no longer does).
3519 [Detail: two cases were found: (a) a repeated subpattern containing an
3520 extended class; (b) a recursive reference to a subpattern that followed a
3521 previous extended class. It wasn't skipping over the extended class
3522 correctly when UTF-8 mode was not set.]
3523
3524 5. A negated single-character class was not being recognized as fixed-length
3525 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3526 compile error "lookbehind assertion is not fixed length".
3527
3528 6. The RunPerlTest auxiliary script was showing an unexpected difference
3529 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3530 write a Perl script that can interpret lines of an input file either as
3531 byte characters or as UTF-8, which is what "perltest" was being required to
3532 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3533 can't do is switch easily at run time between having the "use utf8;" pragma
3534 or not. In the end, I fudged it by using the RunPerlTest script to insert
3535 "use utf8;" explicitly for the UTF-8 tests.
3536
3537 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3538 the end of the subject string, contrary to the documentation and to what
3539 Perl does. This was true of both matching functions. Now it matches only at
3540 the start of the subject and immediately after *internal* newlines.
3541
3542 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3543 a pointer to an int instead of a pointer to an unsigned long int. This
3544 caused problems on 64-bit systems.
3545
3546 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3547 instance of the 'standard' template library not being so standard".
3548
3549 10. There was no check on the number of named subpatterns nor the maximum
3550 length of a subpattern name. The product of these values is used to compute
3551 the size of the memory block for a compiled pattern. By supplying a very
3552 long subpattern name and a large number of named subpatterns, the size
3553 computation could be caused to overflow. This is now prevented by limiting
3554 the length of names to 32 characters, and the number of named subpatterns
3555 to 10,000.
3556
3557 11. Subpatterns that are repeated with specific counts have to be replicated in
3558 the compiled pattern. The size of memory for this was computed from the
3559 length of the subpattern and the repeat count. The latter is limited to
3560 65535, but there was no limit on the former, meaning that integer overflow
3561 could in principle occur. The compiled length of a repeated subpattern is
3562 now limited to 30,000 bytes in order to prevent this.
3563
3564 12. Added the optional facility to have named substrings with the same name.
3565
3566 13. Added the ability to use a named substring as a condition, using the
3567 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3568 are numbers (not recommended). Forward references are permitted.
3569
3570 14. Added forward references in named backreferences (if you see what I mean).
3571
3572 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3573 pattern could run off the end of the subject. For example, the pattern
3574 "(?s)(.{1,5})"8 did this with the subject "ab".
3575
3576 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3577 PCRE_CASELESS was set when matching characters that were quantified with ?
3578 or *.
3579
3580 17. A character class other than a single negated character that had a minimum
3581 but no maximum quantifier - for example [ab]{6,} - was not handled
3582 correctly by pce_dfa_exec(). It would match only one character.
3583
3584 18. A valid (though odd) pattern that looked like a POSIX character
3585 class but used an invalid character after [ (for example [[,abc,]]) caused
3586 pcre_compile() to give the error "Failed: internal error: code overflow" or
3587 in some cases to crash with a glibc free() error. This could even happen if
3588 the pattern terminated after [[ but there just happened to be a sequence of
3589 letters, a binary zero, and a closing ] in the memory that followed.
3590
3591 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3592 over the years. Originally (before any Unicode support), just the bottom 8
3593 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3594 output from "man perlunicode" includes this:
3595
3596 The regular expression compiler produces polymorphic opcodes. That
3597 is, the pattern adapts to the data and automatically switches to
3598 the Unicode character scheme when presented with Unicode data--or
3599 instead uses a traditional byte scheme when presented with byte
3600 data.
3601
3602 Sadly, a wide octal escape does not cause a switch, and in a string with
3603 no other multibyte characters, these octal escapes are treated as before.
3604 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3605 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3606 Unicode string.
3607
3608 I have not perpetrated such confusion in PCRE. Up till now, it took just
3609 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3610 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3611 translate to the appropriate multibyte character.
3612
3613 29. Applied some refactoring to reduce the number of warnings from Microsoft
3614 and Borland compilers. This has included removing the fudge introduced
3615 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3616 a warning about an unused variable.
3617
3618 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3619 characters since release 4.0, because Perl (from release 5.004) does not.
3620 [Or at least, is documented not to: some releases seem to be in conflict
3621 with the documentation.] However, when a pattern was studied with
3622 pcre_study() and all its branches started with \s, PCRE still included VT
3623 as a possible starting character. Of course, this did no harm; it just
3624 caused an unnecessary match attempt.
3625
3626 22. Removed a now-redundant internal flag bit that recorded the fact that case
3627 dependency changed within the pattern. This was once needed for "required
3628 byte" processing, but is no longer used. This recovers a now-scarce options
3629 bit. Also moved the least significant internal flag bit to the most-
3630 significant bit of the word, which was not previously used (hangover from
3631 the days when it was an int rather than a uint) to free up another bit for
3632 the future.
3633
3634 23. Added support for CRLF line endings as well as CR and LF. As well as the
3635 default being selectable at build time, it can now be changed at runtime
3636 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3637 specify that it is scanning data with non-default line endings.
3638
3639 24. Changed the definition of CXXLINK to make it agree with the definition of
3640 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3641
3642 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3643 recursions. This makes a big different to stack usage for some patterns.
3644
3645 26. If a subpattern containing a named recursion or subroutine reference such
3646 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3647 the space required for the compiled pattern went wrong and gave too small a
3648 value. Depending on the environment, this could lead to "Failed: internal
3649 error: code overflow at offset 49" or "glibc detected double free or
3650 corruption" errors.
3651
3652 27. Applied patches from Google (a) to support the new newline modes and (b) to
3653 advance over multibyte UTF-8 characters in GlobalReplace.
3654
3655 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3656 difference for some implementation of PCRE in some Windows version.
3657
3658 29. Added some extra testing facilities to pcretest:
3659
3660 \q<number> in a data line sets the "match limit" value
3661 \Q<number> in a data line sets the "match recursion limt" value
3662 -S <number> sets the stack size, where <number> is in megabytes
3663
3664 The -S option isn't available for Windows.
3665
3666
3667 Version 6.6 06-Feb-06
3668 ---------------------
3669
3670 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3671 in pcreposix.h. I have copied the definition from pcre.h.
3672
3673 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3674 because pcre.h is no longer a built file.
3675
3676 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3677 not normally included in the compiled code.
3678
3679
3680 Version 6.5 01-Feb-06
3681 ---------------------
3682
3683 1. When using the partial match feature with pcre_dfa_exec(), it was not
3684 anchoring the second and subsequent partial matches at the new starting
3685 point. This could lead to incorrect results. For example, with the pattern
3686 /1234/, partially matching against "123" and then "a4" gave a match.
3687
3688 2. Changes to pcregrep:
3689
3690 (a) All non-match returns from pcre_exec() were being treated as failures
3691 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3692 error message is output. Some extra information is given for the
3693 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3694 probably the only errors that are likely to be caused by users (by
3695 specifying a regex that has nested indefinite repeats, for instance).
3696 If there are more than 20 of these errors, pcregrep is abandoned.
3697
3698 (b) A binary zero was treated as data while matching, but terminated the
3699 output line if it was written out. This has been fixed: binary zeroes
3700 are now no different to any other data bytes.
3701
3702 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3703 used to set a locale for matching. The --locale=xxxx long option has
3704 been added (no short equivalent) to specify a locale explicitly on the
3705 pcregrep command, overriding the environment variables.
3706
3707 (d) When -B was used with -n, some line numbers in the output were one less
3708 than they should have been.
3709
3710 (e) Added the -o (--only-matching) option.
3711
3712 (f) If -A or -C was used with -c (count only), some lines of context were
3713 accidentally printed for the final match.
3714
3715 (g) Added the -H (--with-filename) option.
3716
3717 (h) The combination of options -rh failed to suppress file names for files
3718 that were found from directory arguments.
3719
3720 (i) Added the -D (--devices) and -d (--directories) options.
3721
3722 (j) Added the -F (--fixed-strings) option.
3723
3724 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3725
3726 (l) Added the --colo(u)r option.
3727
3728 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3729 is not present by default.
3730
3731 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3732 items such as (?R) or (?1), when the recursion could match a number of
3733 alternatives. If it matched one of the alternatives, but subsequently,
3734 outside the recursion, there was a failure, the code tried to back up into
3735 the recursion. However, because of the way PCRE is implemented, this is not
3736 possible, and the result was an incorrect result from the match.
3737
3738 In order to prevent this happening, the specification of recursion has
3739 been changed so that all such subpatterns are automatically treated as
3740 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3741
3742 4. I had overlooked the fact that, in some locales, there are characters for
3743 which isalpha() is true but neither isupper() nor islower() are true. In
3744 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3745 and ordfeminine) are like this. This affected the treatment of \w and \W
3746 when they appeared in character classes, but not when they appeared outside
3747 a character class. The bit map for "word" characters is now created
3748 separately from the results of isalnum() instead of just taking it from the
3749 upper, lower, and digit maps. (Plus the underscore character, of course.)
3750
3751 5. The above bug also affected the handling of POSIX character classes such as
3752 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3753 permanent tables. Instead, the bit maps for such a class were previously
3754 created as the appropriate unions of the upper, lower, and digit bitmaps.
3755 Now they are created by subtraction from the [[:word:]] class, which has
3756 its own bitmap.
3757
3758 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3759 It is created by subtracting the vertical space characters (\x09, \x0a,
3760 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3761 subtraction was done in the overall bitmap for a character class, meaning
3762 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3763 be recognized. This bug has been fixed.
3764
3765 7. Patches from the folks at Google:
3766
3767 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3768 real life, but is still worth protecting against".
3769
3770 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3771 regular expressions".
3772
3773 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3774 have it.
3775
3776 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3777 "configure" and the latter not, in order to fix a problem somebody had
3778 with compiling the Arg class on HP-UX.
3779
3780 (e) Improve the error-handling of the C++ wrapper a little bit.
3781
3782 (f) New tests for checking recursion limiting.
3783
3784 8. The pcre_memmove() function, which is used only if the environment does not
3785 have a standard memmove() function (and is therefore rarely compiled),
3786 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3787 returning a result (though PCRE never actually uses the result).
3788
3789 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3790 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3791 returned instead of calling malloc() with an overflowing number that would
3792 most likely cause subsequent chaos.
3793
3794 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3795
3796 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3797 with this option is matched, the nmatch and pmatch options of regexec() are
3798 ignored.
3799
3800 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3801 provided in case anyone wants to the the POSIX interface with UTF-8
3802 strings.
3803
3804 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3805 C++ linking (needed for some HP-UX environments).
3806
3807 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3808 (unused parameter) and in the pcre_printint() function (omitted "default"
3809 switch label when the default is to do nothing).
3810
3811 15. Added some code to make it possible, when PCRE is compiled as a C++
3812 library, to replace subject pointers for pcre_exec() with a smart pointer
3813 class, thus making it possible to process discontinuous strings.
3814
3815 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3816 much the same function. They were added by different people who were trying
3817 to make PCRE easy to compile on non-Unix systems. It has been suggested
3818 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3819 for compiling on Windows systems. I have therefore replaced it with
3820 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3821 defaults to "extern" for C or "extern C" for C++, which works fine on
3822 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3823 SCOPE with something explicit in config.h. In addition:
3824
3825 (a) pcreposix.h still had just "extern" instead of either of these macros;
3826 I have replaced it with PCRE_DATA_SCOPE.
3827
3828 (b) Functions such as _pcre_xclass(), which are internal to the library,
3829 but external in the C sense, all had PCRE_EXPORT in their definitions.
3830 This is apparently wrong for the Windows case, so I have removed it.
3831 (It makes no difference on Unix-like systems.)
3832
3833 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3834 of recursive calls to match(). This is different to MATCH_LIMIT because
3835 that limits the total number of calls to match(), not all of which increase
3836 the depth of recursion. Limiting the recursion depth limits the amount of
3837 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3838 when PCRE is compiled, and changed at run time. A patch from Google adds
3839 this functionality to the C++ interface.
3840
3841 18. Changes to the handling of Unicode character properties:
3842
3843 (a) Updated the table to Unicode 4.1.0.
3844
3845 (b) Recognize characters that are not in the table as "Cn" (undefined).
3846
3847 (c) I revised the way the table is implemented to a much improved format
3848 which includes recognition of ranges. It now supports the ranges that
3849 are defined in UnicodeData.txt, and it also amalgamates other
3850 characters into ranges. This has reduced the number of entries in the
3851 table from around 16,000 to around 3,000, thus reducing its size
3852 considerably. I realized I did not need to use a tree structure after
3853 all - a binary chop search is just as efficient. Having reduced the
3854 number of entries, I extended their size from 6 bytes to 8 bytes to
3855 allow for more data.
3856
3857 (d) Added support for Unicode script names via properties such as \p{Han}.
3858
3859 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3860 matching that character.
3861
3862 20. When matching a repeated Unicode property with a minimum greater than zero,
3863 (for example \pL{2,}), PCRE could look past the end of the subject if it
3864 reached it while seeking the minimum number of characters. This could
3865 happen only if some of the characters were more than one byte long, because
3866 there is a check for at least the minimum number of bytes.
3867
3868 21. Refactored the implementation of \p and \P so as to be more general, to
3869 allow for more different types of property in future. This has changed the
3870 compiled form incompatibly. Anybody with saved compiled patterns that use
3871 \p or \P will have to recompile them.
3872
3873 22. Added "Any" and "L&" to the supported property types.
3874
3875 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3876 but give a compile time error if the value is greater than 0xff.
3877
3878 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3879 accidentally not being installed or uninstalled.
3880
3881 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3882 made were to insert the current release number. This seemed silly, because
3883 it made things harder for people building PCRE on systems that don't run
3884 "configure". I have turned pcre.h into a distributed file, no longer built
3885 by "configure", with the version identification directly included. There is
3886 no longer a pcre.h.in file.
3887
3888 However, this change necessitated a change to the pcre-config script as
3889 well. It is built from pcre-config.in, and one of the substitutions was the
3890 release number. I have updated configure.ac so that ./configure now finds
3891 the release number by grepping pcre.h.
3892
3893 26. Added the ability to run the tests under valgrind.
3894
3895
3896 Version 6.4 05-Sep-05
3897 ---------------------
3898
3899 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3900 "--" to be printed when multiple files were scanned, even when none of the
3901 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3902 consider it to be a bug, and have restored the previous behaviour.
3903
3904 2. A couple of code tidies to get rid of compiler warnings.
3905
3906 3. The pcretest program used to cheat by referring to symbols in the library
3907 whose names begin with _pcre_. These are internal symbols that are not
3908 really supposed to be visible externally, and in some environments it is
3909 possible to suppress them. The cheating is now confined to including
3910 certain files from the library's source, which is a bit cleaner.
3911
3912 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3913 file's purpose clearer.
3914
3915 5. Reorganized pcre_ucp_findchar().
3916
3917
3918 Version 6.3 15-Aug-05
3919 ---------------------
3920
3921 1. The file libpcre.pc.in did not have general read permission in the tarball.
3922
3923 2. There were some problems when building without C++ support:
3924
3925 (a) If C++ support was not built, "make install" and "make test" still
3926 tried to test it.
3927
3928 (b) There were problems when the value of CXX was explicitly set. Some
3929 changes have been made to try to fix these, and ...
3930
3931 (c) --disable-cpp can now be used to explicitly disable C++ support.
3932
3933 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3934 backslash in a target when C++ was disabled. This confuses some
3935 versions of "make", apparently. Using an intermediate variable solves
3936 this. (Same for CPP_LOBJ.)
3937
3938 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3939 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3940 necessary on certain architectures.
3941
3942 4. Added a setting of -export-symbols-regex to the link command to remove
3943 those symbols that are exported in the C sense, but actually are local
3944 within the library, and not documented. Their names all begin with
3945 "_pcre_". This is not a perfect job, because (a) we have to except some
3946 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3947 available (and never for static libraries). I have made a note to try to
3948 find a way round (a) in the future.
3949
3950
3951 Version 6.2 01-Aug-05
3952 ---------------------
3953
3954 1. There was no test for integer overflow of quantifier values. A construction
3955 such as {1111111111111111} would give undefined results. What is worse, if
3956 a minimum quantifier for a parenthesized subpattern overflowed and became
3957 negative, the calculation of the memory size went wrong. This could have
3958 led to memory overwriting.
3959
3960 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3961
3962 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3963 operating environments where this matters.
3964
3965 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3966 PCRE options from within the C++ wrapper.
3967
3968 5. Named capturing subpatterns were not being correctly counted when a pattern
3969 was compiled. This caused two problems: (a) If there were more than 100
3970 such subpatterns, the calculation of the memory needed for the whole
3971 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3972 back references of the form \12, where the number was greater than 9, were
3973 not recognized as back references, even though there were sufficient
3974 previous subpatterns.
3975
3976 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3977 versions of gcc, e.g. 2.95.4.
3978
3979
3980 Version 6.1 21-Jun-05
3981 ---------------------
3982
3983 1. There was one reference to the variable "posix" in pcretest.c that was not
3984 surrounded by "#if !defined NOPOSIX".
3985
3986 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3987 the cross-check on the old pcre_info() function, for the benefit of the
3988 cut-down version of PCRE that is currently imported into Exim.
3989
3990 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3991 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3992 patterns that start (?i) but I don't think that matters. The use of (?i) is
3993 just an example; this all applies to the other options as well.
3994
3995 4. Since libtool seems to echo the compile commands it is issuing, the output
3996 from "make" can be reduced a bit by putting "@" in front of each libtool
3997 compile command.
3998
3999 5. Patch from the folks at Google for configure.in to be a bit more thorough
4000 in checking for a suitable C++ installation before trying to compile the
4001 C++ stuff. This should fix a reported problem when a compiler was present,
4002 but no suitable headers.
4003
4004 6. The man pages all had just "PCRE" as their title. I have changed them to
4005 be the relevant file name. I have also arranged that these names are
4006 retained in the file doc/pcre.txt, which is a concatenation in text format
4007 of all the man pages except the little individual ones for each function.
4008
4009 7. The NON-UNIX-USE file had not been updated for the different set of source
4010 files that come with release 6. I also added a few comments about the C++
4011 wrapper.
4012
4013
4014 Version 6.0 07-Jun-05
4015 ---------------------
4016
4017 1. Some minor internal re-organization to help with my DFA experiments.
4018
4019 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4020 didn't matter for the library itself when fully configured, but did matter
4021 when compiling without UCP support, or within Exim, where the ucp files are
4022 not imported.
4023
4024 3. Refactoring of the library code to split up the various functions into
4025 different source modules. The addition of the new DFA matching code (see
4026 below) to a single monolithic source would have made it really too
4027 unwieldy, quite apart from causing all the code to be include in a
4028 statically linked application, when only some functions are used. This is
4029 relevant even without the DFA addition now that patterns can be compiled in
4030 one application and matched in another.
4031
4032 The downside of splitting up is that there have to be some external
4033 functions and data tables that are used internally in different modules of
4034 the library but which are not part of the API. These have all had their
4035 names changed to start with "_pcre_" so that they are unlikely to clash
4036 with other external names.
4037
4038 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4039 a different (DFA) algorithm. Although it is slower than the original
4040 function, it does have some advantages for certain types of matching
4041 problem.
4042
4043 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4044 including restarting after a partial match.
4045
4046 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4047 defined when compiling for Windows was sent to me. I have put it into the
4048 code, though I have no means of testing or verifying it.
4049
4050 7. Added the pcre_refcount() auxiliary function.
4051
4052 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4053 match before or at the first newline in the subject string. In pcretest,
4054 the /f option on a pattern can be used to set this.
4055
4056 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4057 would behave wrongly. This has been present in PCRE since release 4.0.
4058
4059 10. A number of changes to the pcregrep command:
4060
4061 (a) Refactored how -x works; insert ^(...)$ instead of setting
4062 PCRE_ANCHORED and checking the length, in preparation for adding
4063 something similar for -w.
4064
4065 (b) Added the -w (match as a word) option.
4066
4067 (c) Refactored the way lines are read and buffered so as to have more
4068 than one at a time available.
4069
4070 (d) Implemented a pcregrep test script.
4071
4072 (e) Added the -M (multiline match) option. This allows patterns to match
4073 over several lines of the subject. The buffering ensures that at least
4074 8K, or the rest of the document (whichever is the shorter) is available
4075 for matching (and similarly the previous 8K for lookbehind assertions).
4076
4077 (f) Changed the --help output so that it now says
4078
4079 -w, --word-regex(p)
4080
4081 instead of two lines, one with "regex" and the other with "regexp"
4082 because that confused at least one person since the short forms are the
4083 same. (This required a bit of code, as the output is generated
4084 automatically from a table. It wasn't just a text change.)
4085
4086 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4087 option but starts with a hyphen. Could be a pattern or a path name
4088 starting with a hyphen, for instance.
4089
4090 (h) "-" can be given as a file name to represent stdin.
4091
4092 (i) When file names are being printed, "(standard input)" is used for
4093 the standard input, for compatibility with GNU grep. Previously
4094 "<stdin>" was used.
4095
4096 (j) The option --label=xxx can be used to supply a name to be used for
4097 stdin when file names are being printed. There is no short form.
4098
4099 (k) Re-factored the options decoding logic because we are going to add
4100 two more options that take data. Such options can now be given in four
4101 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4102
4103 (l) Added the -A, -B, and -C options for requesting that lines of context
4104 around matches be printed.
4105
4106 (m) Added the -L option to print the names of files that do not contain
4107 any matching lines, that is, the complement of -l.
4108
4109 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4110 continue to scan other files.
4111
4112 (o) The -s option was incorrectly implemented. For compatibility with other
4113 greps, it now suppresses the error message for a non-existent or non-
4114 accessible file (but not the return code). There is a new option called
4115 -q that suppresses the output of matching lines, which was what -s was
4116 previously doing.
4117
4118 (p) Added --include and --exclude options to specify files for inclusion
4119 and exclusion when recursing.
4120
4121 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4122 Hopefully, it now does.
4123
4124 12. Missing cast in pcre_study().
4125
4126 13. Added an "uninstall" target to the makefile.
4127
4128 14. Replaced "extern" in the function prototypes in Makefile.in with
4129 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4130 world, but is set differently for Windows.
4131
4132 15. Added a second compiling function called pcre_compile2(). The only
4133 difference is that it has an extra argument, which is a pointer to an
4134 integer error code. When there is a compile-time failure, this is set
4135 non-zero, in addition to the error test pointer being set to point to an
4136 error message. The new argument may be NULL if no error number is required
4137 (but then you may as well call pcre_compile(), which is now just a
4138 wrapper). This facility is provided because some applications need a
4139 numeric error indication, but it has also enabled me to tidy up the way
4140 compile-time errors are handled in the POSIX wrapper.
4141
4142 16. Added VPATH=.libs to the makefile; this should help when building with one
4143 prefix path and installing with another. (Or so I'm told by someone who
4144 knows more about this stuff than I do.)
4145
4146 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4147 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4148 match everything, including newlines. This is not POSIX-compatible, but
4149 somebody wanted the feature. From pcretest it can be activated by using
4150 both the P and the s flags.
4151
4152 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4153
4154 19. libpcre.pc was being incorrectly installed as executable.
4155
4156 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4157 it now also looks for '\r' so that it will work unmodified on Windows.
4158
4159 21. Added Google's contributed C++ wrapper to the distribution.
4160
4161 22. Added some untidy missing memory free() calls in pcretest, to keep
4162 Electric Fence happy when testing.
4163
4164
4165
4166 Version 5.0 13-Sep-04
4167 ---------------------
4168
4169 1. Internal change: literal characters are no longer packed up into items
4170 containing multiple characters in a single byte-string. Each character
4171 is now matched using a separate opcode. However, there may be more than one
4172 byte in the character in UTF-8 mode.
4173
4174 2. The pcre_callout_block structure has two new fields: pattern_position and
4175 next_item_length. These contain the offset in the pattern to the next match
4176 item, and its length, respectively.
4177
4178 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4179 insertion of callouts before each pattern item. Added the /C option to
4180 pcretest to make use of this.
4181
4182 4. On the advice of a Windows user, the lines
4183
4184 #if defined(_WIN32) || defined(WIN32)
4185 _setmode( _fileno( stdout ), 0x8000 );
4186 #endif /* defined(_WIN32) || defined(WIN32) */
4187
4188 have been added to the source of pcretest. This apparently does useful
4189 magic in relation to line terminators.
4190
4191 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4192 for the benefit of those environments where the "b" makes a difference.
4193
4194 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4195 to know about it. I have put a hack into configure.in that adds in code
4196 to set GCC=yes if CC=icc. This seems to end up at a point in the
4197 generated configure script that is early enough to affect the setting of
4198 compiler options, which is what is needed, but I have no means of testing
4199 whether it really works. (The user who reported this had patched the
4200 generated configure script, which of course I cannot do.)
4201
4202 LATER: After change 22 below (new libtool files), the configure script
4203 seems to know about icc (and also ecc). Therefore, I have commented out
4204 this hack in configure.in.
4205
4206 7. Added support for pkg-config (2 patches were sent in).
4207
4208 8. Negated POSIX character classes that used a combination of internal tables
4209 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4210 [[:^ascii]]. Typically, they would match almost any characters. The other
4211 POSIX classes were not broken in this way.
4212
4213 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4214 to find the match, as PCRE was deluded into thinking that the match had to
4215 start at the start point or following a newline. The same bug applied to
4216 patterns with negative forward assertions or any backward assertions
4217 preceding ".*" at the start, unless the pattern required a fixed first
4218 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4219
4220 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4221 starting at the last subject character, bytes beyond the end of the subject
4222 string were read.
4223
4224 11. Renamed the variable "class" as "classbits" to make life easier for C++
4225 users. (Previously there was a macro definition, but it apparently wasn't
4226 enough.)
4227
4228 12. Added the new field "tables" to the extra data so that tables can be passed
4229 in at exec time, or the internal tables can be re-selected. This allows
4230 a compiled regex to be saved and re-used at a later time by a different
4231 program that might have everything at different addresses.
4232
4233 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4234 -R library as well as a -L library.
4235
4236 14. The debugging options of pcretest (-d on the command line or D on a
4237 pattern) showed incorrect output for anything following an extended class
4238 that contained multibyte characters and which was followed by a quantifier.
4239
4240 15. Added optional support for general category Unicode character properties
4241 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4242 support. It adds about 90K to the size of the library. The meanings of the
4243 inbuilt class escapes such as \d and \s have NOT been changed.
4244
4245 16. Updated pcredemo.c to include calls to free() to release the memory for the
4246 compiled pattern.
4247
4248 17. The generated file chartables.c was being created in the source directory
4249 instead of in the building directory. This caused the build to fail if the
4250 source directory was different from the building directory, and was
4251 read-only.
4252
4253 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4254 file. No doubt somebody will tell me if they don't make sense... Also added
4255 Dan Mooney's comments about building on OpenVMS.
4256
4257 19. Added support for partial matching via the PCRE_PARTIAL option for
4258 pcre_exec() and the \P data escape in pcretest.
4259
4260 20. Extended pcretest with 3 new pattern features:
4261
4262 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4263 write the compiled pattern to the file whose name is "rest-of-line".
4264 This is a straight binary dump of the data, with the saved pointer to
4265 the character tables forced to be NULL. The study data, if any, is
4266 written too. After writing, pcretest reads a new pattern.
4267
4268 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4269 compiled pattern from the given file. There must not be any
4270 occurrences of "<" in the file name (pretty unlikely); if there are,
4271 pcretest will instead treat the initial "<" as a pattern delimiter.
4272 After reading in the pattern, pcretest goes on to read data lines as
4273 usual.
4274
4275 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4276 and 16-bit fields in a compiled pattern, to simulate a pattern that
4277 was compiled on a host of opposite endianness.
4278
4279 21. The pcre-exec() function can now cope with patterns that were compiled on
4280 hosts of opposite endianness, with this restriction:
4281
4282 As for any compiled expression that is saved and used later, the tables
4283 pointer field cannot be preserved; the extra_data field in the arguments
4284 to pcre_exec() should be used to pass in a tables address if a value
4285 other than the default internal tables were used at compile time.
4286
4287 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4288 now diagnosed as an error. Previously, most of the time, a negative number
4289 would have been treated as zero, but if in addition "ovector" was passed as
4290 NULL, a crash could occur.
4291
4292 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4293 new versions from the libtool 1.5 distribution (the last one is a copy of
4294 a file called libtool.m4). This seems to have fixed the need to patch
4295 "configure" to support Darwin 1.3 (which I used to do). However, I still
4296 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4297 workstation).
4298
4299 24. Changed the PCRE licence to be the more standard "BSD" licence.
4300
4301
4302 Version 4.5 01-Dec-03
4303 ---------------------
4304
4305 1. There has been some re-arrangement of the code for the match() function so
4306 that it can be compiled in a version that does not call itself recursively.
4307 Instead, it keeps those local variables that need separate instances for
4308 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4309 needs to "recurse". Keeping track of where control must go is done by means
4310 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4311 hide most of the details from the main code, and operates only if
4312 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4313 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4314 operating.
4315
4316 To make it easier for callers to provide specially tailored get/free
4317 functions for this usage, two new functions, pcre_stack_malloc, and
4318 pcre_stack_free, are used. They are always called in strict stacking order,
4319 and the size of block requested is always the same.
4320
4321 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4322 PCRE has been compiled to use the stack or the heap for recursion. The
4323 -C option of pcretest uses this to show which version is compiled.
4324
4325 A new data escape \S, is added to pcretest; it causes the amounts of store
4326 obtained and freed by both kinds of malloc/free at match time to be added
4327 to the output.
4328
4329 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4330 what's available on my current Linux desktop machine.
4331
4332 3. When matching a UTF-8 string, the test for a valid string at the start has
4333 been extended. If start_offset is not zero, PCRE now checks that it points
4334 to a byte that is the start of a UTF-8 character. If not, it returns
4335 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4336 this is necessary because there may be backward assertions in the pattern.
4337 When matching the same subject several times, it may save resources to use
4338 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4339
4340 4. The code for checking the validity of UTF-8 strings has been tightened so
4341 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4342 containing "overlong sequences".
4343
4344 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4345 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4346 should have been "&", but it just so happened that all the cases this let
4347 through by mistake were picked up later in the function.
4348
4349 6. I had used a variable called "isblank" - this is a C99 function, causing
4350 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4351
4352 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4353 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4354 so the version is shown; (c) stop "make test" echoing "./RunTest".
4355
4356 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4357
4358 9. The prototype for memmove() for systems that don't have it was using
4359 size_t, but the inclusion of the header that defines size_t was later. I've
4360 moved the #includes for the C headers earlier to avoid this.
4361
4362 10. Added some adjustments to the code to make it easier to compiler on certain
4363 special systems:
4364
4365 (a) Some "const" qualifiers were missing.
4366 (b) Added the macro EXPORT before all exported functions; by default this
4367 is defined to be empty.
4368 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4369 that it reads its output file name as an argument instead of writing
4370 to the standard output and assuming this can be redirected.
4371
4372 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4373 class containing characters with values greater than 255, PCRE compilation
4374 went into a loop.
4375
4376 12. A recursive reference to a subpattern that was within another subpattern
4377 that had a minimum quantifier of zero caused PCRE to crash. For example,
4378 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4379 recursion. If the recursively-called subpattern itself had a zero repeat,
4380 that was OK.
4381
4382 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4383 buffer into which it was copied (for escape processing) was still set at
4384 1024, so long lines caused crashes.
4385
4386 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4387 "internal error: code overflow...". This applied to any character class
4388 that was followed by a possessive quantifier.
4389
4390 15. Modified the Makefile to add libpcre.la as a prerequisite for
4391 libpcreposix.la because I was told this is needed for a parallel build to
4392 work.
4393
4394 16. If a pattern that contained .* following optional items at the start was
4395 studied, the wrong optimizing data was generated, leading to matching
4396 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4397 matching string must start with a or b or c. The correct conclusion for
4398 this pattern is that a match can start with any character.
4399
4400
4401 Version 4.4 13-Aug-03
4402 ---------------------
4403
4404 1. In UTF-8 mode, a character class containing characters with values between
4405 127 and 255 was not handled correctly if the compiled pattern was studied.
4406 In fixing this, I have also improved the studying algorithm for such
4407 classes (slightly).
4408
4409 2. Three internal functions had redundant arguments passed to them. Removal
4410 might give a very teeny performance improvement.
4411
4412 3. Documentation bug: the value of the capture_top field in a callout is *one
4413 more than* the number of the hightest numbered captured substring.
4414
4415 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4416 in incorrectly linking with a previously installed version. They now link
4417 explicitly with libpcre.la.
4418
4419 5. configure.in no longer needs to recognize Cygwin specially.
4420
4421 6. A problem in pcre.in for Windows platforms is fixed.
4422
4423 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4424 pcretest, it used to include the size of the study block as part of its
4425 output. Unfortunately, the structure contains a field that has a different
4426 size on different hardware architectures. This meant that the tests that
4427 showed this size failed. As the block is currently always of a fixed size,
4428 this information isn't actually particularly useful in pcretest output, so
4429 I have just removed it.
4430
4431 8. Three pre-processor statements accidentally did not start in column 1.
4432 Sadly, there are *still* compilers around that complain, even though
4433 standard C has not required this for well over a decade. Sigh.
4434
4435 9. In pcretest, the code for checking callouts passed small integers in the
4436 callout_data field, which is a void * field. However, some picky compilers
4437 complained about the casts involved for this on 64-bit systems. Now
4438 pcretest passes the address of the small integer instead, which should get
4439 rid of the warnings.
4440
4441 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4442 both compile and run time, and gives an error if an invalid UTF-8 sequence
4443 is found. There is a option for disabling this check in cases where the
4444 string is known to be correct and/or the maximum performance is wanted.
4445
4446 11. In response to a bug report, I changed one line in Makefile.in from
4447
4448 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4449 to
4450 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4451
4452 to look similar to other lines, but I have no way of telling whether this
4453 is the right thing to do, as I do not use Windows. No doubt I'll get told
4454 if it's wrong...
4455
4456
4457 Version 4.3 21-May-03
4458 ---------------------
4459
4460 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4461 Makefile.
4462
4463 2. Some refactoring to improve the quality of the code:
4464
4465 (i) The utf8_table... variables are now declared "const".
4466
4467 (ii) The code for \cx, which used the "case flipping" table to upper case
4468 lower case letters, now just substracts 32. This is ASCII-specific,
4469 but the whole concept of \cx is ASCII-specific, so it seems
4470 reasonable.
4471
4472 (iii) PCRE was using its character types table to recognize decimal and
4473 hexadecimal digits in the pattern. This is silly, because it handles
4474 only 0-9, a-f, and A-F, but the character types table is locale-
4475 specific, which means strange things might happen. A private
4476 table is now used for this - though it costs 256 bytes, a table is
4477 much faster than multiple explicit tests. Of course, the standard
4478 character types table is still used for matching digits in subject
4479 strings against \d.
4480
4481 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4482 ending in _t are). So I've renamed it as ESC_tee.
4483
4484 3. The first argument for regexec() in the POSIX wrapper should have been
4485 defined as "const".
4486
4487 4. Changed pcretest to use malloc() for its buffers so that they can be
4488 Electric Fenced for debugging.
4489
4490 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4491 to read one or more bytes before the start of the subject string. Often this
4492 had no effect on PCRE's behaviour, but in some circumstances it could
4493 provoke a segmentation fault.
4494
4495 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4496 to try to read one or more bytes before the start of the subject string.
4497
4498 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4499 UTF-8 support could misbehave in various ways if the subject string
4500 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4501 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4502 back over UTF-8 characters.)
4503
4504
4505 Version 4.2 14-Apr-03
4506 ---------------------
4507
4508 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4509
4510 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4511 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4512 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4513 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4514 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4515 and BUILD_EXEEXT
4516 Note: automatic setting of the BUILD variables is not yet working
4517 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4518 compile-time but not at link-time
4519 [LINK]: use for linking executables only
4520 make different versions for Windows and non-Windows
4521 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4522 libraries
4523 [LINK_FOR_BUILD]: new variable
4524 [OBJEXT]: use throughout
4525 [EXEEXT]: use throughout
4526 <winshared>: new target
4527 <wininstall>: new target
4528 <dftables.o>: use native compiler
4529 <dftables>: use native linker
4530 <install>: handle Windows platform correctly
4531 <clean>: ditto
4532 <check>: ditto
4533 copy DLL to top builddir before testing
4534
4535 As part of these changes, -no-undefined was removed again. This was reported
4536 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4537 in any case.
4538
4539 3. Some tidies to get rid of compiler warnings:
4540
4541 . In the match_data structure, match_limit was an unsigned long int, whereas
4542 match_call_count was an int. I've made them both unsigned long ints.
4543
4544 . In pcretest the fact that a const uschar * doesn't automatically cast to
4545 a void * provoked a warning.
4546
4547 . Turning on some more compiler warnings threw up some "shadow" variables
4548 and a few more missing casts.
4549
4550 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4551 option, a class that contained a single character with a value between 128
4552 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4553
4554 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4555 option, a class that contained several characters, but with at least one
4556 whose value was between 128 and 255 caused PCRE to crash.
4557
4558
4559 Version 4.1 12-Mar-03
4560 ---------------------
4561
4562 1. Compiling with gcc -pedantic found a couple of places where casts were
4563 needed, and a string in dftables.c that was longer than standard compilers are
4564 required to support.
4565
4566 2. Compiling with Sun's compiler found a few more places where the code could
4567 be tidied up in order to avoid warnings.
4568
4569 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4570 first of these names is deprecated in the latest Autoconf in favour of the name
4571 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4572 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4573 analogy I have changed it to CFLAGS_FOR_BUILD.
4574
4575 4. Added -no-undefined to the linking command in the Makefile, because this is
4576 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4577 linking step for the pcreposix library.
4578
4579 5. PCRE was failing to diagnose the case of two named groups with the same
4580 name.
4581
4582 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4583 literal character that is needed in the subject for a match, and scans along to
4584 ensure that it is present before embarking on the full matching process. This
4585 saves time in cases of nested unlimited repeats that are never going to match.
4586 Problem: the scan can take a lot of time if the subject is very long (e.g.
4587 megabytes), thus penalizing straightforward matches. It is now done only if the
4588 amount of subject to be scanned is less than 1000 bytes.
4589
4590 7. A lesser problem with the same optimization is that it was recording the
4591 first character of an anchored pattern as "needed", thus provoking a search
4592 right along the subject, even when the first match of the pattern was going to
4593 fail. The "needed" character is now not set for anchored patterns, unless it
4594 follows something in the pattern that is of non-fixed length. Thus, it still
4595 fulfils its original purpose of finding quick non-matches in cases of nested
4596 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4597
4598
4599 Version 4.0 17-Feb-03
4600 ---------------------
4601
4602 1. If a comment in an extended regex that started immediately after a meta-item
4603 extended to the end of string, PCRE compiled incorrect data. This could lead to
4604 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4605
4606 2. Moved to autoconf 2.53 and libtool 1.4.2.
4607
4608 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4609 the special perltest8 script is no longer needed - all the tests can be run
4610 from a single perltest script.
4611
4612 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4613 by \s. It has now been removed in PCRE. This means it isn't recognized as
4614 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4615 class [:space:] *does* include VT, thereby creating a mess.
4616
4617 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4618 space and tab.
4619
4620 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4621 its new features into the main test script, reducing the number of scripts.
4622
4623 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4624 were backward compatible, and made the (?i) apply to the whole pattern, as if
4625 /i were given. Now it behaves more logically, and applies the option setting
4626 only to what follows. PCRE has been changed to follow suit. However, if it
4627 finds options settings right at the start of the pattern, it extracts them into
4628 the global options, as before. Thus, they show up in the info data.
4629
4630 8. Added support for the \Q...\E escape sequence. Characters in between are
4631 treated as literals. This is slightly different from Perl in that $ and @ are
4632 also handled as literals inside the quotes. In Perl, they will cause variable
4633 interpolation. Note the following examples:
4634
4635 Pattern PCRE matches Perl matches
4636
4637 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4638 \Qabc\$xyz\E abc\$xyz abc\$xyz
4639 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4640
4641 For compatibility with Perl, \Q...\E sequences are recognized inside character
4642 classes as well as outside them.
4643
4644 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4645 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4646 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4647 signed/unsigned warnings.
4648
4649 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4650 option for pcretest, so I've replaced it by a simple function that does just
4651 that job.
4652
4653 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4654 "pcregrep -".
4655
4656 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4657 Java package. This provides some syntactic sugar for simple cases of what my
4658 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4659 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4660 item, you can use this simplified notation. Note that only makes sense with
4661 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4662 greediness, whatever the setting of the PCRE_UNGREEDY option.
4663
4664 13. A change of greediness default within a pattern was not taking effect at
4665 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4666 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4667 was abstracted outside.
4668
4669 14. PCRE now supports the \G assertion. It is true when the current matching
4670 position is at the start point of the match. This differs from \A when the
4671 starting offset is non-zero. Used with the /g option of pcretest (or similar
4672 code), it works in the same way as it does for Perl's /g option. If all
4673 alternatives of a regex begin with \G, the expression is anchored to the start
4674 match position, and the "anchored" flag is set in the compiled expression.
4675
4676 15. Some bugs concerning the handling of certain option changes within patterns
4677 have been fixed. These applied to options other than (?ims). For example,
4678 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4679 been the other way round. Some of this was related to change 7 above.
4680
4681 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4682 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4683 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4684 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4685
4686 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4687 mode. Unlike ".", it always matches newline, whatever the setting of
4688 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4689 assertions. Perl allows it, but it doesn't (in general) work because it can't
4690 calculate the length of the lookbehind. At least, that's the case for Perl
4691 5.8.0 - I've been told they are going to document that it doesn't work in
4692 future.
4693
4694 18. Added an error diagnosis for escapes that PCRE does not support: these are
4695 \L, \l, \N, \P, \p, \U, \u, and \X.
4696
4697 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4698 reading past the end of the pattern in cases such as /[abcd/.
4699
4700 20. PCRE was getting more memory than necessary for patterns with classes that
4701 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4702
4703 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4704 compiling PCRE for use with Virtual Pascal.
4705
4706 22. Small fix to the Makefile to make it work properly if the build is done
4707 outside the source tree.
4708
4709 23. Added a new extension: a condition to go with recursion. If a conditional
4710 subpattern starts with (?(R) the "true" branch is used if recursion has
4711 happened, whereas the "false" branch is used only at the top level.
4712
4713 24. When there was a very long string of literal characters (over 255 bytes
4714 without UTF support, over 250 bytes with UTF support), the computation of how
4715 much memory was required could be incorrect, leading to segfaults or other
4716 strange effects.
4717
4718 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4719 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4720 there was a subsequent back reference to those brackets. This meant that, for
4721 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4722 possible to check for precisely this case. All we can do is abandon the
4723 optimization if .* occurs inside capturing brackets when there are any back
4724 references whatsoever. (See below for a better fix that came later.)
4725
4726 26. The handling of the optimization for finding the first character of a
4727 non-anchored pattern, and for finding a character that is required later in the
4728 match were failing in some cases. This didn't break the matching; it just
4729 failed to optimize when it could. The way this is done has been re-implemented.
4730
4731 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4732
4733 28. Added a new feature that provides some of the functionality that Perl
4734 provides with (?{...}). The facility is termed a "callout". The way it is done
4735 in PCRE is for the caller to provide an optional function, by setting
4736 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4737 global variable. By default it is unset, which disables all calling out. To get
4738 the function called, the regex must include (?C) at appropriate points. This
4739 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4740 This provides a means of identifying different callout points. When PCRE
4741 reaches such a point in the regex, if pcre_callout has been set, the external
4742 function is called. It is provided with data in a structure called
4743 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4744 matching continues; if it returns a non-zero value, the match at the current
4745 point fails. However, backtracking will occur if possible. [This was changed
4746 later and other features added - see item 49 below.]
4747
4748 29. pcretest is upgraded to test the callout functionality. It provides a
4749 callout function that displays information. By default, it shows the start of
4750 the match and the current position in the text. There are some new data escapes
4751 to vary what happens:
4752
4753 \C+ in addition, show current contents of captured substrings
4754 \C- do not supply a callout function
4755 \C!n return 1 when callout number n is reached
4756 \C!n!m return 1 when callout number n is reached for the mth time
4757
4758 30. If pcregrep was called with the -l option and just a single file name, it
4759 output "<stdin>" if a match was found, instead of the file name.
4760
4761 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4762 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4763 pcre_exec(). This saves a malloc/free per call. The default value of
4764 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4765 when configuring.
4766
4767 32. The default maximum size of a compiled pattern is 64K. There have been a
4768 few cases of people hitting this limit. The code now uses macros to handle the
4769 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4770 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4771 configuring. Tests 2 and 5 work only with 2-byte links because they output
4772 debugging information about compiled patterns.
4773
4774 33. Internal code re-arrangements:
4775
4776 (a) Moved the debugging function for printing out a compiled regex into
4777 its own source file (printint.c) and used #include to pull it into
4778 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4779 separate copies.
4780
4781 (b) Defined the list of op-code names for debugging as a macro in
4782 internal.h so that it is next to the definition of the opcodes.
4783
4784 (c) Defined a table of op-code lengths for simpler skipping along compiled
4785 code. This is again a macro in internal.h so that it is next to the
4786 definition of the opcodes.
4787
4788 34. Added support for recursive calls to individual subpatterns, along the
4789 lines of Robin Houston's patch (but implemented somewhat differently).
4790
4791 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4792 allow it to read and process whole directories in Win32. This code was
4793 contributed by Lionel Fourquaux; it has not been tested by me.
4794
4795 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4796 used to name a group. Names consist of alphanumerics and underscores, and must
4797 be unique. Back references use the syntax (?P=name) and recursive calls use
4798 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4799 numbers. The function pcre_fullinfo() can be used after compilation to extract
4800 a name/number map. There are three relevant calls:
4801
4802 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4803 PCRE_INFO_NAMECOUNT yields the number of entries
4804 PCRE_INFO_NAMETABLE yields a pointer to the map.
4805
4806 The map is a vector of fixed-size entries. The size of each entry depends on
4807 the length of the longest name used. The first two bytes of each entry are the
4808 group number, most significant byte first. There follows the corresponding
4809 name, zero terminated. The names are in alphabetical order.
4810
4811 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4812 case instead of 255. Making it the same both with and without UTF-8 support
4813 means that the same test output works with both.
4814
4815 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4816 calling malloc() with a zero argument.
4817
4818 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4819 optimization. I've improved things by keeping a bitmap of backreferences with
4820 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4821 fact referenced, the optimization can be applied. It is unlikely that a
4822 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4823 the match to follow \n) will appear inside brackets with a number greater than
4824 31, but if it does, any back reference > 31 suppresses the optimization.
4825
4826 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4827 of disabling numbered capturing parentheses. Any opening parenthesis that is
4828 not followed by ? behaves as if it were followed by ?: but named parentheses
4829 can still be used for capturing (and they will acquire numbers in the usual
4830 way).
4831
4832 41. Redesigned the return codes from the match() function into yes/no/error so
4833 that errors can be passed back from deep inside the nested calls. A malloc
4834 failure while inside a recursive subpattern call now causes the
4835 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4836
4837 42. It is now possible to set a limit on the number of times the match()
4838 function is called in a call to pcre_exec(). This facility makes it possible to
4839 limit the amount of recursion and backtracking, though not in a directly
4840 obvious way, because the match() function is used in a number of different
4841 circumstances. The count starts from zero for each position in the subject
4842 string (for non-anchored patterns). The default limit is, for compatibility, a
4843 large number, namely 10 000 000. You can change this in two ways:
4844
4845 (a) When configuring PCRE before making, you can use --with-match-limit=n
4846 to set a default value for the compiled library.
4847
4848 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4849 a different value is set. See 45 below.
4850
4851 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4852
4853 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4854 of things that can be changed at compile time. The first argument specifies
4855 what is wanted and the second points to where the information is to be placed.
4856 The current list of available information is:
4857
4858 PCRE_CONFIG_UTF8
4859
4860 The output is an integer that is set to one if UTF-8 support is available;
4861 otherwise it is set to zero.
4862
4863 PCRE_CONFIG_NEWLINE
4864
4865 The output is an integer that it set to the value of the code that is used for
4866 newline. It is either LF (10) or CR (13).
4867
4868 PCRE_CONFIG_LINK_SIZE
4869
4870 The output is an integer that contains the number of bytes used for internal
4871 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4872
4873 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4874
4875 The output is an integer that contains the threshold above which the POSIX
4876 interface uses malloc() for output vectors. See item 31 above.
4877
4878 PCRE_CONFIG_MATCH_LIMIT
4879
4880 The output is an unsigned integer that contains the default limit of the number
4881 of match() calls in a pcre_exec() execution. See 42 above.
4882
4883 44. pcretest has been upgraded by the addition of the -C option. This causes it
4884 to extract all the available output from the new pcre_config() function, and to
4885 output it. The program then exits immediately.
4886
4887 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4888 order to support additional features. One way would have been to define
4889 pcre_exec2() (for example) with extra arguments, but this would not have been
4890 extensible, and would also have required all calls to the original function to
4891 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4892 is used for passing in "extra" data from pcre_study().
4893
4894 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4895 contains the following fields:
4896
4897 flags a bitmap indicating which of the following fields are set
4898 study_data opaque data from pcre_study()
4899 match_limit a way of specifying a limit on match() calls for a specific
4900 call to pcre_exec()
4901 callout_data data for callouts (see 49 below)
4902
4903 The flag bits are also defined in pcre.h, and are
4904
4905 PCRE_EXTRA_STUDY_DATA
4906 PCRE_EXTRA_MATCH_LIMIT
4907 PCRE_EXTRA_CALLOUT_DATA
4908
4909 The pcre_study() function now returns one of these new pcre_extra blocks, with
4910 the actual study data pointed to by the study_data field, and the
4911 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4912 before. That is, this change is entirely upwards-compatible and requires no
4913 change to existing code.
4914
4915 If you want to pass in additional data to pcre_exec(), you can either place it
4916 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4917 block.
4918
4919 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4920 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4921 times with different match limits, until it finds the minimum value needed for
4922 pcre_exec() to complete. The value is then output. This can be instructive; for
4923 most simple matches the number is quite small, but for pathological cases it
4924 gets very large very quickly.
4925
4926 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4927 returns the size of the data block pointed to by the study_data field in a
4928 pcre_extra block, that is, the value that was passed as the argument to
4929 pcre_malloc() when PCRE was getting memory in which to place the information
4930 created by pcre_study(). The fourth argument should point to a size_t variable.
4931 pcretest has been extended so that this information is shown after a successful
4932 pcre_study() call when information about the compiled regex is being displayed.
4933
4934 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4935 because what follows is always an absolute path. (Later: it turns out that this
4936 is more than cosmetic for MinGW, because it doesn't like empty path
4937 components.)
4938
4939 49. Some changes have been made to the callout feature (see 28 above):
4940
4941 (i) A callout function now has three choices for what it returns:
4942
4943 0 => success, carry on matching
4944 > 0 => failure at this point, but backtrack if possible
4945 < 0 => serious error, return this value from pcre_exec()
4946
4947 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4948 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4949 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4950 use by callout functions. It will never be used by PCRE itself.
4951
4952 (ii) The pcre_extra structure (see 45 above) has a void * field called
4953 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4954 pcre_callout_block structure has a field of the same name. The contents of
4955 the field passed in the pcre_extra structure are passed to the callout
4956 function in the corresponding field in the callout block. This makes it
4957 easier to use the same callout-containing regex from multiple threads. For
4958 testing, the pcretest program has a new data escape
4959
4960 \C*n pass the number n (may be negative) as callout_data
4961
4962 If the callout function in pcretest receives a non-zero value as
4963 callout_data, it returns that value.
4964
4965 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4966 there were some redundant $(CFLAGS) in commands that are now specified as
4967 $(LINK), which already includes $(CFLAGS).
4968
4969 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4970 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4971 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4972 one-byte characters throughout. Note that case-insensitive matching applies
4973 only to characters whose values are less than 256. PCRE doesn't support the
4974 notion of cases for higher-valued characters.
4975
4976 (i) A character class whose characters are all within 0-255 is handled as
4977 a bit map, and the map is inverted for negative classes. Previously, a
4978 character > 255 always failed to match such a class; however it should
4979 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4980
4981 (ii) A negated character class with a single character < 255 is coded as
4982 "not this character" (OP_NOT). This wasn't working properly when the test
4983 character was multibyte, either singly or repeated.
4984
4985 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4986 mode, for example: \x{100}{2,3}.
4987
4988 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4989 singly or repeated) now correctly test multibyte characters. However,
4990 PCRE doesn't recognize any characters with values greater than 255 as
4991 digits, spaces, or word characters. Such characters always match \D, \S,
4992 and \W, and never match \d, \s, or \w.
4993
4994 (v) Classes may now contain characters and character ranges with values
4995 greater than 255. For example: [ab\x{100}-\x{400}].
4996
4997 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4998 PCRE in UTF-8 mode.
4999
5000 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5001 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5002 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5003 value.)
5004
5005 53. The single man page has become too large. I have therefore split it up into
5006 a number of separate man pages. These also give rise to individual HTML pages;
5007 these are now put in a separate directory, and there is an index.html page that
5008 lists them all. Some hyperlinking between the pages has been installed.
5009
5010 54. Added convenience functions for handling named capturing parentheses.
5011
5012 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5013 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5014 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5015 are faulted.
5016
5017 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5018 calling configure. These values are used when compiling the dftables.c program
5019 which is run to generate the source of the default character tables. They
5020 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5021 you will need to set these values.
5022
5023 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5024
5025
5026 Version 3.9 02-Jan-02
5027 ---------------------
5028
5029 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5030
5031 2. If --disable-static was given, the building process failed when trying to
5032 build pcretest and pcregrep. (For some reason it was using libtool to compile
5033 them, which is not right, as they aren't part of the library.)
5034
5035
5036 Version 3.8 18-Dec-01
5037 ---------------------
5038
5039 1. The experimental UTF-8 code was completely screwed up. It was packing the
5040 bytes in the wrong order. How dumb can you get?
5041
5042
5043 Version 3.7 29-Oct-01
5044 ---------------------
5045
5046 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5047 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5048 this didn't happen under Solaris 8, where I normally test things.
5049
5050 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5051 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5052 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5053 this fix an hour or so after the initial 3.7 release.)
5054
5055
5056 Version 3.6 23-Oct-01
5057 ---------------------
5058
5059 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5060 offsets passed as NULL with zero offset count.
5061
5062 2. The config.guess and config.sub files had not been updated when I moved to
5063 the latest autoconf.
5064
5065
5066 Version 3.5 15-Aug-01
5067 ---------------------
5068
5069 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5070 had been forgotten.
5071
5072 2. By using declared but undefined structures, we can avoid using "void"
5073 definitions in pcre.h while keeping the internal definitions of the structures
5074 private.
5075
5076 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5077 user point of view, this means that both static and shared libraries are built
5078 by default, but this can be individually controlled. More of the work of
5079 handling this static/shared cases is now inside libtool instead of PCRE's make
5080 file.
5081
5082 4. The pcretest utility is now installed along with pcregrep because it is
5083 useful for users (to test regexs) and by doing this, it automatically gets
5084 relinked by libtool. The documentation has been turned into a man page, so
5085 there are now .1, .txt, and .html versions in /doc.
5086
5087 5. Upgrades to pcregrep:
5088 (i) Added long-form option names like gnu grep.
5089 (ii) Added --help to list all options with an explanatory phrase.
5090 (iii) Added -r, --recursive to recurse into sub-directories.
5091 (iv) Added -f, --file to read patterns from a file.
5092
5093 6. pcre_exec() was referring to its "code" argument before testing that
5094 argument for NULL (and giving an error if it was NULL).
5095
5096 7. Upgraded Makefile.in to allow for compiling in a different directory from
5097 the source directory.
5098
5099 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5100 options bits, the pointer it was passed was to an int instead of to an unsigned
5101 long int. This mattered only on 64-bit systems.
5102
5103 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5104 generated) instead of pcre.in, which it its source. Also made the same change
5105 in several of the .c files.
5106
5107 10. A new release of gcc defines printf() as a macro, which broke pcretest
5108 because it had an ifdef in the middle of a string argument for printf(). Fixed
5109 by using separate calls to printf().
5110
5111 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5112 script, to force use of CR or LF instead of \n in the source. On non-Unix
5113 systems, the value can be set in config.h.
5114
5115 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5116 absolute limit. Changed the text of the error message to make this clear, and
5117 likewise updated the man page.
5118
5119 13. The limit of 99 on the number of capturing subpatterns has been removed.
5120 The new limit is 65535, which I hope will not be a "real" limit.
5121
5122
5123 Version 3.4 22-Aug-00
5124 ---------------------
5125
5126 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5127
5128 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5129
5130
5131 Version 3.3 01-Aug-00
5132 ---------------------
5133
5134 1. If an octal character was given, but the value was greater than \377, it
5135 was not getting masked to the least significant bits, as documented. This could
5136 lead to crashes in some systems.
5137
5138 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5139 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5140
5141 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5142 These just pass their arguments on to (pcre_free)(), but they are provided
5143 because some uses of PCRE bind it to non-C systems that can call its functions,
5144 but cannot call free() or pcre_free() directly.
5145
5146 4. Add "make test" as a synonym for "make check". Corrected some comments in
5147 the Makefile.
5148
5149 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5150 Makefile.
5151
5152 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5153 command called pgrep for grepping around the active processes.
5154
5155 7. Added the beginnings of support for UTF-8 character strings.
5156
5157 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5158 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5159 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5160 out for the ar command.)
5161
5162
5163 Version 3.2 12-May-00
5164 ---------------------
5165
5166 This is purely a bug fixing release.
5167
5168 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5169 of ZA. This was just one example of several cases that could provoke this bug,
5170 which was introduced by change 9 of version 2.00. The code for breaking
5171 infinite loops after an iteration that matches an empty string was't working
5172 correctly.
5173
5174 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5175 when matched against abbab (for example). After matching an empty string, it
5176 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5177 caused it to match further down the string than it should.
5178
5179 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5180 was there because it doesn't seem to be needed, and it causes trouble on some
5181 systems, as it is not a Standard C header. It has been removed.
5182
5183 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5184 were reported on the Macintosh. The changes were from
5185
5186 while ((c = *(++ptr)) != 0 && c != '\n');
5187 to
5188 while ((c = *(++ptr)) != 0 && c != '\n') ;
5189
5190 Totally extraordinary, but if that's what it takes...
5191
5192 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5193 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5194 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5195 assumes the way PCRE uses memmove() (always moving upwards).
5196
5197 6. PCRE is being used in one environment where strchr() is not available. There
5198 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5199 faster code anyway.
5200
5201
5202 Version 3.1 09-Feb-00
5203 ---------------------
5204
5205 The only change in this release is the fixing of some bugs in Makefile.in for
5206 the "install" target:
5207
5208 (1) It was failing to install pcreposix.h.
5209
5210 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5211
5212
5213 Version 3.0 01-Feb-00
5214 ---------------------
5215
5216 1. Add support for the /+ modifier to perltest (to output $` like it does in
5217 pcretest).
5218
5219 2. Add support for the /g modifier to perltest.
5220
5221 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5222 matches null strings.
5223
5224 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5225 pattern. Perl treats empty patterns specially - it reuses the most recent
5226 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5227 effect.
5228
5229 5. The POSIX interface was broken in that it was just handing over the POSIX
5230 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5231 required a bigger vector, with some working space on the end. This means that
5232 the POSIX wrapper now has to get and free some memory, and copy the results.
5233
5234 6. Added some simple autoconf support, placing the test data and the
5235 documentation in separate directories, re-organizing some of the
5236 information files, and making it build pcre-config (a GNU standard). Also added
5237 libtool support for building PCRE as a shared library, which is now the
5238 default.
5239
5240 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5241 09 are not valid octal constants. Single digits will be used for minor values
5242 less than 10.
5243
5244 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5245 existing programs that set these in the POSIX interface can use PCRE without
5246 modification.
5247
5248 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5249 return all that pcre_info() returns, plus additional data. The pcre_info()
5250 function is retained for compatibility, but is considered to be obsolete.
5251
5252 10. Added experimental recursion feature (?R) to handle one common case that
5253 Perl 5.6 will be able to do with (?p{...}).
5254
5255 11. Added support for POSIX character classes like [:alpha:], which Perl is
5256 adopting.
5257
5258
5259 Version 2.08 31-Aug-99
5260 ----------------------
5261
5262 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5263 trying to match at the startoffset position, but instead was moving forward to
5264 the next newline as if a previous match had failed.
5265
5266 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5267 and could get into a loop if a null string was matched other than at the start
5268 of the subject.
5269
5270 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5271 be distinguished at compile time, and for completeness also added PCRE_DATE.
5272
5273 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5274 in GnuWin32 environments.
5275
5276
5277 Version 2.07 29-Jul-99
5278 ----------------------
5279
5280 1. The documentation is now supplied in plain text form and HTML as well as in
5281 the form of man page sources.
5282
5283 2. C++ compilers don't like assigning (void *) values to other pointer types.
5284 In particular this affects malloc(). Although there is no problem in Standard
5285 C, I've put in casts to keep C++ compilers happy.
5286
5287 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5288 should be (const char *).
5289
5290 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5291 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5292 However, I haven't made this a standard facility. The documentation doesn't
5293 mention it, and the Makefile doesn't support it.
5294
5295 5. The Makefile now contains an "install" target, with editable destinations at
5296 the top of the file. The pcretest program is not installed.
5297
5298 6. pgrep -V now gives the PCRE version number and date.
5299
5300 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5301 causing the entire string to be ignored, instead of just the last character.
5302
5303 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5304 non-matching string, it can take a very, very long time, even for strings of
5305 quite modest length, because of the nested recursion. PCRE now does better in
5306 some of these cases. It does this by remembering the last required literal
5307 character in the pattern, and pre-searching the subject to ensure it is present
5308 before running the real match. In other words, it applies a heuristic to detect
5309 some types of certain failure quickly, and in the above example, if presented
5310 with a string that has no trailing " it gives "no match" very quickly.
5311
5312 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5313 other alternatives are tried instead.
5314
5315
5316 Version 2.06 09-Jun-99
5317 ----------------------
5318
5319 1. Change pcretest's output for amount of store used to show just the code
5320 space, because the remainder (the data block) varies in size between 32-bit and
5321 64-bit systems.
5322
5323 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5324 start matching at. This allows lookbehinds to work when searching for multiple
5325 occurrences in a string.
5326
5327 3. Added additional options to pcretest for testing multiple occurrences:
5328
5329 /+ outputs the rest of the string that follows a match
5330 /g loops for multiple occurrences, using the new startoffset argument
5331 /G loops for multiple occurrences by passing an incremented pointer
5332
5333 4. PCRE wasn't doing the "first character" optimization for patterns starting
5334 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5335 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5336 the letter 'x'. On long subject strings, this gives a significant speed-up.
5337
5338
5339 Version 2.05 21-Apr-99
5340 ----------------------
5341
5342 1. Changed the type of magic_number from int to long int so that it works
5343 properly on 16-bit systems.
5344
5345 2. Fixed a bug which caused patterns starting with .* not to work correctly
5346 when the subject string contained newline characters. PCRE was assuming
5347 anchoring for such patterns in all cases, which is not correct because .* will
5348 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5349 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5350 must be retried after every newline in the subject.
5351
5352
5353 Version 2.04 18-Feb-99
5354 ----------------------
5355
5356 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5357 computation of the store needed to hold the pattern was incorrect (too large).
5358 If such patterns were nested a few deep, this could multiply and become a real
5359 problem.
5360
5361 2. Added /M option to pcretest to show the memory requirement of a specific
5362 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5363
5364 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5365 compiled in such a way that the backtracking after subsequent failure was
5366 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5367 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5368
5369
5370 Version 2.03 02-Feb-99
5371 ----------------------
5372
5373 1. Fixed typo and small mistake in man page.
5374
5375 2. Added 4th condition (GPL supersedes if conflict) and created separate
5376 LICENCE file containing the conditions.
5377
5378 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5379 Perl, that is the internal \ allows the delimiter to be included in the
5380 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5381 the final delimiter, add \ to the end of the pattern (to test the error).
5382
5383 4. Added the convenience functions for extracting substrings after a successful
5384 match. Updated pcretest to make it able to test these functions.
5385
5386
5387 Version 2.02 14-Jan-99
5388 ----------------------
5389
5390 1. Initialized the working variables associated with each extraction so that
5391 their saving and restoring doesn't refer to uninitialized store.
5392
5393 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5394 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5395 fix the problem.
5396
5397 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5398 calls, and wasn't printing the correct value for compiling calls. Increased the
5399 default value of LOOPREPEAT, and the number of significant figures in the
5400 times.
5401
5402 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5403
5404 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5405 a building problem on Windows NT with a FAT file system.
5406
5407
5408 Version 2.01 21-Oct-98
5409 ----------------------
5410
5411 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5412 to character tables built by pcre_maketables() in the current locale. If NULL
5413 is passed, the default tables are used.
5414
5415
5416 Version 2.00 24-Sep-98
5417 ----------------------
5418
5419 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5420 it any more.
5421
5422 2. Allow quantification of (?>) groups, and make it work correctly.
5423
5424 3. The first character computation wasn't working for (?>) groups.
5425
5426 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5427 end of the subject) and add 5.005's \z, which really does match only at the
5428 very end of the subject.
5429
5430 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5431
5432 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5433 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5434 localized options. All options to pcre_study() were also removed.
5435
5436 7. Add other new features from 5.005:
5437
5438 $(?<= positive lookbehind
5439 $(?<! negative lookbehind
5440 (?imsx-imsx) added the unsetting capability
5441 such a setting is global if at outer level; local otherwise
5442 (?imsx-imsx:) non-capturing groups with option setting
5443 (?(cond)re|re) conditional pattern matching
5444
5445 A backreference to itself in a repeated group matches the previous
5446 captured string.
5447
5448 8. General tidying up of studying (both automatic and via "study")
5449 consequential on the addition of new assertions.
5450
5451 9. As in 5.005, unlimited repeated groups that could match an empty substring
5452 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5453 runtime if any iteration does actually match an empty substring.
5454
5455 10. Include the RunTest script in the distribution.
5456
5457 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5458 discrepancies, some of which were old and were also with respect to 5.004. They
5459 have now been fixed.
5460
5461
5462 Version 1.09 28-Apr-98
5463 ----------------------
5464
5465 1. A negated single character class followed by a quantifier with a minimum
5466 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5467 program crashes, or just wrong answers. This did not apply to negated classes
5468 containing more than one character, or to minima other than one.
5469
5470
5471 Version 1.08 27-Mar-98
5472 ----------------------
5473
5474 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5475
5476 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5477 latter must appear before anything that relies on it in the pattern.
5478
5479
5480 Version 1.07 16-Feb-98
5481 ----------------------
5482
5483 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5484 repeat of a potentially empty string).
5485
5486
5487 Version 1.06 23-Jan-98
5488 ----------------------
5489
5490 1. Added Markus Oberhumer's little patches for C++.
5491
5492 2. Literal strings longer than 255 characters were broken.
5493
5494
5495 Version 1.05 23-Dec-97
5496 ----------------------
5497
5498 1. Negated character classes containing more than one character were failing if
5499 PCRE_CASELESS was set at run time.
5500
5501
5502 Version 1.04 19-Dec-97
5503 ----------------------
5504
5505 1. Corrected the man page, where some "const" qualifiers had been omitted.
5506
5507 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5508 input syntax.
5509
5510 3. Fixed memory leak which occurred when a regex with back references was
5511 matched with an offsets vector that wasn't big enough. The temporary memory
5512 that is used in this case wasn't being freed if the match failed.
5513
5514 4. Tidied pcretest to ensure it frees memory that it gets.
5515
5516 5. Temporary memory was being obtained in the case where the passed offsets
5517 vector was exactly big enough.
5518
5519 6. Corrected definition of offsetof() from change 5 below.
5520
5521 7. I had screwed up change 6 below and broken the rules for the use of
5522 setjmp(). Now fixed.
5523
5524
5525 Version 1.03 18-Dec-97
5526 ----------------------
5527
5528 1. A erroneous regex with a missing opening parenthesis was correctly
5529 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5530 on some systems.
5531
5532 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5533 it was reported that one broken compiler failed on the former because "code" is
5534 also an independent variable.
5535
5536 3. The erroneous regex a[]b caused an array overrun reference.
5537
5538 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5539 fail on data ending with that character. (It was going on too far, and checking
5540 the next character, typically a binary zero.) This was specific to the
5541 optimized code for single-character negative classes.
5542
5543 5. Added a contributed patch from the TIN world which does the following:
5544
5545 + Add an undef for memmove, in case the the system defines a macro for it.
5546
5547 + Add a definition of offsetof(), in case there isn't one. (I don't know
5548 the reason behind this - offsetof() is part of the ANSI standard - but
5549 it does no harm).
5550
5551 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5552 most of the places where whitespace preceded '#'. I have given up and
5553 allowed the remaining 2 cases to be at the margin.
5554
5555 + Rename some variables in pcre to eliminate shadowing. This seems very
5556 pedantic, but does no harm, of course.
5557
5558 6. Moved the call to setjmp() into its own function, to get rid of warnings
5559 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5560
5561 7. Constructs such as \d{8,} were compiling into the equivalent of
5562 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5563 outcome, but in this particular case used more store than had been allocated,
5564 which caused the bug to be discovered because it threw up an internal error.
5565
5566 8. The debugging code in both pcre and pcretest for outputting the compiled
5567 form of a regex was going wrong in the case of back references followed by
5568 curly-bracketed repeats.
5569
5570
5571 Version 1.02 12-Dec-97
5572 ----------------------
5573
5574 1. Typos in pcre.3 and comments in the source fixed.
5575
5576 2. Applied a contributed patch to get rid of places where it used to remove
5577 'const' from variables, and fixed some signed/unsigned and uninitialized
5578 variable warnings.
5579
5580 3. Added the "runtest" target to Makefile.
5581
5582 4. Set default compiler flag to -O2 rather than just -O.
5583
5584
5585 Version 1.01 19-Nov-97
5586 ----------------------
5587
5588 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5589 like /([ab]*)*/, that is, for classes with more than one character in them.
5590
5591 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5592 as /((?>a*))*/ (a PCRE_EXTRA facility).
5593
5594
5595 Version 1.00 18-Nov-97
5596 ----------------------
5597
5598 1. Added compile-time macros to support systems such as SunOS4 which don't have
5599 memmove() or strerror() but have other things that can be used instead.
5600
5601 2. Arranged that "make clean" removes the executables.
5602
5603
5604 Version 0.99 27-Oct-97
5605 ----------------------
5606
5607 1. Fixed bug in code for optimizing classes with only one character. It was
5608 initializing a 32-byte map regardless, which could cause it to run off the end
5609 of the memory it had got.
5610
5611 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5612
5613
5614 Version 0.98 22-Oct-97
5615 ----------------------
5616
5617 1. Fixed bug in code for handling temporary memory usage when there are more
5618 back references than supplied space in the ovector. This could cause segfaults.
5619
5620
5621 Version 0.97 21-Oct-97
5622 ----------------------
5623
5624 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5625
5626 2. Optimized negated single characters not to use a bit map.
5627
5628 3. Brought error texts together as macro definitions; clarified some of them;
5629 fixed one that was wrong - it said "range out of order" when it meant "invalid
5630 escape sequence".
5631
5632 4. Changed some char * arguments to const char *.
5633
5634 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5635
5636 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5637 pcretest.
5638
5639
5640 Version 0.96 16-Oct-97
5641 ----------------------
5642
5643 1. Added a simple "pgrep" utility to the distribution.
5644
5645 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5646 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5647 where "ddd" means "one or more decimal digits".
5648
5649 3. Fixed serious bug. If a pattern had a back reference, but the call to
5650 pcre_exec() didn't supply a large enough ovector to record the related
5651 identifying subpattern, the match always failed. PCRE now remembers the number
5652 of the largest back reference, and gets some temporary memory in which to save
5653 the offsets during matching if necessary, in order to ensure that
5654 backreferences always work.
5655
5656 4. Increased the compatibility with Perl in a number of ways:
5657
5658 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5659 to request this handling. The option can be set at compile or exec time.
5660
5661 (b) $ matches before a terminating newline by default; an option
5662 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5663 mode). The option can be set at compile or exec time.
5664
5665 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5666 the same as Perl's. If the decimal number it represents is less than 10
5667 or there aren't that many previous left capturing parentheses, an octal
5668 escape is read. Inside a character class, it's always an octal escape,
5669 even if it is a single digit.
5670
5671 (d) An escaped but undefined alphabetic character is taken as a literal,
5672 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5673 escapes.
5674
5675 (e) {0} is now permitted. (The previous item is removed from the compiled
5676 pattern).
5677
5678 5. Changed all the names of code files so that the basic parts are no longer
5679 than 10 characters, and abolished the teeny "globals.c" file.
5680
5681 6. Changed the handling of character classes; they are now done with a 32-byte
5682 bit map always.
5683
5684 7. Added the -d and /D options to pcretest to make it possible to look at the
5685 internals of compilation without having to recompile pcre.
5686
5687
5688 Version 0.95 23-Sep-97
5689 ----------------------
5690
5691 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5692 \x20 at the start of a run of normal characters. These were being treated as
5693 real characters, instead of the source characters being re-checked.
5694
5695
5696 Version 0.94 18-Sep-97
5697 ----------------------
5698
5699 1. The functions are now thread-safe, with the caveat that the global variables
5700 containing pointers to malloc() and free() or alternative functions are the
5701 same for all threads.
5702
5703 2. Get pcre_study() to generate a bitmap of initial characters for non-
5704 anchored patterns when this is possible, and use it if passed to pcre_exec().
5705
5706
5707 Version 0.93 15-Sep-97
5708 ----------------------
5709
5710 1. /(b)|(:+)/ was computing an incorrect first character.
5711
5712 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5713 but not actually doing anything yet.
5714
5715 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5716 as Perl does (e.g. [-az] or [az-]).
5717
5718 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5719 all possible positions.
5720
5721 5. Split up into different modules to avoid including unneeded functions in a
5722 compiled binary. However, compile and exec are still in one module. The "study"
5723 function is split off.
5724
5725 6. The character tables are now in a separate module whose source is generated
5726 by an auxiliary program - but can then be edited by hand if required. There are
5727 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5728 toupper() in the code.
5729
5730 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5731 make them global. Abolish the function for setting them, as the caller can now
5732 set them directly.
5733
5734
5735 Version 0.92 11-Sep-97
5736 ----------------------
5737
5738 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5739 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5740
5741 2. Caseless matching was not working in character classes if the characters in
5742 the pattern were in upper case.
5743
5744 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5745
5746 4. Make PCRE_ANCHORED public and accept as a compile option.
5747
5748 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5749 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5750 pass them.
5751
5752 6. Give an error if bad option bits passed at compile or run time.
5753
5754 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5755 pcretest to cause it to pass that flag.
5756
5757 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5758 options, and the first character, if set.
5759
5760 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5761
5762
5763 Version 0.91 10-Sep-97
5764 ----------------------
5765
5766 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5767 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5768
5769 2. PCRE was looping on encountering an indefinitely repeated back reference to
5770 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5771 Perl does - treats the match as successful.
5772
5773 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5