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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1677 - (show annotations)
Mon Jan 23 16:35:09 2017 UTC (2 years, 9 months ago) by ph10
File size: 291938 byte(s)
Error occurred while calculating annotation data.
Fix typo in CMakeLists.txt that affects only MSVC.
1 ChangeLog for PCRE
2 ------------------
3
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
6
7 Version 8.41
8 ------------
9
10 1. Fixed typo in CMakeLists.txt (wrong number of arguments for
11 PCRE_STATIC_RUNTIME (affects MSVC only).
12
13
14 Version 8.40 11-January-2017
15 ----------------------------
16
17 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
18 the match extended over a line boundary.
19
20 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
21 MSVC static compilation, putting the first patch under a new option.
22
23 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
24
25 4. Ignore "show all captures" (/=) for DFA matching.
26
27 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
28
29 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
30 without PCRE_UCP set, a negative character type such as \D in a positive
31 class should cause all characters greater than 255 to match, whatever else
32 is in the class. There was a bug that caused this not to happen if a
33 Unicode property item was added to such a class, for example [\D\P{Nd}] or
34 [\W\pL].
35
36 7. When pcretest was outputing information from a callout, the caret indicator
37 for the current position in the subject line was incorrect if it was after
38 an escape sequence for a character whose code point was greater than
39 \x{ff}.
40
41 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
42 the conditional was interpreted as a reference to capturing group 1 instead
43 of a test for recursion. Any group whose name began with R was
44 misinterpreted in this way. (The reference interpretation should only
45 happen if the group's name is precisely "R".)
46
47 9. A number of bugs have been mended relating to match start-up optimizations
48 when the first thing in a pattern is a positive lookahead. These all
49 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
50
51 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
52 both an initial 'X' and a following 'X'.
53 (b) Some patterns starting with an assertion that started with .* were
54 incorrectly optimized as having to match at the start of the subject or
55 after a newline. There are cases where this is not true, for example,
56 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
57 start with spaces. Starting .* in an assertion is no longer taken as an
58 indication of matching at the start (or after a newline).
59
60
61 Version 8.39 14-June-2016
62 -------------------------
63
64 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
65 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
66 misbehaved. This bug was found by the LLVM fuzzer.
67
68 2. Similar to the above, if an isolated \E was present between an item and its
69 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
70 bug was found by the LLVM fuzzer.
71
72 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
73 working correctly in UCP mode.
74
75 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
76 was set when the pmatch argument was NULL. It now returns REG_INVARG.
77
78 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
79
80 6. An empty \Q\E sequence between an item and its qualifier caused
81 pcre_compile() to misbehave when auto callouts were enabled. This bug was
82 found by the LLVM fuzzer.
83
84 7. If a pattern that was compiled with PCRE_EXTENDED started with white
85 space or a #-type comment that was followed by (?-x), which turns off
86 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
87 pcre_compile() assumed that (?-x) applied to the whole pattern and
88 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
89
90 8. A call of pcre_copy_named_substring() for a named substring whose number
91 was greater than the space in the ovector could cause a crash.
92
93 9. Yet another buffer overflow bug involved duplicate named groups with a
94 group that reset capture numbers (compare 8.38/7 below). Once again, I have
95 just allowed for more memory, even if not needed. (A proper fix is
96 implemented in PCRE2, but it involves a lot of refactoring.)
97
98 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
99 start of the match to be earlier than the end.
100
101 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
102
103 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
104 assertion, caused pcretest to generate incorrect output, and also to read
105 uninitialized memory (detected by ASAN or valgrind).
106
107 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
108 nested set of parentheses of sufficient size caused an overflow of the
109 compiling workspace (which was diagnosed, but of course is not desirable).
110
111 14. And yet another buffer overflow bug involving duplicate named groups, this
112 time nested, with a nested back reference. Yet again, I have just allowed
113 for more memory, because anything more needs all the refactoring that has
114 been done for PCRE2. An example pattern that provoked this bug is:
115 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
116 registered as CVE-2016-1283.
117
118 15. pcretest went into a loop if global matching was requested with an ovector
119 size less than 2. It now gives an error message. This bug was found by
120 afl-fuzz.
121
122 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
123 ("assertion expected") when (?(?C) was not followed by an opening
124 parenthesis.
125
126 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
127 actually affect anything, by sheer luck.
128
129 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
130 static compilation.
131
132 19. Modified the RunTest script to incorporate a valgrind suppressions file so
133 that certain errors, provoked by the SSE2 instruction set when JIT is used,
134 are ignored.
135
136 20. A racing condition is fixed in JIT reported by Mozilla.
137
138 21. Minor code refactor to avoid "array subscript is below array bounds"
139 compiler warning.
140
141 22. Minor code refactor to avoid "left shift of negative number" warning.
142
143 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
144 UCP support.
145
146 24. Refactor to avoid compiler warnings in pcrecpp.cc.
147
148 25. Refactor to fix a typo in pcre_jit_test.c
149
150 26. Patch to support compiling pcrecpp.cc with Intel compiler.
151
152
153 Version 8.38 23-November-2015
154 -----------------------------
155
156 1. If a group that contained a recursive back reference also contained a
157 forward reference subroutine call followed by a non-forward-reference
158 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
159 compile correct code, leading to undefined behaviour or an internally
160 detected error. This bug was discovered by the LLVM fuzzer.
161
162 2. Quantification of certain items (e.g. atomic back references) could cause
163 incorrect code to be compiled when recursive forward references were
164 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
165 This bug was discovered by the LLVM fuzzer.
166
167 3. A repeated conditional group whose condition was a reference by name caused
168 a buffer overflow if there was more than one group with the given name.
169 This bug was discovered by the LLVM fuzzer.
170
171 4. A recursive back reference by name within a group that had the same name as
172 another group caused a buffer overflow. For example:
173 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
174
175 5. A forward reference by name to a group whose number is the same as the
176 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
177 a buffer overflow at compile time. This bug was discovered by the LLVM
178 fuzzer.
179
180 6. A lookbehind assertion within a set of mutually recursive subpatterns could
181 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
182
183 7. Another buffer overflow bug involved duplicate named groups with a
184 reference between their definition, with a group that reset capture
185 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
186 fixed by always allowing for more memory, even if not needed. (A proper fix
187 is implemented in PCRE2, but it involves more refactoring.)
188
189 8. There was no check for integer overflow in subroutine calls such as (?123).
190
191 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
192 being treated as a literal 'l' instead of causing an error.
193
194 10. There was a buffer overflow if pcre_exec() was called with an ovector of
195 size 1. This bug was found by american fuzzy lop.
196
197 11. If a non-capturing group containing a conditional group that could match
198 an empty string was repeated, it was not identified as matching an empty
199 string itself. For example: /^(?:(?(1)x|)+)+$()/.
200
201 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
202 \a and \e in test subject lines.
203
204 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
205 instead of the EBCDIC value.
206
207 14. The handling of \c in an EBCDIC environment has been revised so that it is
208 now compatible with the specification in Perl's perlebcdic page.
209
210 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
211 ASCII/Unicode. This has now been added to the list of characters that are
212 recognized as white space in EBCDIC.
213
214 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
215 error (correctly) when used outside a class, but did not give an error
216 within a class.
217
218 17. \h within a class was incorrectly compiled in EBCDIC environments.
219
220 18. A pattern with an unmatched closing parenthesis that contained a backward
221 assertion which itself contained a forward reference caused buffer
222 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
223
224 19. JIT should return with error when the compiled pattern requires more stack
225 space than the maximum.
226
227 20. A possessively repeated conditional group that could match an empty string,
228 for example, /(?(R))*+/, was incorrectly compiled.
229
230 21. Fix infinite recursion in the JIT compiler when certain patterns such as
231 /(?:|a|){100}x/ are analysed.
232
233 22. Some patterns with character classes involving [: and \\ were incorrectly
234 compiled and could cause reading from uninitialized memory or an incorrect
235 error diagnosis.
236
237 23. Pathological patterns containing many nested occurrences of [: caused
238 pcre_compile() to run for a very long time.
239
240 24. A conditional group with only one branch has an implicit empty alternative
241 branch and must therefore be treated as potentially matching an empty
242 string.
243
244 25. If (?R was followed by - or + incorrect behaviour happened instead of a
245 diagnostic.
246
247 26. Arrange to give up on finding the minimum matching length for overly
248 complex patterns.
249
250 27. Similar to (4) above: in a pattern with duplicated named groups and an
251 occurrence of (?| it is possible for an apparently non-recursive back
252 reference to become recursive if a later named group with the relevant
253 number is encountered. This could lead to a buffer overflow. Wen Guanxing
254 from Venustech ADLAB discovered this bug.
255
256 28. If pcregrep was given the -q option with -c or -l, or when handling a
257 binary file, it incorrectly wrote output to stdout.
258
259 29. The JIT compiler did not restore the control verb head in case of *THEN
260 control verbs. This issue was found by Karl Skomski with a custom LLVM
261 fuzzer.
262
263 30. Error messages for syntax errors following \g and \k were giving inaccurate
264 offsets in the pattern.
265
266 31. Added a check for integer overflow in conditions (?(<digits>) and
267 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
268 fuzzer.
269
270 32. Handling recursive references such as (?2) when the reference is to a group
271 later in the pattern uses code that is very hacked about and error-prone.
272 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
273 give an internal error if it is obvious that compiling has gone wrong.
274
275 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
276 This issue was found by Karl Skomski with a custom LLVM fuzzer.
277
278 34. The JIT compiler should restore the control chain for empty possessive
279 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
280
281 35. Match limit check added to JIT recursion. This issue was found by Karl
282 Skomski with a custom LLVM fuzzer.
283
284 36. Yet another case similar to 27 above has been circumvented by an
285 unconditional allocation of extra memory. This issue is fixed "properly" in
286 PCRE2 by refactoring the way references are handled. Wen Guanxing
287 from Venustech ADLAB discovered this bug.
288
289 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
290 with a custom LLVM fuzzer.
291
292 38. Fixed a corner case of range optimization in JIT.
293
294 39. An incorrect error "overran compiling workspace" was given if there were
295 exactly enough group forward references such that the last one extended
296 into the workspace safety margin. The next one would have expanded the
297 workspace. The test for overflow was not including the safety margin.
298
299 40. A match limit issue is fixed in JIT which was found by Karl Skomski
300 with a custom LLVM fuzzer.
301
302 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
303 work under Windows. (Why has it taken so long for anyone to notice?)
304
305 42. In a character class such as [\W\p{Any}] where both a negative-type escape
306 ("not a word character") and a property escape were present, the property
307 escape was being ignored.
308
309 43. Fix crash caused by very long (*MARK) or (*THEN) names.
310
311 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
312 by a single ASCII character in a class item, was incorrectly compiled in
313 UCP mode. The POSIX class got lost, but only if the single character
314 followed it.
315
316 45. [:punct:] in UCP mode was matching some characters in the range 128-255
317 that should not have been matched.
318
319 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
320 class, all characters with code points greater than 255 are in the class.
321 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
322 such as \w are turned into Unicode properties), wide characters were not
323 correctly handled, and could fail to match.
324
325
326 Version 8.37 28-April-2015
327 --------------------------
328
329 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
330 for those parentheses to be closed with whatever has been captured so far.
331 However, it was failing to mark any other groups between the hightest
332 capture so far and the currrent group as "unset". Thus, the ovector for
333 those groups contained whatever was previously there. An example is the
334 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
335
336 2. If an assertion condition was quantified with a minimum of zero (an odd
337 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
338
339 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
340 unrecognized modifier, a crash could occur.
341
342 4. An attempt to do global matching in pcretest with a zero-length ovector
343 caused a crash.
344
345 5. Fixed a memory leak during matching that could occur for a subpattern
346 subroutine call (recursive or otherwise) if the number of captured groups
347 that had to be saved was greater than ten.
348
349 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
350 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
351 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
352
353 7. A UTF pattern containing a "not" match of a non-ASCII character and a
354 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
355
356 8. When a pattern is compiled, it remembers the highest back reference so that
357 when matching, if the ovector is too small, extra memory can be obtained to
358 use instead. A conditional subpattern whose condition is a check on a
359 capture having happened, such as, for example in the pattern
360 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
361 setting the highest backreference number. This mattered only if pcre_exec()
362 was called with an ovector that was too small to hold the capture, and there
363 was no other kind of back reference (a situation which is probably quite
364 rare). The effect of the bug was that the condition was always treated as
365 FALSE when the capture could not be consulted, leading to a incorrect
366 behaviour by pcre_exec(). This bug has been fixed.
367
368 9. A reference to a duplicated named group (either a back reference or a test
369 for being set in a conditional) that occurred in a part of the pattern where
370 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
371 to be incorrectly calculated, leading to overwriting.
372
373 10. A mutually recursive set of back references such as (\2)(\1) caused a
374 segfault at study time (while trying to find the minimum matching length).
375 The infinite loop is now broken (with the minimum length unset, that is,
376 zero).
377
378 11. If an assertion that was used as a condition was quantified with a minimum
379 of zero, matching went wrong. In particular, if the whole group had
380 unlimited repetition and could match an empty string, a segfault was
381 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
382 assertions to be quantified, but not if they are being used as conditions,
383 so the above pattern is faulted by Perl. PCRE has now been changed so that
384 it also rejects such patterns.
385
386 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
387 failed to allow the zero-repeat case if pcre2_exec() was called with an
388 ovector too small to capture the group.
389
390 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
391 Red Hat Product Security:
392
393 (a) A crash if /K and /F were both set with the option to save the compiled
394 pattern.
395
396 (b) Another crash if the option to print captured substrings in a callout
397 was combined with setting a null ovector, for example \O\C+ as a subject
398 string.
399
400 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
401 forward reference repeated a large (but limited) number of times within a
402 repeated outer group that has a zero minimum quantifier, caused incorrect
403 code to be compiled, leading to the error "internal error:
404 previously-checked referenced subpattern not found" when an incorrect
405 memory address was read. This bug was reported as "heap overflow",
406 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
407 CVE-2015-2325.
408
409 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
410 call within a group that also contained a recursive back reference caused
411 incorrect code to be compiled. This bug was reported as "heap overflow",
412 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
413 number CVE-2015-2326.
414
415 24. Computing the size of the JIT read-only data in advance has been a source
416 of various issues, and new ones are still appear unfortunately. To fix
417 existing and future issues, size computation is eliminated from the code,
418 and replaced by on-demand memory allocation.
419
420 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
421 adjacent to the end of the range, and the range contained characters with
422 more than one other case, caused incorrect behaviour when compiled in UTF
423 mode. In that example, the range a-j was left out of the class.
424
425 26. Fix JIT compilation of conditional blocks, which assertion
426 is converted to (*FAIL). E.g: /(?(?!))/.
427
428 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
429 discovered by the LLVM fuzzer.
430
431 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
432 when this assertion was used as a condition, for example (?(?!)a|b). In
433 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
434 error about an unsupported item.
435
436 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
437 possessification code could take exponential time to complete. A recursion
438 depth limit of 1000 has been imposed to limit the resources used by this
439 optimization.
440
441 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
442 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
443 because \S ensures they are all in the class. The code for doing this was
444 interacting badly with the code for computing the amount of space needed to
445 compile the pattern, leading to a buffer overflow. This bug was discovered
446 by the LLVM fuzzer.
447
448 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
449 other kinds of group caused stack overflow at compile time. This bug was
450 discovered by the LLVM fuzzer.
451
452 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
453 between a subroutine call and its quantifier was incorrectly compiled,
454 leading to buffer overflow or other errors. This bug was discovered by the
455 LLVM fuzzer.
456
457 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
458 assertion after (?(. The code was failing to check the character after
459 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
460 was discovered by the LLVM fuzzer.
461
462 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
463 a fixed maximum following a group that contains a subroutine reference was
464 incorrectly compiled and could trigger buffer overflow. This bug was
465 discovered by the LLVM fuzzer.
466
467 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
468 caused a stack overflow instead of the diagnosis of a non-fixed length
469 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
470
471 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
472 (e.g. /(?<=\Ka)/) could make pcregrep loop.
473
474 37. There was a similar problem to 36 in pcretest for global matches.
475
476 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
477 and a subsequent item in the pattern caused a non-match, backtracking over
478 the repeated \X did not stop, but carried on past the start of the subject,
479 causing reference to random memory and/or a segfault. There were also some
480 other cases where backtracking after \C could crash. This set of bugs was
481 discovered by the LLVM fuzzer.
482
483 39. The function for finding the minimum length of a matching string could take
484 a very long time if mutual recursion was present many times in a pattern,
485 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
486 method has been implemented. This infelicity was discovered by the LLVM
487 fuzzer.
488
489 40. Static linking against the PCRE library using the pkg-config module was
490 failing on missing pthread symbols.
491
492
493 Version 8.36 26-September-2014
494 ------------------------------
495
496 1. Got rid of some compiler warnings in the C++ modules that were shown up by
497 -Wmissing-field-initializers and -Wunused-parameter.
498
499 2. The tests for quantifiers being too big (greater than 65535) were being
500 applied after reading the number, and stupidly assuming that integer
501 overflow would give a negative number. The tests are now applied as the
502 numbers are read.
503
504 3. Tidy code in pcre_exec.c where two branches that used to be different are
505 now the same.
506
507 4. The JIT compiler did not generate match limit checks for certain
508 bracketed expressions with quantifiers. This may lead to exponential
509 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
510 issue should be resolved now.
511
512 5. Fixed an issue, which occures when nested alternatives are optimized
513 with table jumps.
514
515 6. Inserted two casts and changed some ints to size_t in the light of some
516 reported 64-bit compiler warnings (Bugzilla 1477).
517
518 7. Fixed a bug concerned with zero-minimum possessive groups that could match
519 an empty string, which sometimes were behaving incorrectly in the
520 interpreter (though correctly in the JIT matcher). This pcretest input is
521 an example:
522
523 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
524 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
525
526 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
527 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
528 for an empty string was breaking the inner loop and carrying on at a lower
529 level, when possessive repeated groups should always return to a higher
530 level as they have no backtrack points in them. The empty string test now
531 occurs at the outer level.
532
533 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
534 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
535
536 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
537 doesn't).
538
539 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
540 to hold a pattern to be incorrectly computed (too small) when there were
541 named back references to duplicated names. This could cause "internal
542 error: code overflow" or "double free or corruption" or other memory
543 handling errors.
544
545 11. When named subpatterns had the same prefixes, back references could be
546 confused. For example, in this pattern:
547
548 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
549
550 the reference to 'Name' was incorrectly treated as a reference to a
551 duplicate name.
552
553 12. A pattern such as /^s?c/mi8 where the optional character has more than
554 one "other case" was incorrectly compiled such that it would only try to
555 match starting at "c".
556
557 13. When a pattern starting with \s was studied, VT was not included in the
558 list of possible starting characters; this should have been part of the
559 8.34/18 patch.
560
561 14. If a character class started [\Qx]... where x is any character, the class
562 was incorrectly terminated at the ].
563
564 15. If a pattern that started with a caseless match for a character with more
565 than one "other case" was studied, PCRE did not set up the starting code
566 unit bit map for the list of possible characters. Now it does. This is an
567 optimization improvement, not a bug fix.
568
569 16. The Unicode data tables have been updated to Unicode 7.0.0.
570
571 17. Fixed a number of memory leaks in pcregrep.
572
573 18. Avoid a compiler warning (from some compilers) for a function call with
574 a cast that removes "const" from an lvalue by using an intermediate
575 variable (to which the compiler does not object).
576
577 19. Incorrect code was compiled if a group that contained an internal recursive
578 back reference was optional (had quantifier with a minimum of zero). This
579 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
580 caused segmentation faults because of stack overflows at compile time.
581
582 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
583 group that is quantified with an indefinite repeat, caused a compile-time
584 loop which used up all the system stack and provoked a segmentation fault.
585 This was not the same bug as 19 above.
586
587 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
588 Patch by Mike Frysinger.
589
590
591 Version 8.35 04-April-2014
592 --------------------------
593
594 1. A new flag is set, when property checks are present in an XCLASS.
595 When this flag is not set, PCRE can perform certain optimizations
596 such as studying these XCLASS-es.
597
598 2. The auto-possessification of character sets were improved: a normal
599 and an extended character set can be compared now. Furthermore
600 the JIT compiler optimizes more character set checks.
601
602 3. Got rid of some compiler warnings for potentially uninitialized variables
603 that show up only when compiled with -O2.
604
605 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
606 of a match later then the end of the match. The pcretest program was not
607 handling the case sensibly - it was outputting from the start to the next
608 binary zero. It now reports this situation in a message, and outputs the
609 text from the end to the start.
610
611 5. Fast forward search is improved in JIT. Instead of the first three
612 characters, any three characters with fixed position can be searched.
613 Search order: first, last, middle.
614
615 6. Improve character range checks in JIT. Characters are read by an inprecise
616 function now, which returns with an unknown value if the character code is
617 above a certain threshold (e.g: 256). The only limitation is that the value
618 must be bigger than the threshold as well. This function is useful when
619 the characters above the threshold are handled in the same way.
620
621 7. The macros whose names start with RAWUCHAR are placeholders for a future
622 mode in which only the bottom 21 bits of 32-bit data items are used. To
623 make this more memorable for those maintaining the code, the names have
624 been changed to start with UCHAR21, and an extensive comment has been added
625 to their definition.
626
627 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
628 to the export list in Makefile.am (they were accidentally omitted from the
629 8.34 tarball).
630
631 9. The informational output from pcretest used the phrase "starting byte set"
632 which is inappropriate for the 16-bit and 32-bit libraries. As the output
633 for "first char" and "need char" really means "non-UTF-char", I've changed
634 "byte" to "char", and slightly reworded the output. The documentation about
635 these values has also been (I hope) clarified.
636
637 10. Another JIT related optimization: use table jumps for selecting the correct
638 backtracking path, when more than four alternatives are present inside a
639 bracket.
640
641 11. Empty match is not possible, when the minimum length is greater than zero,
642 and there is no \K in the pattern. JIT should avoid empty match checks in
643 such cases.
644
645 12. In a caseless character class with UCP support, when a character with more
646 than one alternative case was not the first character of a range, not all
647 the alternative cases were added to the class. For example, s and \x{17f}
648 are both alternative cases for S: the class [RST] was handled correctly,
649 but [R-T] was not.
650
651 13. The configure.ac file always checked for pthread support when JIT was
652 enabled. This is not used in Windows, so I have put this test inside a
653 check for the presence of windows.h (which was already tested for).
654
655 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
656 The algorithm provides a way to skip certain starting offsets, and usually
657 faster than linear prefix searches.
658
659 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
660 as for 'fr_FR' and 'french'. For some reason, however, it then used the
661 Windows-specific input and output files, which have 'french' screwed in.
662 So this could never have worked. One of the problems with locales is that
663 they aren't always the same. I have now updated RunTest so that it checks
664 the output of the locale test (test 3) against three different output
665 files, and it allows the test to pass if any one of them matches. With luck
666 this should make the test pass on some versions of Solaris where it was
667 failing. Because of the uncertainty, the script did not used to stop if
668 test 3 failed; it now does. If further versions of a French locale ever
669 come to light, they can now easily be added.
670
671 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
672 there was a message during ./configure, but it did not stop. This now
673 provokes an error. The invalid example in README has been corrected.
674 If a value less than the minimum is given, the minimum value has always
675 been used, but now a warning is given.
676
677 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
678 was a bug in the test system, which is now fixed. Also, the list of various
679 configurations that are tested for each release did not have one with both
680 16/32 bits and --enable-bar-anycrlf. It now does.
681
682 18. pcretest was missing "-C bsr" for displaying the \R default setting.
683
684 19. Little endian PowerPC systems are supported now by the JIT compiler.
685
686 20. The fast forward newline mechanism could enter to an infinite loop on
687 certain invalid UTF-8 input. Although we don't support these cases
688 this issue can be fixed by a performance optimization.
689
690 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
691 not take account if existing stack usage. There is now a new global
692 variable called pcre_stack_guard that can be set to point to an external
693 function to check stack availability. It is called at the start of
694 processing every parenthesized group.
695
696 22. A typo in the code meant that in ungreedy mode the max/min qualifier
697 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
698 not match "ab".
699
700 23. When UTF was disabled, the JIT program reported some incorrect compile
701 errors. These messages are silenced now.
702
703 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
704 compiler.
705
706 25. Change all the temporary files used in RunGrepTest to be different to those
707 used by RunTest so that the tests can be run simultaneously, for example by
708 "make -j check".
709
710
711 Version 8.34 15-December-2013
712 -----------------------------
713
714 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
715 executable memory. Patch inspired by Carsten Klein.
716
717 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
718 this macro is never tested and has no effect, because the work to support
719 coverage involves only compiling and linking options and special targets in
720 the Makefile. The comment in config.h implied that defining the macro would
721 enable coverage support, which is totally false. There was also support for
722 setting this macro in the CMake files (my fault, I just copied it from
723 configure). SUPPORT_GCOV has now been removed.
724
725 3. Make a small performance improvement in strlen16() and strlen32() in
726 pcretest.
727
728 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
729 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
730 these cases). There was also one in pcretest.c.
731
732 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
733
734 6. In UTF mode, the code for checking whether a group could match an empty
735 string (which is used for indefinitely repeated groups to allow for
736 breaking an infinite loop) was broken when the group contained a repeated
737 negated single-character class with a character that occupied more than one
738 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
739 UTF-8 mode). The effect was undefined: the group might or might not be
740 deemed as matching an empty string, or the program might have crashed.
741
742 7. The code for checking whether a group could match an empty string was not
743 recognizing that \h, \H, \v, \V, and \R must match a character.
744
745 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
746 an empty string. If it can, pcretest shows this in its information output.
747
748 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
749 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
750 matched by pcre_exec() without using JIT:
751
752 (a) If the rest of the pattern did not match after a maximal run of
753 grapheme clusters, the code for backing up to try with fewer of them
754 did not always back up over a full grapheme when characters that do not
755 have the modifier quality were involved, e.g. Hangul syllables.
756
757 (b) If the match point in a subject started with modifier character, and
758 there was no match, the code could incorrectly back up beyond the match
759 point, and potentially beyond the first character in the subject,
760 leading to a segfault or an incorrect match result.
761
762 10. A conditional group with an assertion condition could lead to PCRE
763 recording an incorrect first data item for a match if no other first data
764 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
765 first data item, and therefore matched "ca" after "c" instead of at the
766 start.
767
768 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
769 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
770
771 12. The source of pcregrep now includes z/OS-specific code so that it can be
772 compiled for z/OS as part of the special z/OS distribution.
773
774 13. Added the -T and -TM options to pcretest.
775
776 14. The code in pcre_compile.c for creating the table of named capturing groups
777 has been refactored. Instead of creating the table dynamically during the
778 actual compiling pass, the information is remembered during the pre-compile
779 pass (on the stack unless there are more than 20 named groups, in which
780 case malloc() is used) and the whole table is created before the actual
781 compile happens. This has simplified the code (it is now nearly 150 lines
782 shorter) and prepared the way for better handling of references to groups
783 with duplicate names.
784
785 15. A back reference to a named subpattern when there is more than one of the
786 same name now checks them in the order in which they appear in the pattern.
787 The first one that is set is used for the reference. Previously only the
788 first one was inspected. This change makes PCRE more compatible with Perl.
789
790 16. Unicode character properties were updated from Unicode 6.3.0.
791
792 17. The compile-time code for auto-possessification has been refactored, based
793 on a patch by Zoltan Herczeg. It now happens after instead of during
794 compilation. The code is cleaner, and more cases are handled. The option
795 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
796 options in pcretest are provided to set it. It can also be set by
797 (*NO_AUTO_POSSESS) at the start of a pattern.
798
799 18. The character VT has been added to the default ("C" locale) set of
800 characters that match \s and are generally treated as white space,
801 following this same change in Perl 5.18. There is now no difference between
802 "Perl space" and "POSIX space". Whether VT is treated as white space in
803 other locales depends on the locale.
804
805 19. The code for checking named groups as conditions, either for being set or
806 for being recursed, has been refactored (this is related to 14 and 15
807 above). Processing unduplicated named groups should now be as fast at
808 numerical groups, and processing duplicated groups should be faster than
809 before.
810
811 20. Two patches to the CMake build system, by Alexander Barkov:
812
813 (1) Replace the "source" command by "." in CMakeLists.txt because
814 "source" is a bash-ism.
815
816 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
817 without these the CMake build does not work on Solaris.
818
819 21. Perl has changed its handling of \8 and \9. If there is no previously
820 encountered capturing group of those numbers, they are treated as the
821 literal characters 8 and 9 instead of a binary zero followed by the
822 literals. PCRE now does the same.
823
824 22. Following Perl, added \o{} to specify codepoints in octal, making it
825 possible to specify values greater than 0777 and also making them
826 unambiguous.
827
828 23. Perl now gives an error for missing closing braces after \x{... instead of
829 treating the string as literal. PCRE now does the same.
830
831 24. RunTest used to grumble if an inappropriate test was selected explicitly,
832 but just skip it when running all tests. This make it awkward to run ranges
833 of tests when one of them was inappropriate. Now it just skips any
834 inappropriate tests, as it always did when running all tests.
835
836 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
837 character types such as \d or \w, too many callouts were inserted, and the
838 data that they returned was rubbish.
839
840 26. In UCP mode, \s was not matching two of the characters that Perl matches,
841 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
842 were matched by \h. The code has now been refactored so that the lists of
843 the horizontal and vertical whitespace characters used for \h and \v (which
844 are defined only in one place) are now also used for \s.
845
846 27. Add JIT support for the 64 bit TileGX architecture.
847 Patch by Jiong Wang (Tilera Corporation).
848
849 28. Possessive quantifiers for classes (both explicit and automatically
850 generated) now use special opcodes instead of wrapping in ONCE brackets.
851
852 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
853 (because it's meaningless), this was not happening when PCRE_CASELESS was
854 set. Not wrong, but inefficient.
855
856 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
857 properties for \w, \d, etc) is present in a test regex. Otherwise if the
858 test contains no characters greater than 255, Perl doesn't realise it
859 should be using Unicode semantics.
860
861 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
862 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
863 does in Unicode mode.
864
865 32. Added the "forbid" facility to pcretest so that putting tests into the
866 wrong test files can sometimes be quickly detected.
867
868 33. There is now a limit (default 250) on the depth of nesting of parentheses.
869 This limit is imposed to control the amount of system stack used at compile
870 time. It can be changed at build time by --with-parens-nest-limit=xxx or
871 the equivalent in CMake.
872
873 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
874 errors. Perl warns for these when in warning mode, but PCRE has no facility
875 for giving warnings.
876
877 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
878 However, this was not working for (?!) because it is optimized to (*FAIL),
879 for which PCRE does not allow quantifiers. The optimization is now disabled
880 when a quantifier follows (?!). I can't see any use for this, but it makes
881 things uniform.
882
883 36. Perl no longer allows group names to start with digits, so I have made this
884 change also in PCRE. It simplifies the code a bit.
885
886 37. In extended mode, Perl ignores spaces before a + that indicates a
887 possessive quantifier. PCRE allowed a space before the quantifier, but not
888 before the possessive +. It now does.
889
890 38. The use of \K (reset reported match start) within a repeated possessive
891 group such as (a\Kb)*+ was not working.
892
893 40. Document that the same character tables must be used at compile time and
894 run time, and that the facility to pass tables to pcre_exec() and
895 pcre_dfa_exec() is for use only with saved/restored patterns.
896
897 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
898 features for Builds with MSVC:
899
900 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
901 builds). The .rc files can be used to set FileDescription and many other
902 attributes.
903
904 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
905 This allows higher-level build scripts which want .pdb files to avoid
906 hard-coding the exact files needed."
907
908 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
909 mean "start of word" and "end of word", respectively, as a transition aid.
910
911 43. A minimizing repeat of a class containing codepoints greater than 255 in
912 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
913 compiled to use the heap for recursion.
914
915 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
916 is configured.
917
918
919 Version 8.33 28-May-2013
920 ------------------------
921
922 1. Added 'U' to some constants that are compared to unsigned integers, to
923 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
924 variables that are added to signed variables, to ensure the result is
925 signed and can be negated.
926
927 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
928 CMake config files.
929
930 3. Revise the creation of config.h.generic so that all boolean macros are
931 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
932 overriding via -D on the command line possible.
933
934 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
935 to unsigned int is reported to make a quite noticeable speed difference in
936 a specific Windows environment. Testing on Linux did also appear to show
937 some benefit (and it is clearly not harmful). Also fixed the definition of
938 Xop which should be unsigned.
939
940 5. Related to (4), changing the definition of the intermediate variable cc
941 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
942 improvements.
943
944 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
945 unnecessary spaces.
946
947 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
948 and later.
949
950 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
951
952 9. Optimizing fast_forward_start_bits in JIT.
953
954 10. Adding support for callouts in JIT, and fixing some issues revealed
955 during this work. Namely:
956
957 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
958
959 (b) Minimum length was not checked before the matching is started.
960
961 11. The value of capture_last that is passed to callouts was incorrect in some
962 cases when there was a capture on one path that was subsequently abandoned
963 after a backtrack. Also, the capture_last value is now reset after a
964 recursion, since all captures are also reset in this case.
965
966 12. The interpreter no longer returns the "too many substrings" error in the
967 case when an overflowing capture is in a branch that is subsequently
968 abandoned after a backtrack.
969
970 13. In the pathological case when an offset vector of size 2 is used, pcretest
971 now prints out the matched string after a yield of 0 or 1.
972
973 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
974 Only supported by the JIT compiler at the moment.
975
976 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
977
978 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
979 offset of the starting point of the matching process, provided the offsets
980 vector is large enough.
981
982 17. The \A escape now records a lookbehind value of 1, though its execution
983 does not actually inspect the previous character. This is to ensure that,
984 in partial multi-segment matching, at least one character from the old
985 segment is retained when a new segment is processed. Otherwise, if there
986 are no lookbehinds in the pattern, \A might match incorrectly at the start
987 of a new segment.
988
989 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
990
991 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
992 gives some modest performance improvement in 8-bit mode.
993
994 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
995 be expressed in certain programming languages using Universal Character
996 Names.
997
998 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
999 which points out that "non characters" are not "characters that may not
1000 appear in Unicode strings" but rather "characters that are reserved for
1001 internal use and have only local meaning".
1002
1003 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
1004 there was a conditional group that depended on an assertion, if the
1005 assertion was false, the callout that immediately followed the alternation
1006 in the condition was skipped when pcre_exec() was used for matching.
1007
1008 23. Allow an explicit callout to be inserted before an assertion that is the
1009 condition for a conditional group, for compatibility with automatic
1010 callouts, which always insert a callout at this point.
1011
1012 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1013 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1014
1015 25. (*PRUNE) is now supported by the JIT compiler.
1016
1017 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1018
1019 27. Fix the case where there are two or more SKIPs with arguments that may be
1020 ignored.
1021
1022 28. (*SKIP) is now supported by the JIT compiler.
1023
1024 29. (*THEN) is now supported by the JIT compiler.
1025
1026 30. Update RunTest with additional test selector options.
1027
1028 31. The way PCRE handles backtracking verbs has been changed in two ways.
1029
1030 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1031 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1032 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1033 rather obscure rules do not always do the same thing.
1034
1035 (2) Previously, backtracking verbs were confined within assertions. This is
1036 no longer the case for positive assertions, except for (*ACCEPT). Again,
1037 this sometimes improves Perl compatibility, and sometimes does not.
1038
1039 32. A number of tests that were in test 2 because Perl did things differently
1040 have been moved to test 1, because either Perl or PCRE has changed, and
1041 these tests are now compatible.
1042
1043 32. Backtracking control verbs are now handled in the same way in JIT and
1044 interpreter.
1045
1046 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1047 contained a forward subroutine reference caused a compile error.
1048
1049 34. Auto-detect and optimize limited repetitions in JIT.
1050
1051 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1052 blocking (*UTF) etc.
1053
1054 36. In the interpreter, maximizing pattern repetitions for characters and
1055 character types now use tail recursion, which reduces stack usage.
1056
1057 37. The value of the max lookbehind was not correctly preserved if a compiled
1058 and saved regex was reloaded on a host of different endianness.
1059
1060 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1061 of the compiled pattern block, expand the flags field from 16 to 32 bits
1062 because it was almost full.
1063
1064 39. Try madvise first before posix_madvise.
1065
1066 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1067 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1068 original change locked out all matches of empty strings. This has been
1069 changed so that one match of an empty string per line is recognized.
1070 Subsequent searches on the same line (for colouring or for --only-matching,
1071 for example) do not recognize empty strings.
1072
1073 41. Applied a user patch to fix a number of spelling mistakes in comments.
1074
1075 42. Data lines longer than 65536 caused pcretest to crash.
1076
1077 43. Clarified the data type for length and startoffset arguments for pcre_exec
1078 and pcre_dfa_exec in the function-specific man pages, where they were
1079 explicitly stated to be in bytes, never having been updated. I also added
1080 some clarification to the pcreapi man page.
1081
1082 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1083 a segmentation fault.
1084
1085
1086 Version 8.32 30-November-2012
1087 -----------------------------
1088
1089 1. Improved JIT compiler optimizations for first character search and single
1090 character iterators.
1091
1092 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1093 Patch by Daniel Richard G.
1094
1095 3. Single character iterator optimizations in the JIT compiler.
1096
1097 4. Improved JIT compiler optimizations for character ranges.
1098
1099 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1100 Reported by Giuseppe D'Angelo.
1101
1102 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1103 of a line, was being set incorrectly in cases where .* appeared inside
1104 atomic brackets at the start of a pattern, or where there was a subsequent
1105 *PRUNE or *SKIP.
1106
1107 7. Improved instruction cache flush for POWER/PowerPC.
1108 Patch by Daniel Richard G.
1109
1110 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1111 grep:
1112
1113 (a) There is now no limit to the number of patterns to be matched.
1114
1115 (b) An error is given if a pattern is too long.
1116
1117 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1118 are now supported.
1119
1120 (d) --exclude-from and --include-from (multiple use) have been added.
1121
1122 (e) Exclusions and inclusions now apply to all files and directories, not
1123 just to those obtained from scanning a directory recursively.
1124
1125 (f) Multiple uses of -f and --file-list are now supported.
1126
1127 (g) In a Windows environment, the default for -d has been changed from
1128 "read" (the GNU grep default) to "skip", because otherwise the presence
1129 of a directory in the file list provokes an error.
1130
1131 (h) The documentation has been revised and clarified in places.
1132
1133 9. Improve the matching speed of capturing brackets.
1134
1135 10. Changed the meaning of \X so that it now matches a Unicode extended
1136 grapheme cluster.
1137
1138 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1139 out POSIX threads when JIT support is configured.
1140
1141 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1142
1143 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1144 structure could go wrong in environments where size_t is not the same size
1145 as int.
1146
1147 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1148 set.
1149
1150 15. The EBCDIC support had decayed; later updates to the code had included
1151 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1152 general tidy up of EBCDIC-related issues, and the documentation was also
1153 not quite right. There is now a test that can be run on ASCII systems to
1154 check some of the EBCDIC-related things (but is it not a full test).
1155
1156 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1157 in a small tidy to the code.
1158
1159 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1160
1161 18. If the --only-matching (-o) option in pcregrep is specified multiple
1162 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1163 substrings matched by the 1st and 2nd capturing parentheses. A separating
1164 string can be specified by --om-separator (default empty).
1165
1166 19. Improving the first n character searches.
1167
1168 20. Turn case lists for horizontal and vertical white space into macros so that
1169 they are defined only once.
1170
1171 21. This set of changes together give more compatible Unicode case-folding
1172 behaviour for characters that have more than one other case when UCP
1173 support is available.
1174
1175 (a) The Unicode property table now has offsets into a new table of sets of
1176 three or more characters that are case-equivalent. The MultiStage2.py
1177 script that generates these tables (the pcre_ucd.c file) now scans
1178 CaseFolding.txt instead of UnicodeData.txt for character case
1179 information.
1180
1181 (b) The code for adding characters or ranges of characters to a character
1182 class has been abstracted into a generalized function that also handles
1183 case-independence. In UTF-mode with UCP support, this uses the new data
1184 to handle characters with more than one other case.
1185
1186 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1187 whose other case is greater than 256 are now correctly matched
1188 caselessly. Previously, the high codepoint matched the low one, but not
1189 vice versa.
1190
1191 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1192 of the new class addition function, using character lists defined as
1193 macros alongside the case definitions of 20 above.
1194
1195 (e) Caseless back references now work with characters that have more than
1196 one other case.
1197
1198 (f) General caseless matching of characters with more than one other case
1199 is supported.
1200
1201 22. Unicode character properties were updated from Unicode 6.2.0
1202
1203 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1204
1205 24. Add support for 32-bit character strings, and UTF-32
1206
1207 25. Major JIT compiler update (code refactoring and bugfixing).
1208 Experimental Sparc 32 support is added.
1209
1210 26. Applied a modified version of Daniel Richard G's patch to create
1211 pcre.h.generic and config.h.generic by "make" instead of in the
1212 PrepareRelease script.
1213
1214 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1215 pcre_compile.c when checking for a zero character.
1216
1217 28. Introducing a native interface for JIT. Through this interface, the compiled
1218 machine code can be directly executed. The purpose of this interface is to
1219 provide fast pattern matching, so several sanity checks are not performed.
1220 However, feature tests are still performed. The new interface provides
1221 1.4x speedup compared to the old one.
1222
1223 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1224 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1225 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1226
1227 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1228 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1229 the "old" RFC 2279). Instead, it ended up passing a negative length to
1230 pcre_exec().
1231
1232 31. Add support for GCC's visibility feature to hide internal functions.
1233
1234 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1235 "unknown -C option" after outputting 0 or 1.
1236
1237 33. There is now support for generating a code coverage report for the test
1238 suite in environments where gcc is the compiler and lcov is installed. This
1239 is mainly for the benefit of the developers.
1240
1241 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1242 unaddressable using valgrind annotations, allowing valgrind to detect
1243 invalid memory accesses. This is mainly for the benefit of the developers.
1244
1245 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1246
1247 26. Give configure error if --enable-cpp but no C++ compiler found.
1248
1249
1250 Version 8.31 06-July-2012
1251 -------------------------
1252
1253 1. Fixing a wrong JIT test case and some compiler warnings.
1254
1255 2. Removed a bashism from the RunTest script.
1256
1257 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1258 to unsigned type, result still unsigned" that was given by an MS compiler
1259 on encountering the code "-sizeof(xxx)".
1260
1261 4. Partial matching support is added to the JIT compiler.
1262
1263 5. Fixed several bugs concerned with partial matching of items that consist
1264 of more than one character:
1265
1266 (a) /^(..)\1/ did not partially match "aba" because checking references was
1267 done on an "all or nothing" basis. This also applied to repeated
1268 references.
1269
1270 (b) \R did not give a hard partial match if \r was found at the end of the
1271 subject.
1272
1273 (c) \X did not give a hard partial match after matching one or more
1274 characters at the end of the subject.
1275
1276 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1277 a partial match for the string "\r".
1278
1279 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1280 a partial match for a CR character at the end of the subject string.
1281
1282 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1283 running pcretest, the text "(JIT)" added to the output whenever JIT is
1284 actually used to run the match.
1285
1286 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1287 or /S+[+] with a digit between 1 and 7.
1288
1289 8. OP_NOT now supports any UTF character not just single-byte ones.
1290
1291 9. (*MARK) control verb is now supported by the JIT compiler.
1292
1293 10. The command "./RunTest list" lists the available tests without actually
1294 running any of them. (Because I keep forgetting what they all are.)
1295
1296 11. Add PCRE_INFO_MAXLOOKBEHIND.
1297
1298 12. Applied a (slightly modified) user-supplied patch that improves performance
1299 when the heap is used for recursion (compiled with --disable-stack-for-
1300 recursion). Instead of malloc and free for each heap frame each time a
1301 logical recursion happens, frames are retained on a chain and re-used where
1302 possible. This sometimes gives as much as 30% improvement.
1303
1304 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1305 call.
1306
1307 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1308
1309 15. It is now possible to link pcretest with libedit as an alternative to
1310 libreadline.
1311
1312 16. (*COMMIT) control verb is now supported by the JIT compiler.
1313
1314 17. The Unicode data tables have been updated to Unicode 6.1.0.
1315
1316 18. Added --file-list option to pcregrep.
1317
1318 19. Added binary file support to pcregrep, including the -a, --binary-files,
1319 -I, and --text options.
1320
1321 20. The madvise function is renamed for posix_madvise for QNX compatibility
1322 reasons. Fixed by Giuseppe D'Angelo.
1323
1324 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1325
1326 22. Changed the option for creating symbolic links for 16-bit man pages from
1327 -s to -sf so that re-installing does not cause issues.
1328
1329 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1330
1331 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1332 matches in certain environments (the workspace was not being correctly
1333 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1334 some of the workspace data at the beginning of a restart.
1335
1336 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1337 was not doing so when it should - probably a typo introduced by SVN 528
1338 (change 8.10/14).
1339
1340 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1341 \w+ when the character tables indicated that \x{c4} was a word character.
1342 There were several related cases, all because the tests for doing a table
1343 lookup were testing for characters less than 127 instead of 255.
1344
1345 27. If a pattern contains capturing parentheses that are not used in a match,
1346 their slots in the ovector are set to -1. For those that are higher than
1347 any matched groups, this happens at the end of processing. In the case when
1348 there were back references that the ovector was too small to contain
1349 (causing temporary malloc'd memory to be used during matching), and the
1350 highest capturing number was not used, memory off the end of the ovector
1351 was incorrectly being set to -1. (It was using the size of the temporary
1352 memory instead of the true size.)
1353
1354 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1355 ovector size, it uses memory at the end of the block that it has got.
1356
1357 29. Check for an overlong MARK name and give an error at compile time. The
1358 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1359
1360 30. JIT compiler update.
1361
1362 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1363 Rill for the patch.
1364
1365 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1366
1367 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1368
1369 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1370 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1371 was enabled.
1372
1373 35. Improve JIT code generation for greedy plus quantifier.
1374
1375 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1376 1 to "aa" instead of to an empty string. The bug affected repeated groups
1377 that could potentially match an empty string.
1378
1379 37. Optimizing single character iterators in JIT.
1380
1381 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1382 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1383 codepoints that are too big for the mode are faulted, and in a UTF mode,
1384 disallowed codepoints are also faulted.
1385
1386 39. If PCRE was compiled with UTF support, in three places in the DFA
1387 matcher there was code that should only have been obeyed in UTF mode, but
1388 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1389 processing when bytes with values greater than 127 were present. In 16-bit
1390 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1391 both cases the values are those that cannot be the first data item in a UTF
1392 character. The three items that might have provoked this were recursions,
1393 possessively repeated groups, and atomic groups.
1394
1395 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1396 and pcregrep, because some OS require shared objects to be explicitly
1397 passed to ld, causing the link step to fail if they are not.
1398
1399 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1400 mode, patterns that started with \h* or \R* might be incorrectly matched.
1401
1402
1403 Version 8.30 04-February-2012
1404 -----------------------------
1405
1406 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1407 name is defined in ctype.h.
1408
1409 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1410 only in quite long subpatterns.
1411
1412 3. Removed the function pcre_info(), which has been obsolete and deprecated
1413 since it was replaced by pcre_fullinfo() in February 2000.
1414
1415 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1416 match a (*MARK), and the match failed at the start of the subject, a
1417 reference to memory before the start of the subject could occur. This bug
1418 was introduced by fix 17 of release 8.21.
1419
1420 5. A reference to an unset group with zero minimum repetition was giving
1421 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1422 /(another)?(\1?)test/ matched against "hello world test". This bug was
1423 introduced in release 8.13.
1424
1425 6. Add support for 16-bit character strings (a large amount of work involving
1426 many changes and refactorings).
1427
1428 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1429 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1430 from a file.
1431
1432 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1433 rounding is not applied in this particular case).
1434
1435 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1436 if they appear, or are escaped, in patterns.
1437
1438 10. Get rid of a number of -Wunused-but-set-variable warnings.
1439
1440 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1441 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1442 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1443 also returns the mark "x". This bug applied to capturing parentheses,
1444 non-capturing parentheses, and atomic parentheses. It also applied to some
1445 assertions.
1446
1447 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1448 information out of configure.ac instead of relying on pcre.h.generic, which
1449 is not stored in the repository.
1450
1451 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1452 -lreadline.
1453
1454 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1455
1456 15. Applied Graycode's patch to put the top-level frame on the stack rather
1457 than the heap when not using the stack for recursion. This gives a
1458 performance improvement in many cases when recursion is not deep.
1459
1460 16. Experimental code added to "pcretest -C" to output the stack frame size.
1461
1462
1463 Version 8.21 12-Dec-2011
1464 ------------------------
1465
1466 1. Updating the JIT compiler.
1467
1468 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1469 are added as well.
1470
1471 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1472 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1473 calling _pcre_jit_exec. Some extra comments are added.
1474
1475 4. (*MARK) settings inside atomic groups that do not contain any capturing
1476 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1477 was introduced by change 18 for 8.20.
1478
1479 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1480 ECMA-262 standard.
1481
1482 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1483 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1484 This bug was probably introduced by change 9 of 8.13.
1485
1486 7. While fixing 6 above, I noticed that a number of other items were being
1487 incorrectly rejected as "not fixed length". This arose partly because newer
1488 opcodes had not been added to the fixed-length checking code. I have (a)
1489 corrected the bug and added tests for these items, and (b) arranged for an
1490 error to occur if an unknown opcode is encountered while checking for fixed
1491 length instead of just assuming "not fixed length". The items that were
1492 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1493 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1494 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1495
1496 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1497 being incorrectly compiled and would have given unpredicatble results.
1498
1499 9. A possessively repeated subpattern with minimum repeat count greater than
1500 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1501 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1502 the first (A) could occur when it should not.
1503
1504 10. Add a cast and remove a redundant test from the code.
1505
1506 11. JIT should use pcre_malloc/pcre_free for allocation.
1507
1508 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1509 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1510 is anything other than /usr, -L is still shown).
1511
1512 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1513
1514 14. Perl does not support \N without a following name in a [] class; PCRE now
1515 also gives an error.
1516
1517 15. If a forward reference was repeated with an upper limit of around 2000,
1518 it caused the error "internal error: overran compiling workspace". The
1519 maximum number of forward references (including repeats) was limited by the
1520 internal workspace, and dependent on the LINK_SIZE. The code has been
1521 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1522 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1523 of around 200,000 forward references. While doing this, I also speeded up
1524 the filling in of repeated forward references.
1525
1526 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1527 incorrectly expecting the subject to contain another "a" after the start.
1528
1529 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1530 in the match, the SKIP should be ignored. This was not happening; instead
1531 the SKIP was being treated as NOMATCH. For patterns such as
1532 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1533 tested.
1534
1535 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1536 now much more compatible with Perl, in particular in cases where the result
1537 is a non-match for a non-anchored pattern. For example, if
1538 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1539 "m", where previously it did not return a name. A side effect of this
1540 change is that for partial matches, the last encountered mark name is
1541 returned, as for non matches. A number of tests that were previously not
1542 Perl-compatible have been moved into the Perl-compatible test files. The
1543 refactoring has had the pleasing side effect of removing one argument from
1544 the match() function, thus reducing its stack requirements.
1545
1546 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1547 subsequent uses of /S (without +) incorrectly behaved like /S+.
1548
1549 21. Retrieve executable code size support for the JIT compiler and fixing
1550 some warnings.
1551
1552 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1553 not work when the shorter character appeared right at the end of the
1554 subject string.
1555
1556 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1557 systems.
1558
1559 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1560 output it when the /M option is used in pcretest.
1561
1562 25. The CheckMan script was not being included in the distribution. Also, added
1563 an explicit "perl" to run Perl scripts from the PrepareRelease script
1564 because this is reportedly needed in Windows.
1565
1566 26. If study data was being save in a file and studying had not found a set of
1567 "starts with" bytes for the pattern, the data written to the file (though
1568 never used) was taken from uninitialized memory and so caused valgrind to
1569 complain.
1570
1571 27. Updated RunTest.bat as provided by Sheri Pierce.
1572
1573 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1574
1575 29. Computation of memory usage for the table of capturing group names was
1576 giving an unnecessarily large value.
1577
1578
1579 Version 8.20 21-Oct-2011
1580 ------------------------
1581
1582 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1583 a POSIX class. After further experiments with Perl, which convinced me that
1584 Perl has bugs and confusions, a closing square bracket is no longer allowed
1585 in a POSIX name. This bug also affected patterns with classes that started
1586 with full stops.
1587
1588 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1589 captured substring, but while checking the failing first alternative,
1590 substring 1 is temporarily captured. If the output vector supplied to
1591 pcre_exec() was not big enough for this capture, the yield of the function
1592 was still zero ("insufficient space for captured substrings"). This cannot
1593 be totally fixed without adding another stack variable, which seems a lot
1594 of expense for a edge case. However, I have improved the situation in cases
1595 such as /(a)(b)x|abc/ matched against "abc", where the return code
1596 indicates that fewer than the maximum number of slots in the ovector have
1597 been set.
1598
1599 3. Related to (2) above: when there are more back references in a pattern than
1600 slots in the output vector, pcre_exec() uses temporary memory during
1601 matching, and copies in the captures as far as possible afterwards. It was
1602 using the entire output vector, but this conflicts with the specification
1603 that only 2/3 is used for passing back captured substrings. Now it uses
1604 only the first 2/3, for compatibility. This is, of course, another edge
1605 case.
1606
1607 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1608 main code base, and can be used by building with --enable-jit. When this is
1609 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1610 runtime --no-jit option is given.
1611
1612 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1613 ovector, the return from the function was zero, implying that there were
1614 other matches that did not fit. The correct "exactly full" value is now
1615 returned.
1616
1617 6. If a subpattern that was called recursively or as a subroutine contained
1618 (*PRUNE) or any other control that caused it to give a non-standard return,
1619 invalid errors such as "Error -26 (nested recursion at the same subject
1620 position)" or even infinite loops could occur.
1621
1622 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1623 computing the minimum length on reaching *ACCEPT, and so ended up with the
1624 wrong value of 1 rather than 0. Further investigation indicates that
1625 computing a minimum subject length in the presence of *ACCEPT is difficult
1626 (think back references, subroutine calls), and so I have changed the code
1627 so that no minimum is registered for a pattern that contains *ACCEPT.
1628
1629 8. If (*THEN) was present in the first (true) branch of a conditional group,
1630 it was not handled as intended. [But see 16 below.]
1631
1632 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1633 Sheri Pierce.
1634
1635 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1636 the first byte in a match must be "a".
1637
1638 11. Change 17 for 8.13 increased the recursion depth for patterns like
1639 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1640 pattern contains any instances of (*THEN). If it does not, the old
1641 optimizations are restored. It would be nice to do this on a per-group
1642 basis, but at the moment that is not feasible.
1643
1644 12. In some environments, the output of pcretest -C is CRLF terminated. This
1645 broke RunTest's code that checks for the link size. A single white space
1646 character after the value is now allowed for.
1647
1648 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1649 For "fr", it uses the Windows-specific input and output files.
1650
1651 14. If (*THEN) appeared in a group that was called recursively or as a
1652 subroutine, it did not work as intended. [But see next item.]
1653
1654 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1655 pattern fragments (but not containing any | characters). If A and B are
1656 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1657 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1658 D. In other words, Perl considers parentheses that do not contain any |
1659 characters to be part of a surrounding alternative, whereas PCRE was
1660 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1661 differently. PCRE now behaves in the same way as Perl, except in the case
1662 of subroutine/recursion calls such as (?1) which have in any case always
1663 been different (but PCRE had them first :-).
1664
1665 16. Related to 15 above: Perl does not treat the | in a conditional group as
1666 creating alternatives. Such a group is treated in the same way as an
1667 ordinary group without any | characters when processing (*THEN). PCRE has
1668 been changed to match Perl's behaviour.
1669
1670 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1671 RunGrepTest script failed.
1672
1673 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1674 inevitable for groups that contain captures, but it can lead to a lot of
1675 stack use in large patterns. The old behaviour has been restored for atomic
1676 groups that do not contain any capturing parentheses.
1677
1678 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1679 suppress the check for a minimum subject length at run time. (If it was
1680 given to pcre_exec() or pcre_dfa_exec() it did work.)
1681
1682 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1683 fail to work when decoding hex characters in data strings in EBCDIC
1684 environments.
1685
1686 21. It appears that in at least one Mac OS environment, the isxdigit() function
1687 is implemented as a macro that evaluates to its argument more than once,
1688 contravening the C 90 Standard (I haven't checked a later standard). There
1689 was an instance in pcretest which caused it to go wrong when processing
1690 \x{...} escapes in subject strings. The has been rewritten to avoid using
1691 things like p++ in the argument of isxdigit().
1692
1693
1694 Version 8.13 16-Aug-2011
1695 ------------------------
1696
1697 1. The Unicode data tables have been updated to Unicode 6.0.0.
1698
1699 2. Two minor typos in pcre_internal.h have been fixed.
1700
1701 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1702 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1703 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1704
1705 4. There were a number of related bugs in the code for matching backrefences
1706 caselessly in UTF-8 mode when codes for the characters concerned were
1707 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1708 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1709 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1710 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1711 2-byte code at the end of the subject (it thought there wasn't enough data
1712 left).
1713
1714 5. Comprehensive information about what went wrong is now returned by
1715 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1716 as the output vector has at least 2 elements. The offset of the start of
1717 the failing character and a reason code are placed in the vector.
1718
1719 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1720 now returned is for the first byte of the failing character, instead of the
1721 last byte inspected. This is an incompatible change, but I hope it is small
1722 enough not to be a problem. It makes the returned offset consistent with
1723 pcre_exec() and pcre_dfa_exec().
1724
1725 7. pcretest now gives a text phrase as well as the error number when
1726 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1727 failure, the offset and reason code are output.
1728
1729 8. When \R was used with a maximizing quantifier it failed to skip backwards
1730 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1731 back over a single character (\n). This seems wrong (because it treated the
1732 two characters as a single entity when going forwards), conflicts with the
1733 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1734 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1735 has been changed.
1736
1737 9. Some internal refactoring has changed the processing so that the handling
1738 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1739 time (the PCRE_DOTALL option was changed this way some time ago: version
1740 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1741 which was always a bit of a fudge. It also means that there is one less
1742 argument for the match() function, which reduces its stack requirements
1743 slightly. This change also fixes an incompatibility with Perl: the pattern
1744 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1745
1746 10. More internal refactoring has drastically reduced the number of recursive
1747 calls to match() for possessively repeated groups such as (abc)++ when
1748 using pcre_exec().
1749
1750 11. While implementing 10, a number of bugs in the handling of groups were
1751 discovered and fixed:
1752
1753 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1754 (a|)*(?1) gave a compile-time internal error.
1755 ((a|)+)+ did not notice that the outer group could match an empty string.
1756 (^a|^)+ was not marked as anchored.
1757 (.*a|.*)+ was not marked as matching at start or after a newline.
1758
1759 12. Yet more internal refactoring has removed another argument from the match()
1760 function. Special calls to this function are now indicated by setting a
1761 value in a variable in the "match data" data block.
1762
1763 13. Be more explicit in pcre_study() instead of relying on "default" for
1764 opcodes that mean there is no starting character; this means that when new
1765 ones are added and accidentally left out of pcre_study(), testing should
1766 pick them up.
1767
1768 14. The -s option of pcretest has been documented for ages as being an old
1769 synonym of -m (show memory usage). I have changed it to mean "force study
1770 for every regex", that is, assume /S for every regex. This is similar to -i
1771 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1772 using it. It makes it easier to run collections of tests with and without
1773 study enabled, and thereby test pcre_study() more easily. All the standard
1774 tests are now run with and without -s (but some patterns can be marked as
1775 "never study" - see 20 below).
1776
1777 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1778 restoration of the capturing data to the outer values was not happening
1779 correctly.
1780
1781 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1782 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1783 pattern had matched an empty string, and so incorrectly returned a no
1784 match.
1785
1786 17. There was optimizing code for the last branch of non-capturing parentheses,
1787 and also for the obeyed branch of a conditional subexpression, which used
1788 tail recursion to cut down on stack usage. Unfortunately, now that there is
1789 the possibility of (*THEN) occurring in these branches, tail recursion is
1790 no longer possible because the return has to be checked for (*THEN). These
1791 two optimizations have therefore been removed. [But see 8.20/11 above.]
1792
1793 18. If a pattern containing \R was studied, it was assumed that \R always
1794 matched two bytes, thus causing the minimum subject length to be
1795 incorrectly computed because \R can also match just one byte.
1796
1797 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1798 was incorrectly computed.
1799
1800 20. If /S is present twice on a test pattern in pcretest input, it now
1801 *disables* studying, thereby overriding the use of -s on the command line
1802 (see 14 above). This is necessary for one or two tests to keep the output
1803 identical in both cases.
1804
1805 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1806 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1807
1808 22. When an atomic group that contained a capturing parenthesis was
1809 successfully matched, but the branch in which it appeared failed, the
1810 capturing was not being forgotten if a higher numbered group was later
1811 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1812 group 1 to "a", when in fact it should be unset. This applied to multi-
1813 branched capturing and non-capturing groups, repeated or not, and also to
1814 positive assertions (capturing in negative assertions does not happen
1815 in PCRE) and also to nested atomic groups.
1816
1817 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1818 subject after a captured substring, to make it easier to tell which of a
1819 number of identical substrings has been captured.
1820
1821 24. The way atomic groups are processed by pcre_exec() has been changed so that
1822 if they are repeated, backtracking one repetition now resets captured
1823 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1824 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1825 "aaa". Previously, it would have been "a". As part of this code
1826 refactoring, the way recursive calls are handled has also been changed.
1827
1828 25. If an assertion condition captured any substrings, they were not passed
1829 back unless some other capturing happened later. For example, if
1830 (?(?=(a))a) was matched against "a", no capturing was returned.
1831
1832 26. When studying a pattern that contained subroutine calls or assertions,
1833 the code for finding the minimum length of a possible match was handling
1834 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1835 group 1 called group 2 while simultaneously a separate group 2 called group
1836 1). A stack overflow occurred in this case. I have fixed this by limiting
1837 the recursion depth to 10.
1838
1839 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1840 Fortmann. This supports explicit test numbers on the command line, and has
1841 argument validation and error reporting.
1842
1843 28. An instance of \X with an unlimited repeat could fail if at any point the
1844 first character it looked at was a mark character.
1845
1846 29. Some minor code refactoring concerning Unicode properties and scripts
1847 should reduce the stack requirement of match() slightly.
1848
1849 30. Added the '=' option to pcretest to check the setting of unused capturing
1850 slots at the end of the pattern, which are documented as being -1, but are
1851 not included in the return count.
1852
1853 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1854 compiled something random. Now it gives a compile-time error (as does
1855 Perl).
1856
1857 32. A *MARK encountered during the processing of a positive assertion is now
1858 recorded and passed back (compatible with Perl).
1859
1860 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1861 had alternative anchored branches, the search for a second match in a line
1862 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1863 matched the line "0102" twice. The same bug affected patterns that started
1864 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1865 twice.
1866
1867 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1868 does, and because of capturing effects, quantifying parenthesized
1869 assertions may at times be useful. Quantifiers are now allowed for
1870 parenthesized assertions.
1871
1872 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1873
1874 36. \g was being checked for fancy things in a character class, when it should
1875 just be a literal "g".
1876
1877 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1878 appearance of a nested POSIX class supersedes an apparent external class.
1879 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1880 unescaped square brackets may also appear as part of class names. For
1881 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1882 more like Perl. (But see 8.20/1 above.)
1883
1884 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1885 was because it thought it was \N{name}, which is not supported).
1886
1887 39. Add minix to OS list not supporting the -S option in pcretest.
1888
1889 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1890 cannot analyze patterns in sufficient detail to catch mutual recursions
1891 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1892 subgroup is called recursively as a subpattern for a second time at the
1893 same position in the subject string. In previous releases this might have
1894 been caught by the recursion limit, or it might have run out of stack.
1895
1896 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1897 happen only once. PCRE was, however incorrectly giving a compile time error
1898 "recursive call could loop indefinitely" because it cannot analyze the
1899 pattern in sufficient detail. The compile time test no longer happens when
1900 PCRE is compiling a conditional subpattern, but actual runaway loops are
1901 now caught at runtime (see 40 above).
1902
1903 42. It seems that Perl allows any characters other than a closing parenthesis
1904 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1905 has been changed to be the same.
1906
1907 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1908 as not to get warnings when autogen.sh is called. Also changed
1909 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1910
1911 44. To help people who use pcregrep to scan files containing exceedingly long
1912 lines, the following changes have been made:
1913
1914 (a) The default value of the buffer size parameter has been increased from
1915 8K to 20K. (The actual buffer used is three times this size.)
1916
1917 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1918 PCRE is built.
1919
1920 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1921 to be set at run time.
1922
1923 (d) Numerical values in pcregrep options can be followed by K or M, for
1924 example --buffer-size=50K.
1925
1926 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1927 given and the return code is set to 2.
1928
1929 45. Add a pointer to the latest mark to the callout data block.
1930
1931 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1932 partial match of an empty string instead of no match. This was specific to
1933 the use of ".".
1934
1935 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1936 complete match instead of a partial match. This bug was dependent on both
1937 the PCRE_UTF8 and PCRE_DOTALL options being set.
1938
1939 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1940 starting byte set, because \b was not being ignored.
1941
1942
1943 Version 8.12 15-Jan-2011
1944 ------------------------
1945
1946 1. Fixed some typos in the markup of the man pages, and wrote a script that
1947 checks for such things as part of the documentation building process.
1948
1949 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1950 --match-limit and --recursion-limit options (added for 8.11). In
1951 particular, this made one of the standard tests fail. (The integer value
1952 went into the wrong half of a long int.)
1953
1954 3. If the --colour option was given to pcregrep with -v (invert match), it
1955 did strange things, either producing crazy output, or crashing. It should,
1956 of course, ignore a request for colour when reporting lines that do not
1957 match.
1958
1959 4. Another pcregrep bug caused similar problems if --colour was specified with
1960 -M (multiline) and the pattern match finished with a line ending.
1961
1962 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1963 matched in multiline mode, the following line was shown as part of the
1964 match. This seems wrong, so I have changed it.
1965
1966 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1967 the check for further matches in the same line (so they could be coloured)
1968 to overrun the end of the current line. If another match was found, it was
1969 incorrectly shown (and then shown again when found in the next line).
1970
1971 7. If pcregrep was compiled under Windows, there was a reference to the
1972 function pcregrep_exit() before it was defined. I am assuming this was
1973 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1974 reported by a user. I've moved the definition above the reference.
1975
1976
1977 Version 8.11 10-Dec-2010
1978 ------------------------
1979
1980 1. (*THEN) was not working properly if there were untried alternatives prior
1981 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1982 backtracked to try for "b" instead of moving to the next alternative branch
1983 at the same level (in this case, to look for "c"). The Perl documentation
1984 is clear that when (*THEN) is backtracked onto, it goes to the "next
1985 alternative in the innermost enclosing group".
1986
1987 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1988 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1989 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1990 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1991 (*THEN).
1992
1993 3. If \s appeared in a character class, it removed the VT character from
1994 the class, even if it had been included by some previous item, for example
1995 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1996 of \s, but is part of the POSIX "space" class.)
1997
1998 4. A partial match never returns an empty string (because you can always
1999 match an empty string at the end of the subject); however the checking for
2000 an empty string was starting at the "start of match" point. This has been
2001 changed to the "earliest inspected character" point, because the returned
2002 data for a partial match starts at this character. This means that, for
2003 example, /(?<=abc)def/ gives a partial match for the subject "abc"
2004 (previously it gave "no match").
2005
2006 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2007 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2008 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2009 has an implication that the given string is incomplete (because a partial
2010 match is preferred over a full match). For this reason, these items now
2011 give a partial match in this situation. [Aside: previously, the one case
2012 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2013 match rather than a full match, which was wrong by the old rules, but is
2014 now correct.]
2015
2016 6. There was a bug in the handling of #-introduced comments, recognized when
2017 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2018 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2019 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2020 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2021 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2022 places in pcre_compile().
2023
2024 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2025 comments when looking ahead for named forward references to subpatterns,
2026 the only newline sequence it recognized was NL. It now handles newlines
2027 according to the set newline convention.
2028
2029 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2030 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2031 cater for a lack of strerror(). These oversights have been fixed.
2032
2033 9. Added --match-limit and --recursion-limit to pcregrep.
2034
2035 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2036
2037 11. When the -o option was used, pcregrep was setting a return code of 1, even
2038 when matches were found, and --line-buffered was not being honoured.
2039
2040 12. Added an optional parentheses number to the -o and --only-matching options
2041 of pcregrep.
2042
2043 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2044 can match an empty string. The code to do it in pcretest and pcredemo
2045 needed fixing:
2046
2047 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2048 only one byte after an empty string match just before CRLF (this case
2049 just got forgotten; "any" and "anycrlf" were OK).
2050
2051 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2052 mode when an empty string match preceded an ASCII character followed by
2053 a non-ASCII character. (The code for advancing by one character rather
2054 than one byte was nonsense.)
2055
2056 (c) The pcredemo.c sample program did not have any code at all to handle
2057 the cases when CRLF is a valid newline sequence.
2058
2059 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2060 as a starting offset was within the subject string. There is now a new
2061 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2062 negative or greater than the length of the string. In order to test this,
2063 pcretest is extended to allow the setting of negative starting offsets.
2064
2065 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2066 starting offset points to the beginning of a UTF-8 character was
2067 unnecessarily clumsy. I tidied it up.
2068
2069 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2070 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2071
2072 17. Nobody had reported that the --include_dir option, which was added in
2073 release 7.7 should have been called --include-dir (hyphen, not underscore)
2074 for compatibility with GNU grep. I have changed it to --include-dir, but
2075 left --include_dir as an undocumented synonym, and the same for
2076 --exclude-dir, though that is not available in GNU grep, at least as of
2077 release 2.5.4.
2078
2079 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2080 characters from a string of bytes) have been redefined so as not to use
2081 loops, in order to improve performance in some environments. At the same
2082 time, I abstracted some of the common code into auxiliary macros to save
2083 repetition (this should not affect the compiled code).
2084
2085 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2086 compile-time error is now given if \c is not followed by an ASCII
2087 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2088 different, and any byte value is allowed.)
2089
2090 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2091 START_OPTIMIZE option, which is now allowed at compile time - but just
2092 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2093 to pcregrep and other applications that have no direct access to PCRE
2094 options. The new /Y option in pcretest sets this option when calling
2095 pcre_compile().
2096
2097 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2098 back references. Groups containing recursive back references were forced to
2099 be atomic by that change, but in the case of named groups, the amount of
2100 memory required was incorrectly computed, leading to "Failed: internal
2101 error: code overflow". This has been fixed.
2102
2103 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2104 pcretest.c, to avoid build problems in some Borland environments.
2105
2106
2107 Version 8.10 25-Jun-2010
2108 ------------------------
2109
2110 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2111 THEN.
2112
2113 2. (*ACCEPT) was not working when inside an atomic group.
2114
2115 3. Inside a character class, \B is treated as a literal by default, but
2116 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2117 causes the error). The code is unchanged, but I tidied the documentation.
2118
2119 4. Inside a character class, PCRE always treated \R and \X as literals,
2120 whereas Perl faults them if its -w option is set. I have changed PCRE so
2121 that it faults them when PCRE_EXTRA is set.
2122
2123 5. Added support for \N, which always matches any character other than
2124 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2125
2126 6. When compiling pcregrep with newer versions of gcc which may have
2127 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2128 declared with attribute warn_unused_result" were given. Just casting the
2129 result to (void) does not stop the warnings; a more elaborate fudge is
2130 needed. I've used a macro to implement this.
2131
2132 7. Minor change to pcretest.c to avoid a compiler warning.
2133
2134 8. Added four artifical Unicode properties to help with an option to make
2135 \s etc use properties (see next item). The new properties are: Xan
2136 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2137
2138 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2139 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2140 this option. Modified pcretest to add /W to test this facility. Added
2141 REG_UCP to make it available via the POSIX interface.
2142
2143 10. Added --line-buffered to pcregrep.
2144
2145 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2146 studied, and the match started with a letter with a code point greater than
2147 127 whose first byte was different to the first byte of the other case of
2148 the letter, the other case of this starting letter was not recognized
2149 (#976).
2150
2151 12. If a pattern that was studied started with a repeated Unicode property
2152 test, for example, \p{Nd}+, there was the theoretical possibility of
2153 setting up an incorrect bitmap of starting bytes, but fortunately it could
2154 not have actually happened in practice until change 8 above was made (it
2155 added property types that matched character-matching opcodes).
2156
2157 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2158 possible starting bytes for non-anchored patterns.
2159
2160 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2161 \R, and also a number of cases that involve Unicode properties, both
2162 explicit and implicit when PCRE_UCP is set.
2163
2164 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2165 input, it could crash or give wrong results if characters with values
2166 greater than 0xc0 were present in the subject string. (Detail: it assumed
2167 UTF-8 input when processing these items.)
2168
2169 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2170 size_t is 64-bit (#991).
2171
2172 17. Added a check for running out of memory when PCRE is compiled with
2173 --disable-stack-for-recursion (#990).
2174
2175 18. If the last data line in a file for pcretest does not have a newline on
2176 the end, a newline was missing in the output.
2177
2178 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2179 less than 128) in its various bitmaps. However, there is a facility for
2180 generating tables according to the current locale when PCRE is compiled. It
2181 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2182 characters, are recognized by isspace() and therefore were getting set in
2183 these tables, and indeed these tables seem to approximate to ISO 8859. This
2184 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2185 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2186 which of course cannot start UTF-8 characters. I have changed the code so
2187 that only real ASCII characters (less than 128) and the correct starting
2188 bytes for UTF-8 encodings are set for characters greater than 127 when in
2189 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2190 altogether.)
2191
2192 20. Added the /T option to pcretest so as to be able to run tests with non-
2193 standard character tables, thus making it possible to include the tests
2194 used for 19 above in the standard set of tests.
2195
2196 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2197 reference to a subpattern the other side of a comment that contains an
2198 opening parenthesis caused either an internal compiling error, or a
2199 reference to the wrong subpattern.
2200
2201
2202 Version 8.02 19-Mar-2010
2203 ------------------------
2204
2205 1. The Unicode data tables have been updated to Unicode 5.2.0.
2206
2207 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2208 configured.
2209
2210 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2211 original author of that file, following a query about its status.
2212
2213 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2214 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2215
2216 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2217 quantifier applied to a forward-referencing subroutine call, could compile
2218 incorrect code or give the error "internal error: previously-checked
2219 referenced subpattern not found".
2220
2221 6. Both MS Visual Studio and Symbian OS have problems with initializing
2222 variables to point to external functions. For these systems, therefore,
2223 pcre_malloc etc. are now initialized to local functions that call the
2224 relevant global functions.
2225
2226 7. There were two entries missing in the vectors called coptable and poptable
2227 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2228 I've fixed the data, and added a kludgy way of testing at compile time that
2229 the lengths are correct (equal to the number of opcodes).
2230
2231 8. Following on from 7, I added a similar kludge to check the length of the
2232 eint vector in pcreposix.c.
2233
2234 9. Error texts for pcre_compile() are held as one long string to avoid too
2235 much relocation at load time. To find a text, the string is searched,
2236 counting zeros. There was no check for running off the end of the string,
2237 which could happen if a new error number was added without updating the
2238 string.
2239
2240 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2241
2242 11. \K was not working if it appeared in an atomic group or in a group that
2243 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2244 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2245 the assertion is positive, but not if it is negative.
2246
2247 12. Change 11 fortuitously reduced the size of the stack frame used in the
2248 "match()" function of pcre_exec.c by one pointer. Forthcoming
2249 implementation of support for (*MARK) will need an extra pointer on the
2250 stack; I have reserved it now, so that the stack frame size does not
2251 decrease.
2252
2253 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2254 item in branch that calls a recursion is a subroutine call - as in the
2255 second branch in the above example - was incorrectly given the compile-
2256 time error "recursive call could loop indefinitely" because pcre_compile()
2257 was not correctly checking the subroutine for matching a non-empty string.
2258
2259 14. The checks for overrunning compiling workspace could trigger after an
2260 overrun had occurred. This is a "should never occur" error, but it can be
2261 triggered by pathological patterns such as hundreds of nested parentheses.
2262 The checks now trigger 100 bytes before the end of the workspace.
2263
2264 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2265
2266
2267 Version 8.01 19-Jan-2010
2268 ------------------------
2269
2270 1. If a pattern contained a conditional subpattern with only one branch (in
2271 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2272 computed the wrong minimum data length (which is of course zero for such
2273 subpatterns). This could cause incorrect "no match" results.
2274
2275 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2276 the pattern is reset in the first branch, pcre_compile() failed with
2277 "internal error: code overflow at offset...". This happened only when
2278 the reset was to the original external option setting. (An optimization
2279 abstracts leading options settings into an external setting, which was the
2280 cause of this.)
2281
2282 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2283 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2284 assertion pattern did not match (meaning that the assertion was true), it
2285 was incorrectly treated as false if the SKIP had been reached during the
2286 matching. This also applied to assertions used as conditions.
2287
2288 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2289 assertion subpattern, including such a pattern used as a condition,
2290 unpredictable results occurred, instead of the error return
2291 PCRE_ERROR_DFA_UITEM.
2292
2293 5. The C++ GlobalReplace function was not working like Perl for the special
2294 situation when an empty string is matched. It now does the fancy magic
2295 stuff that is necessary.
2296
2297 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2298 removed. (These were left over from very, very early versions of PCRE.)
2299
2300 7. Some cosmetic changes to the code to make life easier when compiling it
2301 as part of something else:
2302
2303 (a) Change DEBUG to PCRE_DEBUG.
2304
2305 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2306 called "current" as "current_branch", to prevent a collision with the
2307 Linux macro when compiled as a kernel module.
2308
2309 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2310 prevent a collision with the Linux macro when compiled as a kernel
2311 module.
2312
2313 8. In pcre_compile() there are some checks for integer overflows that used to
2314 cast potentially large values to (double). This has been changed to that
2315 when building, a check for int64_t is made, and if it is found, it is used
2316 instead, thus avoiding the use of floating point arithmetic. (There is no
2317 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2318 double.
2319
2320 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2321 2005 (difference between two addresses compared to an unsigned value).
2322
2323 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2324 custom one, because of the following reported problem in Windows:
2325
2326 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2327 under Win32.
2328 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2329 therefore missing the function definition.
2330 - The compiler thus generates a "C" signature for the test function.
2331 - The linker fails to find the "C" function.
2332 - PCRE fails to configure if asked to do so against libbz2.
2333
2334 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2335 messages were output:
2336
2337 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2338 rerunning libtoolize, to keep the correct libtool macros in-tree.
2339 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2340
2341 I have done both of these things.
2342
2343 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2344 most of the time, it *can* run out if it is given a pattern that contains a
2345 runaway infinite recursion. I updated the discussion in the pcrestack man
2346 page.
2347
2348 13. Now that we have gone to the x.xx style of version numbers, the minor
2349 version may start with zero. Using 08 or 09 is a bad idea because users
2350 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2351 interpreted as invalid octal numbers. I've updated the previous comment in
2352 configure.ac, and also added a check that gives an error if 08 or 09 are
2353 used.
2354
2355 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2356 causing partial matching to fail when the end of the subject matched \W
2357 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2358
2359 15. There were some discrepancies between the declarations in pcre_internal.h
2360 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2361 their definitions. The declarations used "const uschar *" and the
2362 definitions used USPTR. Even though USPTR is normally defined as "const
2363 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2364 reported that: "This difference in casting confuses some C++ compilers, for
2365 example, SunCC recognizes above declarations as different functions and
2366 generates broken code for hbpcre." I have changed the declarations to use
2367 USPTR.
2368
2369 16. GNU libtool is named differently on some systems. The autogen.sh script now
2370 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2371 (FreeBSD).
2372
2373 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2374 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2375 comment: "Figure out how to create a longlong from a string: strtoll and
2376 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2377 instance, but it only takes 2 args instead of 3!"
2378
2379 18. A subtle bug concerned with back references has been fixed by a change of
2380 specification, with a corresponding code fix. A pattern such as
2381 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2382 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2383 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2384 same bug. Such groups have to be quantified to be useful, or contained
2385 inside another quantified group. (If there's no repetition, the reference
2386 can never match.) The problem arises because, having left the group and
2387 moved on to the rest of the pattern, a later failure that backtracks into
2388 the group uses the captured value from the final iteration of the group
2389 rather than the correct earlier one. I have fixed this in PCRE by forcing
2390 any group that contains a reference to itself to be an atomic group; that
2391 is, there cannot be any backtracking into it once it has completed. This is
2392 similar to recursive and subroutine calls.
2393
2394
2395 Version 8.00 19-Oct-09
2396 ----------------------
2397
2398 1. The table for translating pcre_compile() error codes into POSIX error codes
2399 was out-of-date, and there was no check on the pcre_compile() error code
2400 being within the table. This could lead to an OK return being given in
2401 error.
2402
2403 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2404 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2405 in a Windows environment.
2406
2407 3. The pcregrep --count option prints the count for each file even when it is
2408 zero, as does GNU grep. However, pcregrep was also printing all files when
2409 --files-with-matches was added. Now, when both options are given, it prints
2410 counts only for those files that have at least one match. (GNU grep just
2411 prints the file name in this circumstance, but including the count seems
2412 more useful - otherwise, why use --count?) Also ensured that the
2413 combination -clh just lists non-zero counts, with no names.
2414
2415 4. The long form of the pcregrep -F option was incorrectly implemented as
2416 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2417 but it seems right to fix it, and I didn't think it was worth preserving
2418 the old behaviour.
2419
2420 5. The command line items --regex=pattern and --regexp=pattern were not
2421 recognized by pcregrep, which required --regex pattern or --regexp pattern
2422 (with a space rather than an '='). The man page documented the '=' forms,
2423 which are compatible with GNU grep; these now work.
2424
2425 6. No libpcreposix.pc file was created for pkg-config; there was just
2426 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2427
2428 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2429 when UCP support is not needed, by modifying the Python script that
2430 generates it from Unicode data files. This should not matter if the module
2431 is correctly used as a library, but I received one complaint about 50K of
2432 unwanted data. My guess is that the person linked everything into his
2433 program rather than using a library. Anyway, it does no harm.
2434
2435 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2436 was a minimum greater than 1 for a wide character in a possessive
2437 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2438 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2439 character. Chaos in the form of incorrect output or a compiling loop could
2440 result.
2441
2442 9. The restrictions on what a pattern can contain when partial matching is
2443 requested for pcre_exec() have been removed. All patterns can now be
2444 partially matched by this function. In addition, if there are at least two
2445 slots in the offset vector, the offset of the earliest inspected character
2446 for the match and the offset of the end of the subject are set in them when
2447 PCRE_ERROR_PARTIAL is returned.
2448
2449 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2450 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2451 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2452 and may be more useful for multi-segment matching.
2453
2454 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2455 used to be given if ever the end of the subject was reached; now it is
2456 given only if matching could not proceed because another character was
2457 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2458 string "Z", which now yields "no match" instead of "partial match". In the
2459 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2460 final character ended with (*FAIL).
2461
2462 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2463 if the pattern had a "must contain" character that was already found in the
2464 earlier partial match, unless partial matching was again requested. For
2465 example, with the pattern /dog.(body)?/, the "must contain" character is
2466 "g". If the first part-match was for the string "dog", restarting with
2467 "sbody" failed. This bug has been fixed.
2468
2469 13. The string returned by pcre_dfa_exec() after a partial match has been
2470 changed so that it starts at the first inspected character rather than the
2471 first character of the match. This makes a difference only if the pattern
2472 starts with a lookbehind assertion or \b or \B (\K is not supported by
2473 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2474 matching functions compatible, and I think it's the right thing to do.
2475
2476 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2477 so that the demonstration program is easily available in environments where
2478 PCRE has not been installed from source.
2479
2480 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2481 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2482 library.
2483
2484 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2485 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2486 is not the first non-POSIX option to be added. Clearly some people find
2487 these options useful.
2488
2489 17. If a caller to the POSIX matching function regexec() passes a non-zero
2490 value for nmatch with a NULL value for pmatch, the value of
2491 nmatch is forced to zero.
2492
2493 18. RunGrepTest did not have a test for the availability of the -u option of
2494 the diff command, as RunTest does. It now checks in the same way as
2495 RunTest, and also checks for the -b option.
2496
2497 19. If an odd number of negated classes containing just a single character
2498 interposed, within parentheses, between a forward reference to a named
2499 subpattern and the definition of the subpattern, compilation crashed with
2500 an internal error, complaining that it could not find the referenced
2501 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2502 [The bug was that it was starting one character too far in when skipping
2503 over the character class, thus treating the ] as data rather than
2504 terminating the class. This meant it could skip too much.]
2505
2506 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2507 /g option in pcretest when the pattern contains \K, which makes it possible
2508 to have an empty string match not at the start, even when the pattern is
2509 anchored. Updated pcretest and pcredemo to use this option.
2510
2511 21. If the maximum number of capturing subpatterns in a recursion was greater
2512 than the maximum at the outer level, the higher number was returned, but
2513 with unset values at the outer level. The correct (outer level) value is
2514 now given.
2515
2516 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2517 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2518 make it do so. The string so far is captured, making this feature
2519 compatible with Perl.
2520
2521 23. The tests have been re-organized, adding tests 11 and 12, to make it
2522 possible to check the Perl 5.10 features against Perl 5.10.
2523
2524 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2525 pattern matches a fixed length string. PCRE did not allow this; now it
2526 does. Neither allows recursion.
2527
2528 25. I finally figured out how to implement a request to provide the minimum
2529 length of subject string that was needed in order to match a given pattern.
2530 (It was back references and recursion that I had previously got hung up
2531 on.) This code has now been added to pcre_study(); it finds a lower bound
2532 to the length of subject needed. It is not necessarily the greatest lower
2533 bound, but using it to avoid searching strings that are too short does give
2534 some useful speed-ups. The value is available to calling programs via
2535 pcre_fullinfo().
2536
2537 26. While implementing 25, I discovered to my embarrassment that pcretest had
2538 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2539 study optimizations had never been tested with that matching function.
2540 Oops. What is worse, even when it was passed study data, there was a bug in
2541 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2542 were also very few tests of studied patterns with pcre_dfa_exec().
2543
2544 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2545 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2546 on the other side of the coin, they are no longer allowed to have different
2547 names, because these cannot be distinguished in PCRE, and this has caused
2548 confusion. (This is a difference from Perl.)
2549
2550 28. When duplicate subpattern names are present (necessarily with different
2551 numbers, as required by 27 above), and a test is made by name in a
2552 conditional pattern, either for a subpattern having been matched, or for
2553 recursion in such a pattern, all the associated numbered subpatterns are
2554 tested, and the overall condition is true if the condition is true for any
2555 one of them. This is the way Perl works, and is also more like the way
2556 testing by number works.
2557
2558
2559 Version 7.9 11-Apr-09
2560 ---------------------
2561
2562 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2563 (pcretest), all targets were linked against these libraries. This included
2564 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2565 libraries. This caused unwanted dependencies to be created. This problem
2566 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2567 pcretest is linked with readline.
2568
2569 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2570 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2571 moved outside it again, because FALSE and TRUE are already defined in AIX,
2572 but BOOL is not.
2573
2574 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2575 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2576
2577 4. The pcregrep documentation said spaces were inserted as well as colons (or
2578 hyphens) following file names and line numbers when outputting matching
2579 lines. This is not true; no spaces are inserted. I have also clarified the
2580 wording for the --colour (or --color) option.
2581
2582 5. In pcregrep, when --colour was used with -o, the list of matching strings
2583 was not coloured; this is different to GNU grep, so I have changed it to be
2584 the same.
2585
2586 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2587 each matching line was coloured. Now it goes on to look for further matches
2588 of any of the test patterns, which is the same behaviour as GNU grep.
2589
2590 7. A pattern that could match an empty string could cause pcregrep to loop; it
2591 doesn't make sense to accept an empty string match in pcregrep, so I have
2592 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2593 seems to be how GNU grep behaves. [But see later change 40 for release
2594 8.33.]
2595
2596 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2597 start or after a newline", because the conditional assertion was not being
2598 correctly handled. The rule now is that both the assertion and what follows
2599 in the first alternative must satisfy the test.
2600
2601 9. If auto-callout was enabled in a pattern with a conditional group whose
2602 condition was an assertion, PCRE could crash during matching, both with
2603 pcre_exec() and pcre_dfa_exec().
2604
2605 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2606 used for matching.
2607
2608 11. Unicode property support in character classes was not working for
2609 characters (bytes) greater than 127 when not in UTF-8 mode.
2610
2611 12. Added the -M command line option to pcretest.
2612
2613 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2614
2615 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2616
2617 16. Added comments and documentation about mis-use of no_arg in the C++
2618 wrapper.
2619
2620 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2621 from Martin Jerabek that uses macro names for all relevant character and
2622 string constants.
2623
2624 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2625 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2626 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2627 these, but not everybody uses configure.
2628
2629 19. A conditional group that had only one branch was not being correctly
2630 recognized as an item that could match an empty string. This meant that an
2631 enclosing group might also not be so recognized, causing infinite looping
2632 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2633 with the subject "ab", where knowledge that the repeated group can match
2634 nothing is needed in order to break the loop.
2635
2636 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2637 exec(), but without supplying a callout function, matching went wrong.
2638
2639 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2640 leak if the size of the offset vector was greater than 30. When the vector
2641 is smaller, the saved offsets during recursion go onto a local stack
2642 vector, but for larger vectors malloc() is used. It was failing to free
2643 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2644 error, in fact).
2645
2646 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2647 heapframe that is used only when UTF-8 support is enabled. This caused no
2648 problem, but was untidy.
2649
2650 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2651 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2652 included within another project.
2653
2654 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2655 slightly modified by me:
2656
2657 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2658 not building pcregrep.
2659
2660 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2661 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2662
2663 25. Forward references, both numeric and by name, in patterns that made use of
2664 duplicate group numbers, could behave incorrectly or give incorrect errors,
2665 because when scanning forward to find the reference group, PCRE was not
2666 taking into account the duplicate group numbers. A pattern such as
2667 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2668
2669 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2670 the feature of a custom pointer more persuasive (as requested by a user).
2671
2672 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2673 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2674 pre-defined.
2675
2676 28. Added support for (*UTF8) at the start of a pattern.
2677
2678 29. Arrange for flags added by the "release type" setting in CMake to be shown
2679 in the configuration summary.
2680
2681
2682 Version 7.8 05-Sep-08
2683 ---------------------
2684
2685 1. Replaced UCP searching code with optimized version as implemented for Ad
2686 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2687 stage table and inline lookup instead of a function, giving speed ups of 2
2688 to 5 times on some simple patterns that I tested. Permission was given to
2689 distribute the MultiStage2.py script that generates the tables (it's not in
2690 the tarball, but is in the Subversion repository).
2691
2692 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2693 scripts.
2694
2695 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2696 a group with a zero qualifier. The result of the study could be incorrect,
2697 or the function might crash, depending on the pattern.
2698
2699 4. Caseless matching was not working for non-ASCII characters in back
2700 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2701 It now works when Unicode Property Support is available.
2702
2703 5. In pcretest, an escape such as \x{de} in the data was always generating
2704 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2705 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2706 truncation.
2707
2708 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2709
2710 7. Added two (int) casts to pcregrep when printing the difference of two
2711 pointers, in case they are 64-bit values.
2712
2713 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2714 test 2 if it fails.
2715
2716 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2717 and a #define of that name to empty if it is not externally set. This is to
2718 allow users of MSVC to set it if necessary.
2719
2720 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2721 the convenience functions in the pcre_get.c source file.
2722
2723 11. An option change at the start of a pattern that had top-level alternatives
2724 could cause overwriting and/or a crash. This command provoked a crash in
2725 some environments:
2726
2727 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2728
2729 This potential security problem was recorded as CVE-2008-2371.
2730
2731 12. For a pattern where the match had to start at the beginning or immediately
2732 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2733 pcre_dfa_exec() could read past the end of the passed subject if there was
2734 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2735 pcretest so that it places the subject at the end of its malloc-ed buffer.
2736
2737 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2738 exec() might read past the end of the data buffer in UTF-8 mode.
2739
2740 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2741 the data contained the byte 0x85 as part of a UTF-8 character within its
2742 first line. This applied both to normal and DFA matching.
2743
2744 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2745 /^[^d]*?$/8 failed to match "abc".
2746
2747 16. Added a missing copyright notice to pcrecpp_internal.h.
2748
2749 17. Make it more clear in the documentation that values returned from
2750 pcre_exec() in ovector are byte offsets, not character counts.
2751
2752 18. Tidied a few places to stop certain compilers from issuing warnings.
2753
2754 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2755 supplied by Stefan Weber. I made a further small update for 7.8 because
2756 there is a change of source arrangements: the pcre_searchfuncs.c module is
2757 replaced by pcre_ucd.c.
2758
2759
2760 Version 7.7 07-May-08
2761 ---------------------
2762
2763 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2764 a string to a long long, pretend we don't even have a long long." This is
2765 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2766
2767 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2768 pre-7.6 versions, which defined a global no_arg variable instead of putting
2769 it in the RE class. (See also #8 below.)
2770
2771 3. Remove a line of dead code, identified by coverity and reported by Nuno
2772 Lopes.
2773
2774 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2775
2776 (1) The include/exclude patterns were being applied to the whole pathnames
2777 of files, instead of just to the final components.
2778
2779 (2) If there was more than one level of directory, the subdirectories were
2780 skipped unless they satisfied the include/exclude conditions. This is
2781 inconsistent with GNU grep (and could even be seen as contrary to the
2782 pcregrep specification - which I improved to make it absolutely clear).
2783 The action now is always to scan all levels of directory, and just
2784 apply the include/exclude patterns to regular files.
2785
2786 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2787 --exclude_dir in the tests to avoid scanning .svn directories.
2788
2789 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2790 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2791 doesn't support NULs in patterns.
2792
2793 7. Added some missing "const"s to declarations of static tables in
2794 pcre_compile.c and pcre_dfa_exec.c.
2795
2796 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2797 caused by fix #2 above. (Subsequently also a second patch to fix the
2798 first patch. And a third patch - this was a messy problem.)
2799
2800 9. Applied Craig's patch to remove the use of push_back().
2801
2802 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2803 matching function regexec().
2804
2805 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2806 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2807 references. PCRE supports relative numbers with this syntax (I don't think
2808 Oniguruma does).
2809
2810 12. Previously, a group with a zero repeat such as (...){0} was completely
2811 omitted from the compiled regex. However, this means that if the group
2812 was called as a subroutine from elsewhere in the pattern, things went wrong
2813 (an internal error was given). Such groups are now left in the compiled
2814 pattern, with a new opcode that causes them to be skipped at execution
2815 time.
2816
2817 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2818 to the way PCRE behaves:
2819
2820 (a) A lone ] character is dis-allowed (Perl treats it as data).
2821
2822 (b) A back reference to an unmatched subpattern matches an empty string
2823 (Perl fails the current match path).
2824
2825 (c) A data ] in a character class must be notated as \] because if the
2826 first data character in a class is ], it defines an empty class. (In
2827 Perl it is not possible to have an empty class.) The empty class []
2828 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2829 The negative empty class [^] matches any one character, independently
2830 of the DOTALL setting.
2831
2832 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2833 non-existent subpattern following a character class starting with ']' and
2834 containing () gave an internal compiling error instead of "reference to
2835 non-existent subpattern". Fortunately, when the pattern did exist, the
2836 compiled code was correct. (When scanning forwards to check for the
2837 existence of the subpattern, it was treating the data ']' as terminating
2838 the class, so got the count wrong. When actually compiling, the reference
2839 was subsequently set up correctly.)
2840
2841 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2842 it was being rejected as not supported by pcre_dfa_exec(), even though
2843 other assertions are supported. I have made pcre_dfa_exec() support
2844 (*FAIL).
2845
2846 16. The implementation of 13c above involved the invention of a new opcode,
2847 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2848 cannot be changed at match time, I realized I could make a small
2849 improvement to matching performance by compiling OP_ALLANY instead of
2850 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2851 on the OP_ANY path.
2852
2853 17. Compiling pcretest on Windows with readline support failed without the
2854 following two fixes: (1) Make the unistd.h include conditional on
2855 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2856
2857 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2858 ncurses library to be included for pcretest when ReadLine support is
2859 requested, but also to allow for it to be overridden. This patch came from
2860 Daniel Bergström.
2861
2862 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2863 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2864 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2865 spotting this.
2866
2867
2868 Version 7.6 28-Jan-08
2869 ---------------------
2870
2871 1. A character class containing a very large number of characters with
2872 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2873 overflow.
2874
2875 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2876 HAVE_LONG_LONG is not defined.
2877
2878 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2879 bring them up to date and include new features. This patch includes:
2880
2881 - Fixed PH's badly added libz and libbz2 support.
2882 - Fixed a problem with static linking.
2883 - Added pcredemo. [But later removed - see 7 below.]
2884 - Fixed dftables problem and added an option.
2885 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2886 HAVE_LONG_LONG.
2887 - Added readline support for pcretest.
2888 - Added an listing of the option settings after cmake has run.
2889
2890 4. A user submitted a patch to Makefile that makes it easy to create
2891 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2892 Makefile.am that cause it to include this special target, without
2893 affecting anything else. Note that the same mingw target plus all
2894 the other distribution libraries and programs are now supported
2895 when configuring with CMake (see 6 below) instead of with
2896 Configure/Make.
2897
2898 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2899 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2900 exported in the Windows port". It has not yet been confirmed that the patch
2901 solves the problem, but it does no harm.
2902
2903 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2904 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2905 with CMake, and also correct the comment about stack recursion.
2906
2907 7. Remove the automatic building of pcredemo from the ./configure system and
2908 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2909 of a program that users should build themselves after PCRE is installed, so
2910 building it automatically is not really right. What is more, it gave
2911 trouble in some build environments.
2912
2913 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2914
2915
2916 Version 7.5 10-Jan-08
2917 ---------------------
2918
2919 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2920 values in parens when parsing an RE using the C++ wrapper."
2921
2922 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2923 Characters greater than 255 were excluded from the class instead of being
2924 included.
2925
2926 3. The same bug as (2) above applied to negated POSIX classes such as
2927 [:^space:].
2928
2929 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2930 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2931 I have changed it.
2932
2933 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2934 first named subpattern) and a construct such as (?&a) would reference the
2935 first named subpattern whose name started with "a" (in other words, the
2936 length check was missing). Both these problems are fixed. "Subpattern name
2937 expected" is now given for (?&) (a zero-length name), and this patch also
2938 makes it give the same error for \k'' (previously it complained that that
2939 was a reference to a non-existent subpattern).
2940
2941 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2942 this is right because (?- can be followed by option settings as well as by
2943 digits. I have, however, made the messages clearer.
2944
2945 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2946 than the number used in the conditional) now cause a compile-time error.
2947 This is actually not compatible with Perl, which accepts such patterns, but
2948 treats the conditional as always being FALSE (as PCRE used to), but it
2949 seems to me that giving a diagnostic is better.
2950
2951 8. Change "alphameric" to the more common word "alphanumeric" in comments
2952 and messages.
2953
2954 9. Fix two occurrences of "backslash" in comments that should have been
2955 "backspace".
2956
2957 10. Remove two redundant lines of code that can never be obeyed (their function
2958 was moved elsewhere).
2959
2960 11. The program that makes PCRE's Unicode character property table had a bug
2961 which caused it to generate incorrect table entries for sequences of
2962 characters that have the same character type, but are in different scripts.
2963 It amalgamated them into a single range, with the script of the first of
2964 them. In other words, some characters were in the wrong script. There were
2965 thirteen such cases, affecting characters in the following ranges:
2966
2967 U+002b0 - U+002c1
2968 U+0060c - U+0060d
2969 U+0061e - U+00612
2970 U+0064b - U+0065e
2971 U+0074d - U+0076d
2972 U+01800 - U+01805
2973 U+01d00 - U+01d77
2974 U+01d9b - U+01dbf
2975 U+0200b - U+0200f
2976 U+030fc - U+030fe
2977 U+03260 - U+0327f
2978 U+0fb46 - U+0fbb1
2979 U+10450 - U+1049d
2980
2981 12. The -o option (show only the matching part of a line) for pcregrep was not
2982 compatible with GNU grep in that, if there was more than one match in a
2983 line, it showed only the first of them. It now behaves in the same way as
2984 GNU grep.
2985
2986 13. If the -o and -v options were combined for pcregrep, it printed a blank
2987 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2988 does the same. The return code can be used to tell if there were any
2989 non-matching lines.
2990
2991 14. Added --file-offsets and --line-offsets to pcregrep.
2992
2993 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2994 infinitely looping recursion. The bug was that positive lookaheads were not
2995 being skipped when checking for a possible empty match (negative lookaheads
2996 and both kinds of lookbehind were skipped).
2997
2998 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2999 inclusion of <windows.h> to before rather than after the definition of
3000 INVALID_FILE_ATTRIBUTES (patch from David Byron).
3001
3002 17. Specifying a possessive quantifier with a specific limit for a Unicode
3003 character property caused pcre_compile() to compile bad code, which led at
3004 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
3005 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
3006 caused the error; without that there was no problem.
3007
3008 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3009
3010 19. Added --enable-pcretest-libreadline.
3011
3012 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3013 RE::GlobalReplace(). As a result, the number of replacements returned was
3014 double what it should be. I removed one of the increments, but Craig sent a
3015 later patch that removed the other one (the right fix) and added unit tests
3016 that check the return values (which was not done before).
3017
3018 21. Several CMake things:
3019
3020 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3021 the names libpcre and libpcreposix, not just pcre and pcreposix.
3022
3023 (2) The above change means that pcretest and pcregrep are now correctly
3024 linked with the newly-built libraries, not previously installed ones.
3025
3026 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
3027
3028 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3029 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3030 UTF-8 newline character). The key issue is that the pattern starts .*;
3031 this means that the match must be either at the beginning, or after a
3032 newline. The bug was in the code for advancing after a failed match and
3033 checking that the new position followed a newline. It was not taking
3034 account of UTF-8 characters correctly.
3035
3036 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3037 character classes. PCRE was not treating the sequence [:...:] as a
3038 character class unless the ... were all letters. Perl, however, seems to
3039 allow any characters between [: and :], though of course it rejects as
3040 unknown any "names" that contain non-letters, because all the known class
3041 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3042 for example, whereas PCRE did not - it did not recognize a POSIX character
3043 class. This seemed a bit dangerous, so the code has been changed to be
3044 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3045 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3046 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3047 Perl does, and where it didn't before.
3048
3049 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3050 Windows environments %n is disabled by default.
3051
3052
3053 Version 7.4 21-Sep-07
3054 ---------------------
3055
3056 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3057 means that a class such as [\s] counted as "explicit reference to CR or
3058 LF". That isn't really right - the whole point of the change was to try to
3059 help when there was an actual mention of one of the two characters. So now
3060 the change happens only if \r or \n (or a literal CR or LF) character is
3061 encountered.
3062
3063 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3064 of both had grown to the point where there were only 3 bits left.
3065 Fortunately, there was spare space in the data structure, and so I have
3066 moved the internal flags into a new 16-bit field to free up more option
3067 bits.
3068
3069 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3070 but did not set the internal JCHANGED flag - either of these is enough to
3071 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3072 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3073 start sets both bits.
3074
3075 4. Added options (at build time, compile time, exec time) to change \R from
3076 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3077
3078 5. doc/pcresyntax.html was missing from the distribution.
3079
3080 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3081 compatibility, even though it is no longer used.
3082
3083 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3084 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3085 windows.h file is present (where different names are used). [This was
3086 reversed later after testing - see 16 below.]
3087
3088 8. Changed all #include <config.h> to #include "config.h". There were also
3089 some further <pcre.h> cases that I changed to "pcre.h".
3090
3091 9. When pcregrep was used with the --colour option, it missed the line ending
3092 sequence off the lines that it output.
3093
3094 10. It was pointed out to me that arrays of string pointers cause lots of
3095 relocations when a shared library is dynamically loaded. A technique of
3096 using a single long string with a table of offsets can drastically reduce
3097 these. I have refactored PCRE in four places to do this. The result is
3098 dramatic:
3099
3100 Originally: 290
3101 After changing UCP table: 187
3102 After changing error message table: 43
3103 After changing table of "verbs" 36
3104 After changing table of Posix names 22
3105
3106 Thanks to the folks working on Gregex for glib for this insight.
3107
3108 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3109 unicode-properties was also set.
3110
3111 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3112
3113 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3114 checked only for CRLF.
3115
3116 14. Added casts to pcretest.c to avoid compiler warnings.
3117
3118 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3119
3120 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3121 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3122 entirely. This removes changes made in 7 above.
3123
3124 17. The CMake files have been updated, and there is now more information about
3125 building with CMake in the NON-UNIX-USE document.
3126
3127
3128 Version 7.3 28-Aug-07
3129 ---------------------
3130
3131 1. In the rejigging of the build system that eventually resulted in 7.1, the
3132 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3133 brackets there is not right, since it causes compilers to look for an
3134 installed pcre.h, not the version that is in the source that is being
3135 compiled (which of course may be different). I have changed it back to:
3136
3137 #include "pcre.h"
3138
3139 I have a vague recollection that the change was concerned with compiling in
3140 different directories, but in the new build system, that is taken care of
3141 by the VPATH setting the Makefile.
3142
3143 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3144 when the subject happened to end in the byte 0x85 (e.g. if the last
3145 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3146 characters but of course it shouldn't be taken as a newline when it is part
3147 of another character. The bug was that, for an unlimited repeat of . in
3148 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3149 characters when looking for a newline.
3150
3151 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3152
3153 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3154 in debug output.
3155
3156 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3157 long printing in the pcrecpp unittest when running under MinGW.
3158
3159 6. ESC_K was left out of the EBCDIC table.
3160
3161 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3162 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3163 limit also applies to "virtual nesting" when a pattern is recursive, and in
3164 this case 1000 isn't so big. I have been able to remove this limit at the
3165 expense of backing off one optimization in certain circumstances. Normally,
3166 when pcre_exec() would call its internal match() function recursively and
3167 immediately return the result unconditionally, it uses a "tail recursion"
3168 feature to save stack. However, when a subpattern that can match an empty
3169 string has an unlimited repetition quantifier, it no longer makes this
3170 optimization. That gives it a stack frame in which to save the data for
3171 checking that an empty string has been matched. Previously this was taken
3172 from the 1000-entry workspace that had been reserved. So now there is no
3173 explicit limit, but more stack is used.
3174
3175 8. Applied Daniel's patches to solve problems with the import/export magic
3176 syntax that is required for Windows, and which was going wrong for the
3177 pcreposix and pcrecpp parts of the library. These were overlooked when this
3178 problem was solved for the main library.
3179
3180 9. There were some crude static tests to avoid integer overflow when computing
3181 the size of patterns that contain repeated groups with explicit upper
3182 limits. As the maximum quantifier is 65535, the maximum group length was
3183 set at 30,000 so that the product of these two numbers did not overflow a
3184 32-bit integer. However, it turns out that people want to use groups that
3185 are longer than 30,000 bytes (though not repeat them that many times).
3186 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3187 made it possible to implement the integer overflow checks in a much more
3188 dynamic way, which I have now done. The artificial limitation on group
3189 length has been removed - we now have only the limit on the total length of
3190 the compiled pattern, which depends on the LINK_SIZE setting.
3191
3192 10. Fixed a bug in the documentation for get/copy named substring when
3193 duplicate names are permitted. If none of the named substrings are set, the
3194 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3195 empty string.
3196
3197 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3198 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3199 because the ] is interpreted as the first data character and the
3200 terminating ] is not found. PCRE has been made compatible with Perl in this
3201 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3202 cause memory overwriting.
3203
3204 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3205 string has been matched (to stop an infinite loop). It was not recognizing
3206 a conditional subpattern that could match an empty string if that
3207 subpattern was within another subpattern. For example, it looped when
3208 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3209 condition was not nested. This bug has been fixed.
3210
3211 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3212 past the start of the subject in the presence of bytes with the top bit
3213 set, for example "\x8aBCD".
3214
3215 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3216 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3217
3218 14. Optimized (?!) to (*FAIL).
3219
3220 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3221 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3222 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3223 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3224 does: it's just the validity check that is more restrictive.
3225
3226 16. Inserted checks for integer overflows during escape sequence (backslash)
3227 processing, and also fixed erroneous offset values for syntax errors during
3228 backslash processing.
3229
3230 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3231 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3232
3233 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3234 caused an overrun.
3235
3236 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3237 something other than just ASCII characters) inside a group that had an
3238 unlimited repeat caused a loop at compile time (while checking to see
3239 whether the group could match an empty string).
3240
3241 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3242 [\P{Any}] did so. (Error in the code for printing property names.)
3243
3244 21. An orphan \E inside a character class could cause a crash.
3245
3246 22. A repeated capturing bracket such as (A)? could cause a wild memory
3247 reference during compilation.
3248
3249 23. There are several functions in pcre_compile() that scan along a compiled
3250 expression for various reasons (e.g. to see if it's fixed length for look
3251 behind). There were bugs in these functions when a repeated \p or \P was
3252 present in the pattern. These operators have additional parameters compared
3253 with \d, etc, and these were not being taken into account when moving along
3254 the compiled data. Specifically:
3255
3256 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3257 length.
3258
3259 (b) An item such as \pL+ within a repeated group could cause crashes or
3260 loops.
3261
3262 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3263 "reference to non-existent subpattern" error.
3264
3265 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3266
3267 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3268 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3269
3270 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3271
3272 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3273 character were causing crashes (broken optimization).
3274
3275 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3276 \p or \P) caused a compile-time loop.
3277
3278 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3279 break. For example, the unstudied pattern [\r\n]A does not match the string
3280 "\r\nA" because change 7.0/46 below moves the current point on by two
3281 characters after failing to match at the start. However, the pattern \nA
3282 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3283 the same is true. There doesn't seem any very clean way out of this, but
3284 what I have chosen to do makes the common cases work: PCRE now takes note
3285 of whether there can be an explicit match for \r or \n anywhere in the
3286 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3287 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3288 pattern has explicit CR or LF references.
3289
3290 29. Added (*CR) etc for changing newline setting at start of pattern.
3291
3292
3293 Version 7.2 19-Jun-07
3294 ---------------------
3295
3296 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3297 which is apparently normally available under Windows.
3298
3299 2. Re-jig the pcregrep tests with different newline settings in an attempt
3300 to make them independent of the local environment's newline setting.
3301
3302 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3303
3304 4. Some of the "internals" tests were previously cut out when the link size
3305 was not 2, because the output contained actual offsets. The recent new
3306 "Z" feature of pcretest means that these can be cut out, making the tests
3307 usable with all link sizes.
3308
3309 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3310 stack recursion. This gives a massive performance boost under BSD, but just
3311 a small improvement under Linux. However, it saves one field in the frame
3312 in all cases.
3313
3314 6. Added more features from the forthcoming Perl 5.10:
3315
3316 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3317 recursion call. It refers to the nth most recently opened parentheses.
3318
3319 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3320 to be opened parentheses.
3321
3322 (c) Conditions that refer to capturing parentheses can be specified
3323 relatively, for example, (?(-2)... or (?(+3)...
3324
3325 (d) \K resets the start of the current match so that everything before
3326 is not part of it.
3327
3328 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3329
3330 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3331 reference syntax.
3332
3333 (g) (?| introduces a group in which the numbering of parentheses in each
3334 alternative starts with the same number.
3335
3336 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3337
3338 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3339 PCRE_INFO_JCHANGED.
3340
3341 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3342 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3343 for detecting groups that can match an empty string.
3344
3345 9. A pattern with a very large number of alternatives (more than several
3346 hundred) was running out of internal workspace during the pre-compile
3347 phase, where pcre_compile() figures out how much memory will be needed. A
3348 bit of new cunning has reduced the workspace needed for groups with
3349 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3350 workspace instead of running out of the 4096 that are available.
3351
3352 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3353
3354 11. Applied patch from Google to remove an optimization that didn't quite work.
3355 The report of the bug said:
3356
3357 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3358 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3359 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3360
3361 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3362 it matched the wrong number of bytes.
3363
3364
3365 Version 7.1 24-Apr-07
3366 ---------------------
3367
3368 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3369 that is more "standard", making use of automake and other Autotools. There
3370 is some re-arrangement of the files and adjustment of comments consequent
3371 on this.
3372
3373 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3374 for recursive directory scanning broke on some systems because the files
3375 are not scanned in any specific order and on different systems the order
3376 was different. A call to "sort" has been inserted into RunGrepTest for the
3377 approprate test as a short-term fix. In the longer term there may be an
3378 alternative.
3379
3380 3. I had an email from Eric Raymond about problems translating some of PCRE's
3381 man pages to HTML (despite the fact that I distribute HTML pages, some
3382 people do their own conversions for various reasons). The problems
3383 concerned the use of low-level troff macros .br and .in. I have therefore
3384 removed all such uses from the man pages (some were redundant, some could
3385 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3386 HTML has been updated to handle .nf/.fi and to complain if it encounters
3387 .br or .in.
3388
3389 4. Updated comments in configure.ac that get placed in config.h.in and also
3390 arranged for config.h to be included in the distribution, with the name
3391 config.h.generic, for the benefit of those who have to compile without
3392 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3393
3394 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3395 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3396 makevp.bat for the current PCRE, using the additional files
3397 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3398
3399 6. A Windows user reported a minor discrepancy with test 2, which turned out
3400 to be caused by a trailing space on an input line that had got lost in his
3401 copy. The trailing space was an accident, so I've just removed it.
3402
3403 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3404 that is needed.
3405
3406 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3407 as "const" (a) because they are and (b) because it helps the PHP
3408 maintainers who have recently made a script to detect big data structures
3409 in the php code that should be moved to the .rodata section. I remembered
3410 to update Builducptable as well, so it won't revert if ucptable.h is ever
3411 re-created.
3412
3413 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3414 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3415 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3416 support is not required. This saves 1.5-2K of code, which is important in
3417 some applications.
3418
3419 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3420 so as not to refer to the tables, even though these functions will never be
3421 called when UTF-8 support is disabled. Otherwise there are problems with a
3422 shared library.
3423
3424 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3425
3426 (a) It was defining its arguments as char * instead of void *.
3427
3428 (b) It was assuming that all moves were upwards in memory; this was true
3429 a long time ago when I wrote it, but is no longer the case.
3430
3431 The emulated memove() is provided for those environments that have neither
3432 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3433 is clearly not the case, as these two bugs were recently reported.
3434
3435 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3436 and Detrail to create the HTML documentation, the .txt form of the man
3437 pages, and it removes trailing spaces from listed files. It also creates
3438 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3439 case, it wraps all the #defines with #ifndefs. This script should be run
3440 before "make dist".
3441
3442 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3443 with Unicode property support.
3444
3445 (a) For a maximizing quantifier, if the two different cases of the
3446 character were of different lengths in their UTF-8 codings (there are
3447 some cases like this - I found 11), and the matching function had to
3448 back up over a mixture of the two cases, it incorrectly assumed they
3449 were both the same length.
3450
3451 (b) When PCRE was configured to use the heap rather than the stack for
3452 recursion during matching, it was not correctly preserving the data for
3453 the other case of a UTF-8 character when checking ahead for a match
3454 while processing a minimizing repeat. If the check also involved
3455 matching a wide character, but failed, corruption could cause an
3456 erroneous result when trying to check for a repeat of the original
3457 character.
3458
3459 13. Some tidying changes to the testing mechanism:
3460
3461 (a) The RunTest script now detects the internal link size and whether there
3462 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3463 values substituted by "configure". (The RunGrepTest script already did
3464 this for UTF-8.) The configure.ac script no longer substitutes the
3465 relevant variables.
3466
3467 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3468 with length and offset values. This means that the output is different
3469 for different internal link sizes. Test 2 is skipped for link sizes
3470 other than 2 because of this, bypassing the problem. Unfortunately,
3471 there was also a test in test 3 (the locale tests) that used /B and
3472 failed for link sizes other than 2. Rather than cut the whole test out,
3473 I have added a new /Z option to pcretest that replaces the length and
3474 offset values with spaces. This is now used to make test 3 independent
3475 of link size. (Test 2 will be tidied up later.)
3476
3477 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3478 segmentation fault instead of returning the appropriate error message.
3479
3480 15. In multiline mode when the newline sequence was set to "any", the pattern
3481 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3482 This doesn't seem right; it now treats the CRLF combination as the line
3483 ending, and so does not match in that case. It's only a pattern such as ^$
3484 that would hit this one: something like ^ABC$ would have failed after \r
3485 and then tried again after \r\n.
3486
3487 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3488 in an attempt to make files that differ only in their line terminators
3489 compare equal. This works on Linux.
3490
3491 17. Under certain error circumstances pcregrep might try to free random memory
3492 as it exited. This is now fixed, thanks to valgrind.
3493
3494 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3495 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3496 was because its rules for how to advance for /g after matching an empty
3497 string at the end of a line did not allow for this case. They now check for
3498 it specially.
3499
3500 20. pcretest is supposed to handle patterns and data of any length, by
3501 extending its buffers when necessary. It was getting this wrong when the
3502 buffer for a data line had to be extended.
3503
3504 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3505 CRLF as a newline sequence.
3506
3507 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3508 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3509 I have nevertheless tidied it up.
3510
3511 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3512
3513 24. Added a man page for pcre-config.
3514
3515
3516 Version 7.0 19-Dec-06
3517 ---------------------
3518
3519 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3520 moving to gcc 4.1.1.
3521
3522 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3523 sys/time.h, which is documented as needed for this function. It doesn't
3524 seem to matter on Linux, but it showed up on some releases of OS X.
3525
3526 3. It seems that there are systems where bytes whose values are greater than
3527 127 match isprint() in the "C" locale. The "C" locale should be the
3528 default when a C program starts up. In most systems, only ASCII printing
3529 characters match isprint(). This difference caused the output from pcretest
3530 to vary, making some of the tests fail. I have changed pcretest so that:
3531
3532 (a) When it is outputting text in the compiled version of a pattern, bytes
3533 other than 32-126 are always shown as hex escapes.
3534
3535 (b) When it is outputting text that is a matched part of a subject string,
3536 it does the same, unless a different locale has been set for the match
3537 (using the /L modifier). In this case, it uses isprint() to decide.
3538
3539 4. Fixed a major bug that caused incorrect computation of the amount of memory
3540 required for a compiled pattern when options that changed within the
3541 pattern affected the logic of the preliminary scan that determines the
3542 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3543 that the computed length was too small. The symptoms of this bug were
3544 either the PCRE error "internal error: code overflow" from pcre_compile(),
3545 or a glibc crash with a message such as "pcretest: free(): invalid next
3546 size (fast)". Examples of patterns that provoked this bug (shown in
3547 pcretest format) are:
3548
3549 /(?-x: )/x
3550 /(?x)(?-x: \s*#\s*)/
3551 /((?i)[\x{c0}])/8
3552 /(?i:[\x{c0}])/8
3553
3554 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3555 is now done differently.
3556
3557 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3558 wrapper classes; (b) implement a new function in the C++ scanner that is
3559 more efficient than the old way of doing things because it avoids levels of
3560 recursion in the regex matching; (c) add a paragraph to the documentation
3561 for the FullMatch() function.
3562
3563 6. The escape sequence \n was being treated as whatever was defined as
3564 "newline". Not only was this contrary to the documentation, which states
3565 that \n is character 10 (hex 0A), but it also went horribly wrong when
3566 "newline" was defined as CRLF. This has been fixed.
3567
3568 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3569 was being set to -1 for the "end of line" case (supposedly a value that no
3570 character can have). Though this value is never used (the check for end of
3571 line is "zero bytes in current character"), it caused compiler complaints.
3572 I've changed it to 0xffffffff.
3573
3574 8. In pcre_version.c, the version string was being built by a sequence of
3575 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3576 string (as it is for production releases) called a macro with an empty
3577 argument. The C standard says the result of this is undefined. The gcc
3578 compiler treats it as an empty string (which was what was wanted) but it is
3579 reported that Visual C gives an error. The source has been hacked around to
3580 avoid this problem.
3581
3582 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3583 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3584 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3585 of them did).
3586
3587 10. Originally, pcretest opened its input and output without "b"; then I was
3588 told that "b" was needed in some environments, so it was added for release
3589 5.0 to both the input and output. (It makes no difference on Unix-like
3590 systems.) Later I was told that it is wrong for the input on Windows. I've
3591 now abstracted the modes into two macros, to make it easier to fiddle with
3592 them, and removed "b" from the input mode under Windows.
3593
3594 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3595
3596 12. Added -help and --help to pcretest as an official way of being reminded
3597 of the options.
3598
3599 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3600 and pcrecpp.cc because they annoy compilers at high warning levels.
3601
3602 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3603
3604 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3605 scripts are not C programs :-) and which was not noticed because it works
3606 on Linux.
3607
3608 16. pcretest is supposed to handle any length of pattern and data line (as one
3609 line or as a continued sequence of lines) by extending its input buffer if
3610 necessary. This feature was broken for very long pattern lines, leading to
3611 a string of junk being passed to pcre_compile() if the pattern was longer
3612 than about 50K.
3613
3614 17. I have done a major re-factoring of the way pcre_compile() computes the
3615 amount of memory needed for a compiled pattern. Previously, there was code
3616 that made a preliminary scan of the pattern in order to do this. That was
3617 OK when PCRE was new, but as the facilities have expanded, it has become
3618 harder and harder to keep it in step with the real compile phase, and there
3619 have been a number of bugs (see for example, 4 above). I have now found a
3620 cunning way of running the real compile function in a "fake" mode that
3621 enables it to compute how much memory it would need, while actually only
3622 ever using a few hundred bytes of working memory and without too many
3623 tests of the mode. This should make future maintenance and development
3624 easier. A side effect of this work is that the limit of 200 on the nesting
3625 depth of parentheses has been removed (though this was never a serious
3626 limitation, I suspect). However, there is a downside: pcre_compile() now
3627 runs more slowly than before (30% or more, depending on the pattern). I
3628 hope this isn't a big issue. There is no effect on runtime performance.
3629
3630 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3631 newline (only possible for the last line of a file) and it was a
3632 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3633
3634 19. Added additional timing features to pcretest. (1) The -tm option now times
3635 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3636 separate command line item, by a number that specifies the number of
3637 repeats to use when timing. The default is 50000; this gives better
3638 precision, but takes uncomfortably long for very large patterns.
3639
3640 20. Extended pcre_study() to be more clever in cases where a branch of a
3641 subpattern has no definite first character. For example, (a*|b*)[cd] would
3642 previously give no result from pcre_study(). Now it recognizes that the
3643 first character must be a, b, c, or d.
3644
3645 21. There was an incorrect error "recursive call could loop indefinitely" if
3646 a subpattern (or the entire pattern) that was being tested for matching an
3647 empty string contained only one non-empty item after a nested subpattern.
3648 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3649 incorrectly, because the \d was being skipped in the check.
3650
3651 22. The pcretest program now has a new pattern option /B and a command line
3652 option -b, which is equivalent to adding /B to every pattern. This causes
3653 it to show the compiled bytecode, without the additional information that
3654 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3655 is the same as /B/I).
3656
3657 23. A new optimization is now able automatically to treat some sequences such
3658 as a*b as a*+b. More specifically, if something simple (such as a character
3659 or a simple class like \d) has an unlimited quantifier, and is followed by
3660 something that cannot possibly match the quantified thing, the quantifier
3661 is automatically "possessified".
3662
3663 24. A recursive reference to a subpattern whose number was greater than 39
3664 went wrong under certain circumstances in UTF-8 mode. This bug could also
3665 have affected the operation of pcre_study().
3666
3667 25. Realized that a little bit of performance could be had by replacing
3668 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3669
3670 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3671
3672 27. Possessive quantifiers such as a++ were previously implemented by turning
3673 them into atomic groups such as ($>a+). Now they have their own opcodes,
3674 which improves performance. This includes the automatically created ones
3675 from 23 above.
3676
3677 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3678 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3679 the first matched character to be a colon. This applied both to named and
3680 numbered groups.
3681
3682 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3683
3684 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3685 building PCRE on Windows easier, so I have included it in the distribution.
3686
3687 31. There is now a check in pcretest against a ridiculously large number being
3688 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3689 loop, the loop is abandoned.
3690
3691 32. Forward references to subpatterns in conditions such as (?(2)...) where
3692 subpattern 2 is defined later cause pcre_compile() to search forwards in
3693 the pattern for the relevant set of parentheses. This search went wrong
3694 when there were unescaped parentheses in a character class, parentheses
3695 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3696
3697 33. "Subroutine" calls and backreferences were previously restricted to
3698 referencing subpatterns earlier in the regex. This restriction has now
3699 been removed.
3700
3701 34. Added a number of extra features that are going to be in Perl 5.10. On the
3702 whole, these are just syntactic alternatives for features that PCRE had
3703 previously implemented using the Python syntax or my own invention. The
3704 other formats are all retained for compatibility.
3705
3706 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3707 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3708 also .NET compatible.
3709
3710 (b) A recursion or subroutine call to a named group can now be defined as
3711 (?&name) as well as (?P>name).
3712
3713 (c) A backreference to a named group can now be defined as \k<name> or
3714 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3715 5.10, are also .NET compatible.
3716
3717 (d) A conditional reference to a named group can now use the syntax
3718 (?(<name>) or (?('name') as well as (?(name).
3719
3720 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3721 groups (named and numbered) that are never evaluated inline, but can be
3722 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3723 is always false. There may be only one alternative in such a group.
3724
3725 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3726 as the simple (?(R). The condition is true only if the most recent
3727 recursion is that of the given number or name. It does not search out
3728 through the entire recursion stack.
3729
3730 (g) The escape \gN or \g{N} has been added, where N is a positive or
3731 negative number, specifying an absolute or relative reference.
3732
3733 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3734 some "unreachable code" warnings.
3735
3736 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3737 things, this adds five new scripts.
3738
3739 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3740 There were also incompatibilities regarding the handling of \Q..\E inside
3741 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3742 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3743
3744 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3745 matches an empty string, and forcibly breaks the loop. There were bugs in
3746 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3747 against aaaa the result was just "a" rather than "aaaa", for example. Two
3748 separate and independent bugs (that affected different cases) have been
3749 fixed.
3750
3751 39. Refactored the code to abolish the use of different opcodes for small
3752 capturing bracket numbers. This is a tidy that I avoided doing when I
3753 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3754 The new approach is not only tidier, it makes it possible to reduce the
3755 memory needed to fix the previous bug (38).
3756
3757 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3758 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3759 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3760 mode.
3761
3762 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3763 report.
3764
3765 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3766 copy construction and assignment in the C++ wrapper.
3767
3768 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3769 couple of bugs that could have given wrong results in the "--newline=crlf"
3770 case.
3771
3772 44. Added a number of casts and did some reorganization of signed/unsigned int
3773 variables following suggestions from Dair Grant. Also renamed the variable
3774 "this" as "item" because it is a C++ keyword.
3775
3776 45. Arranged for dftables to add
3777
3778 #include "pcre_internal.h"
3779
3780 to pcre_chartables.c because without it, gcc 4.x may remove the array
3781 definition from the final binary if PCRE is built into a static library and
3782 dead code stripping is activated.
3783
3784 46. For an unanchored pattern, if a match attempt fails at the start of a
3785 newline sequence, and the newline setting is CRLF or ANY, and the next two
3786 characters are CRLF, advance by two characters instead of one.
3787
3788
3789 Version 6.7 04-Jul-06
3790 ---------------------
3791
3792 1. In order to handle tests when input lines are enormously long, pcretest has
3793 been re-factored so that it automatically extends its buffers when
3794 necessary. The code is crude, but this _is_ just a test program. The
3795 default size has been increased from 32K to 50K.
3796
3797 2. The code in pcre_study() was using the value of the re argument before
3798 testing it for NULL. (Of course, in any sensible call of the function, it
3799 won't be NULL.)
3800
3801 3. The memmove() emulation function in pcre_internal.h, which is used on
3802 systems that lack both memmove() and bcopy() - that is, hardly ever -
3803 was missing a "static" storage class specifier.
3804
3805 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3806 containing an extended class (one that cannot be represented by a bitmap
3807 because it contains high-valued characters or Unicode property items, e.g.
3808 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3809 pattern, but PCRE should not loop if you do not (it no longer does).
3810 [Detail: two cases were found: (a) a repeated subpattern containing an
3811 extended class; (b) a recursive reference to a subpattern that followed a
3812 previous extended class. It wasn't skipping over the extended class
3813 correctly when UTF-8 mode was not set.]
3814
3815 5. A negated single-character class was not being recognized as fixed-length
3816 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3817 compile error "lookbehind assertion is not fixed length".
3818
3819 6. The RunPerlTest auxiliary script was showing an unexpected difference
3820 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3821 write a Perl script that can interpret lines of an input file either as
3822 byte characters or as UTF-8, which is what "perltest" was being required to
3823 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3824 can't do is switch easily at run time between having the "use utf8;" pragma
3825 or not. In the end, I fudged it by using the RunPerlTest script to insert
3826 "use utf8;" explicitly for the UTF-8 tests.
3827
3828 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3829 the end of the subject string, contrary to the documentation and to what
3830 Perl does. This was true of both matching functions. Now it matches only at
3831 the start of the subject and immediately after *internal* newlines.
3832
3833 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3834 a pointer to an int instead of a pointer to an unsigned long int. This
3835 caused problems on 64-bit systems.
3836
3837 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3838 instance of the 'standard' template library not being so standard".
3839
3840 10. There was no check on the number of named subpatterns nor the maximum
3841 length of a subpattern name. The product of these values is used to compute
3842 the size of the memory block for a compiled pattern. By supplying a very
3843 long subpattern name and a large number of named subpatterns, the size
3844 computation could be caused to overflow. This is now prevented by limiting
3845 the length of names to 32 characters, and the number of named subpatterns
3846 to 10,000.
3847
3848 11. Subpatterns that are repeated with specific counts have to be replicated in
3849 the compiled pattern. The size of memory for this was computed from the
3850 length of the subpattern and the repeat count. The latter is limited to
3851 65535, but there was no limit on the former, meaning that integer overflow
3852 could in principle occur. The compiled length of a repeated subpattern is
3853 now limited to 30,000 bytes in order to prevent this.
3854
3855 12. Added the optional facility to have named substrings with the same name.
3856
3857 13. Added the ability to use a named substring as a condition, using the
3858 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3859 are numbers (not recommended). Forward references are permitted.
3860
3861 14. Added forward references in named backreferences (if you see what I mean).
3862
3863 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3864 pattern could run off the end of the subject. For example, the pattern
3865 "(?s)(.{1,5})"8 did this with the subject "ab".
3866
3867 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3868 PCRE_CASELESS was set when matching characters that were quantified with ?
3869 or *.
3870
3871 17. A character class other than a single negated character that had a minimum
3872 but no maximum quantifier - for example [ab]{6,} - was not handled
3873 correctly by pce_dfa_exec(). It would match only one character.
3874
3875 18. A valid (though odd) pattern that looked like a POSIX character
3876 class but used an invalid character after [ (for example [[,abc,]]) caused
3877 pcre_compile() to give the error "Failed: internal error: code overflow" or
3878 in some cases to crash with a glibc free() error. This could even happen if
3879 the pattern terminated after [[ but there just happened to be a sequence of
3880 letters, a binary zero, and a closing ] in the memory that followed.
3881
3882 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3883 over the years. Originally (before any Unicode support), just the bottom 8
3884 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3885 output from "man perlunicode" includes this:
3886
3887 The regular expression compiler produces polymorphic opcodes. That
3888 is, the pattern adapts to the data and automatically switches to
3889 the Unicode character scheme when presented with Unicode data--or
3890 instead uses a traditional byte scheme when presented with byte
3891 data.
3892
3893 Sadly, a wide octal escape does not cause a switch, and in a string with
3894 no other multibyte characters, these octal escapes are treated as before.
3895 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3896 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3897 Unicode string.
3898
3899 I have not perpetrated such confusion in PCRE. Up till now, it took just
3900 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3901 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3902 translate to the appropriate multibyte character.
3903
3904 29. Applied some refactoring to reduce the number of warnings from Microsoft
3905 and Borland compilers. This has included removing the fudge introduced
3906 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3907 a warning about an unused variable.
3908
3909 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3910 characters since release 4.0, because Perl (from release 5.004) does not.
3911 [Or at least, is documented not to: some releases seem to be in conflict
3912 with the documentation.] However, when a pattern was studied with
3913 pcre_study() and all its branches started with \s, PCRE still included VT
3914 as a possible starting character. Of course, this did no harm; it just
3915 caused an unnecessary match attempt.
3916
3917 22. Removed a now-redundant internal flag bit that recorded the fact that case
3918 dependency changed within the pattern. This was once needed for "required
3919 byte" processing, but is no longer used. This recovers a now-scarce options
3920 bit. Also moved the least significant internal flag bit to the most-
3921 significant bit of the word, which was not previously used (hangover from
3922 the days when it was an int rather than a uint) to free up another bit for
3923 the future.
3924
3925 23. Added support for CRLF line endings as well as CR and LF. As well as the
3926 default being selectable at build time, it can now be changed at runtime
3927 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3928 specify that it is scanning data with non-default line endings.
3929
3930 24. Changed the definition of CXXLINK to make it agree with the definition of
3931 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3932
3933 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3934 recursions. This makes a big different to stack usage for some patterns.
3935
3936 26. If a subpattern containing a named recursion or subroutine reference such
3937 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3938 the space required for the compiled pattern went wrong and gave too small a
3939 value. Depending on the environment, this could lead to "Failed: internal
3940 error: code overflow at offset 49" or "glibc detected double free or
3941 corruption" errors.
3942
3943 27. Applied patches from Google (a) to support the new newline modes and (b) to
3944 advance over multibyte UTF-8 characters in GlobalReplace.
3945
3946 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3947 difference for some implementation of PCRE in some Windows version.
3948
3949 29. Added some extra testing facilities to pcretest:
3950
3951 \q<number> in a data line sets the "match limit" value
3952 \Q<number> in a data line sets the "match recursion limt" value
3953 -S <number> sets the stack size, where <number> is in megabytes
3954
3955 The -S option isn't available for Windows.
3956
3957
3958 Version 6.6 06-Feb-06
3959 ---------------------
3960
3961 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3962 in pcreposix.h. I have copied the definition from pcre.h.
3963
3964 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3965 because pcre.h is no longer a built file.
3966
3967 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3968 not normally included in the compiled code.
3969
3970
3971 Version 6.5 01-Feb-06
3972 ---------------------
3973
3974 1. When using the partial match feature with pcre_dfa_exec(), it was not
3975 anchoring the second and subsequent partial matches at the new starting
3976 point. This could lead to incorrect results. For example, with the pattern
3977 /1234/, partially matching against "123" and then "a4" gave a match.
3978
3979 2. Changes to pcregrep:
3980
3981 (a) All non-match returns from pcre_exec() were being treated as failures
3982 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3983 error message is output. Some extra information is given for the
3984 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3985 probably the only errors that are likely to be caused by users (by
3986 specifying a regex that has nested indefinite repeats, for instance).
3987 If there are more than 20 of these errors, pcregrep is abandoned.
3988
3989 (b) A binary zero was treated as data while matching, but terminated the
3990 output line if it was written out. This has been fixed: binary zeroes
3991 are now no different to any other data bytes.
3992
3993 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3994 used to set a locale for matching. The --locale=xxxx long option has
3995 been added (no short equivalent) to specify a locale explicitly on the
3996 pcregrep command, overriding the environment variables.
3997
3998 (d) When -B was used with -n, some line numbers in the output were one less
3999 than they should have been.
4000
4001 (e) Added the -o (--only-matching) option.
4002
4003 (f) If -A or -C was used with -c (count only), some lines of context were
4004 accidentally printed for the final match.
4005
4006 (g) Added the -H (--with-filename) option.
4007
4008 (h) The combination of options -rh failed to suppress file names for files
4009 that were found from directory arguments.
4010
4011 (i) Added the -D (--devices) and -d (--directories) options.
4012
4013 (j) Added the -F (--fixed-strings) option.
4014
4015 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4016
4017 (l) Added the --colo(u)r option.
4018
4019 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4020 is not present by default.
4021
4022 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4023 items such as (?R) or (?1), when the recursion could match a number of
4024 alternatives. If it matched one of the alternatives, but subsequently,
4025 outside the recursion, there was a failure, the code tried to back up into
4026 the recursion. However, because of the way PCRE is implemented, this is not
4027 possible, and the result was an incorrect result from the match.
4028
4029 In order to prevent this happening, the specification of recursion has
4030 been changed so that all such subpatterns are automatically treated as
4031 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4032
4033 4. I had overlooked the fact that, in some locales, there are characters for
4034 which isalpha() is true but neither isupper() nor islower() are true. In
4035 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4036 and ordfeminine) are like this. This affected the treatment of \w and \W
4037 when they appeared in character classes, but not when they appeared outside
4038 a character class. The bit map for "word" characters is now created
4039 separately from the results of isalnum() instead of just taking it from the
4040 upper, lower, and digit maps. (Plus the underscore character, of course.)
4041
4042 5. The above bug also affected the handling of POSIX character classes such as
4043 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4044 permanent tables. Instead, the bit maps for such a class were previously
4045 created as the appropriate unions of the upper, lower, and digit bitmaps.
4046 Now they are created by subtraction from the [[:word:]] class, which has
4047 its own bitmap.
4048
4049 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4050 It is created by subtracting the vertical space characters (\x09, \x0a,
4051 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4052 subtraction was done in the overall bitmap for a character class, meaning
4053 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4054 be recognized. This bug has been fixed.
4055
4056 7. Patches from the folks at Google:
4057
4058 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4059 real life, but is still worth protecting against".
4060
4061 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4062 regular expressions".
4063
4064 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4065 have it.
4066
4067 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4068 "configure" and the latter not, in order to fix a problem somebody had
4069 with compiling the Arg class on HP-UX.
4070
4071 (e) Improve the error-handling of the C++ wrapper a little bit.
4072
4073 (f) New tests for checking recursion limiting.
4074
4075 8. The pcre_memmove() function, which is used only if the environment does not
4076 have a standard memmove() function (and is therefore rarely compiled),
4077 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4078 returning a result (though PCRE never actually uses the result).
4079
4080 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4081 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4082 returned instead of calling malloc() with an overflowing number that would
4083 most likely cause subsequent chaos.
4084
4085 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4086
4087 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4088 with this option is matched, the nmatch and pmatch options of regexec() are
4089 ignored.
4090
4091 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4092 provided in case anyone wants to the the POSIX interface with UTF-8
4093 strings.
4094
4095 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4096 C++ linking (needed for some HP-UX environments).
4097
4098 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4099 (unused parameter) and in the pcre_printint() function (omitted "default"
4100 switch label when the default is to do nothing).
4101
4102 15. Added some code to make it possible, when PCRE is compiled as a C++
4103 library, to replace subject pointers for pcre_exec() with a smart pointer
4104 class, thus making it possible to process discontinuous strings.
4105
4106 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4107 much the same function. They were added by different people who were trying
4108 to make PCRE easy to compile on non-Unix systems. It has been suggested
4109 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4110 for compiling on Windows systems. I have therefore replaced it with
4111 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4112 defaults to "extern" for C or "extern C" for C++, which works fine on
4113 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4114 SCOPE with something explicit in config.h. In addition:
4115
4116 (a) pcreposix.h still had just "extern" instead of either of these macros;
4117 I have replaced it with PCRE_DATA_SCOPE.
4118
4119 (b) Functions such as _pcre_xclass(), which are internal to the library,
4120 but external in the C sense, all had PCRE_EXPORT in their definitions.
4121 This is apparently wrong for the Windows case, so I have removed it.
4122 (It makes no difference on Unix-like systems.)
4123
4124 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4125 of recursive calls to match(). This is different to MATCH_LIMIT because
4126 that limits the total number of calls to match(), not all of which increase
4127 the depth of recursion. Limiting the recursion depth limits the amount of
4128 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4129 when PCRE is compiled, and changed at run time. A patch from Google adds
4130 this functionality to the C++ interface.
4131
4132 18. Changes to the handling of Unicode character properties:
4133
4134 (a) Updated the table to Unicode 4.1.0.
4135
4136 (b) Recognize characters that are not in the table as "Cn" (undefined).
4137
4138 (c) I revised the way the table is implemented to a much improved format
4139 which includes recognition of ranges. It now supports the ranges that
4140 are defined in UnicodeData.txt, and it also amalgamates other
4141 characters into ranges. This has reduced the number of entries in the
4142 table from around 16,000 to around 3,000, thus reducing its size
4143 considerably. I realized I did not need to use a tree structure after
4144 all - a binary chop search is just as efficient. Having reduced the
4145 number of entries, I extended their size from 6 bytes to 8 bytes to
4146 allow for more data.
4147
4148 (d) Added support for Unicode script names via properties such as \p{Han}.
4149
4150 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4151 matching that character.
4152
4153 20. When matching a repeated Unicode property with a minimum greater than zero,
4154 (for example \pL{2,}), PCRE could look past the end of the subject if it
4155 reached it while seeking the minimum number of characters. This could
4156 happen only if some of the characters were more than one byte long, because
4157 there is a check for at least the minimum number of bytes.
4158
4159 21. Refactored the implementation of \p and \P so as to be more general, to
4160 allow for more different types of property in future. This has changed the
4161 compiled form incompatibly. Anybody with saved compiled patterns that use
4162 \p or \P will have to recompile them.
4163
4164 22. Added "Any" and "L&" to the supported property types.
4165
4166 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4167 but give a compile time error if the value is greater than 0xff.
4168
4169 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4170 accidentally not being installed or uninstalled.
4171
4172 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4173 made were to insert the current release number. This seemed silly, because
4174 it made things harder for people building PCRE on systems that don't run
4175 "configure". I have turned pcre.h into a distributed file, no longer built
4176 by "configure", with the version identification directly included. There is
4177 no longer a pcre.h.in file.
4178
4179 However, this change necessitated a change to the pcre-config script as
4180 well. It is built from pcre-config.in, and one of the substitutions was the
4181 release number. I have updated configure.ac so that ./configure now finds
4182 the release number by grepping pcre.h.
4183
4184 26. Added the ability to run the tests under valgrind.
4185
4186
4187 Version 6.4 05-Sep-05
4188 ---------------------
4189
4190 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4191 "--" to be printed when multiple files were scanned, even when none of the
4192 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4193 consider it to be a bug, and have restored the previous behaviour.
4194
4195 2. A couple of code tidies to get rid of compiler warnings.
4196
4197 3. The pcretest program used to cheat by referring to symbols in the library
4198 whose names begin with _pcre_. These are internal symbols that are not
4199 really supposed to be visible externally, and in some environments it is
4200 possible to suppress them. The cheating is now confined to including
4201 certain files from the library's source, which is a bit cleaner.
4202
4203 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4204 file's purpose clearer.
4205
4206 5. Reorganized pcre_ucp_findchar().
4207
4208
4209 Version 6.3 15-Aug-05
4210 ---------------------
4211
4212 1. The file libpcre.pc.in did not have general read permission in the tarball.
4213
4214 2. There were some problems when building without C++ support:
4215
4216 (a) If C++ support was not built, "make install" and "make test" still
4217 tried to test it.
4218
4219 (b) There were problems when the value of CXX was explicitly set. Some
4220 changes have been made to try to fix these, and ...
4221
4222 (c) --disable-cpp can now be used to explicitly disable C++ support.
4223
4224 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4225 backslash in a target when C++ was disabled. This confuses some
4226 versions of "make", apparently. Using an intermediate variable solves
4227 this. (Same for CPP_LOBJ.)
4228
4229 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4230 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4231 necessary on certain architectures.
4232
4233 4. Added a setting of -export-symbols-regex to the link command to remove
4234 those symbols that are exported in the C sense, but actually are local
4235 within the library, and not documented. Their names all begin with
4236 "_pcre_". This is not a perfect job, because (a) we have to except some
4237 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4238 available (and never for static libraries). I have made a note to try to
4239 find a way round (a) in the future.
4240
4241
4242 Version 6.2 01-Aug-05
4243 ---------------------
4244
4245 1. There was no test for integer overflow of quantifier values. A construction
4246 such as {1111111111111111} would give undefined results. What is worse, if
4247 a minimum quantifier for a parenthesized subpattern overflowed and became
4248 negative, the calculation of the memory size went wrong. This could have
4249 led to memory overwriting.
4250
4251 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4252
4253 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4254 operating environments where this matters.
4255
4256 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4257 PCRE options from within the C++ wrapper.
4258
4259 5. Named capturing subpatterns were not being correctly counted when a pattern
4260 was compiled. This caused two problems: (a) If there were more than 100
4261 such subpatterns, the calculation of the memory needed for the whole
4262 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4263 back references of the form \12, where the number was greater than 9, were
4264 not recognized as back references, even though there were sufficient
4265 previous subpatterns.
4266
4267 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4268 versions of gcc, e.g. 2.95.4.
4269
4270
4271 Version 6.1 21-Jun-05
4272 ---------------------
4273
4274 1. There was one reference to the variable "posix" in pcretest.c that was not
4275 surrounded by "#if !defined NOPOSIX".
4276
4277 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4278 the cross-check on the old pcre_info() function, for the benefit of the
4279 cut-down version of PCRE that is currently imported into Exim.
4280
4281 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4282 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4283 patterns that start (?i) but I don't think that matters. The use of (?i) is
4284 just an example; this all applies to the other options as well.
4285
4286 4. Since libtool seems to echo the compile commands it is issuing, the output
4287 from "make" can be reduced a bit by putting "@" in front of each libtool
4288 compile command.
4289
4290 5. Patch from the folks at Google for configure.in to be a bit more thorough
4291 in checking for a suitable C++ installation before trying to compile the
4292 C++ stuff. This should fix a reported problem when a compiler was present,
4293 but no suitable headers.
4294
4295 6. The man pages all had just "PCRE" as their title. I have changed them to
4296 be the relevant file name. I have also arranged that these names are
4297 retained in the file doc/pcre.txt, which is a concatenation in text format
4298 of all the man pages except the little individual ones for each function.
4299
4300 7. The NON-UNIX-USE file had not been updated for the different set of source
4301 files that come with release 6. I also added a few comments about the C++
4302 wrapper.
4303
4304
4305 Version 6.0 07-Jun-05
4306 ---------------------
4307
4308 1. Some minor internal re-organization to help with my DFA experiments.
4309
4310 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4311 didn't matter for the library itself when fully configured, but did matter
4312 when compiling without UCP support, or within Exim, where the ucp files are
4313 not imported.
4314
4315 3. Refactoring of the library code to split up the various functions into
4316 different source modules. The addition of the new DFA matching code (see
4317 below) to a single monolithic source would have made it really too
4318 unwieldy, quite apart from causing all the code to be include in a
4319 statically linked application, when only some functions are used. This is
4320 relevant even without the DFA addition now that patterns can be compiled in
4321 one application and matched in another.
4322
4323 The downside of splitting up is that there have to be some external
4324 functions and data tables that are used internally in different modules of
4325 the library but which are not part of the API. These have all had their
4326 names changed to start with "_pcre_" so that they are unlikely to clash
4327 with other external names.
4328
4329 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4330 a different (DFA) algorithm. Although it is slower than the original
4331 function, it does have some advantages for certain types of matching
4332 problem.
4333
4334 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4335 including restarting after a partial match.
4336
4337 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4338 defined when compiling for Windows was sent to me. I have put it into the
4339 code, though I have no means of testing or verifying it.
4340
4341 7. Added the pcre_refcount() auxiliary function.
4342
4343 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4344 match before or at the first newline in the subject string. In pcretest,
4345 the /f option on a pattern can be used to set this.
4346
4347 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4348 would behave wrongly. This has been present in PCRE since release 4.0.
4349
4350 10. A number of changes to the pcregrep command:
4351
4352 (a) Refactored how -x works; insert ^(...)$ instead of setting
4353 PCRE_ANCHORED and checking the length, in preparation for adding
4354 something similar for -w.
4355
4356 (b) Added the -w (match as a word) option.
4357
4358 (c) Refactored the way lines are read and buffered so as to have more
4359 than one at a time available.
4360
4361 (d) Implemented a pcregrep test script.
4362
4363 (e) Added the -M (multiline match) option. This allows patterns to match
4364 over several lines of the subject. The buffering ensures that at least
4365 8K, or the rest of the document (whichever is the shorter) is available
4366 for matching (and similarly the previous 8K for lookbehind assertions).
4367
4368 (f) Changed the --help output so that it now says
4369
4370 -w, --word-regex(p)
4371
4372 instead of two lines, one with "regex" and the other with "regexp"
4373 because that confused at least one person since the short forms are the
4374 same. (This required a bit of code, as the output is generated
4375 automatically from a table. It wasn't just a text change.)
4376
4377 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4378 option but starts with a hyphen. Could be a pattern or a path name
4379 starting with a hyphen, for instance.
4380
4381 (h) "-" can be given as a file name to represent stdin.
4382
4383 (i) When file names are being printed, "(standard input)" is used for
4384 the standard input, for compatibility with GNU grep. Previously
4385 "<stdin>" was used.
4386
4387 (j) The option --label=xxx can be used to supply a name to be used for
4388 stdin when file names are being printed. There is no short form.
4389
4390 (k) Re-factored the options decoding logic because we are going to add
4391 two more options that take data. Such options can now be given in four
4392 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4393
4394 (l) Added the -A, -B, and -C options for requesting that lines of context
4395 around matches be printed.
4396
4397 (m) Added the -L option to print the names of files that do not contain
4398 any matching lines, that is, the complement of -l.
4399
4400 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4401 continue to scan other files.
4402
4403 (o) The -s option was incorrectly implemented. For compatibility with other
4404 greps, it now suppresses the error message for a non-existent or non-
4405 accessible file (but not the return code). There is a new option called
4406 -q that suppresses the output of matching lines, which was what -s was
4407 previously doing.
4408
4409 (p) Added --include and --exclude options to specify files for inclusion
4410 and exclusion when recursing.
4411
4412 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4413 Hopefully, it now does.
4414
4415 12. Missing cast in pcre_study().
4416
4417 13. Added an "uninstall" target to the makefile.
4418
4419 14. Replaced "extern" in the function prototypes in Makefile.in with
4420 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4421 world, but is set differently for Windows.
4422
4423 15. Added a second compiling function called pcre_compile2(). The only
4424 difference is that it has an extra argument, which is a pointer to an
4425 integer error code. When there is a compile-time failure, this is set
4426 non-zero, in addition to the error test pointer being set to point to an
4427 error message. The new argument may be NULL if no error number is required
4428 (but then you may as well call pcre_compile(), which is now just a
4429 wrapper). This facility is provided because some applications need a
4430 numeric error indication, but it has also enabled me to tidy up the way
4431 compile-time errors are handled in the POSIX wrapper.
4432
4433 16. Added VPATH=.libs to the makefile; this should help when building with one
4434 prefix path and installing with another. (Or so I'm told by someone who
4435 knows more about this stuff than I do.)
4436
4437 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4438 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4439 match everything, including newlines. This is not POSIX-compatible, but
4440 somebody wanted the feature. From pcretest it can be activated by using
4441 both the P and the s flags.
4442
4443 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4444
4445 19. libpcre.pc was being incorrectly installed as executable.
4446
4447 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4448 it now also looks for '\r' so that it will work unmodified on Windows.
4449
4450 21. Added Google's contributed C++ wrapper to the distribution.
4451
4452 22. Added some untidy missing memory free() calls in pcretest, to keep
4453 Electric Fence happy when testing.
4454
4455
4456
4457 Version 5.0 13-Sep-04
4458 ---------------------
4459
4460 1. Internal change: literal characters are no longer packed up into items
4461 containing multiple characters in a single byte-string. Each character
4462 is now matched using a separate opcode. However, there may be more than one
4463 byte in the character in UTF-8 mode.
4464
4465 2. The pcre_callout_block structure has two new fields: pattern_position and
4466 next_item_length. These contain the offset in the pattern to the next match
4467 item, and its length, respectively.
4468
4469 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4470 insertion of callouts before each pattern item. Added the /C option to
4471 pcretest to make use of this.
4472
4473 4. On the advice of a Windows user, the lines
4474
4475 #if defined(_WIN32) || defined(WIN32)
4476 _setmode( _fileno( stdout ), 0x8000 );
4477 #endif /* defined(_WIN32) || defined(WIN32) */
4478
4479 have been added to the source of pcretest. This apparently does useful
4480 magic in relation to line terminators.
4481
4482 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4483 for the benefit of those environments where the "b" makes a difference.
4484
4485 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4486 to know about it. I have put a hack into configure.in that adds in code
4487 to set GCC=yes if CC=icc. This seems to end up at a point in the
4488 generated configure script that is early enough to affect the setting of
4489 compiler options, which is what is needed, but I have no means of testing
4490 whether it really works. (The user who reported this had patched the
4491 generated configure script, which of course I cannot do.)
4492
4493 LATER: After change 22 below (new libtool files), the configure script
4494 seems to know about icc (and also ecc). Therefore, I have commented out
4495 this hack in configure.in.
4496
4497 7. Added support for pkg-config (2 patches were sent in).
4498
4499 8. Negated POSIX character classes that used a combination of internal tables
4500 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4501 [[:^ascii]]. Typically, they would match almost any characters. The other
4502 POSIX classes were not broken in this way.
4503
4504 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4505 to find the match, as PCRE was deluded into thinking that the match had to
4506 start at the start point or following a newline. The same bug applied to
4507 patterns with negative forward assertions or any backward assertions
4508 preceding ".*" at the start, unless the pattern required a fixed first
4509 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4510
4511 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4512 starting at the last subject character, bytes beyond the end of the subject
4513 string were read.
4514
4515 11. Renamed the variable "class" as "classbits" to make life easier for C++
4516 users. (Previously there was a macro definition, but it apparently wasn't
4517 enough.)
4518
4519 12. Added the new field "tables" to the extra data so that tables can be passed
4520 in at exec time, or the internal tables can be re-selected. This allows
4521 a compiled regex to be saved and re-used at a later time by a different
4522 program that might have everything at different addresses.
4523
4524 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4525 -R library as well as a -L library.
4526
4527 14. The debugging options of pcretest (-d on the command line or D on a
4528 pattern) showed incorrect output for anything following an extended class
4529 that contained multibyte characters and which was followed by a quantifier.
4530
4531 15. Added optional support for general category Unicode character properties
4532 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4533 support. It adds about 90K to the size of the library. The meanings of the
4534 inbuilt class escapes such as \d and \s have NOT been changed.
4535
4536 16. Updated pcredemo.c to include calls to free() to release the memory for the
4537 compiled pattern.
4538
4539 17. The generated file chartables.c was being created in the source directory
4540 instead of in the building directory. This caused the build to fail if the
4541 source directory was different from the building directory, and was
4542 read-only.
4543
4544 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4545 file. No doubt somebody will tell me if they don't make sense... Also added
4546 Dan Mooney's comments about building on OpenVMS.
4547
4548 19. Added support for partial matching via the PCRE_PARTIAL option for
4549 pcre_exec() and the \P data escape in pcretest.
4550
4551 20. Extended pcretest with 3 new pattern features:
4552
4553 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4554 write the compiled pattern to the file whose name is "rest-of-line".
4555 This is a straight binary dump of the data, with the saved pointer to
4556 the character tables forced to be NULL. The study data, if any, is
4557 written too. After writing, pcretest reads a new pattern.
4558
4559 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4560 compiled pattern from the given file. There must not be any
4561 occurrences of "<" in the file name (pretty unlikely); if there are,
4562 pcretest will instead treat the initial "<" as a pattern delimiter.
4563 After reading in the pattern, pcretest goes on to read data lines as
4564 usual.
4565
4566 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4567 and 16-bit fields in a compiled pattern, to simulate a pattern that
4568 was compiled on a host of opposite endianness.
4569
4570 21. The pcre-exec() function can now cope with patterns that were compiled on
4571 hosts of opposite endianness, with this restriction:
4572
4573 As for any compiled expression that is saved and used later, the tables
4574 pointer field cannot be preserved; the extra_data field in the arguments
4575 to pcre_exec() should be used to pass in a tables address if a value
4576 other than the default internal tables were used at compile time.
4577
4578 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4579 now diagnosed as an error. Previously, most of the time, a negative number
4580 would have been treated as zero, but if in addition "ovector" was passed as
4581 NULL, a crash could occur.
4582
4583 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4584 new versions from the libtool 1.5 distribution (the last one is a copy of
4585 a file called libtool.m4). This seems to have fixed the need to patch
4586 "configure" to support Darwin 1.3 (which I used to do). However, I still
4587 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4588 workstation).
4589
4590 24. Changed the PCRE licence to be the more standard "BSD" licence.
4591
4592
4593 Version 4.5 01-Dec-03
4594 ---------------------
4595
4596 1. There has been some re-arrangement of the code for the match() function so
4597 that it can be compiled in a version that does not call itself recursively.
4598 Instead, it keeps those local variables that need separate instances for
4599 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4600 needs to "recurse". Keeping track of where control must go is done by means
4601 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4602 hide most of the details from the main code, and operates only if
4603 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4604 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4605 operating.
4606
4607 To make it easier for callers to provide specially tailored get/free
4608 functions for this usage, two new functions, pcre_stack_malloc, and
4609 pcre_stack_free, are used. They are always called in strict stacking order,
4610 and the size of block requested is always the same.
4611
4612 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4613 PCRE has been compiled to use the stack or the heap for recursion. The
4614 -C option of pcretest uses this to show which version is compiled.
4615
4616 A new data escape \S, is added to pcretest; it causes the amounts of store
4617 obtained and freed by both kinds of malloc/free at match time to be added
4618 to the output.
4619
4620 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4621 what's available on my current Linux desktop machine.
4622
4623 3. When matching a UTF-8 string, the test for a valid string at the start has
4624 been extended. If start_offset is not zero, PCRE now checks that it points
4625 to a byte that is the start of a UTF-8 character. If not, it returns
4626 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4627 this is necessary because there may be backward assertions in the pattern.
4628 When matching the same subject several times, it may save resources to use
4629 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4630
4631 4. The code for checking the validity of UTF-8 strings has been tightened so
4632 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4633 containing "overlong sequences".
4634
4635 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4636 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4637 should have been "&", but it just so happened that all the cases this let
4638 through by mistake were picked up later in the function.
4639
4640 6. I had used a variable called "isblank" - this is a C99 function, causing
4641 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4642
4643 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4644 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4645 so the version is shown; (c) stop "make test" echoing "./RunTest".
4646
4647 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4648
4649 9. The prototype for memmove() for systems that don't have it was using
4650 size_t, but the inclusion of the header that defines size_t was later. I've
4651 moved the #includes for the C headers earlier to avoid this.
4652
4653 10. Added some adjustments to the code to make it easier to compiler on certain
4654 special systems:
4655
4656 (a) Some "const" qualifiers were missing.
4657 (b) Added the macro EXPORT before all exported functions; by default this
4658 is defined to be empty.
4659 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4660 that it reads its output file name as an argument instead of writing
4661 to the standard output and assuming this can be redirected.
4662
4663 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4664 class containing characters with values greater than 255, PCRE compilation
4665 went into a loop.
4666
4667 12. A recursive reference to a subpattern that was within another subpattern
4668 that had a minimum quantifier of zero caused PCRE to crash. For example,
4669 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4670 recursion. If the recursively-called subpattern itself had a zero repeat,
4671 that was OK.
4672
4673 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4674 buffer into which it was copied (for escape processing) was still set at
4675 1024, so long lines caused crashes.
4676
4677 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4678 "internal error: code overflow...". This applied to any character class
4679 that was followed by a possessive quantifier.
4680
4681 15. Modified the Makefile to add libpcre.la as a prerequisite for
4682 libpcreposix.la because I was told this is needed for a parallel build to
4683 work.
4684
4685 16. If a pattern that contained .* following optional items at the start was
4686 studied, the wrong optimizing data was generated, leading to matching
4687 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4688 matching string must start with a or b or c. The correct conclusion for
4689 this pattern is that a match can start with any character.
4690
4691
4692 Version 4.4 13-Aug-03
4693 ---------------------
4694
4695 1. In UTF-8 mode, a character class containing characters with values between
4696 127 and 255 was not handled correctly if the compiled pattern was studied.
4697 In fixing this, I have also improved the studying algorithm for such
4698 classes (slightly).
4699
4700 2. Three internal functions had redundant arguments passed to them. Removal
4701 might give a very teeny performance improvement.
4702
4703 3. Documentation bug: the value of the capture_top field in a callout is *one
4704 more than* the number of the hightest numbered captured substring.
4705
4706 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4707 in incorrectly linking with a previously installed version. They now link
4708 explicitly with libpcre.la.
4709
4710 5. configure.in no longer needs to recognize Cygwin specially.
4711
4712 6. A problem in pcre.in for Windows platforms is fixed.
4713
4714 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4715 pcretest, it used to include the size of the study block as part of its
4716 output. Unfortunately, the structure contains a field that has a different
4717 size on different hardware architectures. This meant that the tests that
4718 showed this size failed. As the block is currently always of a fixed size,
4719 this information isn't actually particularly useful in pcretest output, so
4720 I have just removed it.
4721
4722 8. Three pre-processor statements accidentally did not start in column 1.
4723 Sadly, there are *still* compilers around that complain, even though
4724 standard C has not required this for well over a decade. Sigh.
4725
4726 9. In pcretest, the code for checking callouts passed small integers in the
4727 callout_data field, which is a void * field. However, some picky compilers
4728 complained about the casts involved for this on 64-bit systems. Now
4729 pcretest passes the address of the small integer instead, which should get
4730 rid of the warnings.
4731
4732 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4733 both compile and run time, and gives an error if an invalid UTF-8 sequence
4734 is found. There is a option for disabling this check in cases where the
4735 string is known to be correct and/or the maximum performance is wanted.
4736
4737 11. In response to a bug report, I changed one line in Makefile.in from
4738
4739 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4740 to
4741 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4742
4743 to look similar to other lines, but I have no way of telling whether this
4744 is the right thing to do, as I do not use Windows. No doubt I'll get told
4745 if it's wrong...
4746
4747
4748 Version 4.3 21-May-03
4749 ---------------------
4750
4751 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4752 Makefile.
4753
4754 2. Some refactoring to improve the quality of the code:
4755
4756 (i) The utf8_table... variables are now declared "const".
4757
4758 (ii) The code for \cx, which used the "case flipping" table to upper case
4759 lower case letters, now just substracts 32. This is ASCII-specific,
4760 but the whole concept of \cx is ASCII-specific, so it seems
4761 reasonable.
4762
4763 (iii) PCRE was using its character types table to recognize decimal and
4764 hexadecimal digits in the pattern. This is silly, because it handles
4765 only 0-9, a-f, and A-F, but the character types table is locale-
4766 specific, which means strange things might happen. A private
4767 table is now used for this - though it costs 256 bytes, a table is
4768 much faster than multiple explicit tests. Of course, the standard
4769 character types table is still used for matching digits in subject
4770 strings against \d.
4771
4772 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4773 ending in _t are). So I've renamed it as ESC_tee.
4774
4775 3. The first argument for regexec() in the POSIX wrapper should have been
4776 defined as "const".
4777
4778 4. Changed pcretest to use malloc() for its buffers so that they can be
4779 Electric Fenced for debugging.
4780
4781 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4782 to read one or more bytes before the start of the subject string. Often this
4783 had no effect on PCRE's behaviour, but in some circumstances it could
4784 provoke a segmentation fault.
4785
4786 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4787 to try to read one or more bytes before the start of the subject string.
4788
4789 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4790 UTF-8 support could misbehave in various ways if the subject string
4791 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4792 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4793 back over UTF-8 characters.)
4794
4795
4796 Version 4.2 14-Apr-03
4797 ---------------------
4798
4799 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4800
4801 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4802 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4803 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4804 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4805 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4806 and BUILD_EXEEXT
4807 Note: automatic setting of the BUILD variables is not yet working
4808 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4809 compile-time but not at link-time
4810 [LINK]: use for linking executables only
4811 make different versions for Windows and non-Windows
4812 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4813 libraries
4814 [LINK_FOR_BUILD]: new variable
4815 [OBJEXT]: use throughout
4816 [EXEEXT]: use throughout
4817 <winshared>: new target
4818 <wininstall>: new target
4819 <dftables.o>: use native compiler
4820 <dftables>: use native linker
4821 <install>: handle Windows platform correctly
4822 <clean>: ditto
4823 <check>: ditto
4824 copy DLL to top builddir before testing
4825
4826 As part of these changes, -no-undefined was removed again. This was reported
4827 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4828 in any case.
4829
4830 3. Some tidies to get rid of compiler warnings:
4831
4832 . In the match_data structure, match_limit was an unsigned long int, whereas
4833 match_call_count was an int. I've made them both unsigned long ints.
4834
4835 . In pcretest the fact that a const uschar * doesn't automatically cast to
4836 a void * provoked a warning.
4837
4838 . Turning on some more compiler warnings threw up some "shadow" variables
4839 and a few more missing casts.
4840
4841 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4842 option, a class that contained a single character with a value between 128
4843 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4844
4845 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4846 option, a class that contained several characters, but with at least one
4847 whose value was between 128 and 255 caused PCRE to crash.
4848
4849
4850 Version 4.1 12-Mar-03
4851 ---------------------
4852
4853 1. Compiling with gcc -pedantic found a couple of places where casts were
4854 needed, and a string in dftables.c that was longer than standard compilers are
4855 required to support.
4856
4857 2. Compiling with Sun's compiler found a few more places where the code could
4858 be tidied up in order to avoid warnings.
4859
4860 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4861 first of these names is deprecated in the latest Autoconf in favour of the name
4862 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4863 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4864 analogy I have changed it to CFLAGS_FOR_BUILD.
4865
4866 4. Added -no-undefined to the linking command in the Makefile, because this is
4867 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4868 linking step for the pcreposix library.
4869
4870 5. PCRE was failing to diagnose the case of two named groups with the same
4871 name.
4872
4873 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4874 literal character that is needed in the subject for a match, and scans along to
4875 ensure that it is present before embarking on the full matching process. This
4876 saves time in cases of nested unlimited repeats that are never going to match.
4877 Problem: the scan can take a lot of time if the subject is very long (e.g.
4878 megabytes), thus penalizing straightforward matches. It is now done only if the
4879 amount of subject to be scanned is less than 1000 bytes.
4880
4881 7. A lesser problem with the same optimization is that it was recording the
4882 first character of an anchored pattern as "needed", thus provoking a search
4883 right along the subject, even when the first match of the pattern was going to
4884 fail. The "needed" character is now not set for anchored patterns, unless it
4885 follows something in the pattern that is of non-fixed length. Thus, it still
4886 fulfils its original purpose of finding quick non-matches in cases of nested
4887 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4888
4889
4890 Version 4.0 17-Feb-03
4891 ---------------------
4892
4893 1. If a comment in an extended regex that started immediately after a meta-item
4894 extended to the end of string, PCRE compiled incorrect data. This could lead to
4895 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4896
4897 2. Moved to autoconf 2.53 and libtool 1.4.2.
4898
4899 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4900 the special perltest8 script is no longer needed - all the tests can be run
4901 from a single perltest script.
4902
4903 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4904 by \s. It has now been removed in PCRE. This means it isn't recognized as
4905 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4906 class [:space:] *does* include VT, thereby creating a mess.
4907
4908 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4909 space and tab.
4910
4911 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4912 its new features into the main test script, reducing the number of scripts.
4913
4914 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4915 were backward compatible, and made the (?i) apply to the whole pattern, as if
4916 /i were given. Now it behaves more logically, and applies the option setting
4917 only to what follows. PCRE has been changed to follow suit. However, if it
4918 finds options settings right at the start of the pattern, it extracts them into
4919 the global options, as before. Thus, they show up in the info data.
4920
4921 8. Added support for the \Q...\E escape sequence. Characters in between are
4922 treated as literals. This is slightly different from Perl in that $ and @ are
4923 also handled as literals inside the quotes. In Perl, they will cause variable
4924 interpolation. Note the following examples:
4925
4926 Pattern PCRE matches Perl matches
4927
4928 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4929 \Qabc\$xyz\E abc\$xyz abc\$xyz
4930 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4931
4932 For compatibility with Perl, \Q...\E sequences are recognized inside character
4933 classes as well as outside them.
4934
4935 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4936 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4937 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4938 signed/unsigned warnings.
4939
4940 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4941 option for pcretest, so I've replaced it by a simple function that does just
4942 that job.
4943
4944 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4945 "pcregrep -".
4946
4947 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4948 Java package. This provides some syntactic sugar for simple cases of what my
4949 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4950 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4951 item, you can use this simplified notation. Note that only makes sense with
4952 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4953 greediness, whatever the setting of the PCRE_UNGREEDY option.
4954
4955 13. A change of greediness default within a pattern was not taking effect at
4956 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4957 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4958 was abstracted outside.
4959
4960 14. PCRE now supports the \G assertion. It is true when the current matching
4961 position is at the start point of the match. This differs from \A when the
4962 starting offset is non-zero. Used with the /g option of pcretest (or similar
4963 code), it works in the same way as it does for Perl's /g option. If all
4964 alternatives of a regex begin with \G, the expression is anchored to the start
4965 match position, and the "anchored" flag is set in the compiled expression.
4966
4967 15. Some bugs concerning the handling of certain option changes within patterns
4968 have been fixed. These applied to options other than (?ims). For example,
4969 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4970 been the other way round. Some of this was related to change 7 above.
4971
4972 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4973 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4974 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4975 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4976
4977 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4978 mode. Unlike ".", it always matches newline, whatever the setting of
4979 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4980 assertions. Perl allows it, but it doesn't (in general) work because it can't
4981 calculate the length of the lookbehind. At least, that's the case for Perl
4982 5.8.0 - I've been told they are going to document that it doesn't work in
4983 future.
4984
4985 18. Added an error diagnosis for escapes that PCRE does not support: these are
4986 \L, \l, \N, \P, \p, \U, \u, and \X.
4987
4988 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4989 reading past the end of the pattern in cases such as /[abcd/.
4990
4991 20. PCRE was getting more memory than necessary for patterns with classes that
4992 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4993
4994 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4995 compiling PCRE for use with Virtual Pascal.
4996
4997 22. Small fix to the Makefile to make it work properly if the build is done
4998 outside the source tree.
4999
5000 23. Added a new extension: a condition to go with recursion. If a conditional
5001 subpattern starts with (?(R) the "true" branch is used if recursion has
5002 happened, whereas the "false" branch is used only at the top level.
5003
5004 24. When there was a very long string of literal characters (over 255 bytes
5005 without UTF support, over 250 bytes with UTF support), the computation of how
5006 much memory was required could be incorrect, leading to segfaults or other
5007 strange effects.
5008
5009 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5010 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5011 there was a subsequent back reference to those brackets. This meant that, for
5012 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5013 possible to check for precisely this case. All we can do is abandon the
5014 optimization if .* occurs inside capturing brackets when there are any back
5015 references whatsoever. (See below for a better fix that came later.)
5016
5017 26. The handling of the optimization for finding the first character of a
5018 non-anchored pattern, and for finding a character that is required later in the
5019 match were failing in some cases. This didn't break the matching; it just
5020 failed to optimize when it could. The way this is done has been re-implemented.
5021
5022 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5023
5024 28. Added a new feature that provides some of the functionality that Perl
5025 provides with (?{...}). The facility is termed a "callout". The way it is done
5026 in PCRE is for the caller to provide an optional function, by setting
5027 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5028 global variable. By default it is unset, which disables all calling out. To get
5029 the function called, the regex must include (?C) at appropriate points. This
5030 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5031 This provides a means of identifying different callout points. When PCRE
5032 reaches such a point in the regex, if pcre_callout has been set, the external
5033 function is called. It is provided with data in a structure called
5034 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5035 matching continues; if it returns a non-zero value, the match at the current
5036 point fails. However, backtracking will occur if possible. [This was changed
5037 later and other features added - see item 49 below.]
5038
5039 29. pcretest is upgraded to test the callout functionality. It provides a
5040 callout function that displays information. By default, it shows the start of
5041 the match and the current position in the text. There are some new data escapes
5042 to vary what happens:
5043
5044 \C+ in addition, show current contents of captured substrings
5045 \C- do not supply a callout function
5046 \C!n return 1 when callout number n is reached
5047 \C!n!m return 1 when callout number n is reached for the mth time
5048
5049 30. If pcregrep was called with the -l option and just a single file name, it
5050 output "<stdin>" if a match was found, instead of the file name.
5051
5052 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5053 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5054 pcre_exec(). This saves a malloc/free per call. The default value of
5055 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5056 when configuring.
5057
5058 32. The default maximum size of a compiled pattern is 64K. There have been a
5059 few cases of people hitting this limit. The code now uses macros to handle the
5060 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5061 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5062 configuring. Tests 2 and 5 work only with 2-byte links because they output
5063 debugging information about compiled patterns.
5064
5065 33. Internal code re-arrangements:
5066
5067 (a) Moved the debugging function for printing out a compiled regex into
5068 its own source file (printint.c) and used #include to pull it into
5069 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5070 separate copies.
5071
5072 (b) Defined the list of op-code names for debugging as a macro in
5073 internal.h so that it is next to the definition of the opcodes.
5074
5075 (c) Defined a table of op-code lengths for simpler skipping along compiled
5076 code. This is again a macro in internal.h so that it is next to the
5077 definition of the opcodes.
5078
5079 34. Added support for recursive calls to individual subpatterns, along the
5080 lines of Robin Houston's patch (but implemented somewhat differently).
5081
5082 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5083 allow it to read and process whole directories in Win32. This code was
5084 contributed by Lionel Fourquaux; it has not been tested by me.
5085
5086 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5087 used to name a group. Names consist of alphanumerics and underscores, and must
5088 be unique. Back references use the syntax (?P=name) and recursive calls use
5089 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5090 numbers. The function pcre_fullinfo() can be used after compilation to extract
5091 a name/number map. There are three relevant calls:
5092
5093 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5094 PCRE_INFO_NAMECOUNT yields the number of entries
5095 PCRE_INFO_NAMETABLE yields a pointer to the map.
5096
5097 The map is a vector of fixed-size entries. The size of each entry depends on
5098 the length of the longest name used. The first two bytes of each entry are the
5099 group number, most significant byte first. There follows the corresponding
5100 name, zero terminated. The names are in alphabetical order.
5101
5102 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5103 case instead of 255. Making it the same both with and without UTF-8 support
5104 means that the same test output works with both.
5105
5106 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5107 calling malloc() with a zero argument.
5108
5109 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5110 optimization. I've improved things by keeping a bitmap of backreferences with
5111 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5112 fact referenced, the optimization can be applied. It is unlikely that a
5113 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5114 the match to follow \n) will appear inside brackets with a number greater than
5115 31, but if it does, any back reference > 31 suppresses the optimization.
5116
5117 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5118 of disabling numbered capturing parentheses. Any opening parenthesis that is
5119 not followed by ? behaves as if it were followed by ?: but named parentheses
5120 can still be used for capturing (and they will acquire numbers in the usual
5121 way).
5122
5123 41. Redesigned the return codes from the match() function into yes/no/error so
5124 that errors can be passed back from deep inside the nested calls. A malloc
5125 failure while inside a recursive subpattern call now causes the
5126 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5127
5128 42. It is now possible to set a limit on the number of times the match()
5129 function is called in a call to pcre_exec(). This facility makes it possible to
5130 limit the amount of recursion and backtracking, though not in a directly
5131 obvious way, because the match() function is used in a number of different
5132 circumstances. The count starts from zero for each position in the subject
5133 string (for non-anchored patterns). The default limit is, for compatibility, a
5134 large number, namely 10 000 000. You can change this in two ways:
5135
5136 (a) When configuring PCRE before making, you can use --with-match-limit=n
5137 to set a default value for the compiled library.
5138
5139 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5140 a different value is set. See 45 below.
5141
5142 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5143
5144 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5145 of things that can be changed at compile time. The first argument specifies
5146 what is wanted and the second points to where the information is to be placed.
5147 The current list of available information is:
5148
5149 PCRE_CONFIG_UTF8
5150
5151 The output is an integer that is set to one if UTF-8 support is available;
5152 otherwise it is set to zero.
5153
5154 PCRE_CONFIG_NEWLINE
5155
5156 The output is an integer that it set to the value of the code that is used for
5157 newline. It is either LF (10) or CR (13).
5158
5159 PCRE_CONFIG_LINK_SIZE
5160
5161 The output is an integer that contains the number of bytes used for internal
5162 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5163
5164 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5165
5166 The output is an integer that contains the threshold above which the POSIX
5167 interface uses malloc() for output vectors. See item 31 above.
5168
5169 PCRE_CONFIG_MATCH_LIMIT
5170
5171 The output is an unsigned integer that contains the default limit of the number
5172 of match() calls in a pcre_exec() execution. See 42 above.
5173
5174 44. pcretest has been upgraded by the addition of the -C option. This causes it
5175 to extract all the available output from the new pcre_config() function, and to
5176 output it. The program then exits immediately.
5177
5178 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5179 order to support additional features. One way would have been to define
5180 pcre_exec2() (for example) with extra arguments, but this would not have been
5181 extensible, and would also have required all calls to the original function to
5182 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5183 is used for passing in "extra" data from pcre_study().
5184
5185 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5186 contains the following fields:
5187
5188 flags a bitmap indicating which of the following fields are set
5189 study_data opaque data from pcre_study()
5190 match_limit a way of specifying a limit on match() calls for a specific
5191 call to pcre_exec()
5192 callout_data data for callouts (see 49 below)
5193
5194 The flag bits are also defined in pcre.h, and are
5195
5196 PCRE_EXTRA_STUDY_DATA
5197 PCRE_EXTRA_MATCH_LIMIT
5198 PCRE_EXTRA_CALLOUT_DATA
5199
5200 The pcre_study() function now returns one of these new pcre_extra blocks, with
5201 the actual study data pointed to by the study_data field, and the
5202 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5203 before. That is, this change is entirely upwards-compatible and requires no
5204 change to existing code.
5205
5206 If you want to pass in additional data to pcre_exec(), you can either place it
5207 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5208 block.
5209
5210 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5211 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5212 times with different match limits, until it finds the minimum value needed for
5213 pcre_exec() to complete. The value is then output. This can be instructive; for
5214 most simple matches the number is quite small, but for pathological cases it
5215 gets very large very quickly.
5216
5217 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5218 returns the size of the data block pointed to by the study_data field in a
5219 pcre_extra block, that is, the value that was passed as the argument to
5220 pcre_malloc() when PCRE was getting memory in which to place the information
5221 created by pcre_study(). The fourth argument should point to a size_t variable.
5222 pcretest has been extended so that this information is shown after a successful
5223 pcre_study() call when information about the compiled regex is being displayed.
5224
5225 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5226 because what follows is always an absolute path. (Later: it turns out that this
5227 is more than cosmetic for MinGW, because it doesn't like empty path
5228 components.)
5229
5230 49. Some changes have been made to the callout feature (see 28 above):
5231
5232 (i) A callout function now has three choices for what it returns:
5233
5234 0 => success, carry on matching
5235 > 0 => failure at this point, but backtrack if possible
5236 < 0 => serious error, return this value from pcre_exec()
5237
5238 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5239 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5240 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5241 use by callout functions. It will never be used by PCRE itself.
5242
5243 (ii) The pcre_extra structure (see 45 above) has a void * field called
5244 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5245 pcre_callout_block structure has a field of the same name. The contents of
5246 the field passed in the pcre_extra structure are passed to the callout
5247 function in the corresponding field in the callout block. This makes it
5248 easier to use the same callout-containing regex from multiple threads. For
5249 testing, the pcretest program has a new data escape
5250
5251 \C*n pass the number n (may be negative) as callout_data
5252
5253 If the callout function in pcretest receives a non-zero value as
5254 callout_data, it returns that value.
5255
5256 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5257 there were some redundant $(CFLAGS) in commands that are now specified as
5258 $(LINK), which already includes $(CFLAGS).
5259
5260 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5261 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5262 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5263 one-byte characters throughout. Note that case-insensitive matching applies
5264 only to characters whose values are less than 256. PCRE doesn't support the
5265 notion of cases for higher-valued characters.
5266
5267 (i) A character class whose characters are all within 0-255 is handled as
5268 a bit map, and the map is inverted for negative classes. Previously, a
5269 character > 255 always failed to match such a class; however it should
5270 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5271
5272 (ii) A negated character class with a single character < 255 is coded as
5273 "not this character" (OP_NOT). This wasn't working properly when the test
5274 character was multibyte, either singly or repeated.
5275
5276 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5277 mode, for example: \x{100}{2,3}.
5278
5279 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5280 singly or repeated) now correctly test multibyte characters. However,
5281 PCRE doesn't recognize any characters with values greater than 255 as
5282 digits, spaces, or word characters. Such characters always match \D, \S,
5283 and \W, and never match \d, \s, or \w.
5284
5285 (v) Classes may now contain characters and character ranges with values
5286 greater than 255. For example: [ab\x{100}-\x{400}].
5287
5288 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5289 PCRE in UTF-8 mode.
5290
5291 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5292 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5293 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5294 value.)
5295
5296 53. The single man page has become too large. I have therefore split it up into
5297 a number of separate man pages. These also give rise to individual HTML pages;
5298 these are now put in a separate directory, and there is an index.html page that
5299 lists them all. Some hyperlinking between the pages has been installed.
5300
5301 54. Added convenience functions for handling named capturing parentheses.
5302
5303 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5304 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5305 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5306 are faulted.
5307
5308 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5309 calling configure. These values are used when compiling the dftables.c program
5310 which is run to generate the source of the default character tables. They
5311 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5312 you will need to set these values.
5313
5314 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5315
5316
5317 Version 3.9 02-Jan-02
5318 ---------------------
5319
5320 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5321
5322 2. If --disable-static was given, the building process failed when trying to
5323 build pcretest and pcregrep. (For some reason it was using libtool to compile
5324 them, which is not right, as they aren't part of the library.)
5325
5326
5327 Version 3.8 18-Dec-01
5328 ---------------------
5329
5330 1. The experimental UTF-8 code was completely screwed up. It was packing the
5331 bytes in the wrong order. How dumb can you get?
5332
5333
5334 Version 3.7 29-Oct-01
5335 ---------------------
5336
5337 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5338 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5339 this didn't happen under Solaris 8, where I normally test things.
5340
5341 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5342 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5343 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5344 this fix an hour or so after the initial 3.7 release.)
5345
5346
5347 Version 3.6 23-Oct-01
5348 ---------------------
5349
5350 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5351 offsets passed as NULL with zero offset count.
5352
5353 2. The config.guess and config.sub files had not been updated when I moved to
5354 the latest autoconf.
5355
5356
5357 Version 3.5 15-Aug-01
5358 ---------------------
5359
5360 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5361 had been forgotten.
5362
5363 2. By using declared but undefined structures, we can avoid using "void"
5364 definitions in pcre.h while keeping the internal definitions of the structures
5365 private.
5366
5367 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5368 user point of view, this means that both static and shared libraries are built
5369 by default, but this can be individually controlled. More of the work of
5370 handling this static/shared cases is now inside libtool instead of PCRE's make
5371 file.
5372
5373 4. The pcretest utility is now installed along with pcregrep because it is
5374 useful for users (to test regexs) and by doing this, it automatically gets
5375 relinked by libtool. The documentation has been turned into a man page, so
5376 there are now .1, .txt, and .html versions in /doc.
5377
5378 5. Upgrades to pcregrep:
5379 (i) Added long-form option names like gnu grep.
5380 (ii) Added --help to list all options with an explanatory phrase.
5381 (iii) Added -r, --recursive to recurse into sub-directories.
5382 (iv) Added -f, --file to read patterns from a file.
5383
5384 6. pcre_exec() was referring to its "code" argument before testing that
5385 argument for NULL (and giving an error if it was NULL).
5386
5387 7. Upgraded Makefile.in to allow for compiling in a different directory from
5388 the source directory.
5389
5390 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5391 options bits, the pointer it was passed was to an int instead of to an unsigned
5392 long int. This mattered only on 64-bit systems.
5393
5394 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5395 generated) instead of pcre.in, which it its source. Also made the same change
5396 in several of the .c files.
5397
5398 10. A new release of gcc defines printf() as a macro, which broke pcretest
5399 because it had an ifdef in the middle of a string argument for printf(). Fixed
5400 by using separate calls to printf().
5401
5402 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5403 script, to force use of CR or LF instead of \n in the source. On non-Unix
5404 systems, the value can be set in config.h.
5405
5406 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5407 absolute limit. Changed the text of the error message to make this clear, and
5408 likewise updated the man page.
5409
5410 13. The limit of 99 on the number of capturing subpatterns has been removed.
5411 The new limit is 65535, which I hope will not be a "real" limit.
5412
5413
5414 Version 3.4 22-Aug-00
5415 ---------------------
5416
5417 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5418
5419 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5420
5421
5422 Version 3.3 01-Aug-00
5423 ---------------------
5424
5425 1. If an octal character was given, but the value was greater than \377, it
5426 was not getting masked to the least significant bits, as documented. This could
5427 lead to crashes in some systems.
5428
5429 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5430 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5431
5432 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5433 These just pass their arguments on to (pcre_free)(), but they are provided
5434 because some uses of PCRE bind it to non-C systems that can call its functions,
5435 but cannot call free() or pcre_free() directly.
5436
5437 4. Add "make test" as a synonym for "make check". Corrected some comments in
5438 the Makefile.
5439
5440 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5441 Makefile.
5442
5443 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5444 command called pgrep for grepping around the active processes.
5445
5446 7. Added the beginnings of support for UTF-8 character strings.
5447
5448 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5449 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5450 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5451 out for the ar command.)
5452
5453
5454 Version 3.2 12-May-00
5455 ---------------------
5456
5457 This is purely a bug fixing release.
5458
5459 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5460 of ZA. This was just one example of several cases that could provoke this bug,
5461 which was introduced by change 9 of version 2.00. The code for breaking
5462 infinite loops after an iteration that matches an empty string was't working
5463 correctly.
5464
5465 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5466 when matched against abbab (for example). After matching an empty string, it
5467 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5468 caused it to match further down the string than it should.
5469
5470 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5471 was there because it doesn't seem to be needed, and it causes trouble on some
5472 systems, as it is not a Standard C header. It has been removed.
5473
5474 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5475 were reported on the Macintosh. The changes were from
5476
5477 while ((c = *(++ptr)) != 0 && c != '\n');
5478 to
5479 while ((c = *(++ptr)) != 0 && c != '\n') ;
5480
5481 Totally extraordinary, but if that's what it takes...
5482
5483 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5484 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5485 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5486 assumes the way PCRE uses memmove() (always moving upwards).
5487
5488 6. PCRE is being used in one environment where strchr() is not available. There
5489 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5490 faster code anyway.
5491
5492
5493 Version 3.1 09-Feb-00
5494 ---------------------
5495
5496 The only change in this release is the fixing of some bugs in Makefile.in for
5497 the "install" target:
5498
5499 (1) It was failing to install pcreposix.h.
5500
5501 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5502
5503
5504 Version 3.0 01-Feb-00
5505 ---------------------
5506
5507 1. Add support for the /+ modifier to perltest (to output $` like it does in
5508 pcretest).
5509
5510 2. Add support for the /g modifier to perltest.
5511
5512 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5513 matches null strings.
5514
5515 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5516 pattern. Perl treats empty patterns specially - it reuses the most recent
5517 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5518 effect.
5519
5520 5. The POSIX interface was broken in that it was just handing over the POSIX
5521 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5522 required a bigger vector, with some working space on the end. This means that
5523 the POSIX wrapper now has to get and free some memory, and copy the results.
5524
5525 6. Added some simple autoconf support, placing the test data and the
5526 documentation in separate directories, re-organizing some of the
5527 information files, and making it build pcre-config (a GNU standard). Also added
5528 libtool support for building PCRE as a shared library, which is now the
5529 default.
5530
5531 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5532 09 are not valid octal constants. Single digits will be used for minor values
5533 less than 10.
5534
5535 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5536 existing programs that set these in the POSIX interface can use PCRE without
5537 modification.
5538
5539 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5540 return all that pcre_info() returns, plus additional data. The pcre_info()
5541 function is retained for compatibility, but is considered to be obsolete.
5542
5543 10. Added experimental recursion feature (?R) to handle one common case that
5544 Perl 5.6 will be able to do with (?p{...}).
5545
5546 11. Added support for POSIX character classes like [:alpha:], which Perl is
5547 adopting.
5548
5549
5550 Version 2.08 31-Aug-99
5551 ----------------------
5552
5553 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5554 trying to match at the startoffset position, but instead was moving forward to
5555 the next newline as if a previous match had failed.
5556
5557 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5558 and could get into a loop if a null string was matched other than at the start
5559 of the subject.
5560
5561 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5562 be distinguished at compile time, and for completeness also added PCRE_DATE.
5563
5564 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5565 in GnuWin32 environments.
5566
5567
5568 Version 2.07 29-Jul-99
5569 ----------------------
5570
5571 1. The documentation is now supplied in plain text form and HTML as well as in
5572 the form of man page sources.
5573
5574 2. C++ compilers don't like assigning (void *) values to other pointer types.
5575 In particular this affects malloc(). Although there is no problem in Standard
5576 C, I've put in casts to keep C++ compilers happy.
5577
5578 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5579 should be (const char *).
5580
5581 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5582 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5583 However, I haven't made this a standard facility. The documentation doesn't
5584 mention it, and the Makefile doesn't support it.
5585
5586 5. The Makefile now contains an "install" target, with editable destinations at
5587 the top of the file. The pcretest program is not installed.
5588
5589 6. pgrep -V now gives the PCRE version number and date.
5590
5591 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5592 causing the entire string to be ignored, instead of just the last character.
5593
5594 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5595 non-matching string, it can take a very, very long time, even for strings of
5596 quite modest length, because of the nested recursion. PCRE now does better in
5597 some of these cases. It does this by remembering the last required literal
5598 character in the pattern, and pre-searching the subject to ensure it is present
5599 before running the real match. In other words, it applies a heuristic to detect
5600 some types of certain failure quickly, and in the above example, if presented
5601 with a string that has no trailing " it gives "no match" very quickly.
5602
5603 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5604 other alternatives are tried instead.
5605
5606
5607 Version 2.06 09-Jun-99
5608 ----------------------
5609
5610 1. Change pcretest's output for amount of store used to show just the code
5611 space, because the remainder (the data block) varies in size between 32-bit and
5612 64-bit systems.
5613
5614 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5615 start matching at. This allows lookbehinds to work when searching for multiple
5616 occurrences in a string.
5617
5618 3. Added additional options to pcretest for testing multiple occurrences:
5619
5620 /+ outputs the rest of the string that follows a match
5621 /g loops for multiple occurrences, using the new startoffset argument
5622 /G loops for multiple occurrences by passing an incremented pointer
5623
5624 4. PCRE wasn't doing the "first character" optimization for patterns starting
5625 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5626 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5627 the letter 'x'. On long subject strings, this gives a significant speed-up.
5628
5629
5630 Version 2.05 21-Apr-99
5631 ----------------------
5632
5633 1. Changed the type of magic_number from int to long int so that it works
5634 properly on 16-bit systems.
5635
5636 2. Fixed a bug which caused patterns starting with .* not to work correctly
5637 when the subject string contained newline characters. PCRE was assuming
5638 anchoring for such patterns in all cases, which is not correct because .* will
5639 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5640 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5641 must be retried after every newline in the subject.
5642
5643
5644 Version 2.04 18-Feb-99
5645 ----------------------
5646
5647 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5648 computation of the store needed to hold the pattern was incorrect (too large).
5649 If such patterns were nested a few deep, this could multiply and become a real
5650 problem.
5651
5652 2. Added /M option to pcretest to show the memory requirement of a specific
5653 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5654
5655 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5656 compiled in such a way that the backtracking after subsequent failure was
5657 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5658 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5659
5660
5661 Version 2.03 02-Feb-99
5662 ----------------------
5663
5664 1. Fixed typo and small mistake in man page.
5665
5666 2. Added 4th condition (GPL supersedes if conflict) and created separate
5667 LICENCE file containing the conditions.
5668
5669 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5670 Perl, that is the internal \ allows the delimiter to be included in the
5671 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5672 the final delimiter, add \ to the end of the pattern (to test the error).
5673
5674 4. Added the convenience functions for extracting substrings after a successful
5675 match. Updated pcretest to make it able to test these functions.
5676
5677
5678 Version 2.02 14-Jan-99
5679 ----------------------
5680
5681 1. Initialized the working variables associated with each extraction so that
5682 their saving and restoring doesn't refer to uninitialized store.
5683
5684 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5685 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5686 fix the problem.
5687
5688 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5689 calls, and wasn't printing the correct value for compiling calls. Increased the
5690 default value of LOOPREPEAT, and the number of significant figures in the
5691 times.
5692
5693 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5694
5695 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5696 a building problem on Windows NT with a FAT file system.
5697
5698
5699 Version 2.01 21-Oct-98
5700 ----------------------
5701
5702 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5703 to character tables built by pcre_maketables() in the current locale. If NULL
5704 is passed, the default tables are used.
5705
5706
5707 Version 2.00 24-Sep-98
5708 ----------------------
5709
5710 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5711 it any more.
5712
5713 2. Allow quantification of (?>) groups, and make it work correctly.
5714
5715 3. The first character computation wasn't working for (?>) groups.
5716
5717 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5718 end of the subject) and add 5.005's \z, which really does match only at the
5719 very end of the subject.
5720
5721 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5722
5723 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5724 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5725 localized options. All options to pcre_study() were also removed.
5726
5727 7. Add other new features from 5.005:
5728
5729 $(?<= positive lookbehind
5730 $(?<! negative lookbehind
5731 (?imsx-imsx) added the unsetting capability
5732 such a setting is global if at outer level; local otherwise
5733 (?imsx-imsx:) non-capturing groups with option setting
5734 (?(cond)re|re) conditional pattern matching
5735
5736 A backreference to itself in a repeated group matches the previous
5737 captured string.
5738
5739 8. General tidying up of studying (both automatic and via "study")
5740 consequential on the addition of new assertions.
5741
5742 9. As in 5.005, unlimited repeated groups that could match an empty substring
5743 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5744 runtime if any iteration does actually match an empty substring.
5745
5746 10. Include the RunTest script in the distribution.
5747
5748 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5749 discrepancies, some of which were old and were also with respect to 5.004. They
5750 have now been fixed.
5751
5752
5753 Version 1.09 28-Apr-98
5754 ----------------------
5755
5756 1. A negated single character class followed by a quantifier with a minimum
5757 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5758 program crashes, or just wrong answers. This did not apply to negated classes
5759 containing more than one character, or to minima other than one.
5760
5761
5762 Version 1.08 27-Mar-98
5763 ----------------------
5764
5765 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5766
5767 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5768 latter must appear before anything that relies on it in the pattern.
5769
5770
5771 Version 1.07 16-Feb-98
5772 ----------------------
5773
5774 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5775 repeat of a potentially empty string).
5776
5777
5778 Version 1.06 23-Jan-98
5779 ----------------------
5780
5781 1. Added Markus Oberhumer's little patches for C++.
5782
5783 2. Literal strings longer than 255 characters were broken.
5784
5785
5786 Version 1.05 23-Dec-97
5787 ----------------------
5788
5789 1. Negated character classes containing more than one character were failing if
5790 PCRE_CASELESS was set at run time.
5791
5792
5793 Version 1.04 19-Dec-97
5794 ----------------------
5795
5796 1. Corrected the man page, where some "const" qualifiers had been omitted.
5797
5798 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5799 input syntax.
5800
5801 3. Fixed memory leak which occurred when a regex with back references was
5802 matched with an offsets vector that wasn't big enough. The temporary memory
5803 that is used in this case wasn't being freed if the match failed.
5804
5805 4. Tidied pcretest to ensure it frees memory that it gets.
5806
5807 5. Temporary memory was being obtained in the case where the passed offsets
5808 vector was exactly big enough.
5809
5810 6. Corrected definition of offsetof() from change 5 below.
5811
5812 7. I had screwed up change 6 below and broken the rules for the use of
5813 setjmp(). Now fixed.
5814
5815
5816 Version 1.03 18-Dec-97
5817 ----------------------
5818
5819 1. A erroneous regex with a missing opening parenthesis was correctly
5820 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5821 on some systems.
5822
5823 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5824 it was reported that one broken compiler failed on the former because "code" is
5825 also an independent variable.
5826
5827 3. The erroneous regex a[]b caused an array overrun reference.
5828
5829 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5830 fail on data ending with that character. (It was going on too far, and checking
5831 the next character, typically a binary zero.) This was specific to the
5832 optimized code for single-character negative classes.
5833
5834 5. Added a contributed patch from the TIN world which does the following:
5835
5836 + Add an undef for memmove, in case the the system defines a macro for it.
5837
5838 + Add a definition of offsetof(), in case there isn't one. (I don't know
5839 the reason behind this - offsetof() is part of the ANSI standard - but
5840 it does no harm).
5841
5842 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5843 most of the places where whitespace preceded '#'. I have given up and
5844 allowed the remaining 2 cases to be at the margin.
5845
5846 + Rename some variables in pcre to eliminate shadowing. This seems very
5847 pedantic, but does no harm, of course.
5848
5849 6. Moved the call to setjmp() into its own function, to get rid of warnings
5850 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5851
5852 7. Constructs such as \d{8,} were compiling into the equivalent of
5853 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5854 outcome, but in this particular case used more store than had been allocated,
5855 which caused the bug to be discovered because it threw up an internal error.
5856
5857 8. The debugging code in both pcre and pcretest for outputting the compiled
5858 form of a regex was going wrong in the case of back references followed by
5859 curly-bracketed repeats.
5860
5861
5862 Version 1.02 12-Dec-97
5863 ----------------------
5864
5865 1. Typos in pcre.3 and comments in the source fixed.
5866
5867 2. Applied a contributed patch to get rid of places where it used to remove
5868 'const' from variables, and fixed some signed/unsigned and uninitialized
5869 variable warnings.
5870
5871 3. Added the "runtest" target to Makefile.
5872
5873 4. Set default compiler flag to -O2 rather than just -O.
5874
5875
5876 Version 1.01 19-Nov-97
5877 ----------------------
5878
5879 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5880 like /([ab]*)*/, that is, for classes with more than one character in them.
5881
5882 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5883 as /((?>a*))*/ (a PCRE_EXTRA facility).
5884
5885
5886 Version 1.00 18-Nov-97
5887 ----------------------
5888
5889 1. Added compile-time macros to support systems such as SunOS4 which don't have
5890 memmove() or strerror() but have other things that can be used instead.
5891
5892 2. Arranged that "make clean" removes the executables.
5893
5894
5895 Version 0.99 27-Oct-97
5896 ----------------------
5897
5898 1. Fixed bug in code for optimizing classes with only one character. It was
5899 initializing a 32-byte map regardless, which could cause it to run off the end
5900 of the memory it had got.
5901
5902 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5903
5904
5905 Version 0.98 22-Oct-97
5906 ----------------------
5907
5908 1. Fixed bug in code for handling temporary memory usage when there are more
5909 back references than supplied space in the ovector. This could cause segfaults.
5910
5911
5912 Version 0.97 21-Oct-97
5913 ----------------------
5914
5915 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5916
5917 2. Optimized negated single characters not to use a bit map.
5918
5919 3. Brought error texts together as macro definitions; clarified some of them;
5920 fixed one that was wrong - it said "range out of order" when it meant "invalid
5921 escape sequence".
5922
5923 4. Changed some char * arguments to const char *.
5924
5925 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5926
5927 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5928 pcretest.
5929
5930
5931 Version 0.96 16-Oct-97
5932 ----------------------
5933
5934 1. Added a simple "pgrep" utility to the distribution.
5935
5936 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5937 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5938 where "ddd" means "one or more decimal digits".
5939
5940 3. Fixed serious bug. If a pattern had a back reference, but the call to
5941 pcre_exec() didn't supply a large enough ovector to record the related
5942 identifying subpattern, the match always failed. PCRE now remembers the number
5943 of the largest back reference, and gets some temporary memory in which to save
5944 the offsets during matching if necessary, in order to ensure that
5945 backreferences always work.
5946
5947 4. Increased the compatibility with Perl in a number of ways:
5948
5949 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5950 to request this handling. The option can be set at compile or exec time.
5951
5952 (b) $ matches before a terminating newline by default; an option
5953 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5954 mode). The option can be set at compile or exec time.
5955
5956 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5957 the same as Perl's. If the decimal number it represents is less than 10
5958 or there aren't that many previous left capturing parentheses, an octal
5959 escape is read. Inside a character class, it's always an octal escape,
5960 even if it is a single digit.
5961
5962 (d) An escaped but undefined alphabetic character is taken as a literal,
5963 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5964 escapes.
5965
5966 (e) {0} is now permitted. (The previous item is removed from the compiled
5967 pattern).
5968
5969 5. Changed all the names of code files so that the basic parts are no longer
5970 than 10 characters, and abolished the teeny "globals.c" file.
5971
5972 6. Changed the handling of character classes; they are now done with a 32-byte
5973 bit map always.
5974
5975 7. Added the -d and /D options to pcretest to make it possible to look at the
5976 internals of compilation without having to recompile pcre.
5977
5978
5979 Version 0.95 23-Sep-97
5980 ----------------------
5981
5982 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5983 \x20 at the start of a run of normal characters. These were being treated as
5984 real characters, instead of the source characters being re-checked.
5985
5986
5987 Version 0.94 18-Sep-97
5988 ----------------------
5989
5990 1. The functions are now thread-safe, with the caveat that the global variables
5991 containing pointers to malloc() and free() or alternative functions are the
5992 same for all threads.
5993
5994 2. Get pcre_study() to generate a bitmap of initial characters for non-
5995 anchored patterns when this is possible, and use it if passed to pcre_exec().
5996
5997
5998 Version 0.93 15-Sep-97
5999 ----------------------
6000
6001 1. /(b)|(:+)/ was computing an incorrect first character.
6002
6003 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
6004 but not actually doing anything yet.
6005
6006 3. Treat "-" characters in classes that cannot be part of ranges as literals,
6007 as Perl does (e.g. [-az] or [az-]).
6008
6009 4. Set the anchored flag if a branch starts with .* or .*? because that tests
6010 all possible positions.
6011
6012 5. Split up into different modules to avoid including unneeded functions in a
6013 compiled binary. However, compile and exec are still in one module. The "study"
6014 function is split off.
6015
6016 6. The character tables are now in a separate module whose source is generated
6017 by an auxiliary program - but can then be edited by hand if required. There are
6018 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
6019 toupper() in the code.
6020
6021 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
6022 make them global. Abolish the function for setting them, as the caller can now
6023 set them directly.
6024
6025
6026 Version 0.92 11-Sep-97
6027 ----------------------
6028
6029 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
6030 (e.g. /a{1,3}/) was broken (I mis-optimized it).
6031
6032 2. Caseless matching was not working in character classes if the characters in
6033 the pattern were in upper case.
6034
6035 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
6036
6037 4. Make PCRE_ANCHORED public and accept as a compile option.
6038
6039 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
6040 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
6041 pass them.
6042
6043 6. Give an error if bad option bits passed at compile or run time.
6044
6045 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
6046 pcretest to cause it to pass that flag.
6047
6048 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
6049 options, and the first character, if set.
6050
6051 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
6052
6053
6054 Version 0.91 10-Sep-97
6055 ----------------------
6056
6057 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
6058 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
6059
6060 2. PCRE was looping on encountering an indefinitely repeated back reference to
6061 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does