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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1664 - (show annotations)
Thu Aug 4 17:15:38 2016 UTC (2 weeks, 6 days ago) by ph10
File size: 290395 byte(s)
Fix character class bug when a Unicode property was present with \D etc in a 
non-ucp character class in a wide character mode.
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 8.40 xx-xxxx-2016
8 -------------------------
9
10 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
11 the match extended over a line boundary.
12
13 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
14 MSVC static compilation, putting the first patch under a new option.
15
16 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
17
18 4. Ignore "show all captures" (/=) for DFA matching.
19
20 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
21
22 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode), without
23 PCRE_UCP set, a negative character type such as \D in a positive class
24 should cause all characters greater than 255 to match, whatever else is in
25 the class. There was a bug that caused this not to happen if a Unicode
26 property item was added to such a class, for example [\D\P{Nd}] or [\W\pL].
27
28
29 Version 8.39 14-June-2016
30 -------------------------
31
32 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
33 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
34 misbehaved. This bug was found by the LLVM fuzzer.
35
36 2. Similar to the above, if an isolated \E was present between an item and its
37 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
38 bug was found by the LLVM fuzzer.
39
40 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
41 working correctly in UCP mode.
42
43 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
44 was set when the pmatch argument was NULL. It now returns REG_INVARG.
45
46 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
47
48 6. An empty \Q\E sequence between an item and its qualifier caused
49 pcre_compile() to misbehave when auto callouts were enabled. This bug was
50 found by the LLVM fuzzer.
51
52 7. If a pattern that was compiled with PCRE_EXTENDED started with white
53 space or a #-type comment that was followed by (?-x), which turns off
54 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
55 pcre_compile() assumed that (?-x) applied to the whole pattern and
56 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
57
58 8. A call of pcre_copy_named_substring() for a named substring whose number
59 was greater than the space in the ovector could cause a crash.
60
61 9. Yet another buffer overflow bug involved duplicate named groups with a
62 group that reset capture numbers (compare 8.38/7 below). Once again, I have
63 just allowed for more memory, even if not needed. (A proper fix is
64 implemented in PCRE2, but it involves a lot of refactoring.)
65
66 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
67 start of the match to be earlier than the end.
68
69 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
70
71 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
72 assertion, caused pcretest to generate incorrect output, and also to read
73 uninitialized memory (detected by ASAN or valgrind).
74
75 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
76 nested set of parentheses of sufficient size caused an overflow of the
77 compiling workspace (which was diagnosed, but of course is not desirable).
78
79 14. And yet another buffer overflow bug involving duplicate named groups, this
80 time nested, with a nested back reference. Yet again, I have just allowed
81 for more memory, because anything more needs all the refactoring that has
82 been done for PCRE2. An example pattern that provoked this bug is:
83 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
84 registered as CVE-2016-1283.
85
86 15. pcretest went into a loop if global matching was requested with an ovector
87 size less than 2. It now gives an error message. This bug was found by
88 afl-fuzz.
89
90 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
91 ("assertion expected") when (?(?C) was not followed by an opening
92 parenthesis.
93
94 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
95 actually affect anything, by sheer luck.
96
97 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
98 static compilation.
99
100 19. Modified the RunTest script to incorporate a valgrind suppressions file so
101 that certain errors, provoked by the SSE2 instruction set when JIT is used,
102 are ignored.
103
104 20. A racing condition is fixed in JIT reported by Mozilla.
105
106 21. Minor code refactor to avoid "array subscript is below array bounds"
107 compiler warning.
108
109 22. Minor code refactor to avoid "left shift of negative number" warning.
110
111 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
112 UCP support.
113
114 24. Refactor to avoid compiler warnings in pcrecpp.cc.
115
116 25. Refactor to fix a typo in pcre_jit_test.c
117
118 26. Patch to support compiling pcrecpp.cc with Intel compiler.
119
120
121 Version 8.38 23-November-2015
122 -----------------------------
123
124 1. If a group that contained a recursive back reference also contained a
125 forward reference subroutine call followed by a non-forward-reference
126 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
127 compile correct code, leading to undefined behaviour or an internally
128 detected error. This bug was discovered by the LLVM fuzzer.
129
130 2. Quantification of certain items (e.g. atomic back references) could cause
131 incorrect code to be compiled when recursive forward references were
132 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
133 This bug was discovered by the LLVM fuzzer.
134
135 3. A repeated conditional group whose condition was a reference by name caused
136 a buffer overflow if there was more than one group with the given name.
137 This bug was discovered by the LLVM fuzzer.
138
139 4. A recursive back reference by name within a group that had the same name as
140 another group caused a buffer overflow. For example:
141 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
142
143 5. A forward reference by name to a group whose number is the same as the
144 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
145 a buffer overflow at compile time. This bug was discovered by the LLVM
146 fuzzer.
147
148 6. A lookbehind assertion within a set of mutually recursive subpatterns could
149 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
150
151 7. Another buffer overflow bug involved duplicate named groups with a
152 reference between their definition, with a group that reset capture
153 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
154 fixed by always allowing for more memory, even if not needed. (A proper fix
155 is implemented in PCRE2, but it involves more refactoring.)
156
157 8. There was no check for integer overflow in subroutine calls such as (?123).
158
159 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
160 being treated as a literal 'l' instead of causing an error.
161
162 10. There was a buffer overflow if pcre_exec() was called with an ovector of
163 size 1. This bug was found by american fuzzy lop.
164
165 11. If a non-capturing group containing a conditional group that could match
166 an empty string was repeated, it was not identified as matching an empty
167 string itself. For example: /^(?:(?(1)x|)+)+$()/.
168
169 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
170 \a and \e in test subject lines.
171
172 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
173 instead of the EBCDIC value.
174
175 14. The handling of \c in an EBCDIC environment has been revised so that it is
176 now compatible with the specification in Perl's perlebcdic page.
177
178 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
179 ASCII/Unicode. This has now been added to the list of characters that are
180 recognized as white space in EBCDIC.
181
182 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
183 error (correctly) when used outside a class, but did not give an error
184 within a class.
185
186 17. \h within a class was incorrectly compiled in EBCDIC environments.
187
188 18. A pattern with an unmatched closing parenthesis that contained a backward
189 assertion which itself contained a forward reference caused buffer
190 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
191
192 19. JIT should return with error when the compiled pattern requires more stack
193 space than the maximum.
194
195 20. A possessively repeated conditional group that could match an empty string,
196 for example, /(?(R))*+/, was incorrectly compiled.
197
198 21. Fix infinite recursion in the JIT compiler when certain patterns such as
199 /(?:|a|){100}x/ are analysed.
200
201 22. Some patterns with character classes involving [: and \\ were incorrectly
202 compiled and could cause reading from uninitialized memory or an incorrect
203 error diagnosis.
204
205 23. Pathological patterns containing many nested occurrences of [: caused
206 pcre_compile() to run for a very long time.
207
208 24. A conditional group with only one branch has an implicit empty alternative
209 branch and must therefore be treated as potentially matching an empty
210 string.
211
212 25. If (?R was followed by - or + incorrect behaviour happened instead of a
213 diagnostic.
214
215 26. Arrange to give up on finding the minimum matching length for overly
216 complex patterns.
217
218 27. Similar to (4) above: in a pattern with duplicated named groups and an
219 occurrence of (?| it is possible for an apparently non-recursive back
220 reference to become recursive if a later named group with the relevant
221 number is encountered. This could lead to a buffer overflow. Wen Guanxing
222 from Venustech ADLAB discovered this bug.
223
224 28. If pcregrep was given the -q option with -c or -l, or when handling a
225 binary file, it incorrectly wrote output to stdout.
226
227 29. The JIT compiler did not restore the control verb head in case of *THEN
228 control verbs. This issue was found by Karl Skomski with a custom LLVM
229 fuzzer.
230
231 30. Error messages for syntax errors following \g and \k were giving inaccurate
232 offsets in the pattern.
233
234 31. Added a check for integer overflow in conditions (?(<digits>) and
235 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
236 fuzzer.
237
238 32. Handling recursive references such as (?2) when the reference is to a group
239 later in the pattern uses code that is very hacked about and error-prone.
240 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
241 give an internal error if it is obvious that compiling has gone wrong.
242
243 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
244 This issue was found by Karl Skomski with a custom LLVM fuzzer.
245
246 34. The JIT compiler should restore the control chain for empty possessive
247 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
248
249 35. Match limit check added to JIT recursion. This issue was found by Karl
250 Skomski with a custom LLVM fuzzer.
251
252 36. Yet another case similar to 27 above has been circumvented by an
253 unconditional allocation of extra memory. This issue is fixed "properly" in
254 PCRE2 by refactoring the way references are handled. Wen Guanxing
255 from Venustech ADLAB discovered this bug.
256
257 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
258 with a custom LLVM fuzzer.
259
260 38. Fixed a corner case of range optimization in JIT.
261
262 39. An incorrect error "overran compiling workspace" was given if there were
263 exactly enough group forward references such that the last one extended
264 into the workspace safety margin. The next one would have expanded the
265 workspace. The test for overflow was not including the safety margin.
266
267 40. A match limit issue is fixed in JIT which was found by Karl Skomski
268 with a custom LLVM fuzzer.
269
270 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
271 work under Windows. (Why has it taken so long for anyone to notice?)
272
273 42. In a character class such as [\W\p{Any}] where both a negative-type escape
274 ("not a word character") and a property escape were present, the property
275 escape was being ignored.
276
277 43. Fix crash caused by very long (*MARK) or (*THEN) names.
278
279 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
280 by a single ASCII character in a class item, was incorrectly compiled in
281 UCP mode. The POSIX class got lost, but only if the single character
282 followed it.
283
284 45. [:punct:] in UCP mode was matching some characters in the range 128-255
285 that should not have been matched.
286
287 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
288 class, all characters with code points greater than 255 are in the class.
289 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
290 such as \w are turned into Unicode properties), wide characters were not
291 correctly handled, and could fail to match.
292
293
294 Version 8.37 28-April-2015
295 --------------------------
296
297 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
298 for those parentheses to be closed with whatever has been captured so far.
299 However, it was failing to mark any other groups between the hightest
300 capture so far and the currrent group as "unset". Thus, the ovector for
301 those groups contained whatever was previously there. An example is the
302 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
303
304 2. If an assertion condition was quantified with a minimum of zero (an odd
305 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
306
307 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
308 unrecognized modifier, a crash could occur.
309
310 4. An attempt to do global matching in pcretest with a zero-length ovector
311 caused a crash.
312
313 5. Fixed a memory leak during matching that could occur for a subpattern
314 subroutine call (recursive or otherwise) if the number of captured groups
315 that had to be saved was greater than ten.
316
317 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
318 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
319 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
320
321 7. A UTF pattern containing a "not" match of a non-ASCII character and a
322 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
323
324 8. When a pattern is compiled, it remembers the highest back reference so that
325 when matching, if the ovector is too small, extra memory can be obtained to
326 use instead. A conditional subpattern whose condition is a check on a
327 capture having happened, such as, for example in the pattern
328 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
329 setting the highest backreference number. This mattered only if pcre_exec()
330 was called with an ovector that was too small to hold the capture, and there
331 was no other kind of back reference (a situation which is probably quite
332 rare). The effect of the bug was that the condition was always treated as
333 FALSE when the capture could not be consulted, leading to a incorrect
334 behaviour by pcre_exec(). This bug has been fixed.
335
336 9. A reference to a duplicated named group (either a back reference or a test
337 for being set in a conditional) that occurred in a part of the pattern where
338 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
339 to be incorrectly calculated, leading to overwriting.
340
341 10. A mutually recursive set of back references such as (\2)(\1) caused a
342 segfault at study time (while trying to find the minimum matching length).
343 The infinite loop is now broken (with the minimum length unset, that is,
344 zero).
345
346 11. If an assertion that was used as a condition was quantified with a minimum
347 of zero, matching went wrong. In particular, if the whole group had
348 unlimited repetition and could match an empty string, a segfault was
349 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
350 assertions to be quantified, but not if they are being used as conditions,
351 so the above pattern is faulted by Perl. PCRE has now been changed so that
352 it also rejects such patterns.
353
354 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
355 failed to allow the zero-repeat case if pcre2_exec() was called with an
356 ovector too small to capture the group.
357
358 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
359 Red Hat Product Security:
360
361 (a) A crash if /K and /F were both set with the option to save the compiled
362 pattern.
363
364 (b) Another crash if the option to print captured substrings in a callout
365 was combined with setting a null ovector, for example \O\C+ as a subject
366 string.
367
368 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
369 forward reference repeated a large (but limited) number of times within a
370 repeated outer group that has a zero minimum quantifier, caused incorrect
371 code to be compiled, leading to the error "internal error:
372 previously-checked referenced subpattern not found" when an incorrect
373 memory address was read. This bug was reported as "heap overflow",
374 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
375 CVE-2015-2325.
376
377 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
378 call within a group that also contained a recursive back reference caused
379 incorrect code to be compiled. This bug was reported as "heap overflow",
380 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
381 number CVE-2015-2326.
382
383 24. Computing the size of the JIT read-only data in advance has been a source
384 of various issues, and new ones are still appear unfortunately. To fix
385 existing and future issues, size computation is eliminated from the code,
386 and replaced by on-demand memory allocation.
387
388 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
389 adjacent to the end of the range, and the range contained characters with
390 more than one other case, caused incorrect behaviour when compiled in UTF
391 mode. In that example, the range a-j was left out of the class.
392
393 26. Fix JIT compilation of conditional blocks, which assertion
394 is converted to (*FAIL). E.g: /(?(?!))/.
395
396 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
397 discovered by the LLVM fuzzer.
398
399 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
400 when this assertion was used as a condition, for example (?(?!)a|b). In
401 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
402 error about an unsupported item.
403
404 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
405 possessification code could take exponential time to complete. A recursion
406 depth limit of 1000 has been imposed to limit the resources used by this
407 optimization.
408
409 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
410 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
411 because \S ensures they are all in the class. The code for doing this was
412 interacting badly with the code for computing the amount of space needed to
413 compile the pattern, leading to a buffer overflow. This bug was discovered
414 by the LLVM fuzzer.
415
416 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
417 other kinds of group caused stack overflow at compile time. This bug was
418 discovered by the LLVM fuzzer.
419
420 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
421 between a subroutine call and its quantifier was incorrectly compiled,
422 leading to buffer overflow or other errors. This bug was discovered by the
423 LLVM fuzzer.
424
425 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
426 assertion after (?(. The code was failing to check the character after
427 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
428 was discovered by the LLVM fuzzer.
429
430 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
431 a fixed maximum following a group that contains a subroutine reference was
432 incorrectly compiled and could trigger buffer overflow. This bug was
433 discovered by the LLVM fuzzer.
434
435 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
436 caused a stack overflow instead of the diagnosis of a non-fixed length
437 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
438
439 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
440 (e.g. /(?<=\Ka)/) could make pcregrep loop.
441
442 37. There was a similar problem to 36 in pcretest for global matches.
443
444 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
445 and a subsequent item in the pattern caused a non-match, backtracking over
446 the repeated \X did not stop, but carried on past the start of the subject,
447 causing reference to random memory and/or a segfault. There were also some
448 other cases where backtracking after \C could crash. This set of bugs was
449 discovered by the LLVM fuzzer.
450
451 39. The function for finding the minimum length of a matching string could take
452 a very long time if mutual recursion was present many times in a pattern,
453 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
454 method has been implemented. This infelicity was discovered by the LLVM
455 fuzzer.
456
457 40. Static linking against the PCRE library using the pkg-config module was
458 failing on missing pthread symbols.
459
460
461 Version 8.36 26-September-2014
462 ------------------------------
463
464 1. Got rid of some compiler warnings in the C++ modules that were shown up by
465 -Wmissing-field-initializers and -Wunused-parameter.
466
467 2. The tests for quantifiers being too big (greater than 65535) were being
468 applied after reading the number, and stupidly assuming that integer
469 overflow would give a negative number. The tests are now applied as the
470 numbers are read.
471
472 3. Tidy code in pcre_exec.c where two branches that used to be different are
473 now the same.
474
475 4. The JIT compiler did not generate match limit checks for certain
476 bracketed expressions with quantifiers. This may lead to exponential
477 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
478 issue should be resolved now.
479
480 5. Fixed an issue, which occures when nested alternatives are optimized
481 with table jumps.
482
483 6. Inserted two casts and changed some ints to size_t in the light of some
484 reported 64-bit compiler warnings (Bugzilla 1477).
485
486 7. Fixed a bug concerned with zero-minimum possessive groups that could match
487 an empty string, which sometimes were behaving incorrectly in the
488 interpreter (though correctly in the JIT matcher). This pcretest input is
489 an example:
490
491 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
492 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
493
494 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
495 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
496 for an empty string was breaking the inner loop and carrying on at a lower
497 level, when possessive repeated groups should always return to a higher
498 level as they have no backtrack points in them. The empty string test now
499 occurs at the outer level.
500
501 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
502 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
503
504 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
505 doesn't).
506
507 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
508 to hold a pattern to be incorrectly computed (too small) when there were
509 named back references to duplicated names. This could cause "internal
510 error: code overflow" or "double free or corruption" or other memory
511 handling errors.
512
513 11. When named subpatterns had the same prefixes, back references could be
514 confused. For example, in this pattern:
515
516 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
517
518 the reference to 'Name' was incorrectly treated as a reference to a
519 duplicate name.
520
521 12. A pattern such as /^s?c/mi8 where the optional character has more than
522 one "other case" was incorrectly compiled such that it would only try to
523 match starting at "c".
524
525 13. When a pattern starting with \s was studied, VT was not included in the
526 list of possible starting characters; this should have been part of the
527 8.34/18 patch.
528
529 14. If a character class started [\Qx]... where x is any character, the class
530 was incorrectly terminated at the ].
531
532 15. If a pattern that started with a caseless match for a character with more
533 than one "other case" was studied, PCRE did not set up the starting code
534 unit bit map for the list of possible characters. Now it does. This is an
535 optimization improvement, not a bug fix.
536
537 16. The Unicode data tables have been updated to Unicode 7.0.0.
538
539 17. Fixed a number of memory leaks in pcregrep.
540
541 18. Avoid a compiler warning (from some compilers) for a function call with
542 a cast that removes "const" from an lvalue by using an intermediate
543 variable (to which the compiler does not object).
544
545 19. Incorrect code was compiled if a group that contained an internal recursive
546 back reference was optional (had quantifier with a minimum of zero). This
547 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
548 caused segmentation faults because of stack overflows at compile time.
549
550 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
551 group that is quantified with an indefinite repeat, caused a compile-time
552 loop which used up all the system stack and provoked a segmentation fault.
553 This was not the same bug as 19 above.
554
555 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
556 Patch by Mike Frysinger.
557
558
559 Version 8.35 04-April-2014
560 --------------------------
561
562 1. A new flag is set, when property checks are present in an XCLASS.
563 When this flag is not set, PCRE can perform certain optimizations
564 such as studying these XCLASS-es.
565
566 2. The auto-possessification of character sets were improved: a normal
567 and an extended character set can be compared now. Furthermore
568 the JIT compiler optimizes more character set checks.
569
570 3. Got rid of some compiler warnings for potentially uninitialized variables
571 that show up only when compiled with -O2.
572
573 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
574 of a match later then the end of the match. The pcretest program was not
575 handling the case sensibly - it was outputting from the start to the next
576 binary zero. It now reports this situation in a message, and outputs the
577 text from the end to the start.
578
579 5. Fast forward search is improved in JIT. Instead of the first three
580 characters, any three characters with fixed position can be searched.
581 Search order: first, last, middle.
582
583 6. Improve character range checks in JIT. Characters are read by an inprecise
584 function now, which returns with an unknown value if the character code is
585 above a certain threshold (e.g: 256). The only limitation is that the value
586 must be bigger than the threshold as well. This function is useful when
587 the characters above the threshold are handled in the same way.
588
589 7. The macros whose names start with RAWUCHAR are placeholders for a future
590 mode in which only the bottom 21 bits of 32-bit data items are used. To
591 make this more memorable for those maintaining the code, the names have
592 been changed to start with UCHAR21, and an extensive comment has been added
593 to their definition.
594
595 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
596 to the export list in Makefile.am (they were accidentally omitted from the
597 8.34 tarball).
598
599 9. The informational output from pcretest used the phrase "starting byte set"
600 which is inappropriate for the 16-bit and 32-bit libraries. As the output
601 for "first char" and "need char" really means "non-UTF-char", I've changed
602 "byte" to "char", and slightly reworded the output. The documentation about
603 these values has also been (I hope) clarified.
604
605 10. Another JIT related optimization: use table jumps for selecting the correct
606 backtracking path, when more than four alternatives are present inside a
607 bracket.
608
609 11. Empty match is not possible, when the minimum length is greater than zero,
610 and there is no \K in the pattern. JIT should avoid empty match checks in
611 such cases.
612
613 12. In a caseless character class with UCP support, when a character with more
614 than one alternative case was not the first character of a range, not all
615 the alternative cases were added to the class. For example, s and \x{17f}
616 are both alternative cases for S: the class [RST] was handled correctly,
617 but [R-T] was not.
618
619 13. The configure.ac file always checked for pthread support when JIT was
620 enabled. This is not used in Windows, so I have put this test inside a
621 check for the presence of windows.h (which was already tested for).
622
623 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
624 The algorithm provides a way to skip certain starting offsets, and usually
625 faster than linear prefix searches.
626
627 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
628 as for 'fr_FR' and 'french'. For some reason, however, it then used the
629 Windows-specific input and output files, which have 'french' screwed in.
630 So this could never have worked. One of the problems with locales is that
631 they aren't always the same. I have now updated RunTest so that it checks
632 the output of the locale test (test 3) against three different output
633 files, and it allows the test to pass if any one of them matches. With luck
634 this should make the test pass on some versions of Solaris where it was
635 failing. Because of the uncertainty, the script did not used to stop if
636 test 3 failed; it now does. If further versions of a French locale ever
637 come to light, they can now easily be added.
638
639 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
640 there was a message during ./configure, but it did not stop. This now
641 provokes an error. The invalid example in README has been corrected.
642 If a value less than the minimum is given, the minimum value has always
643 been used, but now a warning is given.
644
645 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
646 was a bug in the test system, which is now fixed. Also, the list of various
647 configurations that are tested for each release did not have one with both
648 16/32 bits and --enable-bar-anycrlf. It now does.
649
650 18. pcretest was missing "-C bsr" for displaying the \R default setting.
651
652 19. Little endian PowerPC systems are supported now by the JIT compiler.
653
654 20. The fast forward newline mechanism could enter to an infinite loop on
655 certain invalid UTF-8 input. Although we don't support these cases
656 this issue can be fixed by a performance optimization.
657
658 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
659 not take account if existing stack usage. There is now a new global
660 variable called pcre_stack_guard that can be set to point to an external
661 function to check stack availability. It is called at the start of
662 processing every parenthesized group.
663
664 22. A typo in the code meant that in ungreedy mode the max/min qualifier
665 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
666 not match "ab".
667
668 23. When UTF was disabled, the JIT program reported some incorrect compile
669 errors. These messages are silenced now.
670
671 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
672 compiler.
673
674 25. Change all the temporary files used in RunGrepTest to be different to those
675 used by RunTest so that the tests can be run simultaneously, for example by
676 "make -j check".
677
678
679 Version 8.34 15-December-2013
680 -----------------------------
681
682 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
683 executable memory. Patch inspired by Carsten Klein.
684
685 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
686 this macro is never tested and has no effect, because the work to support
687 coverage involves only compiling and linking options and special targets in
688 the Makefile. The comment in config.h implied that defining the macro would
689 enable coverage support, which is totally false. There was also support for
690 setting this macro in the CMake files (my fault, I just copied it from
691 configure). SUPPORT_GCOV has now been removed.
692
693 3. Make a small performance improvement in strlen16() and strlen32() in
694 pcretest.
695
696 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
697 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
698 these cases). There was also one in pcretest.c.
699
700 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
701
702 6. In UTF mode, the code for checking whether a group could match an empty
703 string (which is used for indefinitely repeated groups to allow for
704 breaking an infinite loop) was broken when the group contained a repeated
705 negated single-character class with a character that occupied more than one
706 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
707 UTF-8 mode). The effect was undefined: the group might or might not be
708 deemed as matching an empty string, or the program might have crashed.
709
710 7. The code for checking whether a group could match an empty string was not
711 recognizing that \h, \H, \v, \V, and \R must match a character.
712
713 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
714 an empty string. If it can, pcretest shows this in its information output.
715
716 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
717 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
718 matched by pcre_exec() without using JIT:
719
720 (a) If the rest of the pattern did not match after a maximal run of
721 grapheme clusters, the code for backing up to try with fewer of them
722 did not always back up over a full grapheme when characters that do not
723 have the modifier quality were involved, e.g. Hangul syllables.
724
725 (b) If the match point in a subject started with modifier character, and
726 there was no match, the code could incorrectly back up beyond the match
727 point, and potentially beyond the first character in the subject,
728 leading to a segfault or an incorrect match result.
729
730 10. A conditional group with an assertion condition could lead to PCRE
731 recording an incorrect first data item for a match if no other first data
732 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
733 first data item, and therefore matched "ca" after "c" instead of at the
734 start.
735
736 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
737 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
738
739 12. The source of pcregrep now includes z/OS-specific code so that it can be
740 compiled for z/OS as part of the special z/OS distribution.
741
742 13. Added the -T and -TM options to pcretest.
743
744 14. The code in pcre_compile.c for creating the table of named capturing groups
745 has been refactored. Instead of creating the table dynamically during the
746 actual compiling pass, the information is remembered during the pre-compile
747 pass (on the stack unless there are more than 20 named groups, in which
748 case malloc() is used) and the whole table is created before the actual
749 compile happens. This has simplified the code (it is now nearly 150 lines
750 shorter) and prepared the way for better handling of references to groups
751 with duplicate names.
752
753 15. A back reference to a named subpattern when there is more than one of the
754 same name now checks them in the order in which they appear in the pattern.
755 The first one that is set is used for the reference. Previously only the
756 first one was inspected. This change makes PCRE more compatible with Perl.
757
758 16. Unicode character properties were updated from Unicode 6.3.0.
759
760 17. The compile-time code for auto-possessification has been refactored, based
761 on a patch by Zoltan Herczeg. It now happens after instead of during
762 compilation. The code is cleaner, and more cases are handled. The option
763 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
764 options in pcretest are provided to set it. It can also be set by
765 (*NO_AUTO_POSSESS) at the start of a pattern.
766
767 18. The character VT has been added to the default ("C" locale) set of
768 characters that match \s and are generally treated as white space,
769 following this same change in Perl 5.18. There is now no difference between
770 "Perl space" and "POSIX space". Whether VT is treated as white space in
771 other locales depends on the locale.
772
773 19. The code for checking named groups as conditions, either for being set or
774 for being recursed, has been refactored (this is related to 14 and 15
775 above). Processing unduplicated named groups should now be as fast at
776 numerical groups, and processing duplicated groups should be faster than
777 before.
778
779 20. Two patches to the CMake build system, by Alexander Barkov:
780
781 (1) Replace the "source" command by "." in CMakeLists.txt because
782 "source" is a bash-ism.
783
784 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
785 without these the CMake build does not work on Solaris.
786
787 21. Perl has changed its handling of \8 and \9. If there is no previously
788 encountered capturing group of those numbers, they are treated as the
789 literal characters 8 and 9 instead of a binary zero followed by the
790 literals. PCRE now does the same.
791
792 22. Following Perl, added \o{} to specify codepoints in octal, making it
793 possible to specify values greater than 0777 and also making them
794 unambiguous.
795
796 23. Perl now gives an error for missing closing braces after \x{... instead of
797 treating the string as literal. PCRE now does the same.
798
799 24. RunTest used to grumble if an inappropriate test was selected explicitly,
800 but just skip it when running all tests. This make it awkward to run ranges
801 of tests when one of them was inappropriate. Now it just skips any
802 inappropriate tests, as it always did when running all tests.
803
804 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
805 character types such as \d or \w, too many callouts were inserted, and the
806 data that they returned was rubbish.
807
808 26. In UCP mode, \s was not matching two of the characters that Perl matches,
809 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
810 were matched by \h. The code has now been refactored so that the lists of
811 the horizontal and vertical whitespace characters used for \h and \v (which
812 are defined only in one place) are now also used for \s.
813
814 27. Add JIT support for the 64 bit TileGX architecture.
815 Patch by Jiong Wang (Tilera Corporation).
816
817 28. Possessive quantifiers for classes (both explicit and automatically
818 generated) now use special opcodes instead of wrapping in ONCE brackets.
819
820 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
821 (because it's meaningless), this was not happening when PCRE_CASELESS was
822 set. Not wrong, but inefficient.
823
824 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
825 properties for \w, \d, etc) is present in a test regex. Otherwise if the
826 test contains no characters greater than 255, Perl doesn't realise it
827 should be using Unicode semantics.
828
829 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
830 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
831 does in Unicode mode.
832
833 32. Added the "forbid" facility to pcretest so that putting tests into the
834 wrong test files can sometimes be quickly detected.
835
836 33. There is now a limit (default 250) on the depth of nesting of parentheses.
837 This limit is imposed to control the amount of system stack used at compile
838 time. It can be changed at build time by --with-parens-nest-limit=xxx or
839 the equivalent in CMake.
840
841 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
842 errors. Perl warns for these when in warning mode, but PCRE has no facility
843 for giving warnings.
844
845 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
846 However, this was not working for (?!) because it is optimized to (*FAIL),
847 for which PCRE does not allow quantifiers. The optimization is now disabled
848 when a quantifier follows (?!). I can't see any use for this, but it makes
849 things uniform.
850
851 36. Perl no longer allows group names to start with digits, so I have made this
852 change also in PCRE. It simplifies the code a bit.
853
854 37. In extended mode, Perl ignores spaces before a + that indicates a
855 possessive quantifier. PCRE allowed a space before the quantifier, but not
856 before the possessive +. It now does.
857
858 38. The use of \K (reset reported match start) within a repeated possessive
859 group such as (a\Kb)*+ was not working.
860
861 40. Document that the same character tables must be used at compile time and
862 run time, and that the facility to pass tables to pcre_exec() and
863 pcre_dfa_exec() is for use only with saved/restored patterns.
864
865 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
866 features for Builds with MSVC:
867
868 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
869 builds). The .rc files can be used to set FileDescription and many other
870 attributes.
871
872 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
873 This allows higher-level build scripts which want .pdb files to avoid
874 hard-coding the exact files needed."
875
876 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
877 mean "start of word" and "end of word", respectively, as a transition aid.
878
879 43. A minimizing repeat of a class containing codepoints greater than 255 in
880 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
881 compiled to use the heap for recursion.
882
883 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
884 is configured.
885
886
887 Version 8.33 28-May-2013
888 ------------------------
889
890 1. Added 'U' to some constants that are compared to unsigned integers, to
891 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
892 variables that are added to signed variables, to ensure the result is
893 signed and can be negated.
894
895 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
896 CMake config files.
897
898 3. Revise the creation of config.h.generic so that all boolean macros are
899 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
900 overriding via -D on the command line possible.
901
902 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
903 to unsigned int is reported to make a quite noticeable speed difference in
904 a specific Windows environment. Testing on Linux did also appear to show
905 some benefit (and it is clearly not harmful). Also fixed the definition of
906 Xop which should be unsigned.
907
908 5. Related to (4), changing the definition of the intermediate variable cc
909 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
910 improvements.
911
912 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
913 unnecessary spaces.
914
915 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
916 and later.
917
918 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
919
920 9. Optimizing fast_forward_start_bits in JIT.
921
922 10. Adding support for callouts in JIT, and fixing some issues revealed
923 during this work. Namely:
924
925 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
926
927 (b) Minimum length was not checked before the matching is started.
928
929 11. The value of capture_last that is passed to callouts was incorrect in some
930 cases when there was a capture on one path that was subsequently abandoned
931 after a backtrack. Also, the capture_last value is now reset after a
932 recursion, since all captures are also reset in this case.
933
934 12. The interpreter no longer returns the "too many substrings" error in the
935 case when an overflowing capture is in a branch that is subsequently
936 abandoned after a backtrack.
937
938 13. In the pathological case when an offset vector of size 2 is used, pcretest
939 now prints out the matched string after a yield of 0 or 1.
940
941 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
942 Only supported by the JIT compiler at the moment.
943
944 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
945
946 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
947 offset of the starting point of the matching process, provided the offsets
948 vector is large enough.
949
950 17. The \A escape now records a lookbehind value of 1, though its execution
951 does not actually inspect the previous character. This is to ensure that,
952 in partial multi-segment matching, at least one character from the old
953 segment is retained when a new segment is processed. Otherwise, if there
954 are no lookbehinds in the pattern, \A might match incorrectly at the start
955 of a new segment.
956
957 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
958
959 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
960 gives some modest performance improvement in 8-bit mode.
961
962 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
963 be expressed in certain programming languages using Universal Character
964 Names.
965
966 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
967 which points out that "non characters" are not "characters that may not
968 appear in Unicode strings" but rather "characters that are reserved for
969 internal use and have only local meaning".
970
971 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
972 there was a conditional group that depended on an assertion, if the
973 assertion was false, the callout that immediately followed the alternation
974 in the condition was skipped when pcre_exec() was used for matching.
975
976 23. Allow an explicit callout to be inserted before an assertion that is the
977 condition for a conditional group, for compatibility with automatic
978 callouts, which always insert a callout at this point.
979
980 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
981 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
982
983 25. (*PRUNE) is now supported by the JIT compiler.
984
985 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
986
987 27. Fix the case where there are two or more SKIPs with arguments that may be
988 ignored.
989
990 28. (*SKIP) is now supported by the JIT compiler.
991
992 29. (*THEN) is now supported by the JIT compiler.
993
994 30. Update RunTest with additional test selector options.
995
996 31. The way PCRE handles backtracking verbs has been changed in two ways.
997
998 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
999 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1000 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1001 rather obscure rules do not always do the same thing.
1002
1003 (2) Previously, backtracking verbs were confined within assertions. This is
1004 no longer the case for positive assertions, except for (*ACCEPT). Again,
1005 this sometimes improves Perl compatibility, and sometimes does not.
1006
1007 32. A number of tests that were in test 2 because Perl did things differently
1008 have been moved to test 1, because either Perl or PCRE has changed, and
1009 these tests are now compatible.
1010
1011 32. Backtracking control verbs are now handled in the same way in JIT and
1012 interpreter.
1013
1014 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1015 contained a forward subroutine reference caused a compile error.
1016
1017 34. Auto-detect and optimize limited repetitions in JIT.
1018
1019 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1020 blocking (*UTF) etc.
1021
1022 36. In the interpreter, maximizing pattern repetitions for characters and
1023 character types now use tail recursion, which reduces stack usage.
1024
1025 37. The value of the max lookbehind was not correctly preserved if a compiled
1026 and saved regex was reloaded on a host of different endianness.
1027
1028 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1029 of the compiled pattern block, expand the flags field from 16 to 32 bits
1030 because it was almost full.
1031
1032 39. Try madvise first before posix_madvise.
1033
1034 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1035 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1036 original change locked out all matches of empty strings. This has been
1037 changed so that one match of an empty string per line is recognized.
1038 Subsequent searches on the same line (for colouring or for --only-matching,
1039 for example) do not recognize empty strings.
1040
1041 41. Applied a user patch to fix a number of spelling mistakes in comments.
1042
1043 42. Data lines longer than 65536 caused pcretest to crash.
1044
1045 43. Clarified the data type for length and startoffset arguments for pcre_exec
1046 and pcre_dfa_exec in the function-specific man pages, where they were
1047 explicitly stated to be in bytes, never having been updated. I also added
1048 some clarification to the pcreapi man page.
1049
1050 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1051 a segmentation fault.
1052
1053
1054 Version 8.32 30-November-2012
1055 -----------------------------
1056
1057 1. Improved JIT compiler optimizations for first character search and single
1058 character iterators.
1059
1060 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1061 Patch by Daniel Richard G.
1062
1063 3. Single character iterator optimizations in the JIT compiler.
1064
1065 4. Improved JIT compiler optimizations for character ranges.
1066
1067 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1068 Reported by Giuseppe D'Angelo.
1069
1070 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1071 of a line, was being set incorrectly in cases where .* appeared inside
1072 atomic brackets at the start of a pattern, or where there was a subsequent
1073 *PRUNE or *SKIP.
1074
1075 7. Improved instruction cache flush for POWER/PowerPC.
1076 Patch by Daniel Richard G.
1077
1078 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1079 grep:
1080
1081 (a) There is now no limit to the number of patterns to be matched.
1082
1083 (b) An error is given if a pattern is too long.
1084
1085 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1086 are now supported.
1087
1088 (d) --exclude-from and --include-from (multiple use) have been added.
1089
1090 (e) Exclusions and inclusions now apply to all files and directories, not
1091 just to those obtained from scanning a directory recursively.
1092
1093 (f) Multiple uses of -f and --file-list are now supported.
1094
1095 (g) In a Windows environment, the default for -d has been changed from
1096 "read" (the GNU grep default) to "skip", because otherwise the presence
1097 of a directory in the file list provokes an error.
1098
1099 (h) The documentation has been revised and clarified in places.
1100
1101 9. Improve the matching speed of capturing brackets.
1102
1103 10. Changed the meaning of \X so that it now matches a Unicode extended
1104 grapheme cluster.
1105
1106 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1107 out POSIX threads when JIT support is configured.
1108
1109 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1110
1111 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1112 structure could go wrong in environments where size_t is not the same size
1113 as int.
1114
1115 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1116 set.
1117
1118 15. The EBCDIC support had decayed; later updates to the code had included
1119 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1120 general tidy up of EBCDIC-related issues, and the documentation was also
1121 not quite right. There is now a test that can be run on ASCII systems to
1122 check some of the EBCDIC-related things (but is it not a full test).
1123
1124 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1125 in a small tidy to the code.
1126
1127 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1128
1129 18. If the --only-matching (-o) option in pcregrep is specified multiple
1130 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1131 substrings matched by the 1st and 2nd capturing parentheses. A separating
1132 string can be specified by --om-separator (default empty).
1133
1134 19. Improving the first n character searches.
1135
1136 20. Turn case lists for horizontal and vertical white space into macros so that
1137 they are defined only once.
1138
1139 21. This set of changes together give more compatible Unicode case-folding
1140 behaviour for characters that have more than one other case when UCP
1141 support is available.
1142
1143 (a) The Unicode property table now has offsets into a new table of sets of
1144 three or more characters that are case-equivalent. The MultiStage2.py
1145 script that generates these tables (the pcre_ucd.c file) now scans
1146 CaseFolding.txt instead of UnicodeData.txt for character case
1147 information.
1148
1149 (b) The code for adding characters or ranges of characters to a character
1150 class has been abstracted into a generalized function that also handles
1151 case-independence. In UTF-mode with UCP support, this uses the new data
1152 to handle characters with more than one other case.
1153
1154 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1155 whose other case is greater than 256 are now correctly matched
1156 caselessly. Previously, the high codepoint matched the low one, but not
1157 vice versa.
1158
1159 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1160 of the new class addition function, using character lists defined as
1161 macros alongside the case definitions of 20 above.
1162
1163 (e) Caseless back references now work with characters that have more than
1164 one other case.
1165
1166 (f) General caseless matching of characters with more than one other case
1167 is supported.
1168
1169 22. Unicode character properties were updated from Unicode 6.2.0
1170
1171 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1172
1173 24. Add support for 32-bit character strings, and UTF-32
1174
1175 25. Major JIT compiler update (code refactoring and bugfixing).
1176 Experimental Sparc 32 support is added.
1177
1178 26. Applied a modified version of Daniel Richard G's patch to create
1179 pcre.h.generic and config.h.generic by "make" instead of in the
1180 PrepareRelease script.
1181
1182 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1183 pcre_compile.c when checking for a zero character.
1184
1185 28. Introducing a native interface for JIT. Through this interface, the compiled
1186 machine code can be directly executed. The purpose of this interface is to
1187 provide fast pattern matching, so several sanity checks are not performed.
1188 However, feature tests are still performed. The new interface provides
1189 1.4x speedup compared to the old one.
1190
1191 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1192 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1193 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1194
1195 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1196 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1197 the "old" RFC 2279). Instead, it ended up passing a negative length to
1198 pcre_exec().
1199
1200 31. Add support for GCC's visibility feature to hide internal functions.
1201
1202 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1203 "unknown -C option" after outputting 0 or 1.
1204
1205 33. There is now support for generating a code coverage report for the test
1206 suite in environments where gcc is the compiler and lcov is installed. This
1207 is mainly for the benefit of the developers.
1208
1209 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1210 unaddressable using valgrind annotations, allowing valgrind to detect
1211 invalid memory accesses. This is mainly for the benefit of the developers.
1212
1213 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1214
1215 26. Give configure error if --enable-cpp but no C++ compiler found.
1216
1217
1218 Version 8.31 06-July-2012
1219 -------------------------
1220
1221 1. Fixing a wrong JIT test case and some compiler warnings.
1222
1223 2. Removed a bashism from the RunTest script.
1224
1225 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1226 to unsigned type, result still unsigned" that was given by an MS compiler
1227 on encountering the code "-sizeof(xxx)".
1228
1229 4. Partial matching support is added to the JIT compiler.
1230
1231 5. Fixed several bugs concerned with partial matching of items that consist
1232 of more than one character:
1233
1234 (a) /^(..)\1/ did not partially match "aba" because checking references was
1235 done on an "all or nothing" basis. This also applied to repeated
1236 references.
1237
1238 (b) \R did not give a hard partial match if \r was found at the end of the
1239 subject.
1240
1241 (c) \X did not give a hard partial match after matching one or more
1242 characters at the end of the subject.
1243
1244 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1245 a partial match for the string "\r".
1246
1247 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1248 a partial match for a CR character at the end of the subject string.
1249
1250 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1251 running pcretest, the text "(JIT)" added to the output whenever JIT is
1252 actually used to run the match.
1253
1254 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1255 or /S+[+] with a digit between 1 and 7.
1256
1257 8. OP_NOT now supports any UTF character not just single-byte ones.
1258
1259 9. (*MARK) control verb is now supported by the JIT compiler.
1260
1261 10. The command "./RunTest list" lists the available tests without actually
1262 running any of them. (Because I keep forgetting what they all are.)
1263
1264 11. Add PCRE_INFO_MAXLOOKBEHIND.
1265
1266 12. Applied a (slightly modified) user-supplied patch that improves performance
1267 when the heap is used for recursion (compiled with --disable-stack-for-
1268 recursion). Instead of malloc and free for each heap frame each time a
1269 logical recursion happens, frames are retained on a chain and re-used where
1270 possible. This sometimes gives as much as 30% improvement.
1271
1272 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1273 call.
1274
1275 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1276
1277 15. It is now possible to link pcretest with libedit as an alternative to
1278 libreadline.
1279
1280 16. (*COMMIT) control verb is now supported by the JIT compiler.
1281
1282 17. The Unicode data tables have been updated to Unicode 6.1.0.
1283
1284 18. Added --file-list option to pcregrep.
1285
1286 19. Added binary file support to pcregrep, including the -a, --binary-files,
1287 -I, and --text options.
1288
1289 20. The madvise function is renamed for posix_madvise for QNX compatibility
1290 reasons. Fixed by Giuseppe D'Angelo.
1291
1292 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1293
1294 22. Changed the option for creating symbolic links for 16-bit man pages from
1295 -s to -sf so that re-installing does not cause issues.
1296
1297 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1298
1299 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1300 matches in certain environments (the workspace was not being correctly
1301 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1302 some of the workspace data at the beginning of a restart.
1303
1304 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1305 was not doing so when it should - probably a typo introduced by SVN 528
1306 (change 8.10/14).
1307
1308 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1309 \w+ when the character tables indicated that \x{c4} was a word character.
1310 There were several related cases, all because the tests for doing a table
1311 lookup were testing for characters less than 127 instead of 255.
1312
1313 27. If a pattern contains capturing parentheses that are not used in a match,
1314 their slots in the ovector are set to -1. For those that are higher than
1315 any matched groups, this happens at the end of processing. In the case when
1316 there were back references that the ovector was too small to contain
1317 (causing temporary malloc'd memory to be used during matching), and the
1318 highest capturing number was not used, memory off the end of the ovector
1319 was incorrectly being set to -1. (It was using the size of the temporary
1320 memory instead of the true size.)
1321
1322 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1323 ovector size, it uses memory at the end of the block that it has got.
1324
1325 29. Check for an overlong MARK name and give an error at compile time. The
1326 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1327
1328 30. JIT compiler update.
1329
1330 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1331 Rill for the patch.
1332
1333 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1334
1335 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1336
1337 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1338 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1339 was enabled.
1340
1341 35. Improve JIT code generation for greedy plus quantifier.
1342
1343 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1344 1 to "aa" instead of to an empty string. The bug affected repeated groups
1345 that could potentially match an empty string.
1346
1347 37. Optimizing single character iterators in JIT.
1348
1349 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1350 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1351 codepoints that are too big for the mode are faulted, and in a UTF mode,
1352 disallowed codepoints are also faulted.
1353
1354 39. If PCRE was compiled with UTF support, in three places in the DFA
1355 matcher there was code that should only have been obeyed in UTF mode, but
1356 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1357 processing when bytes with values greater than 127 were present. In 16-bit
1358 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1359 both cases the values are those that cannot be the first data item in a UTF
1360 character. The three items that might have provoked this were recursions,
1361 possessively repeated groups, and atomic groups.
1362
1363 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1364 and pcregrep, because some OS require shared objects to be explicitly
1365 passed to ld, causing the link step to fail if they are not.
1366
1367 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1368 mode, patterns that started with \h* or \R* might be incorrectly matched.
1369
1370
1371 Version 8.30 04-February-2012
1372 -----------------------------
1373
1374 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1375 name is defined in ctype.h.
1376
1377 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1378 only in quite long subpatterns.
1379
1380 3. Removed the function pcre_info(), which has been obsolete and deprecated
1381 since it was replaced by pcre_fullinfo() in February 2000.
1382
1383 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1384 match a (*MARK), and the match failed at the start of the subject, a
1385 reference to memory before the start of the subject could occur. This bug
1386 was introduced by fix 17 of release 8.21.
1387
1388 5. A reference to an unset group with zero minimum repetition was giving
1389 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1390 /(another)?(\1?)test/ matched against "hello world test". This bug was
1391 introduced in release 8.13.
1392
1393 6. Add support for 16-bit character strings (a large amount of work involving
1394 many changes and refactorings).
1395
1396 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1397 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1398 from a file.
1399
1400 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1401 rounding is not applied in this particular case).
1402
1403 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1404 if they appear, or are escaped, in patterns.
1405
1406 10. Get rid of a number of -Wunused-but-set-variable warnings.
1407
1408 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1409 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1410 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1411 also returns the mark "x". This bug applied to capturing parentheses,
1412 non-capturing parentheses, and atomic parentheses. It also applied to some
1413 assertions.
1414
1415 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1416 information out of configure.ac instead of relying on pcre.h.generic, which
1417 is not stored in the repository.
1418
1419 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1420 -lreadline.
1421
1422 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1423
1424 15. Applied Graycode's patch to put the top-level frame on the stack rather
1425 than the heap when not using the stack for recursion. This gives a
1426 performance improvement in many cases when recursion is not deep.
1427
1428 16. Experimental code added to "pcretest -C" to output the stack frame size.
1429
1430
1431 Version 8.21 12-Dec-2011
1432 ------------------------
1433
1434 1. Updating the JIT compiler.
1435
1436 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1437 are added as well.
1438
1439 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1440 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1441 calling _pcre_jit_exec. Some extra comments are added.
1442
1443 4. (*MARK) settings inside atomic groups that do not contain any capturing
1444 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1445 was introduced by change 18 for 8.20.
1446
1447 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1448 ECMA-262 standard.
1449
1450 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1451 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1452 This bug was probably introduced by change 9 of 8.13.
1453
1454 7. While fixing 6 above, I noticed that a number of other items were being
1455 incorrectly rejected as "not fixed length". This arose partly because newer
1456 opcodes had not been added to the fixed-length checking code. I have (a)
1457 corrected the bug and added tests for these items, and (b) arranged for an
1458 error to occur if an unknown opcode is encountered while checking for fixed
1459 length instead of just assuming "not fixed length". The items that were
1460 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1461 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1462 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1463
1464 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1465 being incorrectly compiled and would have given unpredicatble results.
1466
1467 9. A possessively repeated subpattern with minimum repeat count greater than
1468 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1469 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1470 the first (A) could occur when it should not.
1471
1472 10. Add a cast and remove a redundant test from the code.
1473
1474 11. JIT should use pcre_malloc/pcre_free for allocation.
1475
1476 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1477 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1478 is anything other than /usr, -L is still shown).
1479
1480 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1481
1482 14. Perl does not support \N without a following name in a [] class; PCRE now
1483 also gives an error.
1484
1485 15. If a forward reference was repeated with an upper limit of around 2000,
1486 it caused the error "internal error: overran compiling workspace". The
1487 maximum number of forward references (including repeats) was limited by the
1488 internal workspace, and dependent on the LINK_SIZE. The code has been
1489 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1490 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1491 of around 200,000 forward references. While doing this, I also speeded up
1492 the filling in of repeated forward references.
1493
1494 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1495 incorrectly expecting the subject to contain another "a" after the start.
1496
1497 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1498 in the match, the SKIP should be ignored. This was not happening; instead
1499 the SKIP was being treated as NOMATCH. For patterns such as
1500 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1501 tested.
1502
1503 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1504 now much more compatible with Perl, in particular in cases where the result
1505 is a non-match for a non-anchored pattern. For example, if
1506 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1507 "m", where previously it did not return a name. A side effect of this
1508 change is that for partial matches, the last encountered mark name is
1509 returned, as for non matches. A number of tests that were previously not
1510 Perl-compatible have been moved into the Perl-compatible test files. The
1511 refactoring has had the pleasing side effect of removing one argument from
1512 the match() function, thus reducing its stack requirements.
1513
1514 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1515 subsequent uses of /S (without +) incorrectly behaved like /S+.
1516
1517 21. Retrieve executable code size support for the JIT compiler and fixing
1518 some warnings.
1519
1520 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1521 not work when the shorter character appeared right at the end of the
1522 subject string.
1523
1524 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1525 systems.
1526
1527 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1528 output it when the /M option is used in pcretest.
1529
1530 25. The CheckMan script was not being included in the distribution. Also, added
1531 an explicit "perl" to run Perl scripts from the PrepareRelease script
1532 because this is reportedly needed in Windows.
1533
1534 26. If study data was being save in a file and studying had not found a set of
1535 "starts with" bytes for the pattern, the data written to the file (though
1536 never used) was taken from uninitialized memory and so caused valgrind to
1537 complain.
1538
1539 27. Updated RunTest.bat as provided by Sheri Pierce.
1540
1541 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1542
1543 29. Computation of memory usage for the table of capturing group names was
1544 giving an unnecessarily large value.
1545
1546
1547 Version 8.20 21-Oct-2011
1548 ------------------------
1549
1550 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1551 a POSIX class. After further experiments with Perl, which convinced me that
1552 Perl has bugs and confusions, a closing square bracket is no longer allowed
1553 in a POSIX name. This bug also affected patterns with classes that started
1554 with full stops.
1555
1556 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1557 captured substring, but while checking the failing first alternative,
1558 substring 1 is temporarily captured. If the output vector supplied to
1559 pcre_exec() was not big enough for this capture, the yield of the function
1560 was still zero ("insufficient space for captured substrings"). This cannot
1561 be totally fixed without adding another stack variable, which seems a lot
1562 of expense for a edge case. However, I have improved the situation in cases
1563 such as /(a)(b)x|abc/ matched against "abc", where the return code
1564 indicates that fewer than the maximum number of slots in the ovector have
1565 been set.
1566
1567 3. Related to (2) above: when there are more back references in a pattern than
1568 slots in the output vector, pcre_exec() uses temporary memory during
1569 matching, and copies in the captures as far as possible afterwards. It was
1570 using the entire output vector, but this conflicts with the specification
1571 that only 2/3 is used for passing back captured substrings. Now it uses
1572 only the first 2/3, for compatibility. This is, of course, another edge
1573 case.
1574
1575 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1576 main code base, and can be used by building with --enable-jit. When this is
1577 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1578 runtime --no-jit option is given.
1579
1580 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1581 ovector, the return from the function was zero, implying that there were
1582 other matches that did not fit. The correct "exactly full" value is now
1583 returned.
1584
1585 6. If a subpattern that was called recursively or as a subroutine contained
1586 (*PRUNE) or any other control that caused it to give a non-standard return,
1587 invalid errors such as "Error -26 (nested recursion at the same subject
1588 position)" or even infinite loops could occur.
1589
1590 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1591 computing the minimum length on reaching *ACCEPT, and so ended up with the
1592 wrong value of 1 rather than 0. Further investigation indicates that
1593 computing a minimum subject length in the presence of *ACCEPT is difficult
1594 (think back references, subroutine calls), and so I have changed the code
1595 so that no minimum is registered for a pattern that contains *ACCEPT.
1596
1597 8. If (*THEN) was present in the first (true) branch of a conditional group,
1598 it was not handled as intended. [But see 16 below.]
1599
1600 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1601 Sheri Pierce.
1602
1603 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1604 the first byte in a match must be "a".
1605
1606 11. Change 17 for 8.13 increased the recursion depth for patterns like
1607 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1608 pattern contains any instances of (*THEN). If it does not, the old
1609 optimizations are restored. It would be nice to do this on a per-group
1610 basis, but at the moment that is not feasible.
1611
1612 12. In some environments, the output of pcretest -C is CRLF terminated. This
1613 broke RunTest's code that checks for the link size. A single white space
1614 character after the value is now allowed for.
1615
1616 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1617 For "fr", it uses the Windows-specific input and output files.
1618
1619 14. If (*THEN) appeared in a group that was called recursively or as a
1620 subroutine, it did not work as intended. [But see next item.]
1621
1622 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1623 pattern fragments (but not containing any | characters). If A and B are
1624 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1625 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1626 D. In other words, Perl considers parentheses that do not contain any |
1627 characters to be part of a surrounding alternative, whereas PCRE was
1628 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1629 differently. PCRE now behaves in the same way as Perl, except in the case
1630 of subroutine/recursion calls such as (?1) which have in any case always
1631 been different (but PCRE had them first :-).
1632
1633 16. Related to 15 above: Perl does not treat the | in a conditional group as
1634 creating alternatives. Such a group is treated in the same way as an
1635 ordinary group without any | characters when processing (*THEN). PCRE has
1636 been changed to match Perl's behaviour.
1637
1638 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1639 RunGrepTest script failed.
1640
1641 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1642 inevitable for groups that contain captures, but it can lead to a lot of
1643 stack use in large patterns. The old behaviour has been restored for atomic
1644 groups that do not contain any capturing parentheses.
1645
1646 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1647 suppress the check for a minimum subject length at run time. (If it was
1648 given to pcre_exec() or pcre_dfa_exec() it did work.)
1649
1650 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1651 fail to work when decoding hex characters in data strings in EBCDIC
1652 environments.
1653
1654 21. It appears that in at least one Mac OS environment, the isxdigit() function
1655 is implemented as a macro that evaluates to its argument more than once,
1656 contravening the C 90 Standard (I haven't checked a later standard). There
1657 was an instance in pcretest which caused it to go wrong when processing
1658 \x{...} escapes in subject strings. The has been rewritten to avoid using
1659 things like p++ in the argument of isxdigit().
1660
1661
1662 Version 8.13 16-Aug-2011
1663 ------------------------
1664
1665 1. The Unicode data tables have been updated to Unicode 6.0.0.
1666
1667 2. Two minor typos in pcre_internal.h have been fixed.
1668
1669 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1670 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1671 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1672
1673 4. There were a number of related bugs in the code for matching backrefences
1674 caselessly in UTF-8 mode when codes for the characters concerned were
1675 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1676 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1677 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1678 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1679 2-byte code at the end of the subject (it thought there wasn't enough data
1680 left).
1681
1682 5. Comprehensive information about what went wrong is now returned by
1683 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1684 as the output vector has at least 2 elements. The offset of the start of
1685 the failing character and a reason code are placed in the vector.
1686
1687 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1688 now returned is for the first byte of the failing character, instead of the
1689 last byte inspected. This is an incompatible change, but I hope it is small
1690 enough not to be a problem. It makes the returned offset consistent with
1691 pcre_exec() and pcre_dfa_exec().
1692
1693 7. pcretest now gives a text phrase as well as the error number when
1694 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1695 failure, the offset and reason code are output.
1696
1697 8. When \R was used with a maximizing quantifier it failed to skip backwards
1698 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1699 back over a single character (\n). This seems wrong (because it treated the
1700 two characters as a single entity when going forwards), conflicts with the
1701 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1702 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1703 has been changed.
1704
1705 9. Some internal refactoring has changed the processing so that the handling
1706 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1707 time (the PCRE_DOTALL option was changed this way some time ago: version
1708 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1709 which was always a bit of a fudge. It also means that there is one less
1710 argument for the match() function, which reduces its stack requirements
1711 slightly. This change also fixes an incompatibility with Perl: the pattern
1712 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1713
1714 10. More internal refactoring has drastically reduced the number of recursive
1715 calls to match() for possessively repeated groups such as (abc)++ when
1716 using pcre_exec().
1717
1718 11. While implementing 10, a number of bugs in the handling of groups were
1719 discovered and fixed:
1720
1721 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1722 (a|)*(?1) gave a compile-time internal error.
1723 ((a|)+)+ did not notice that the outer group could match an empty string.
1724 (^a|^)+ was not marked as anchored.
1725 (.*a|.*)+ was not marked as matching at start or after a newline.
1726
1727 12. Yet more internal refactoring has removed another argument from the match()
1728 function. Special calls to this function are now indicated by setting a
1729 value in a variable in the "match data" data block.
1730
1731 13. Be more explicit in pcre_study() instead of relying on "default" for
1732 opcodes that mean there is no starting character; this means that when new
1733 ones are added and accidentally left out of pcre_study(), testing should
1734 pick them up.
1735
1736 14. The -s option of pcretest has been documented for ages as being an old
1737 synonym of -m (show memory usage). I have changed it to mean "force study
1738 for every regex", that is, assume /S for every regex. This is similar to -i
1739 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1740 using it. It makes it easier to run collections of tests with and without
1741 study enabled, and thereby test pcre_study() more easily. All the standard
1742 tests are now run with and without -s (but some patterns can be marked as
1743 "never study" - see 20 below).
1744
1745 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1746 restoration of the capturing data to the outer values was not happening
1747 correctly.
1748
1749 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1750 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1751 pattern had matched an empty string, and so incorrectly returned a no
1752 match.
1753
1754 17. There was optimizing code for the last branch of non-capturing parentheses,
1755 and also for the obeyed branch of a conditional subexpression, which used
1756 tail recursion to cut down on stack usage. Unfortunately, now that there is
1757 the possibility of (*THEN) occurring in these branches, tail recursion is
1758 no longer possible because the return has to be checked for (*THEN). These
1759 two optimizations have therefore been removed. [But see 8.20/11 above.]
1760
1761 18. If a pattern containing \R was studied, it was assumed that \R always
1762 matched two bytes, thus causing the minimum subject length to be
1763 incorrectly computed because \R can also match just one byte.
1764
1765 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1766 was incorrectly computed.
1767
1768 20. If /S is present twice on a test pattern in pcretest input, it now
1769 *disables* studying, thereby overriding the use of -s on the command line
1770 (see 14 above). This is necessary for one or two tests to keep the output
1771 identical in both cases.
1772
1773 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1774 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1775
1776 22. When an atomic group that contained a capturing parenthesis was
1777 successfully matched, but the branch in which it appeared failed, the
1778 capturing was not being forgotten if a higher numbered group was later
1779 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1780 group 1 to "a", when in fact it should be unset. This applied to multi-
1781 branched capturing and non-capturing groups, repeated or not, and also to
1782 positive assertions (capturing in negative assertions does not happen
1783 in PCRE) and also to nested atomic groups.
1784
1785 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1786 subject after a captured substring, to make it easier to tell which of a
1787 number of identical substrings has been captured.
1788
1789 24. The way atomic groups are processed by pcre_exec() has been changed so that
1790 if they are repeated, backtracking one repetition now resets captured
1791 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1792 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1793 "aaa". Previously, it would have been "a". As part of this code
1794 refactoring, the way recursive calls are handled has also been changed.
1795
1796 25. If an assertion condition captured any substrings, they were not passed
1797 back unless some other capturing happened later. For example, if
1798 (?(?=(a))a) was matched against "a", no capturing was returned.
1799
1800 26. When studying a pattern that contained subroutine calls or assertions,
1801 the code for finding the minimum length of a possible match was handling
1802 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1803 group 1 called group 2 while simultaneously a separate group 2 called group
1804 1). A stack overflow occurred in this case. I have fixed this by limiting
1805 the recursion depth to 10.
1806
1807 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1808 Fortmann. This supports explicit test numbers on the command line, and has
1809 argument validation and error reporting.
1810
1811 28. An instance of \X with an unlimited repeat could fail if at any point the
1812 first character it looked at was a mark character.
1813
1814 29. Some minor code refactoring concerning Unicode properties and scripts
1815 should reduce the stack requirement of match() slightly.
1816
1817 30. Added the '=' option to pcretest to check the setting of unused capturing
1818 slots at the end of the pattern, which are documented as being -1, but are
1819 not included in the return count.
1820
1821 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1822 compiled something random. Now it gives a compile-time error (as does
1823 Perl).
1824
1825 32. A *MARK encountered during the processing of a positive assertion is now
1826 recorded and passed back (compatible with Perl).
1827
1828 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1829 had alternative anchored branches, the search for a second match in a line
1830 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1831 matched the line "0102" twice. The same bug affected patterns that started
1832 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1833 twice.
1834
1835 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1836 does, and because of capturing effects, quantifying parenthesized
1837 assertions may at times be useful. Quantifiers are now allowed for
1838 parenthesized assertions.
1839
1840 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1841
1842 36. \g was being checked for fancy things in a character class, when it should
1843 just be a literal "g".
1844
1845 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1846 appearance of a nested POSIX class supersedes an apparent external class.
1847 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1848 unescaped square brackets may also appear as part of class names. For
1849 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1850 more like Perl. (But see 8.20/1 above.)
1851
1852 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1853 was because it thought it was \N{name}, which is not supported).
1854
1855 39. Add minix to OS list not supporting the -S option in pcretest.
1856
1857 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1858 cannot analyze patterns in sufficient detail to catch mutual recursions
1859 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1860 subgroup is called recursively as a subpattern for a second time at the
1861 same position in the subject string. In previous releases this might have
1862 been caught by the recursion limit, or it might have run out of stack.
1863
1864 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1865 happen only once. PCRE was, however incorrectly giving a compile time error
1866 "recursive call could loop indefinitely" because it cannot analyze the
1867 pattern in sufficient detail. The compile time test no longer happens when
1868 PCRE is compiling a conditional subpattern, but actual runaway loops are
1869 now caught at runtime (see 40 above).
1870
1871 42. It seems that Perl allows any characters other than a closing parenthesis
1872 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1873 has been changed to be the same.
1874
1875 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1876 as not to get warnings when autogen.sh is called. Also changed
1877 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1878
1879 44. To help people who use pcregrep to scan files containing exceedingly long
1880 lines, the following changes have been made:
1881
1882 (a) The default value of the buffer size parameter has been increased from
1883 8K to 20K. (The actual buffer used is three times this size.)
1884
1885 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1886 PCRE is built.
1887
1888 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1889 to be set at run time.
1890
1891 (d) Numerical values in pcregrep options can be followed by K or M, for
1892 example --buffer-size=50K.
1893
1894 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1895 given and the return code is set to 2.
1896
1897 45. Add a pointer to the latest mark to the callout data block.
1898
1899 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1900 partial match of an empty string instead of no match. This was specific to
1901 the use of ".".
1902
1903 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1904 complete match instead of a partial match. This bug was dependent on both
1905 the PCRE_UTF8 and PCRE_DOTALL options being set.
1906
1907 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1908 starting byte set, because \b was not being ignored.
1909
1910
1911 Version 8.12 15-Jan-2011
1912 ------------------------
1913
1914 1. Fixed some typos in the markup of the man pages, and wrote a script that
1915 checks for such things as part of the documentation building process.
1916
1917 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1918 --match-limit and --recursion-limit options (added for 8.11). In
1919 particular, this made one of the standard tests fail. (The integer value
1920 went into the wrong half of a long int.)
1921
1922 3. If the --colour option was given to pcregrep with -v (invert match), it
1923 did strange things, either producing crazy output, or crashing. It should,
1924 of course, ignore a request for colour when reporting lines that do not
1925 match.
1926
1927 4. Another pcregrep bug caused similar problems if --colour was specified with
1928 -M (multiline) and the pattern match finished with a line ending.
1929
1930 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1931 matched in multiline mode, the following line was shown as part of the
1932 match. This seems wrong, so I have changed it.
1933
1934 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1935 the check for further matches in the same line (so they could be coloured)
1936 to overrun the end of the current line. If another match was found, it was
1937 incorrectly shown (and then shown again when found in the next line).
1938
1939 7. If pcregrep was compiled under Windows, there was a reference to the
1940 function pcregrep_exit() before it was defined. I am assuming this was
1941 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1942 reported by a user. I've moved the definition above the reference.
1943
1944
1945 Version 8.11 10-Dec-2010
1946 ------------------------
1947
1948 1. (*THEN) was not working properly if there were untried alternatives prior
1949 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1950 backtracked to try for "b" instead of moving to the next alternative branch
1951 at the same level (in this case, to look for "c"). The Perl documentation
1952 is clear that when (*THEN) is backtracked onto, it goes to the "next
1953 alternative in the innermost enclosing group".
1954
1955 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1956 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1957 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1958 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1959 (*THEN).
1960
1961 3. If \s appeared in a character class, it removed the VT character from
1962 the class, even if it had been included by some previous item, for example
1963 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1964 of \s, but is part of the POSIX "space" class.)
1965
1966 4. A partial match never returns an empty string (because you can always
1967 match an empty string at the end of the subject); however the checking for
1968 an empty string was starting at the "start of match" point. This has been
1969 changed to the "earliest inspected character" point, because the returned
1970 data for a partial match starts at this character. This means that, for
1971 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1972 (previously it gave "no match").
1973
1974 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1975 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1976 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1977 has an implication that the given string is incomplete (because a partial
1978 match is preferred over a full match). For this reason, these items now
1979 give a partial match in this situation. [Aside: previously, the one case
1980 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1981 match rather than a full match, which was wrong by the old rules, but is
1982 now correct.]
1983
1984 6. There was a bug in the handling of #-introduced comments, recognized when
1985 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1986 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1987 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1988 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1989 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1990 places in pcre_compile().
1991
1992 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1993 comments when looking ahead for named forward references to subpatterns,
1994 the only newline sequence it recognized was NL. It now handles newlines
1995 according to the set newline convention.
1996
1997 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1998 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1999 cater for a lack of strerror(). These oversights have been fixed.
2000
2001 9. Added --match-limit and --recursion-limit to pcregrep.
2002
2003 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2004
2005 11. When the -o option was used, pcregrep was setting a return code of 1, even
2006 when matches were found, and --line-buffered was not being honoured.
2007
2008 12. Added an optional parentheses number to the -o and --only-matching options
2009 of pcregrep.
2010
2011 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2012 can match an empty string. The code to do it in pcretest and pcredemo
2013 needed fixing:
2014
2015 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2016 only one byte after an empty string match just before CRLF (this case
2017 just got forgotten; "any" and "anycrlf" were OK).
2018
2019 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2020 mode when an empty string match preceded an ASCII character followed by
2021 a non-ASCII character. (The code for advancing by one character rather
2022 than one byte was nonsense.)
2023
2024 (c) The pcredemo.c sample program did not have any code at all to handle
2025 the cases when CRLF is a valid newline sequence.
2026
2027 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2028 as a starting offset was within the subject string. There is now a new
2029 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2030 negative or greater than the length of the string. In order to test this,
2031 pcretest is extended to allow the setting of negative starting offsets.
2032
2033 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2034 starting offset points to the beginning of a UTF-8 character was
2035 unnecessarily clumsy. I tidied it up.
2036
2037 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2038 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2039
2040 17. Nobody had reported that the --include_dir option, which was added in
2041 release 7.7 should have been called --include-dir (hyphen, not underscore)
2042 for compatibility with GNU grep. I have changed it to --include-dir, but
2043 left --include_dir as an undocumented synonym, and the same for
2044 --exclude-dir, though that is not available in GNU grep, at least as of
2045 release 2.5.4.
2046
2047 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2048 characters from a string of bytes) have been redefined so as not to use
2049 loops, in order to improve performance in some environments. At the same
2050 time, I abstracted some of the common code into auxiliary macros to save
2051 repetition (this should not affect the compiled code).
2052
2053 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2054 compile-time error is now given if \c is not followed by an ASCII
2055 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2056 different, and any byte value is allowed.)
2057
2058 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2059 START_OPTIMIZE option, which is now allowed at compile time - but just
2060 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2061 to pcregrep and other applications that have no direct access to PCRE
2062 options. The new /Y option in pcretest sets this option when calling
2063 pcre_compile().
2064
2065 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2066 back references. Groups containing recursive back references were forced to
2067 be atomic by that change, but in the case of named groups, the amount of
2068 memory required was incorrectly computed, leading to "Failed: internal
2069 error: code overflow". This has been fixed.
2070
2071 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2072 pcretest.c, to avoid build problems in some Borland environments.
2073
2074
2075 Version 8.10 25-Jun-2010
2076 ------------------------
2077
2078 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2079 THEN.
2080
2081 2. (*ACCEPT) was not working when inside an atomic group.
2082
2083 3. Inside a character class, \B is treated as a literal by default, but
2084 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2085 causes the error). The code is unchanged, but I tidied the documentation.
2086
2087 4. Inside a character class, PCRE always treated \R and \X as literals,
2088 whereas Perl faults them if its -w option is set. I have changed PCRE so
2089 that it faults them when PCRE_EXTRA is set.
2090
2091 5. Added support for \N, which always matches any character other than
2092 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2093
2094 6. When compiling pcregrep with newer versions of gcc which may have
2095 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2096 declared with attribute warn_unused_result" were given. Just casting the
2097 result to (void) does not stop the warnings; a more elaborate fudge is
2098 needed. I've used a macro to implement this.
2099
2100 7. Minor change to pcretest.c to avoid a compiler warning.
2101
2102 8. Added four artifical Unicode properties to help with an option to make
2103 \s etc use properties (see next item). The new properties are: Xan
2104 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2105
2106 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2107 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2108 this option. Modified pcretest to add /W to test this facility. Added
2109 REG_UCP to make it available via the POSIX interface.
2110
2111 10. Added --line-buffered to pcregrep.
2112
2113 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2114 studied, and the match started with a letter with a code point greater than
2115 127 whose first byte was different to the first byte of the other case of
2116 the letter, the other case of this starting letter was not recognized
2117 (#976).
2118
2119 12. If a pattern that was studied started with a repeated Unicode property
2120 test, for example, \p{Nd}+, there was the theoretical possibility of
2121 setting up an incorrect bitmap of starting bytes, but fortunately it could
2122 not have actually happened in practice until change 8 above was made (it
2123 added property types that matched character-matching opcodes).
2124
2125 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2126 possible starting bytes for non-anchored patterns.
2127
2128 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2129 \R, and also a number of cases that involve Unicode properties, both
2130 explicit and implicit when PCRE_UCP is set.
2131
2132 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2133 input, it could crash or give wrong results if characters with values
2134 greater than 0xc0 were present in the subject string. (Detail: it assumed
2135 UTF-8 input when processing these items.)
2136
2137 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2138 size_t is 64-bit (#991).
2139
2140 17. Added a check for running out of memory when PCRE is compiled with
2141 --disable-stack-for-recursion (#990).
2142
2143 18. If the last data line in a file for pcretest does not have a newline on
2144 the end, a newline was missing in the output.
2145
2146 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2147 less than 128) in its various bitmaps. However, there is a facility for
2148 generating tables according to the current locale when PCRE is compiled. It
2149 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2150 characters, are recognized by isspace() and therefore were getting set in
2151 these tables, and indeed these tables seem to approximate to ISO 8859. This
2152 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2153 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2154 which of course cannot start UTF-8 characters. I have changed the code so
2155 that only real ASCII characters (less than 128) and the correct starting
2156 bytes for UTF-8 encodings are set for characters greater than 127 when in
2157 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2158 altogether.)
2159
2160 20. Added the /T option to pcretest so as to be able to run tests with non-
2161 standard character tables, thus making it possible to include the tests
2162 used for 19 above in the standard set of tests.
2163
2164 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2165 reference to a subpattern the other side of a comment that contains an
2166 opening parenthesis caused either an internal compiling error, or a
2167 reference to the wrong subpattern.
2168
2169
2170 Version 8.02 19-Mar-2010
2171 ------------------------
2172
2173 1. The Unicode data tables have been updated to Unicode 5.2.0.
2174
2175 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2176 configured.
2177
2178 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2179 original author of that file, following a query about its status.
2180
2181 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2182 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2183
2184 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2185 quantifier applied to a forward-referencing subroutine call, could compile
2186 incorrect code or give the error "internal error: previously-checked
2187 referenced subpattern not found".
2188
2189 6. Both MS Visual Studio and Symbian OS have problems with initializing
2190 variables to point to external functions. For these systems, therefore,
2191 pcre_malloc etc. are now initialized to local functions that call the
2192 relevant global functions.
2193
2194 7. There were two entries missing in the vectors called coptable and poptable
2195 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2196 I've fixed the data, and added a kludgy way of testing at compile time that
2197 the lengths are correct (equal to the number of opcodes).
2198
2199 8. Following on from 7, I added a similar kludge to check the length of the
2200 eint vector in pcreposix.c.
2201
2202 9. Error texts for pcre_compile() are held as one long string to avoid too
2203 much relocation at load time. To find a text, the string is searched,
2204 counting zeros. There was no check for running off the end of the string,
2205 which could happen if a new error number was added without updating the
2206 string.
2207
2208 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2209
2210 11. \K was not working if it appeared in an atomic group or in a group that
2211 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2212 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2213 the assertion is positive, but not if it is negative.
2214
2215 12. Change 11 fortuitously reduced the size of the stack frame used in the
2216 "match()" function of pcre_exec.c by one pointer. Forthcoming
2217 implementation of support for (*MARK) will need an extra pointer on the
2218 stack; I have reserved it now, so that the stack frame size does not
2219 decrease.
2220
2221 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2222 item in branch that calls a recursion is a subroutine call - as in the
2223 second branch in the above example - was incorrectly given the compile-
2224 time error "recursive call could loop indefinitely" because pcre_compile()
2225 was not correctly checking the subroutine for matching a non-empty string.
2226
2227 14. The checks for overrunning compiling workspace could trigger after an
2228 overrun had occurred. This is a "should never occur" error, but it can be
2229 triggered by pathological patterns such as hundreds of nested parentheses.
2230 The checks now trigger 100 bytes before the end of the workspace.
2231
2232 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2233
2234
2235 Version 8.01 19-Jan-2010
2236 ------------------------
2237
2238 1. If a pattern contained a conditional subpattern with only one branch (in
2239 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2240 computed the wrong minimum data length (which is of course zero for such
2241 subpatterns). This could cause incorrect "no match" results.
2242
2243 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2244 the pattern is reset in the first branch, pcre_compile() failed with
2245 "internal error: code overflow at offset...". This happened only when
2246 the reset was to the original external option setting. (An optimization
2247 abstracts leading options settings into an external setting, which was the
2248 cause of this.)
2249
2250 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2251 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2252 assertion pattern did not match (meaning that the assertion was true), it
2253 was incorrectly treated as false if the SKIP had been reached during the
2254 matching. This also applied to assertions used as conditions.
2255
2256 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2257 assertion subpattern, including such a pattern used as a condition,
2258 unpredictable results occurred, instead of the error return
2259 PCRE_ERROR_DFA_UITEM.
2260
2261 5. The C++ GlobalReplace function was not working like Perl for the special
2262 situation when an empty string is matched. It now does the fancy magic
2263 stuff that is necessary.
2264
2265 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2266 removed. (These were left over from very, very early versions of PCRE.)
2267
2268 7. Some cosmetic changes to the code to make life easier when compiling it
2269 as part of something else:
2270
2271 (a) Change DEBUG to PCRE_DEBUG.
2272
2273 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2274 called "current" as "current_branch", to prevent a collision with the
2275 Linux macro when compiled as a kernel module.
2276
2277 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2278 prevent a collision with the Linux macro when compiled as a kernel
2279 module.
2280
2281 8. In pcre_compile() there are some checks for integer overflows that used to
2282 cast potentially large values to (double). This has been changed to that
2283 when building, a check for int64_t is made, and if it is found, it is used
2284 instead, thus avoiding the use of floating point arithmetic. (There is no
2285 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2286 double.
2287
2288 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2289 2005 (difference between two addresses compared to an unsigned value).
2290
2291 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2292 custom one, because of the following reported problem in Windows:
2293
2294 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2295 under Win32.
2296 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2297 therefore missing the function definition.
2298 - The compiler thus generates a "C" signature for the test function.
2299 - The linker fails to find the "C" function.
2300 - PCRE fails to configure if asked to do so against libbz2.
2301
2302 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2303 messages were output:
2304
2305 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2306 rerunning libtoolize, to keep the correct libtool macros in-tree.
2307 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2308
2309 I have done both of these things.
2310
2311 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2312 most of the time, it *can* run out if it is given a pattern that contains a
2313 runaway infinite recursion. I updated the discussion in the pcrestack man
2314 page.
2315
2316 13. Now that we have gone to the x.xx style of version numbers, the minor
2317 version may start with zero. Using 08 or 09 is a bad idea because users
2318 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2319 interpreted as invalid octal numbers. I've updated the previous comment in
2320 configure.ac, and also added a check that gives an error if 08 or 09 are
2321 used.
2322
2323 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2324 causing partial matching to fail when the end of the subject matched \W
2325 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2326
2327 15. There were some discrepancies between the declarations in pcre_internal.h
2328 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2329 their definitions. The declarations used "const uschar *" and the
2330 definitions used USPTR. Even though USPTR is normally defined as "const
2331 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2332 reported that: "This difference in casting confuses some C++ compilers, for
2333 example, SunCC recognizes above declarations as different functions and
2334 generates broken code for hbpcre." I have changed the declarations to use
2335 USPTR.
2336
2337 16. GNU libtool is named differently on some systems. The autogen.sh script now
2338 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2339 (FreeBSD).
2340
2341 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2342 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2343 comment: "Figure out how to create a longlong from a string: strtoll and
2344 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2345 instance, but it only takes 2 args instead of 3!"
2346
2347 18. A subtle bug concerned with back references has been fixed by a change of
2348 specification, with a corresponding code fix. A pattern such as
2349 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2350 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2351 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2352 same bug. Such groups have to be quantified to be useful, or contained
2353 inside another quantified group. (If there's no repetition, the reference
2354 can never match.) The problem arises because, having left the group and
2355 moved on to the rest of the pattern, a later failure that backtracks into
2356 the group uses the captured value from the final iteration of the group
2357 rather than the correct earlier one. I have fixed this in PCRE by forcing
2358 any group that contains a reference to itself to be an atomic group; that
2359 is, there cannot be any backtracking into it once it has completed. This is
2360 similar to recursive and subroutine calls.
2361
2362
2363 Version 8.00 19-Oct-09
2364 ----------------------
2365
2366 1. The table for translating pcre_compile() error codes into POSIX error codes
2367 was out-of-date, and there was no check on the pcre_compile() error code
2368 being within the table. This could lead to an OK return being given in
2369 error.
2370
2371 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2372 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2373 in a Windows environment.
2374
2375 3. The pcregrep --count option prints the count for each file even when it is
2376 zero, as does GNU grep. However, pcregrep was also printing all files when
2377 --files-with-matches was added. Now, when both options are given, it prints
2378 counts only for those files that have at least one match. (GNU grep just
2379 prints the file name in this circumstance, but including the count seems
2380 more useful - otherwise, why use --count?) Also ensured that the
2381 combination -clh just lists non-zero counts, with no names.
2382
2383 4. The long form of the pcregrep -F option was incorrectly implemented as
2384 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2385 but it seems right to fix it, and I didn't think it was worth preserving
2386 the old behaviour.
2387
2388 5. The command line items --regex=pattern and --regexp=pattern were not
2389 recognized by pcregrep, which required --regex pattern or --regexp pattern
2390 (with a space rather than an '='). The man page documented the '=' forms,
2391 which are compatible with GNU grep; these now work.
2392
2393 6. No libpcreposix.pc file was created for pkg-config; there was just
2394 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2395
2396 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2397 when UCP support is not needed, by modifying the Python script that
2398 generates it from Unicode data files. This should not matter if the module
2399 is correctly used as a library, but I received one complaint about 50K of
2400 unwanted data. My guess is that the person linked everything into his
2401 program rather than using a library. Anyway, it does no harm.
2402
2403 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2404 was a minimum greater than 1 for a wide character in a possessive
2405 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2406 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2407 character. Chaos in the form of incorrect output or a compiling loop could
2408 result.
2409
2410 9. The restrictions on what a pattern can contain when partial matching is
2411 requested for pcre_exec() have been removed. All patterns can now be
2412 partially matched by this function. In addition, if there are at least two
2413 slots in the offset vector, the offset of the earliest inspected character
2414 for the match and the offset of the end of the subject are set in them when
2415 PCRE_ERROR_PARTIAL is returned.
2416
2417 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2418 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2419 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2420 and may be more useful for multi-segment matching.
2421
2422 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2423 used to be given if ever the end of the subject was reached; now it is
2424 given only if matching could not proceed because another character was
2425 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2426 string "Z", which now yields "no match" instead of "partial match". In the
2427 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2428 final character ended with (*FAIL).
2429
2430 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2431 if the pattern had a "must contain" character that was already found in the
2432 earlier partial match, unless partial matching was again requested. For
2433 example, with the pattern /dog.(body)?/, the "must contain" character is
2434 "g". If the first part-match was for the string "dog", restarting with
2435 "sbody" failed. This bug has been fixed.
2436
2437 13. The string returned by pcre_dfa_exec() after a partial match has been
2438 changed so that it starts at the first inspected character rather than the
2439 first character of the match. This makes a difference only if the pattern
2440 starts with a lookbehind assertion or \b or \B (\K is not supported by
2441 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2442 matching functions compatible, and I think it's the right thing to do.
2443
2444 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2445 so that the demonstration program is easily available in environments where
2446 PCRE has not been installed from source.
2447
2448 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2449 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2450 library.
2451
2452 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2453 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2454 is not the first non-POSIX option to be added. Clearly some people find
2455 these options useful.
2456
2457 17. If a caller to the POSIX matching function regexec() passes a non-zero
2458 value for nmatch with a NULL value for pmatch, the value of
2459 nmatch is forced to zero.
2460
2461 18. RunGrepTest did not have a test for the availability of the -u option of
2462 the diff command, as RunTest does. It now checks in the same way as
2463 RunTest, and also checks for the -b option.
2464
2465 19. If an odd number of negated classes containing just a single character
2466 interposed, within parentheses, between a forward reference to a named
2467 subpattern and the definition of the subpattern, compilation crashed with
2468 an internal error, complaining that it could not find the referenced
2469 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2470 [The bug was that it was starting one character too far in when skipping
2471 over the character class, thus treating the ] as data rather than
2472 terminating the class. This meant it could skip too much.]
2473
2474 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2475 /g option in pcretest when the pattern contains \K, which makes it possible
2476 to have an empty string match not at the start, even when the pattern is
2477 anchored. Updated pcretest and pcredemo to use this option.
2478
2479 21. If the maximum number of capturing subpatterns in a recursion was greater
2480 than the maximum at the outer level, the higher number was returned, but
2481 with unset values at the outer level. The correct (outer level) value is
2482 now given.
2483
2484 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2485 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2486 make it do so. The string so far is captured, making this feature
2487 compatible with Perl.
2488
2489 23. The tests have been re-organized, adding tests 11 and 12, to make it
2490 possible to check the Perl 5.10 features against Perl 5.10.
2491
2492 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2493 pattern matches a fixed length string. PCRE did not allow this; now it
2494 does. Neither allows recursion.
2495
2496 25. I finally figured out how to implement a request to provide the minimum
2497 length of subject string that was needed in order to match a given pattern.
2498 (It was back references and recursion that I had previously got hung up
2499 on.) This code has now been added to pcre_study(); it finds a lower bound
2500 to the length of subject needed. It is not necessarily the greatest lower
2501 bound, but using it to avoid searching strings that are too short does give
2502 some useful speed-ups. The value is available to calling programs via
2503 pcre_fullinfo().
2504
2505 26. While implementing 25, I discovered to my embarrassment that pcretest had
2506 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2507 study optimizations had never been tested with that matching function.
2508 Oops. What is worse, even when it was passed study data, there was a bug in
2509 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2510 were also very few tests of studied patterns with pcre_dfa_exec().
2511
2512 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2513 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2514 on the other side of the coin, they are no longer allowed to have different
2515 names, because these cannot be distinguished in PCRE, and this has caused
2516 confusion. (This is a difference from Perl.)
2517
2518 28. When duplicate subpattern names are present (necessarily with different
2519 numbers, as required by 27 above), and a test is made by name in a
2520 conditional pattern, either for a subpattern having been matched, or for
2521 recursion in such a pattern, all the associated numbered subpatterns are
2522 tested, and the overall condition is true if the condition is true for any
2523 one of them. This is the way Perl works, and is also more like the way
2524 testing by number works.
2525
2526
2527 Version 7.9 11-Apr-09
2528 ---------------------
2529
2530 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2531 (pcretest), all targets were linked against these libraries. This included
2532 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2533 libraries. This caused unwanted dependencies to be created. This problem
2534 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2535 pcretest is linked with readline.
2536
2537 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2538 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2539 moved outside it again, because FALSE and TRUE are already defined in AIX,
2540 but BOOL is not.
2541
2542 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2543 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2544
2545 4. The pcregrep documentation said spaces were inserted as well as colons (or
2546 hyphens) following file names and line numbers when outputting matching
2547 lines. This is not true; no spaces are inserted. I have also clarified the
2548 wording for the --colour (or --color) option.
2549
2550 5. In pcregrep, when --colour was used with -o, the list of matching strings
2551 was not coloured; this is different to GNU grep, so I have changed it to be
2552 the same.
2553
2554 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2555 each matching line was coloured. Now it goes on to look for further matches
2556 of any of the test patterns, which is the same behaviour as GNU grep.
2557
2558 7. A pattern that could match an empty string could cause pcregrep to loop; it
2559 doesn't make sense to accept an empty string match in pcregrep, so I have
2560 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2561 seems to be how GNU grep behaves. [But see later change 40 for release
2562 8.33.]
2563
2564 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2565 start or after a newline", because the conditional assertion was not being
2566 correctly handled. The rule now is that both the assertion and what follows
2567 in the first alternative must satisfy the test.
2568
2569 9. If auto-callout was enabled in a pattern with a conditional group whose
2570 condition was an assertion, PCRE could crash during matching, both with
2571 pcre_exec() and pcre_dfa_exec().
2572
2573 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2574 used for matching.
2575
2576 11. Unicode property support in character classes was not working for
2577 characters (bytes) greater than 127 when not in UTF-8 mode.
2578
2579 12. Added the -M command line option to pcretest.
2580
2581 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2582
2583 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2584
2585 16. Added comments and documentation about mis-use of no_arg in the C++
2586 wrapper.
2587
2588 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2589 from Martin Jerabek that uses macro names for all relevant character and
2590 string constants.
2591
2592 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2593 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2594 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2595 these, but not everybody uses configure.
2596
2597 19. A conditional group that had only one branch was not being correctly
2598 recognized as an item that could match an empty string. This meant that an
2599 enclosing group might also not be so recognized, causing infinite looping
2600 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2601 with the subject "ab", where knowledge that the repeated group can match
2602 nothing is needed in order to break the loop.
2603
2604 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2605 exec(), but without supplying a callout function, matching went wrong.
2606
2607 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2608 leak if the size of the offset vector was greater than 30. When the vector
2609 is smaller, the saved offsets during recursion go onto a local stack
2610 vector, but for larger vectors malloc() is used. It was failing to free
2611 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2612 error, in fact).
2613
2614 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2615 heapframe that is used only when UTF-8 support is enabled. This caused no
2616 problem, but was untidy.
2617
2618 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2619 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2620 included within another project.
2621
2622 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2623 slightly modified by me:
2624
2625 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2626 not building pcregrep.
2627
2628 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2629 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2630
2631 25. Forward references, both numeric and by name, in patterns that made use of
2632 duplicate group numbers, could behave incorrectly or give incorrect errors,
2633 because when scanning forward to find the reference group, PCRE was not
2634 taking into account the duplicate group numbers. A pattern such as
2635 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2636
2637 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2638 the feature of a custom pointer more persuasive (as requested by a user).
2639
2640 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2641 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2642 pre-defined.
2643
2644 28. Added support for (*UTF8) at the start of a pattern.
2645
2646 29. Arrange for flags added by the "release type" setting in CMake to be shown
2647 in the configuration summary.
2648
2649
2650 Version 7.8 05-Sep-08
2651 ---------------------
2652
2653 1. Replaced UCP searching code with optimized version as implemented for Ad
2654 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2655 stage table and inline lookup instead of a function, giving speed ups of 2
2656 to 5 times on some simple patterns that I tested. Permission was given to
2657 distribute the MultiStage2.py script that generates the tables (it's not in
2658 the tarball, but is in the Subversion repository).
2659
2660 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2661 scripts.
2662
2663 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2664 a group with a zero qualifier. The result of the study could be incorrect,
2665 or the function might crash, depending on the pattern.
2666
2667 4. Caseless matching was not working for non-ASCII characters in back
2668 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2669 It now works when Unicode Property Support is available.
2670
2671 5. In pcretest, an escape such as \x{de} in the data was always generating
2672 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2673 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2674 truncation.
2675
2676 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2677
2678 7. Added two (int) casts to pcregrep when printing the difference of two
2679 pointers, in case they are 64-bit values.
2680
2681 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2682 test 2 if it fails.
2683
2684 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2685 and a #define of that name to empty if it is not externally set. This is to
2686 allow users of MSVC to set it if necessary.
2687
2688 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2689 the convenience functions in the pcre_get.c source file.
2690
2691 11. An option change at the start of a pattern that had top-level alternatives
2692 could cause overwriting and/or a crash. This command provoked a crash in
2693 some environments:
2694
2695 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2696
2697 This potential security problem was recorded as CVE-2008-2371.
2698
2699 12. For a pattern where the match had to start at the beginning or immediately
2700 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2701 pcre_dfa_exec() could read past the end of the passed subject if there was
2702 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2703 pcretest so that it places the subject at the end of its malloc-ed buffer.
2704
2705 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2706 exec() might read past the end of the data buffer in UTF-8 mode.
2707
2708 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2709 the data contained the byte 0x85 as part of a UTF-8 character within its
2710 first line. This applied both to normal and DFA matching.
2711
2712 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2713 /^[^d]*?$/8 failed to match "abc".
2714
2715 16. Added a missing copyright notice to pcrecpp_internal.h.
2716
2717 17. Make it more clear in the documentation that values returned from
2718 pcre_exec() in ovector are byte offsets, not character counts.
2719
2720 18. Tidied a few places to stop certain compilers from issuing warnings.
2721
2722 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2723 supplied by Stefan Weber. I made a further small update for 7.8 because
2724 there is a change of source arrangements: the pcre_searchfuncs.c module is
2725 replaced by pcre_ucd.c.
2726
2727
2728 Version 7.7 07-May-08
2729 ---------------------
2730
2731 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2732 a string to a long long, pretend we don't even have a long long." This is
2733 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2734
2735 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2736 pre-7.6 versions, which defined a global no_arg variable instead of putting
2737 it in the RE class. (See also #8 below.)
2738
2739 3. Remove a line of dead code, identified by coverity and reported by Nuno
2740 Lopes.
2741
2742 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2743
2744 (1) The include/exclude patterns were being applied to the whole pathnames
2745 of files, instead of just to the final components.
2746
2747 (2) If there was more than one level of directory, the subdirectories were
2748 skipped unless they satisfied the include/exclude conditions. This is
2749 inconsistent with GNU grep (and could even be seen as contrary to the
2750 pcregrep specification - which I improved to make it absolutely clear).
2751 The action now is always to scan all levels of directory, and just
2752 apply the include/exclude patterns to regular files.
2753
2754 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2755 --exclude_dir in the tests to avoid scanning .svn directories.
2756
2757 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2758 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2759 doesn't support NULs in patterns.
2760
2761 7. Added some missing "const"s to declarations of static tables in
2762 pcre_compile.c and pcre_dfa_exec.c.
2763
2764 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2765 caused by fix #2 above. (Subsequently also a second patch to fix the
2766 first patch. And a third patch - this was a messy problem.)
2767
2768 9. Applied Craig's patch to remove the use of push_back().
2769
2770 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2771 matching function regexec().
2772
2773 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2774 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2775 references. PCRE supports relative numbers with this syntax (I don't think
2776 Oniguruma does).
2777
2778 12. Previously, a group with a zero repeat such as (...){0} was completely
2779 omitted from the compiled regex. However, this means that if the group
2780 was called as a subroutine from elsewhere in the pattern, things went wrong
2781 (an internal error was given). Such groups are now left in the compiled
2782 pattern, with a new opcode that causes them to be skipped at execution
2783 time.
2784
2785 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2786 to the way PCRE behaves:
2787
2788 (a) A lone ] character is dis-allowed (Perl treats it as data).
2789
2790 (b) A back reference to an unmatched subpattern matches an empty string
2791 (Perl fails the current match path).
2792
2793 (c) A data ] in a character class must be notated as \] because if the
2794 first data character in a class is ], it defines an empty class. (In
2795 Perl it is not possible to have an empty class.) The empty class []
2796 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2797 The negative empty class [^] matches any one character, independently
2798 of the DOTALL setting.
2799
2800 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2801 non-existent subpattern following a character class starting with ']' and
2802 containing () gave an internal compiling error instead of "reference to
2803 non-existent subpattern". Fortunately, when the pattern did exist, the
2804 compiled code was correct. (When scanning forwards to check for the
2805 existence of the subpattern, it was treating the data ']' as terminating
2806 the class, so got the count wrong. When actually compiling, the reference
2807 was subsequently set up correctly.)
2808
2809 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2810 it was being rejected as not supported by pcre_dfa_exec(), even though
2811 other assertions are supported. I have made pcre_dfa_exec() support
2812 (*FAIL).
2813
2814 16. The implementation of 13c above involved the invention of a new opcode,
2815 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2816 cannot be changed at match time, I realized I could make a small
2817 improvement to matching performance by compiling OP_ALLANY instead of
2818 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2819 on the OP_ANY path.
2820
2821 17. Compiling pcretest on Windows with readline support failed without the
2822 following two fixes: (1) Make the unistd.h include conditional on
2823 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2824
2825 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2826 ncurses library to be included for pcretest when ReadLine support is
2827 requested, but also to allow for it to be overridden. This patch came from
2828 Daniel Bergström.
2829
2830 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2831 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2832 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2833 spotting this.
2834
2835
2836 Version 7.6 28-Jan-08
2837 ---------------------
2838
2839 1. A character class containing a very large number of characters with
2840 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2841 overflow.
2842
2843 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2844 HAVE_LONG_LONG is not defined.
2845
2846 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2847 bring them up to date and include new features. This patch includes:
2848
2849 - Fixed PH's badly added libz and libbz2 support.
2850 - Fixed a problem with static linking.
2851 - Added pcredemo. [But later removed - see 7 below.]
2852 - Fixed dftables problem and added an option.
2853 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2854 HAVE_LONG_LONG.
2855 - Added readline support for pcretest.
2856 - Added an listing of the option settings after cmake has run.
2857
2858 4. A user submitted a patch to Makefile that makes it easy to create
2859 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2860 Makefile.am that cause it to include this special target, without
2861 affecting anything else. Note that the same mingw target plus all
2862 the other distribution libraries and programs are now supported
2863 when configuring with CMake (see 6 below) instead of with
2864 Configure/Make.
2865
2866 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2867 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2868 exported in the Windows port". It has not yet been confirmed that the patch
2869 solves the problem, but it does no harm.
2870
2871 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2872 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2873 with CMake, and also correct the comment about stack recursion.
2874
2875 7. Remove the automatic building of pcredemo from the ./configure system and
2876 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2877 of a program that users should build themselves after PCRE is installed, so
2878 building it automatically is not really right. What is more, it gave
2879 trouble in some build environments.
2880
2881 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2882
2883
2884 Version 7.5 10-Jan-08
2885 ---------------------
2886
2887 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2888 values in parens when parsing an RE using the C++ wrapper."
2889
2890 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2891 Characters greater than 255 were excluded from the class instead of being
2892 included.
2893
2894 3. The same bug as (2) above applied to negated POSIX classes such as
2895 [:^space:].
2896
2897 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2898 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2899 I have changed it.
2900
2901 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2902 first named subpattern) and a construct such as (?&a) would reference the
2903 first named subpattern whose name started with "a" (in other words, the
2904 length check was missing). Both these problems are fixed. "Subpattern name
2905 expected" is now given for (?&) (a zero-length name), and this patch also
2906 makes it give the same error for \k'' (previously it complained that that
2907 was a reference to a non-existent subpattern).
2908
2909 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2910 this is right because (?- can be followed by option settings as well as by
2911 digits. I have, however, made the messages clearer.
2912
2913 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2914 than the number used in the conditional) now cause a compile-time error.
2915 This is actually not compatible with Perl, which accepts such patterns, but
2916 treats the conditional as always being FALSE (as PCRE used to), but it
2917 seems to me that giving a diagnostic is better.
2918
2919 8. Change "alphameric" to the more common word "alphanumeric" in comments
2920 and messages.
2921
2922 9. Fix two occurrences of "backslash" in comments that should have been
2923 "backspace".
2924
2925 10. Remove two redundant lines of code that can never be obeyed (their function
2926 was moved elsewhere).
2927
2928 11. The program that makes PCRE's Unicode character property table had a bug
2929 which caused it to generate incorrect table entries for sequences of
2930 characters that have the same character type, but are in different scripts.
2931 It amalgamated them into a single range, with the script of the first of
2932 them. In other words, some characters were in the wrong script. There were
2933 thirteen such cases, affecting characters in the following ranges:
2934
2935 U+002b0 - U+002c1
2936 U+0060c - U+0060d
2937 U+0061e - U+00612
2938 U+0064b - U+0065e
2939 U+0074d - U+0076d
2940 U+01800 - U+01805
2941 U+01d00 - U+01d77
2942 U+01d9b - U+01dbf
2943 U+0200b - U+0200f
2944 U+030fc - U+030fe
2945 U+03260 - U+0327f
2946 U+0fb46 - U+0fbb1
2947 U+10450 - U+1049d
2948
2949 12. The -o option (show only the matching part of a line) for pcregrep was not
2950 compatible with GNU grep in that, if there was more than one match in a
2951 line, it showed only the first of them. It now behaves in the same way as
2952 GNU grep.
2953
2954 13. If the -o and -v options were combined for pcregrep, it printed a blank
2955 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2956 does the same. The return code can be used to tell if there were any
2957 non-matching lines.
2958
2959 14. Added --file-offsets and --line-offsets to pcregrep.
2960
2961 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2962 infinitely looping recursion. The bug was that positive lookaheads were not
2963 being skipped when checking for a possible empty match (negative lookaheads
2964 and both kinds of lookbehind were skipped).
2965
2966 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2967 inclusion of <windows.h> to before rather than after the definition of
2968 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2969
2970 17. Specifying a possessive quantifier with a specific limit for a Unicode
2971 character property caused pcre_compile() to compile bad code, which led at
2972 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2973 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2974 caused the error; without that there was no problem.
2975
2976 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2977
2978 19. Added --enable-pcretest-libreadline.
2979
2980 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2981 RE::GlobalReplace(). As a result, the number of replacements returned was
2982 double what it should be. I removed one of the increments, but Craig sent a
2983 later patch that removed the other one (the right fix) and added unit tests
2984 that check the return values (which was not done before).
2985
2986 21. Several CMake things:
2987
2988 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2989 the names libpcre and libpcreposix, not just pcre and pcreposix.
2990
2991 (2) The above change means that pcretest and pcregrep are now correctly
2992 linked with the newly-built libraries, not previously installed ones.
2993
2994 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2995
2996 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2997 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2998 UTF-8 newline character). The key issue is that the pattern starts .*;
2999 this means that the match must be either at the beginning, or after a
3000 newline. The bug was in the code for advancing after a failed match and
3001 checking that the new position followed a newline. It was not taking
3002 account of UTF-8 characters correctly.
3003
3004 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3005 character classes. PCRE was not treating the sequence [:...:] as a
3006 character class unless the ... were all letters. Perl, however, seems to
3007 allow any characters between [: and :], though of course it rejects as
3008 unknown any "names" that contain non-letters, because all the known class
3009 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3010 for example, whereas PCRE did not - it did not recognize a POSIX character
3011 class. This seemed a bit dangerous, so the code has been changed to be
3012 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3013 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3014 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3015 Perl does, and where it didn't before.
3016
3017 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3018 Windows environments %n is disabled by default.
3019
3020
3021 Version 7.4 21-Sep-07
3022 ---------------------
3023
3024 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3025 means that a class such as [\s] counted as "explicit reference to CR or
3026 LF". That isn't really right - the whole point of the change was to try to
3027 help when there was an actual mention of one of the two characters. So now
3028 the change happens only if \r or \n (or a literal CR or LF) character is
3029 encountered.
3030
3031 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3032 of both had grown to the point where there were only 3 bits left.
3033 Fortunately, there was spare space in the data structure, and so I have
3034 moved the internal flags into a new 16-bit field to free up more option
3035 bits.
3036
3037 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3038 but did not set the internal JCHANGED flag - either of these is enough to
3039 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3040 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3041 start sets both bits.
3042
3043 4. Added options (at build time, compile time, exec time) to change \R from
3044 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3045
3046 5. doc/pcresyntax.html was missing from the distribution.
3047
3048 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3049 compatibility, even though it is no longer used.
3050
3051 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3052 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3053 windows.h file is present (where different names are used). [This was
3054 reversed later after testing - see 16 below.]
3055
3056 8. Changed all #include <config.h> to #include "config.h". There were also
3057 some further <pcre.h> cases that I changed to "pcre.h".
3058
3059 9. When pcregrep was used with the --colour option, it missed the line ending
3060 sequence off the lines that it output.
3061
3062 10. It was pointed out to me that arrays of string pointers cause lots of
3063 relocations when a shared library is dynamically loaded. A technique of
3064 using a single long string with a table of offsets can drastically reduce
3065 these. I have refactored PCRE in four places to do this. The result is
3066 dramatic:
3067
3068 Originally: 290
3069 After changing UCP table: 187
3070 After changing error message table: 43
3071 After changing table of "verbs" 36
3072 After changing table of Posix names 22
3073
3074 Thanks to the folks working on Gregex for glib for this insight.
3075
3076 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3077 unicode-properties was also set.
3078
3079 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3080
3081 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3082 checked only for CRLF.
3083
3084 14. Added casts to pcretest.c to avoid compiler warnings.
3085
3086 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3087
3088 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3089 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3090 entirely. This removes changes made in 7 above.
3091
3092 17. The CMake files have been updated, and there is now more information about
3093 building with CMake in the NON-UNIX-USE document.
3094
3095
3096 Version 7.3 28-Aug-07
3097 ---------------------
3098
3099 1. In the rejigging of the build system that eventually resulted in 7.1, the
3100 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3101 brackets there is not right, since it causes compilers to look for an
3102 installed pcre.h, not the version that is in the source that is being
3103 compiled (which of course may be different). I have changed it back to:
3104
3105 #include "pcre.h"
3106
3107 I have a vague recollection that the change was concerned with compiling in
3108 different directories, but in the new build system, that is taken care of
3109 by the VPATH setting the Makefile.
3110
3111 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3112 when the subject happened to end in the byte 0x85 (e.g. if the last
3113 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3114 characters but of course it shouldn't be taken as a newline when it is part
3115 of another character. The bug was that, for an unlimited repeat of . in
3116 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3117 characters when looking for a newline.
3118
3119 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3120
3121 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3122 in debug output.
3123
3124 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3125 long printing in the pcrecpp unittest when running under MinGW.
3126
3127 6. ESC_K was left out of the EBCDIC table.
3128
3129 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3130 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3131 limit also applies to "virtual nesting" when a pattern is recursive, and in
3132 this case 1000 isn't so big. I have been able to remove this limit at the
3133 expense of backing off one optimization in certain circumstances. Normally,
3134 when pcre_exec() would call its internal match() function recursively and
3135 immediately return the result unconditionally, it uses a "tail recursion"
3136 feature to save stack. However, when a subpattern that can match an empty
3137 string has an unlimited repetition quantifier, it no longer makes this
3138 optimization. That gives it a stack frame in which to save the data for
3139 checking that an empty string has been matched. Previously this was taken
3140 from the 1000-entry workspace that had been reserved. So now there is no
3141 explicit limit, but more stack is used.
3142
3143 8. Applied Daniel's patches to solve problems with the import/export magic
3144 syntax that is required for Windows, and which was going wrong for the
3145 pcreposix and pcrecpp parts of the library. These were overlooked when this
3146 problem was solved for the main library.
3147
3148 9. There were some crude static tests to avoid integer overflow when computing
3149 the size of patterns that contain repeated groups with explicit upper
3150 limits. As the maximum quantifier is 65535, the maximum group length was
3151 set at 30,000 so that the product of these two numbers did not overflow a
3152 32-bit integer. However, it turns out that people want to use groups that
3153 are longer than 30,000 bytes (though not repeat them that many times).
3154 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3155 made it possible to implement the integer overflow checks in a much more
3156 dynamic way, which I have now done. The artificial limitation on group
3157 length has been removed - we now have only the limit on the total length of
3158 the compiled pattern, which depends on the LINK_SIZE setting.
3159
3160 10. Fixed a bug in the documentation for get/copy named substring when
3161 duplicate names are permitted. If none of the named substrings are set, the
3162 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3163 empty string.
3164
3165 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3166 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3167 because the ] is interpreted as the first data character and the
3168 terminating ] is not found. PCRE has been made compatible with Perl in this
3169 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3170 cause memory overwriting.
3171
3172 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3173 string has been matched (to stop an infinite loop). It was not recognizing
3174 a conditional subpattern that could match an empty string if that
3175 subpattern was within another subpattern. For example, it looped when
3176 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3177 condition was not nested. This bug has been fixed.
3178
3179 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3180 past the start of the subject in the presence of bytes with the top bit
3181 set, for example "\x8aBCD".
3182
3183 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3184 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3185
3186 14. Optimized (?!) to (*FAIL).
3187
3188 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3189 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3190 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3191 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3192 does: it's just the validity check that is more restrictive.
3193
3194 16. Inserted checks for integer overflows during escape sequence (backslash)
3195 processing, and also fixed erroneous offset values for syntax errors during
3196 backslash processing.
3197
3198 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3199 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3200
3201 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3202 caused an overrun.
3203
3204 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3205 something other than just ASCII characters) inside a group that had an
3206 unlimited repeat caused a loop at compile time (while checking to see
3207 whether the group could match an empty string).
3208
3209 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3210 [\P{Any}] did so. (Error in the code for printing property names.)
3211
3212 21. An orphan \E inside a character class could cause a crash.
3213
3214 22. A repeated capturing bracket such as (A)? could cause a wild memory
3215 reference during compilation.
3216
3217 23. There are several functions in pcre_compile() that scan along a compiled
3218 expression for various reasons (e.g. to see if it's fixed length for look
3219 behind). There were bugs in these functions when a repeated \p or \P was
3220 present in the pattern. These operators have additional parameters compared
3221 with \d, etc, and these were not being taken into account when moving along
3222 the compiled data. Specifically:
3223
3224 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3225 length.
3226
3227 (b) An item such as \pL+ within a repeated group could cause crashes or
3228 loops.
3229
3230 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3231 "reference to non-existent subpattern" error.
3232
3233 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3234
3235 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3236 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3237
3238 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3239
3240 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3241 character were causing crashes (broken optimization).
3242
3243 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3244 \p or \P) caused a compile-time loop.
3245
3246 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3247 break. For example, the unstudied pattern [\r\n]A does not match the string
3248 "\r\nA" because change 7.0/46 below moves the current point on by two
3249 characters after failing to match at the start. However, the pattern \nA
3250 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3251 the same is true. There doesn't seem any very clean way out of this, but
3252 what I have chosen to do makes the common cases work: PCRE now takes note
3253 of whether there can be an explicit match for \r or \n anywhere in the
3254 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3255 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3256 pattern has explicit CR or LF references.
3257
3258 29. Added (*CR) etc for changing newline setting at start of pattern.
3259
3260
3261 Version 7.2 19-Jun-07
3262 ---------------------
3263
3264 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3265 which is apparently normally available under Windows.
3266
3267 2. Re-jig the pcregrep tests with different newline settings in an attempt
3268 to make them independent of the local environment's newline setting.
3269
3270 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3271
3272 4. Some of the "internals" tests were previously cut out when the link size
3273 was not 2, because the output contained actual offsets. The recent new
3274 "Z" feature of pcretest means that these can be cut out, making the tests
3275 usable with all link sizes.
3276
3277 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3278 stack recursion. This gives a massive performance boost under BSD, but just
3279 a small improvement under Linux. However, it saves one field in the frame
3280 in all cases.
3281
3282 6. Added more features from the forthcoming Perl 5.10:
3283
3284 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3285 recursion call. It refers to the nth most recently opened parentheses.
3286
3287 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3288 to be opened parentheses.
3289
3290 (c) Conditions that refer to capturing parentheses can be specified
3291 relatively, for example, (?(-2)... or (?(+3)...
3292
3293 (d) \K resets the start of the current match so that everything before
3294 is not part of it.
3295
3296 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3297
3298 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3299 reference syntax.
3300
3301 (g) (?| introduces a group in which the numbering of parentheses in each
3302 alternative starts with the same number.
3303
3304 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3305
3306 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3307 PCRE_INFO_JCHANGED.
3308
3309 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3310 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3311 for detecting groups that can match an empty string.
3312
3313 9. A pattern with a very large number of alternatives (more than several
3314 hundred) was running out of internal workspace during the pre-compile
3315 phase, where pcre_compile() figures out how much memory will be needed. A
3316 bit of new cunning has reduced the workspace needed for groups with
3317 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3318 workspace instead of running out of the 4096 that are available.
3319
3320 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3321
3322 11. Applied patch from Google to remove an optimization that didn't quite work.
3323 The report of the bug said:
3324
3325 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3326 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3327 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3328
3329 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3330 it matched the wrong number of bytes.
3331
3332
3333 Version 7.1 24-Apr-07
3334 ---------------------
3335
3336 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3337 that is more "standard", making use of automake and other Autotools. There
3338 is some re-arrangement of the files and adjustment of comments consequent
3339 on this.
3340
3341 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3342 for recursive directory scanning broke on some systems because the files
3343 are not scanned in any specific order and on different systems the order
3344 was different. A call to "sort" has been inserted into RunGrepTest for the
3345 approprate test as a short-term fix. In the longer term there may be an
3346 alternative.
3347
3348 3. I had an email from Eric Raymond about problems translating some of PCRE's
3349 man pages to HTML (despite the fact that I distribute HTML pages, some
3350 people do their own conversions for various reasons). The problems
3351 concerned the use of low-level troff macros .br and .in. I have therefore
3352 removed all such uses from the man pages (some were redundant, some could
3353 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3354 HTML has been updated to handle .nf/.fi and to complain if it encounters
3355 .br or .in.
3356
3357 4. Updated comments in configure.ac that get placed in config.h.in and also
3358 arranged for config.h to be included in the distribution, with the name
3359 config.h.generic, for the benefit of those who have to compile without
3360 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3361
3362 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3363 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3364 makevp.bat for the current PCRE, using the additional files
3365 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3366
3367 6. A Windows user reported a minor discrepancy with test 2, which turned out
3368 to be caused by a trailing space on an input line that had got lost in his
3369 copy. The trailing space was an accident, so I've just removed it.
3370
3371 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3372 that is needed.
3373
3374 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3375 as "const" (a) because they are and (b) because it helps the PHP
3376 maintainers who have recently made a script to detect big data structures
3377 in the php code that should be moved to the .rodata section. I remembered
3378 to update Builducptable as well, so it won't revert if ucptable.h is ever
3379 re-created.
3380
3381 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3382 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3383 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3384 support is not required. This saves 1.5-2K of code, which is important in
3385 some applications.
3386
3387 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3388 so as not to refer to the tables, even though these functions will never be
3389 called when UTF-8 support is disabled. Otherwise there are problems with a
3390 shared library.
3391
3392 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3393
3394 (a) It was defining its arguments as char * instead of void *.
3395
3396 (b) It was assuming that all moves were upwards in memory; this was true
3397 a long time ago when I wrote it, but is no longer the case.
3398
3399 The emulated memove() is provided for those environments that have neither
3400 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3401 is clearly not the case, as these two bugs were recently reported.
3402
3403 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3404 and Detrail to create the HTML documentation, the .txt form of the man
3405 pages, and it removes trailing spaces from listed files. It also creates
3406 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3407 case, it wraps all the #defines with #ifndefs. This script should be run
3408 before "make dist".
3409
3410 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3411 with Unicode property support.
3412
3413 (a) For a maximizing quantifier, if the two different cases of the
3414 character were of different lengths in their UTF-8 codings (there are
3415 some cases like this - I found 11), and the matching function had to
3416 back up over a mixture of the two cases, it incorrectly assumed they
3417 were both the same length.
3418
3419 (b) When PCRE was configured to use the heap rather than the stack for
3420 recursion during matching, it was not correctly preserving the data for
3421 the other case of a UTF-8 character when checking ahead for a match
3422 while processing a minimizing repeat. If the check also involved
3423 matching a wide character, but failed, corruption could cause an
3424 erroneous result when trying to check for a repeat of the original
3425 character.
3426
3427 13. Some tidying changes to the testing mechanism:
3428
3429 (a) The RunTest script now detects the internal link size and whether there
3430 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3431 values substituted by "configure". (The RunGrepTest script already did
3432 this for UTF-8.) The configure.ac script no longer substitutes the
3433 relevant variables.
3434
3435 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3436 with length and offset values. This means that the output is different
3437 for different internal link sizes. Test 2 is skipped for link sizes
3438 other than 2 because of this, bypassing the problem. Unfortunately,
3439 there was also a test in test 3 (the locale tests) that used /B and
3440 failed for link sizes other than 2. Rather than cut the whole test out,
3441 I have added a new /Z option to pcretest that replaces the length and
3442 offset values with spaces. This is now used to make test 3 independent
3443 of link size. (Test 2 will be tidied up later.)
3444
3445 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3446 segmentation fault instead of returning the appropriate error message.
3447
3448 15. In multiline mode when the newline sequence was set to "any", the pattern
3449 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3450 This doesn't seem right; it now treats the CRLF combination as the line
3451 ending, and so does not match in that case. It's only a pattern such as ^$
3452 that would hit this one: something like ^ABC$ would have failed after \r
3453 and then tried again after \r\n.
3454
3455 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3456 in an attempt to make files that differ only in their line terminators
3457 compare equal. This works on Linux.
3458
3459 17. Under certain error circumstances pcregrep might try to free random memory
3460 as it exited. This is now fixed, thanks to valgrind.
3461
3462 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3463 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3464 was because its rules for how to advance for /g after matching an empty
3465 string at the end of a line did not allow for this case. They now check for
3466 it specially.
3467
3468 20. pcretest is supposed to handle patterns and data of any length, by
3469 extending its buffers when necessary. It was getting this wrong when the
3470 buffer for a data line had to be extended.
3471
3472 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3473 CRLF as a newline sequence.
3474
3475 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3476 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3477 I have nevertheless tidied it up.
3478
3479 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3480
3481 24. Added a man page for pcre-config.
3482
3483
3484 Version 7.0 19-Dec-06
3485 ---------------------
3486
3487 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3488 moving to gcc 4.1.1.
3489
3490 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3491 sys/time.h, which is documented as needed for this function. It doesn't
3492 seem to matter on Linux, but it showed up on some releases of OS X.
3493
3494 3. It seems that there are systems where bytes whose values are greater than
3495 127 match isprint() in the "C" locale. The "C" locale should be the
3496 default when a C program starts up. In most systems, only ASCII printing
3497 characters match isprint(). This difference caused the output from pcretest
3498 to vary, making some of the tests fail. I have changed pcretest so that:
3499
3500 (a) When it is outputting text in the compiled version of a pattern, bytes
3501 other than 32-126 are always shown as hex escapes.
3502
3503 (b) When it is outputting text that is a matched part of a subject string,
3504 it does the same, unless a different locale has been set for the match
3505 (using the /L modifier). In this case, it uses isprint() to decide.
3506
3507 4. Fixed a major bug that caused incorrect computation of the amount of memory
3508 required for a compiled pattern when options that changed within the
3509 pattern affected the logic of the preliminary scan that determines the
3510 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3511 that the computed length was too small. The symptoms of this bug were
3512 either the PCRE error "internal error: code overflow" from pcre_compile(),
3513 or a glibc crash with a message such as "pcretest: free(): invalid next
3514 size (fast)". Examples of patterns that provoked this bug (shown in
3515 pcretest format) are:
3516
3517 /(?-x: )/x
3518 /(?x)(?-x: \s*#\s*)/
3519 /((?i)[\x{c0}])/8
3520 /(?i:[\x{c0}])/8
3521
3522 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3523 is now done differently.
3524
3525 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3526 wrapper classes; (b) implement a new function in the C++ scanner that is
3527 more efficient than the old way of doing things because it avoids levels of
3528 recursion in the regex matching; (c) add a paragraph to the documentation
3529 for the FullMatch() function.
3530
3531 6. The escape sequence \n was being treated as whatever was defined as
3532 "newline". Not only was this contrary to the documentation, which states
3533 that \n is character 10 (hex 0A), but it also went horribly wrong when
3534 "newline" was defined as CRLF. This has been fixed.
3535
3536 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3537 was being set to -1 for the "end of line" case (supposedly a value that no
3538 character can have). Though this value is never used (the check for end of
3539 line is "zero bytes in current character"), it caused compiler complaints.
3540 I've changed it to 0xffffffff.
3541
3542 8. In pcre_version.c, the version string was being built by a sequence of
3543 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3544 string (as it is for production releases) called a macro with an empty
3545 argument. The C standard says the result of this is undefined. The gcc
3546 compiler treats it as an empty string (which was what was wanted) but it is
3547 reported that Visual C gives an error. The source has been hacked around to
3548 avoid this problem.
3549
3550 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3551 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3552 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3553 of them did).
3554
3555 10. Originally, pcretest opened its input and output without "b"; then I was
3556 told that "b" was needed in some environments, so it was added for release
3557 5.0 to both the input and output. (It makes no difference on Unix-like
3558 systems.) Later I was told that it is wrong for the input on Windows. I've
3559 now abstracted the modes into two macros, to make it easier to fiddle with
3560 them, and removed "b" from the input mode under Windows.
3561
3562 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3563
3564 12. Added -help and --help to pcretest as an official way of being reminded
3565 of the options.
3566
3567 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3568 and pcrecpp.cc because they annoy compilers at high warning levels.
3569
3570 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3571
3572 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3573 scripts are not C programs :-) and which was not noticed because it works
3574 on Linux.
3575
3576 16. pcretest is supposed to handle any length of pattern and data line (as one
3577 line or as a continued sequence of lines) by extending its input buffer if
3578 necessary. This feature was broken for very long pattern lines, leading to
3579 a string of junk being passed to pcre_compile() if the pattern was longer
3580 than about 50K.
3581
3582 17. I have done a major re-factoring of the way pcre_compile() computes the
3583 amount of memory needed for a compiled pattern. Previously, there was code
3584 that made a preliminary scan of the pattern in order to do this. That was
3585 OK when PCRE was new, but as the facilities have expanded, it has become
3586 harder and harder to keep it in step with the real compile phase, and there
3587 have been a number of bugs (see for example, 4 above). I have now found a
3588 cunning way of running the real compile function in a "fake" mode that
3589 enables it to compute how much memory it would need, while actually only
3590 ever using a few hundred bytes of working memory and without too many
3591 tests of the mode. This should make future maintenance and development
3592 easier. A side effect of this work is that the limit of 200 on the nesting
3593 depth of parentheses has been removed (though this was never a serious
3594 limitation, I suspect). However, there is a downside: pcre_compile() now
3595 runs more slowly than before (30% or more, depending on the pattern). I
3596 hope this isn't a big issue. There is no effect on runtime performance.
3597
3598 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3599 newline (only possible for the last line of a file) and it was a
3600 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3601
3602 19. Added additional timing features to pcretest. (1) The -tm option now times
3603 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3604 separate command line item, by a number that specifies the number of
3605 repeats to use when timing. The default is 50000; this gives better
3606 precision, but takes uncomfortably long for very large patterns.
3607
3608 20. Extended pcre_study() to be more clever in cases where a branch of a
3609 subpattern has no definite first character. For example, (a*|b*)[cd] would
3610 previously give no result from pcre_study(). Now it recognizes that the
3611 first character must be a, b, c, or d.
3612
3613 21. There was an incorrect error "recursive call could loop indefinitely" if
3614 a subpattern (or the entire pattern) that was being tested for matching an
3615 empty string contained only one non-empty item after a nested subpattern.
3616 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3617 incorrectly, because the \d was being skipped in the check.
3618
3619 22. The pcretest program now has a new pattern option /B and a command line
3620 option -b, which is equivalent to adding /B to every pattern. This causes
3621 it to show the compiled bytecode, without the additional information that
3622 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3623 is the same as /B/I).
3624
3625 23. A new optimization is now able automatically to treat some sequences such
3626 as a*b as a*+b. More specifically, if something simple (such as a character
3627 or a simple class like \d) has an unlimited quantifier, and is followed by
3628 something that cannot possibly match the quantified thing, the quantifier
3629 is automatically "possessified".
3630
3631 24. A recursive reference to a subpattern whose number was greater than 39
3632 went wrong under certain circumstances in UTF-8 mode. This bug could also
3633 have affected the operation of pcre_study().
3634
3635 25. Realized that a little bit of performance could be had by replacing
3636 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3637
3638 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3639
3640 27. Possessive quantifiers such as a++ were previously implemented by turning
3641 them into atomic groups such as ($>a+). Now they have their own opcodes,
3642 which improves performance. This includes the automatically created ones
3643 from 23 above.
3644
3645 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3646 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3647 the first matched character to be a colon. This applied both to named and
3648 numbered groups.
3649
3650 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3651
3652 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3653 building PCRE on Windows easier, so I have included it in the distribution.
3654
3655 31. There is now a check in pcretest against a ridiculously large number being
3656 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3657 loop, the loop is abandoned.
3658
3659 32. Forward references to subpatterns in conditions such as (?(2)...) where
3660 subpattern 2 is defined later cause pcre_compile() to search forwards in
3661 the pattern for the relevant set of parentheses. This search went wrong
3662 when there were unescaped parentheses in a character class, parentheses
3663 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3664
3665 33. "Subroutine" calls and backreferences were previously restricted to
3666 referencing subpatterns earlier in the regex. This restriction has now
3667 been removed.
3668
3669 34. Added a number of extra features that are going to be in Perl 5.10. On the
3670 whole, these are just syntactic alternatives for features that PCRE had
3671 previously implemented using the Python syntax or my own invention. The
3672 other formats are all retained for compatibility.
3673
3674 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3675 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3676 also .NET compatible.
3677
3678 (b) A recursion or subroutine call to a named group can now be defined as
3679 (?&name) as well as (?P>name).
3680
3681 (c) A backreference to a named group can now be defined as \k<name> or
3682 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3683 5.10, are also .NET compatible.
3684
3685 (d) A conditional reference to a named group can now use the syntax
3686 (?(<name>) or (?('name') as well as (?(name).
3687
3688 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3689 groups (named and numbered) that are never evaluated inline, but can be
3690 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3691 is always false. There may be only one alternative in such a group.
3692
3693 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3694 as the simple (?(R). The condition is true only if the most recent
3695 recursion is that of the given number or name. It does not search out
3696 through the entire recursion stack.
3697
3698 (g) The escape \gN or \g{N} has been added, where N is a positive or
3699 negative number, specifying an absolute or relative reference.
3700
3701 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3702 some "unreachable code" warnings.
3703
3704 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3705 things, this adds five new scripts.
3706
3707 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3708 There were also incompatibilities regarding the handling of \Q..\E inside
3709 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3710 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3711
3712 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3713 matches an empty string, and forcibly breaks the loop. There were bugs in
3714 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3715 against aaaa the result was just "a" rather than "aaaa", for example. Two
3716 separate and independent bugs (that affected different cases) have been
3717 fixed.
3718
3719 39. Refactored the code to abolish the use of different opcodes for small
3720 capturing bracket numbers. This is a tidy that I avoided doing when I
3721 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3722 The new approach is not only tidier, it makes it possible to reduce the
3723 memory needed to fix the previous bug (38).
3724
3725 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3726 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3727 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3728 mode.
3729
3730 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3731 report.
3732
3733 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3734 copy construction and assignment in the C++ wrapper.
3735
3736 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3737 couple of bugs that could have given wrong results in the "--newline=crlf"
3738 case.
3739
3740 44. Added a number of casts and did some reorganization of signed/unsigned int
3741 variables following suggestions from Dair Grant. Also renamed the variable
3742 "this" as "item" because it is a C++ keyword.
3743
3744 45. Arranged for dftables to add
3745
3746 #include "pcre_internal.h"
3747
3748 to pcre_chartables.c because without it, gcc 4.x may remove the array
3749 definition from the final binary if PCRE is built into a static library and
3750 dead code stripping is activated.
3751
3752 46. For an unanchored pattern, if a match attempt fails at the start of a
3753 newline sequence, and the newline setting is CRLF or ANY, and the next two
3754 characters are CRLF, advance by two characters instead of one.
3755
3756
3757 Version 6.7 04-Jul-06
3758 ---------------------
3759
3760 1. In order to handle tests when input lines are enormously long, pcretest has
3761 been re-factored so that it automatically extends its buffers when
3762 necessary. The code is crude, but this _is_ just a test program. The
3763 default size has been increased from 32K to 50K.
3764
3765 2. The code in pcre_study() was using the value of the re argument before
3766 testing it for NULL. (Of course, in any sensible call of the function, it
3767 won't be NULL.)
3768
3769 3. The memmove() emulation function in pcre_internal.h, which is used on
3770 systems that lack both memmove() and bcopy() - that is, hardly ever -
3771 was missing a "static" storage class specifier.
3772
3773 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3774 containing an extended class (one that cannot be represented by a bitmap
3775 because it contains high-valued characters or Unicode property items, e.g.
3776 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3777 pattern, but PCRE should not loop if you do not (it no longer does).
3778 [Detail: two cases were found: (a) a repeated subpattern containing an
3779 extended class; (b) a recursive reference to a subpattern that followed a
3780 previous extended class. It wasn't skipping over the extended class
3781 correctly when UTF-8 mode was not set.]
3782
3783 5. A negated single-character class was not being recognized as fixed-length
3784 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3785 compile error "lookbehind assertion is not fixed length".
3786
3787 6. The RunPerlTest auxiliary script was showing an unexpected difference
3788 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3789 write a Perl script that can interpret lines of an input file either as
3790 byte characters or as UTF-8, which is what "perltest" was being required to
3791 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3792 can't do is switch easily at run time between having the "use utf8;" pragma
3793 or not. In the end, I fudged it by using the RunPerlTest script to insert
3794 "use utf8;" explicitly for the UTF-8 tests.
3795
3796 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3797 the end of the subject string, contrary to the documentation and to what
3798 Perl does. This was true of both matching functions. Now it matches only at
3799 the start of the subject and immediately after *internal* newlines.
3800
3801 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3802 a pointer to an int instead of a pointer to an unsigned long int. This
3803 caused problems on 64-bit systems.
3804
3805 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3806 instance of the 'standard' template library not being so standard".
3807
3808 10. There was no check on the number of named subpatterns nor the maximum
3809 length of a subpattern name. The product of these values is used to compute
3810 the size of the memory block for a compiled pattern. By supplying a very
3811 long subpattern name and a large number of named subpatterns, the size
3812 computation could be caused to overflow. This is now prevented by limiting
3813 the length of names to 32 characters, and the number of named subpatterns
3814 to 10,000.
3815
3816 11. Subpatterns that are repeated with specific counts have to be replicated in
3817 the compiled pattern. The size of memory for this was computed from the
3818 length of the subpattern and the repeat count. The latter is limited to
3819 65535, but there was no limit on the former, meaning that integer overflow
3820 could in principle occur. The compiled length of a repeated subpattern is
3821 now limited to 30,000 bytes in order to prevent this.
3822
3823 12. Added the optional facility to have named substrings with the same name.
3824
3825 13. Added the ability to use a named substring as a condition, using the
3826 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3827 are numbers (not recommended). Forward references are permitted.
3828
3829 14. Added forward references in named backreferences (if you see what I mean).
3830
3831 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3832 pattern could run off the end of the subject. For example, the pattern
3833 "(?s)(.{1,5})"8 did this with the subject "ab".
3834
3835 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3836 PCRE_CASELESS was set when matching characters that were quantified with ?
3837 or *.
3838
3839 17. A character class other than a single negated character that had a minimum
3840 but no maximum quantifier - for example [ab]{6,} - was not handled
3841 correctly by pce_dfa_exec(). It would match only one character.
3842
3843 18. A valid (though odd) pattern that looked like a POSIX character
3844 class but used an invalid character after [ (for example [[,abc,]]) caused
3845 pcre_compile() to give the error "Failed: internal error: code overflow" or
3846 in some cases to crash with a glibc free() error. This could even happen if
3847 the pattern terminated after [[ but there just happened to be a sequence of
3848 letters, a binary zero, and a closing ] in the memory that followed.
3849
3850 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3851 over the years. Originally (before any Unicode support), just the bottom 8
3852 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3853 output from "man perlunicode" includes this:
3854
3855 The regular expression compiler produces polymorphic opcodes. That
3856 is, the pattern adapts to the data and automatically switches to
3857 the Unicode character scheme when presented with Unicode data--or
3858 instead uses a traditional byte scheme when presented with byte
3859 data.
3860
3861 Sadly, a wide octal escape does not cause a switch, and in a string with
3862 no other multibyte characters, these octal escapes are treated as before.
3863 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3864 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3865 Unicode string.
3866
3867 I have not perpetrated such confusion in PCRE. Up till now, it took just
3868 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3869 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3870 translate to the appropriate multibyte character.
3871
3872 29. Applied some refactoring to reduce the number of warnings from Microsoft
3873 and Borland compilers. This has included removing the fudge introduced
3874 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3875 a warning about an unused variable.
3876
3877 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3878 characters since release 4.0, because Perl (from release 5.004) does not.
3879 [Or at least, is documented not to: some releases seem to be in conflict
3880 with the documentation.] However, when a pattern was studied with
3881 pcre_study() and all its branches started with \s, PCRE still included VT
3882 as a possible starting character. Of course, this did no harm; it just
3883 caused an unnecessary match attempt.
3884
3885 22. Removed a now-redundant internal flag bit that recorded the fact that case
3886 dependency changed within the pattern. This was once needed for "required
3887 byte" processing, but is no longer used. This recovers a now-scarce options
3888 bit. Also moved the least significant internal flag bit to the most-
3889 significant bit of the word, which was not previously used (hangover from
3890 the days when it was an int rather than a uint) to free up another bit for
3891 the future.
3892
3893 23. Added support for CRLF line endings as well as CR and LF. As well as the
3894 default being selectable at build time, it can now be changed at runtime
3895 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3896 specify that it is scanning data with non-default line endings.
3897
3898 24. Changed the definition of CXXLINK to make it agree with the definition of
3899 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3900
3901 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3902 recursions. This makes a big different to stack usage for some patterns.
3903
3904 26. If a subpattern containing a named recursion or subroutine reference such
3905 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3906 the space required for the compiled pattern went wrong and gave too small a
3907 value. Depending on the environment, this could lead to "Failed: internal
3908 error: code overflow at offset 49" or "glibc detected double free or
3909 corruption" errors.
3910
3911 27. Applied patches from Google (a) to support the new newline modes and (b) to
3912 advance over multibyte UTF-8 characters in GlobalReplace.
3913
3914 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3915 difference for some implementation of PCRE in some Windows version.
3916
3917 29. Added some extra testing facilities to pcretest:
3918
3919 \q<number> in a data line sets the "match limit" value
3920 \Q<number> in a data line sets the "match recursion limt" value
3921 -S <number> sets the stack size, where <number> is in megabytes
3922
3923 The -S option isn't available for Windows.
3924
3925
3926 Version 6.6 06-Feb-06
3927 ---------------------
3928
3929 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3930 in pcreposix.h. I have copied the definition from pcre.h.
3931
3932 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3933 because pcre.h is no longer a built file.
3934
3935 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3936 not normally included in the compiled code.
3937
3938
3939 Version 6.5 01-Feb-06
3940 ---------------------
3941
3942 1. When using the partial match feature with pcre_dfa_exec(), it was not
3943 anchoring the second and subsequent partial matches at the new starting
3944 point. This could lead to incorrect results. For example, with the pattern
3945 /1234/, partially matching against "123" and then "a4" gave a match.
3946
3947 2. Changes to pcregrep:
3948
3949 (a) All non-match returns from pcre_exec() were being treated as failures
3950 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3951 error message is output. Some extra information is given for the
3952 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3953 probably the only errors that are likely to be caused by users (by
3954 specifying a regex that has nested indefinite repeats, for instance).
3955 If there are more than 20 of these errors, pcregrep is abandoned.
3956
3957 (b) A binary zero was treated as data while matching, but terminated the
3958 output line if it was written out. This has been fixed: binary zeroes
3959 are now no different to any other data bytes.
3960
3961 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3962 used to set a locale for matching. The --locale=xxxx long option has
3963 been added (no short equivalent) to specify a locale explicitly on the
3964 pcregrep command, overriding the environment variables.
3965
3966 (d) When -B was used with -n, some line numbers in the output were one less
3967 than they should have been.
3968
3969 (e) Added the -o (--only-matching) option.
3970
3971 (f) If -A or -C was used with -c (count only), some lines of context were
3972 accidentally printed for the final match.
3973
3974 (g) Added the -H (--with-filename) option.
3975
3976 (h) The combination of options -rh failed to suppress file names for files
3977 that were found from directory arguments.
3978
3979 (i) Added the -D (--devices) and -d (--directories) options.
3980
3981 (j) Added the -F (--fixed-strings) option.
3982
3983 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3984
3985 (l) Added the --colo(u)r option.
3986
3987 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3988 is not present by default.
3989
3990 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3991 items such as (?R) or (?1), when the recursion could match a number of
3992 alternatives. If it matched one of the alternatives, but subsequently,
3993 outside the recursion, there was a failure, the code tried to back up into
3994 the recursion. However, because of the way PCRE is implemented, this is not
3995 possible, and the result was an incorrect result from the match.
3996
3997 In order to prevent this happening, the specification of recursion has
3998 been changed so that all such subpatterns are automatically treated as
3999 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4000
4001 4. I had overlooked the fact that, in some locales, there are characters for
4002 which isalpha() is true but neither isupper() nor islower() are true. In
4003 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4004 and ordfeminine) are like this. This affected the treatment of \w and \W
4005 when they appeared in character classes, but not when they appeared outside
4006 a character class. The bit map for "word" characters is now created
4007 separately from the results of isalnum() instead of just taking it from the
4008 upper, lower, and digit maps. (Plus the underscore character, of course.)
4009
4010 5. The above bug also affected the handling of POSIX character classes such as
4011 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4012 permanent tables. Instead, the bit maps for such a class were previously
4013 created as the appropriate unions of the upper, lower, and digit bitmaps.
4014 Now they are created by subtraction from the [[:word:]] class, which has
4015 its own bitmap.
4016
4017 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4018 It is created by subtracting the vertical space characters (\x09, \x0a,
4019 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4020 subtraction was done in the overall bitmap for a character class, meaning
4021 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4022 be recognized. This bug has been fixed.
4023
4024 7. Patches from the folks at Google:
4025
4026 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4027 real life, but is still worth protecting against".
4028
4029 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4030 regular expressions".
4031
4032 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4033 have it.
4034
4035 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4036 "configure" and the latter not, in order to fix a problem somebody had
4037 with compiling the Arg class on HP-UX.
4038
4039 (e) Improve the error-handling of the C++ wrapper a little bit.
4040
4041 (f) New tests for checking recursion limiting.
4042
4043 8. The pcre_memmove() function, which is used only if the environment does not
4044 have a standard memmove() function (and is therefore rarely compiled),
4045 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4046 returning a result (though PCRE never actually uses the result).
4047
4048 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4049 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4050 returned instead of calling malloc() with an overflowing number that would
4051 most likely cause subsequent chaos.
4052
4053 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4054
4055 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4056 with this option is matched, the nmatch and pmatch options of regexec() are
4057 ignored.
4058
4059 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4060 provided in case anyone wants to the the POSIX interface with UTF-8
4061 strings.
4062
4063 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4064 C++ linking (needed for some HP-UX environments).
4065
4066 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4067 (unused parameter) and in the pcre_printint() function (omitted "default"
4068 switch label when the default is to do nothing).
4069
4070 15. Added some code to make it possible, when PCRE is compiled as a C++
4071 library, to replace subject pointers for pcre_exec() with a smart pointer
4072 class, thus making it possible to process discontinuous strings.
4073
4074 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4075 much the same function. They were added by different people who were trying
4076 to make PCRE easy to compile on non-Unix systems. It has been suggested
4077 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4078 for compiling on Windows systems. I have therefore replaced it with
4079 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4080 defaults to "extern" for C or "extern C" for C++, which works fine on
4081 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4082 SCOPE with something explicit in config.h. In addition:
4083
4084 (a) pcreposix.h still had just "extern" instead of either of these macros;
4085 I have replaced it with PCRE_DATA_SCOPE.
4086
4087 (b) Functions such as _pcre_xclass(), which are internal to the library,
4088 but external in the C sense, all had PCRE_EXPORT in their definitions.
4089 This is apparently wrong for the Windows case, so I have removed it.
4090 (It makes no difference on Unix-like systems.)
4091
4092 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4093 of recursive calls to match(). This is different to MATCH_LIMIT because
4094 that limits the total number of calls to match(), not all of which increase
4095 the depth of recursion. Limiting the recursion depth limits the amount of
4096 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4097 when PCRE is compiled, and changed at run time. A patch from Google adds
4098 this functionality to the C++ interface.
4099
4100 18. Changes to the handling of Unicode character properties:
4101
4102 (a) Updated the table to Unicode 4.1.0.
4103
4104 (b) Recognize characters that are not in the table as "Cn" (undefined).
4105
4106 (c) I revised the way the table is implemented to a much improved format
4107 which includes recognition of ranges. It now supports the ranges that
4108 are defined in UnicodeData.txt, and it also amalgamates other
4109 characters into ranges. This has reduced the number of entries in the
4110 table from around 16,000 to around 3,000, thus reducing its size
4111 considerably. I realized I did not need to use a tree structure after
4112 all - a binary chop search is just as efficient. Having reduced the
4113 number of entries, I extended their size from 6 bytes to 8 bytes to
4114 allow for more data.
4115
4116 (d) Added support for Unicode script names via properties such as \p{Han}.
4117
4118 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4119 matching that character.
4120
4121 20. When matching a repeated Unicode property with a minimum greater than zero,
4122 (for example \pL{2,}), PCRE could look past the end of the subject if it
4123 reached it while seeking the minimum number of characters. This could
4124 happen only if some of the characters were more than one byte long, because
4125 there is a check for at least the minimum number of bytes.
4126
4127 21. Refactored the implementation of \p and \P so as to be more general, to
4128 allow for more different types of property in future. This has changed the
4129 compiled form incompatibly. Anybody with saved compiled patterns that use
4130 \p or \P will have to recompile them.
4131
4132 22. Added "Any" and "L&" to the supported property types.
4133
4134 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4135 but give a compile time error if the value is greater than 0xff.
4136
4137 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4138 accidentally not being installed or uninstalled.
4139
4140 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4141 made were to insert the current release number. This seemed silly, because
4142 it made things harder for people building PCRE on systems that don't run
4143 "configure". I have turned pcre.h into a distributed file, no longer built
4144 by "configure", with the version identification directly included. There is
4145 no longer a pcre.h.in file.
4146
4147 However, this change necessitated a change to the pcre-config script as
4148 well. It is built from pcre-config.in, and one of the substitutions was the
4149 release number. I have updated configure.ac so that ./configure now finds
4150 the release number by grepping pcre.h.
4151
4152 26. Added the ability to run the tests under valgrind.
4153
4154
4155 Version 6.4 05-Sep-05
4156 ---------------------
4157
4158 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4159 "--" to be printed when multiple files were scanned, even when none of the
4160 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4161 consider it to be a bug, and have restored the previous behaviour.
4162
4163 2. A couple of code tidies to get rid of compiler warnings.
4164
4165 3. The pcretest program used to cheat by referring to symbols in the library
4166 whose names begin with _pcre_. These are internal symbols that are not
4167 really supposed to be visible externally, and in some environments it is
4168 possible to suppress them. The cheating is now confined to including
4169 certain files from the library's source, which is a bit cleaner.
4170
4171 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4172 file's purpose clearer.
4173
4174 5. Reorganized pcre_ucp_findchar().
4175
4176
4177 Version 6.3 15-Aug-05
4178 ---------------------
4179
4180 1. The file libpcre.pc.in did not have general read permission in the tarball.
4181
4182 2. There were some problems when building without C++ support:
4183
4184 (a) If C++ support was not built, "make install" and "make test" still
4185 tried to test it.
4186
4187 (b) There were problems when the value of CXX was explicitly set. Some
4188 changes have been made to try to fix these, and ...
4189
4190 (c) --disable-cpp can now be used to explicitly disable C++ support.
4191
4192 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4193 backslash in a target when C++ was disabled. This confuses some
4194 versions of "make", apparently. Using an intermediate variable solves
4195 this. (Same for CPP_LOBJ.)
4196
4197 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4198 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4199 necessary on certain architectures.
4200
4201 4. Added a setting of -export-symbols-regex to the link command to remove
4202 those symbols that are exported in the C sense, but actually are local
4203 within the library, and not documented. Their names all begin with
4204 "_pcre_". This is not a perfect job, because (a) we have to except some
4205 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4206 available (and never for static libraries). I have made a note to try to
4207 find a way round (a) in the future.
4208
4209
4210 Version 6.2 01-Aug-05
4211 ---------------------
4212
4213 1. There was no test for integer overflow of quantifier values. A construction
4214 such as {1111111111111111} would give undefined results. What is worse, if
4215 a minimum quantifier for a parenthesized subpattern overflowed and became
4216 negative, the calculation of the memory size went wrong. This could have
4217 led to memory overwriting.
4218
4219 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4220
4221 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4222 operating environments where this matters.
4223
4224 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4225 PCRE options from within the C++ wrapper.
4226
4227 5. Named capturing subpatterns were not being correctly counted when a pattern
4228 was compiled. This caused two problems: (a) If there were more than 100
4229 such subpatterns, the calculation of the memory needed for the whole
4230 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4231 back references of the form \12, where the number was greater than 9, were
4232 not recognized as back references, even though there were sufficient
4233 previous subpatterns.
4234
4235 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4236 versions of gcc, e.g. 2.95.4.
4237
4238
4239 Version 6.1 21-Jun-05
4240 ---------------------
4241
4242 1. There was one reference to the variable "posix" in pcretest.c that was not
4243 surrounded by "#if !defined NOPOSIX".
4244
4245 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4246 the cross-check on the old pcre_info() function, for the benefit of the
4247 cut-down version of PCRE that is currently imported into Exim.
4248
4249 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4250 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4251 patterns that start (?i) but I don't think that matters. The use of (?i) is
4252 just an example; this all applies to the other options as well.
4253
4254 4. Since libtool seems to echo the compile commands it is issuing, the output
4255 from "make" can be reduced a bit by putting "@" in front of each libtool
4256 compile command.
4257
4258 5. Patch from the folks at Google for configure.in to be a bit more thorough
4259 in checking for a suitable C++ installation before trying to compile the
4260 C++ stuff. This should fix a reported problem when a compiler was present,
4261 but no suitable headers.
4262
4263 6. The man pages all had just "PCRE" as their title. I have changed them to
4264 be the relevant file name. I have also arranged that these names are
4265 retained in the file doc/pcre.txt, which is a concatenation in text format
4266 of all the man pages except the little individual ones for each function.
4267
4268 7. The NON-UNIX-USE file had not been updated for the different set of source
4269 files that come with release 6. I also added a few comments about the C++
4270 wrapper.
4271
4272
4273 Version 6.0 07-Jun-05
4274 ---------------------
4275
4276 1. Some minor internal re-organization to help with my DFA experiments.
4277
4278 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4279 didn't matter for the library itself when fully configured, but did matter
4280 when compiling without UCP support, or within Exim, where the ucp files are
4281 not imported.
4282
4283 3. Refactoring of the library code to split up the various functions into
4284 different source modules. The addition of the new DFA matching code (see
4285 below) to a single monolithic source would have made it really too
4286 unwieldy, quite apart from causing all the code to be include in a
4287 statically linked application, when only some functions are used. This is
4288 relevant even without the DFA addition now that patterns can be compiled in
4289 one application and matched in another.
4290
4291 The downside of splitting up is that there have to be some external
4292 functions and data tables that are used internally in different modules of
4293 the library but which are not part of the API. These have all had their
4294 names changed to start with "_pcre_" so that they are unlikely to clash
4295 with other external names.
4296
4297 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4298 a different (DFA) algorithm. Although it is slower than the original
4299 function, it does have some advantages for certain types of matching
4300 problem.
4301
4302 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4303 including restarting after a partial match.
4304
4305 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4306 defined when compiling for Windows was sent to me. I have put it into the
4307 code, though I have no means of testing or verifying it.
4308
4309 7. Added the pcre_refcount() auxiliary function.
4310
4311 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4312 match before or at the first newline in the subject string. In pcretest,
4313 the /f option on a pattern can be used to set this.
4314
4315 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4316 would behave wrongly. This has been present in PCRE since release 4.0.
4317
4318 10. A number of changes to the pcregrep command:
4319
4320 (a) Refactored how -x works; insert ^(...)$ instead of setting
4321 PCRE_ANCHORED and checking the length, in preparation for adding
4322 something similar for -w.
4323
4324 (b) Added the -w (match as a word) option.
4325
4326 (c) Refactored the way lines are read and buffered so as to have more
4327 than one at a time available.
4328
4329 (d) Implemented a pcregrep test script.
4330
4331 (e) Added the -M (multiline match) option. This allows patterns to match
4332 over several lines of the subject. The buffering ensures that at least
4333 8K, or the rest of the document (whichever is the shorter) is available
4334 for matching (and similarly the previous 8K for lookbehind assertions).
4335
4336 (f) Changed the --help output so that it now says
4337
4338 -w, --word-regex(p)
4339
4340 instead of two lines, one with "regex" and the other with "regexp"
4341 because that confused at least one person since the short forms are the
4342 same. (This required a bit of code, as the output is generated
4343 automatically from a table. It wasn't just a text change.)
4344
4345 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4346 option but starts with a hyphen. Could be a pattern or a path name
4347 starting with a hyphen, for instance.
4348
4349 (h) "-" can be given as a file name to represent stdin.
4350
4351 (i) When file names are being printed, "(standard input)" is used for
4352 the standard input, for compatibility with GNU grep. Previously
4353 "<stdin>" was used.
4354
4355 (j) The option --label=xxx can be used to supply a name to be used for
4356 stdin when file names are being printed. There is no short form.
4357
4358 (k) Re-factored the options decoding logic because we are going to add
4359 two more options that take data. Such options can now be given in four
4360 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4361
4362 (l) Added the -A, -B, and -C options for requesting that lines of context
4363 around matches be printed.
4364
4365 (m) Added the -L option to print the names of files that do not contain
4366 any matching lines, that is, the complement of -l.
4367
4368 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4369 continue to scan other files.
4370
4371 (o) The -s option was incorrectly implemented. For compatibility with other
4372 greps, it now suppresses the error message for a non-existent or non-
4373 accessible file (but not the return code). There is a new option called
4374 -q that suppresses the output of matching lines, which was what -s was
4375 previously doing.
4376
4377 (p) Added --include and --exclude options to specify files for inclusion
4378 and exclusion when recursing.
4379
4380 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4381 Hopefully, it now does.
4382
4383 12. Missing cast in pcre_study().
4384
4385 13. Added an "uninstall" target to the makefile.
4386
4387 14. Replaced "extern" in the function prototypes in Makefile.in with
4388 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4389 world, but is set differently for Windows.
4390
4391 15. Added a second compiling function called pcre_compile2(). The only
4392 difference is that it has an extra argument, which is a pointer to an
4393 integer error code. When there is a compile-time failure, this is set
4394 non-zero, in addition to the error test pointer being set to point to an
4395 error message. The new argument may be NULL if no error number is required
4396 (but then you may as well call pcre_compile(), which is now just a
4397 wrapper). This facility is provided because some applications need a
4398 numeric error indication, but it has also enabled me to tidy up the way
4399 compile-time errors are handled in the POSIX wrapper.
4400
4401 16. Added VPATH=.libs to the makefile; this should help when building with one
4402 prefix path and installing with another. (Or so I'm told by someone who
4403 knows more about this stuff than I do.)
4404
4405 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4406 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4407 match everything, including newlines. This is not POSIX-compatible, but
4408 somebody wanted the feature. From pcretest it can be activated by using
4409 both the P and the s flags.
4410
4411 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4412
4413 19. libpcre.pc was being incorrectly installed as executable.
4414
4415 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4416 it now also looks for '\r' so that it will work unmodified on Windows.
4417
4418 21. Added Google's contributed C++ wrapper to the distribution.
4419
4420 22. Added some untidy missing memory free() calls in pcretest, to keep
4421 Electric Fence happy when testing.
4422
4423
4424
4425 Version 5.0 13-Sep-04
4426 ---------------------
4427
4428 1. Internal change: literal characters are no longer packed up into items
4429 containing multiple characters in a single byte-string. Each character
4430 is now matched using a separate opcode. However, there may be more than one
4431 byte in the character in UTF-8 mode.
4432
4433 2. The pcre_callout_block structure has two new fields: pattern_position and
4434 next_item_length. These contain the offset in the pattern to the next match
4435 item, and its length, respectively.
4436
4437 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4438 insertion of callouts before each pattern item. Added the /C option to
4439 pcretest to make use of this.
4440
4441 4. On the advice of a Windows user, the lines
4442
4443 #if defined(_WIN32) || defined(WIN32)
4444 _setmode( _fileno( stdout ), 0x8000 );
4445 #endif /* defined(_WIN32) || defined(WIN32) */
4446
4447 have been added to the source of pcretest. This apparently does useful
4448 magic in relation to line terminators.
4449
4450 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4451 for the benefit of those environments where the "b" makes a difference.
4452
4453 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4454 to know about it. I have put a hack into configure.in that adds in code
4455 to set GCC=yes if CC=icc. This seems to end up at a point in the
4456 generated configure script that is early enough to affect the setting of
4457 compiler options, which is what is needed, but I have no means of testing
4458 whether it really works. (The user who reported this had patched the
4459 generated configure script, which of course I cannot do.)
4460
4461 LATER: After change 22 below (new libtool files), the configure script
4462 seems to know about icc (and also ecc). Therefore, I have commented out
4463 this hack in configure.in.
4464
4465 7. Added support for pkg-config (2 patches were sent in).
4466
4467 8. Negated POSIX character classes that used a combination of internal tables
4468 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4469 [[:^ascii]]. Typically, they would match almost any characters. The other
4470 POSIX classes were not broken in this way.
4471
4472 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4473 to find the match, as PCRE was deluded into thinking that the match had to
4474 start at the start point or following a newline. The same bug applied to
4475 patterns with negative forward assertions or any backward assertions
4476 preceding ".*" at the start, unless the pattern required a fixed first
4477 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4478
4479 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4480 starting at the last subject character, bytes beyond the end of the subject
4481 string were read.
4482
4483 11. Renamed the variable "class" as "classbits" to make life easier for C++
4484 users. (Previously there was a macro definition, but it apparently wasn't
4485 enough.)
4486
4487 12. Added the new field "tables" to the extra data so that tables can be passed
4488 in at exec time, or the internal tables can be re-selected. This allows
4489 a compiled regex to be saved and re-used at a later time by a different
4490 program that might have everything at different addresses.
4491
4492 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4493 -R library as well as a -L library.
4494
4495 14. The debugging options of pcretest (-d on the command line or D on a
4496 pattern) showed incorrect output for anything following an extended class
4497 that contained multibyte characters and which was followed by a quantifier.
4498
4499 15. Added optional support for general category Unicode character properties
4500 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4501 support. It adds about 90K to the size of the library. The meanings of the
4502 inbuilt class escapes such as \d and \s have NOT been changed.
4503
4504 16. Updated pcredemo.c to include calls to free() to release the memory for the
4505 compiled pattern.
4506
4507 17. The generated file chartables.c was being created in the source directory
4508 instead of in the building directory. This caused the build to fail if the
4509 source directory was different from the building directory, and was
4510 read-only.
4511
4512 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4513 file. No doubt somebody will tell me if they don't make sense... Also added
4514 Dan Mooney's comments about building on OpenVMS.
4515
4516 19. Added support for partial matching via the PCRE_PARTIAL option for
4517 pcre_exec() and the \P data escape in pcretest.
4518
4519 20. Extended pcretest with 3 new pattern features:
4520
4521 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4522 write the compiled pattern to the file whose name is "rest-of-line".
4523 This is a straight binary dump of the data, with the saved pointer to
4524 the character tables forced to be NULL. The study data, if any, is
4525 written too. After writing, pcretest reads a new pattern.
4526
4527 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4528 compiled pattern from the given file. There must not be any
4529 occurrences of "<" in the file name (pretty unlikely); if there are,
4530 pcretest will instead treat the initial "<" as a pattern delimiter.
4531 After reading in the pattern, pcretest goes on to read data lines as
4532 usual.
4533
4534 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4535 and 16-bit fields in a compiled pattern, to simulate a pattern that
4536 was compiled on a host of opposite endianness.
4537
4538 21. The pcre-exec() function can now cope with patterns that were compiled on
4539 hosts of opposite endianness, with this restriction:
4540
4541 As for any compiled expression that is saved and used later, the tables
4542 pointer field cannot be preserved; the extra_data field in the arguments
4543 to pcre_exec() should be used to pass in a tables address if a value
4544 other than the default internal tables were used at compile time.
4545
4546 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4547 now diagnosed as an error. Previously, most of the time, a negative number
4548 would have been treated as zero, but if in addition "ovector" was passed as
4549 NULL, a crash could occur.
4550
4551 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4552 new versions from the libtool 1.5 distribution (the last one is a copy of
4553 a file called libtool.m4). This seems to have fixed the need to patch
4554 "configure" to support Darwin 1.3 (which I used to do). However, I still
4555 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4556 workstation).
4557
4558 24. Changed the PCRE licence to be the more standard "BSD" licence.
4559
4560
4561 Version 4.5 01-Dec-03
4562 ---------------------
4563
4564 1. There has been some re-arrangement of the code for the match() function so
4565 that it can be compiled in a version that does not call itself recursively.
4566 Instead, it keeps those local variables that need separate instances for
4567 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4568 needs to "recurse". Keeping track of where control must go is done by means
4569 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4570 hide most of the details from the main code, and operates only if
4571 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4572 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4573 operating.
4574
4575 To make it easier for callers to provide specially tailored get/free
4576 functions for this usage, two new functions, pcre_stack_malloc, and
4577 pcre_stack_free, are used. They are always called in strict stacking order,
4578 and the size of block requested is always the same.
4579
4580 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4581 PCRE has been compiled to use the stack or the heap for recursion. The
4582 -C option of pcretest uses this to show which version is compiled.
4583
4584 A new data escape \S, is added to pcretest; it causes the amounts of store
4585 obtained and freed by both kinds of malloc/free at match time to be added
4586 to the output.
4587
4588 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4589 what's available on my current Linux desktop machine.
4590
4591 3. When matching a UTF-8 string, the test for a valid string at the start has
4592 been extended. If start_offset is not zero, PCRE now checks that it points
4593 to a byte that is the start of a UTF-8 character. If not, it returns
4594 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4595 this is necessary because there may be backward assertions in the pattern.
4596 When matching the same subject several times, it may save resources to use
4597 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4598
4599 4. The code for checking the validity of UTF-8 strings has been tightened so
4600 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4601 containing "overlong sequences".
4602
4603 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4604 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4605 should have been "&", but it just so happened that all the cases this let
4606 through by mistake were picked up later in the function.
4607
4608 6. I had used a variable called "isblank" - this is a C99 function, causing
4609 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4610
4611 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4612 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4613 so the version is shown; (c) stop "make test" echoing "./RunTest".
4614
4615 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4616
4617 9. The prototype for memmove() for systems that don't have it was using
4618 size_t, but the inclusion of the header that defines size_t was later. I've
4619 moved the #includes for the C headers earlier to avoid this.
4620
4621 10. Added some adjustments to the code to make it easier to compiler on certain
4622 special systems:
4623
4624 (a) Some "const" qualifiers were missing.
4625 (b) Added the macro EXPORT before all exported functions; by default this
4626 is defined to be empty.
4627 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4628 that it reads its output file name as an argument instead of writing
4629 to the standard output and assuming this can be redirected.
4630
4631 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4632 class containing characters with values greater than 255, PCRE compilation
4633 went into a loop.
4634
4635 12. A recursive reference to a subpattern that was within another subpattern
4636 that had a minimum quantifier of zero caused PCRE to crash. For example,
4637 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4638 recursion. If the recursively-called subpattern itself had a zero repeat,
4639 that was OK.
4640
4641 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4642 buffer into which it was copied (for escape processing) was still set at
4643 1024, so long lines caused crashes.
4644
4645 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4646 "internal error: code overflow...". This applied to any character class
4647 that was followed by a possessive quantifier.
4648
4649 15. Modified the Makefile to add libpcre.la as a prerequisite for
4650 libpcreposix.la because I was told this is needed for a parallel build to
4651 work.
4652
4653 16. If a pattern that contained .* following optional items at the start was
4654 studied, the wrong optimizing data was generated, leading to matching
4655 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4656 matching string must start with a or b or c. The correct conclusion for
4657 this pattern is that a match can start with any character.
4658
4659
4660 Version 4.4 13-Aug-03
4661 ---------------------
4662
4663 1. In UTF-8 mode, a character class containing characters with values between
4664 127 and 255 was not handled correctly if the compiled pattern was studied.
4665 In fixing this, I have also improved the studying algorithm for such
4666 classes (slightly).
4667
4668 2. Three internal functions had redundant arguments passed to them. Removal
4669 might give a very teeny performance improvement.
4670
4671 3. Documentation bug: the value of the capture_top field in a callout is *one
4672 more than* the number of the hightest numbered captured substring.
4673
4674 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4675 in incorrectly linking with a previously installed version. They now link
4676 explicitly with libpcre.la.
4677
4678 5. configure.in no longer needs to recognize Cygwin specially.
4679
4680 6. A problem in pcre.in for Windows platforms is fixed.
4681
4682 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4683 pcretest, it used to include the size of the study block as part of its
4684 output. Unfortunately, the structure contains a field that has a different
4685 size on different hardware architectures. This meant that the tests that
4686 showed this size failed. As the block is currently always of a fixed size,
4687 this information isn't actually particularly useful in pcretest output, so
4688 I have just removed it.
4689
4690 8. Three pre-processor statements accidentally did not start in column 1.
4691 Sadly, there are *still* compilers around that complain, even though
4692 standard C has not required this for well over a decade. Sigh.
4693
4694 9. In pcretest, the code for checking callouts passed small integers in the
4695 callout_data field, which is a void * field. However, some picky compilers
4696 complained about the casts involved for this on 64-bit systems. Now
4697 pcretest passes the address of the small integer instead, which should get
4698 rid of the warnings.
4699
4700 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4701 both compile and run time, and gives an error if an invalid UTF-8 sequence
4702 is found. There is a option for disabling this check in cases where the
4703 string is known to be correct and/or the maximum performance is wanted.
4704
4705 11. In response to a bug report, I changed one line in Makefile.in from
4706
4707 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4708 to
4709 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4710
4711 to look similar to other lines, but I have no way of telling whether this
4712 is the right thing to do, as I do not use Windows. No doubt I'll get told
4713 if it's wrong...
4714
4715
4716 Version 4.3 21-May-03
4717 ---------------------
4718
4719 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4720 Makefile.
4721
4722 2. Some refactoring to improve the quality of the code:
4723
4724 (i) The utf8_table... variables are now declared "const".
4725
4726 (ii) The code for \cx, which used the "case flipping" table to upper case
4727 lower case letters, now just substracts 32. This is ASCII-specific,
4728 but the whole concept of \cx is ASCII-specific, so it seems
4729 reasonable.
4730
4731 (iii) PCRE was using its character types table to recognize decimal and
4732 hexadecimal digits in the pattern. This is silly, because it handles
4733 only 0-9, a-f, and A-F, but the character types table is locale-
4734 specific, which means strange things might happen. A private
4735 table is now used for this - though it costs 256 bytes, a table is
4736 much faster than multiple explicit tests. Of course, the standard
4737 character types table is still used for matching digits in subject
4738 strings against \d.
4739
4740 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4741 ending in _t are). So I've renamed it as ESC_tee.
4742
4743 3. The first argument for regexec() in the POSIX wrapper should have been
4744 defined as "const".
4745
4746 4. Changed pcretest to use malloc() for its buffers so that they can be
4747 Electric Fenced for debugging.
4748
4749 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4750 to read one or more bytes before the start of the subject string. Often this
4751 had no effect on PCRE's behaviour, but in some circumstances it could
4752 provoke a segmentation fault.
4753
4754 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4755 to try to read one or more bytes before the start of the subject string.
4756
4757 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4758 UTF-8 support could misbehave in various ways if the subject string
4759 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4760 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4761 back over UTF-8 characters.)
4762
4763
4764 Version 4.2 14-Apr-03
4765 ---------------------
4766
4767 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4768
4769 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4770 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4771 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4772 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4773 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4774 and BUILD_EXEEXT
4775 Note: automatic setting of the BUILD variables is not yet working
4776 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4777 compile-time but not at link-time
4778 [LINK]: use for linking executables only
4779 make different versions for Windows and non-Windows
4780 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4781 libraries
4782 [LINK_FOR_BUILD]: new variable
4783 [OBJEXT]: use throughout
4784 [EXEEXT]: use throughout
4785 <winshared>: new target
4786 <wininstall>: new target
4787 <dftables.o>: use native compiler
4788 <dftables>: use native linker
4789 <install>: handle Windows platform correctly
4790 <clean>: ditto
4791 <check>: ditto
4792 copy DLL to top builddir before testing
4793
4794 As part of these changes, -no-undefined was removed again. This was reported
4795 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4796 in any case.
4797
4798 3. Some tidies to get rid of compiler warnings:
4799
4800 . In the match_data structure, match_limit was an unsigned long int, whereas
4801 match_call_count was an int. I've made them both unsigned long ints.
4802
4803 . In pcretest the fact that a const uschar * doesn't automatically cast to
4804 a void * provoked a warning.
4805
4806 . Turning on some more compiler warnings threw up some "shadow" variables
4807 and a few more missing casts.
4808
4809 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4810 option, a class that contained a single character with a value between 128
4811 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4812
4813 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4814 option, a class that contained several characters, but with at least one
4815 whose value was between 128 and 255 caused PCRE to crash.
4816
4817
4818 Version 4.1 12-Mar-03
4819 ---------------------
4820
4821 1. Compiling with gcc -pedantic found a couple of places where casts were
4822 needed, and a string in dftables.c that was longer than standard compilers are
4823 required to support.
4824
4825 2. Compiling with Sun's compiler found a few more places where the code could
4826 be tidied up in order to avoid warnings.
4827
4828 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4829 first of these names is deprecated in the latest Autoconf in favour of the name
4830 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4831 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4832 analogy I have changed it to CFLAGS_FOR_BUILD.
4833
4834 4. Added -no-undefined to the linking command in the Makefile, because this is
4835 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4836 linking step for the pcreposix library.
4837
4838 5. PCRE was failing to diagnose the case of two named groups with the same
4839 name.
4840
4841 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4842 literal character that is needed in the subject for a match, and scans along to
4843 ensure that it is present before embarking on the full matching process. This
4844 saves time in cases of nested unlimited repeats that are never going to match.
4845 Problem: the scan can take a lot of time if the subject is very long (e.g.
4846 megabytes), thus penalizing straightforward matches. It is now done only if the
4847 amount of subject to be scanned is less than 1000 bytes.
4848
4849 7. A lesser problem with the same optimization is that it was recording the
4850 first character of an anchored pattern as "needed", thus provoking a search
4851 right along the subject, even when the first match of the pattern was going to
4852 fail. The "needed" character is now not set for anchored patterns, unless it
4853 follows something in the pattern that is of non-fixed length. Thus, it still
4854 fulfils its original purpose of finding quick non-matches in cases of nested
4855 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4856
4857
4858 Version 4.0 17-Feb-03
4859 ---------------------
4860
4861 1. If a comment in an extended regex that started immediately after a meta-item
4862 extended to the end of string, PCRE compiled incorrect data. This could lead to
4863 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4864
4865 2. Moved to autoconf 2.53 and libtool 1.4.2.
4866
4867 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4868 the special perltest8 script is no longer needed - all the tests can be run
4869 from a single perltest script.
4870
4871 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4872 by \s. It has now been removed in PCRE. This means it isn't recognized as
4873 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4874 class [:space:] *does* include VT, thereby creating a mess.
4875
4876 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4877 space and tab.
4878
4879 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4880 its new features into the main test script, reducing the number of scripts.
4881
4882 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4883 were backward compatible, and made the (?i) apply to the whole pattern, as if
4884 /i were given. Now it behaves more logically, and applies the option setting
4885 only to what follows. PCRE has been changed to follow suit. However, if it
4886 finds options settings right at the start of the pattern, it extracts them into
4887 the global options, as before. Thus, they show up in the info data.
4888
4889 8. Added support for the \Q...\E escape sequence. Characters in between are
4890 treated as literals. This is slightly different from Perl in that $ and @ are
4891 also handled as literals inside the quotes. In Perl, they will cause variable
4892 interpolation. Note the following examples:
4893
4894 Pattern PCRE matches Perl matches
4895
4896 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4897 \Qabc\$xyz\E abc\$xyz abc\$xyz
4898 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4899
4900 For compatibility with Perl, \Q...\E sequences are recognized inside character
4901 classes as well as outside them.
4902
4903 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4904 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4905 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4906 signed/unsigned warnings.
4907
4908 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4909 option for pcretest, so I've replaced it by a simple function that does just
4910 that job.
4911
4912 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4913 "pcregrep -".
4914
4915 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4916 Java package. This provides some syntactic sugar for simple cases of what my
4917 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4918 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4919 item, you can use this simplified notation. Note that only makes sense with
4920 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4921 greediness, whatever the setting of the PCRE_UNGREEDY option.
4922
4923 13. A change of greediness default within a pattern was not taking effect at
4924 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4925 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4926 was abstracted outside.
4927
4928 14. PCRE now supports the \G assertion. It is true when the current matching
4929 position is at the start point of the match. This differs from \A when the
4930 starting offset is non-zero. Used with the /g option of pcretest (or similar
4931 code), it works in the same way as it does for Perl's /g option. If all
4932 alternatives of a regex begin with \G, the expression is anchored to the start
4933 match position, and the "anchored" flag is set in the compiled expression.
4934
4935 15. Some bugs concerning the handling of certain option changes within patterns
4936 have been fixed. These applied to options other than (?ims). For example,
4937 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4938 been the other way round. Some of this was related to change 7 above.
4939
4940 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4941 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4942 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4943 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4944
4945 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4946 mode. Unlike ".", it always matches newline, whatever the setting of
4947 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4948 assertions. Perl allows it, but it doesn't (in general) work because it can't
4949 calculate the length of the lookbehind. At least, that's the case for Perl
4950 5.8.0 - I've been told they are going to document that it doesn't work in
4951 future.
4952
4953 18. Added an error diagnosis for escapes that PCRE does not support: these are
4954 \L, \l, \N, \P, \p, \U, \u, and \X.
4955
4956 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4957 reading past the end of the pattern in cases such as /[abcd/.
4958
4959 20. PCRE was getting more memory than necessary for patterns with classes that
4960 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4961
4962 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4963 compiling PCRE for use with Virtual Pascal.
4964
4965 22. Small fix to the Makefile to make it work properly if the build is done
4966 outside the source tree.
4967
4968 23. Added a new extension: a condition to go with recursion. If a conditional
4969 subpattern starts with (?(R) the "true" branch is used if recursion has
4970 happened, whereas the "false" branch is used only at the top level.
4971
4972 24. When there was a very long string of literal characters (over 255 bytes
4973 without UTF support, over 250 bytes with UTF support), the computation of how
4974 much memory was required could be incorrect, leading to segfaults or other
4975 strange effects.
4976
4977 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4978 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4979 there was a subsequent back reference to those brackets. This meant that, for
4980 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4981 possible to check for precisely this case. All we can do is abandon the
4982 optimization if .* occurs inside capturing brackets when there are any back
4983 references whatsoever. (See below for a better fix that came later.)
4984
4985 26. The handling of the optimization for finding the first character of a
4986 non-anchored pattern, and for finding a character that is required later in the
4987 match were failing in some cases. This didn't break the matching; it just
4988 failed to optimize when it could. The way this is done has been re-implemented.
4989
4990 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4991
4992 28. Added a new feature that provides some of the functionality that Perl
4993 provides with (?{...}). The facility is termed a "callout". The way it is done
4994 in PCRE is for the caller to provide an optional function, by setting
4995 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4996 global variable. By default it is unset, which disables all calling out. To get
4997 the function called, the regex must include (?C) at appropriate points. This
4998 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4999 This provides a means of identifying different callout points. When PCRE
5000 reaches such a point in the regex, if pcre_callout has been set, the external
5001 function is called. It is provided with data in a structure called
5002 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5003 matching continues; if it returns a non-zero value, the match at the current
5004 point fails. However, backtracking will occur if possible. [This was changed
5005 later and other features added - see item 49 below.]
5006
5007 29. pcretest is upgraded to test the callout functionality. It provides a
5008 callout function that displays information. By default, it shows the start of
5009 the match and the current position in the text. There are some new data escapes
5010 to vary what happens:
5011
5012 \C+ in addition, show current contents of captured substrings
5013 \C- do not supply a callout function
5014 \C!n return 1 when callout number n is reached
5015 \C!n!m return 1 when callout number n is reached for the mth time
5016
5017 30. If pcregrep was called with the -l option and just a single file name, it
5018 output "<stdin>" if a match was found, instead of the file name.
5019
5020 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5021 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5022 pcre_exec(). This saves a malloc/free per call. The default value of
5023 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5024 when configuring.
5025
5026 32. The default maximum size of a compiled pattern is 64K. There have been a
5027 few cases of people hitting this limit. The code now uses macros to handle the
5028 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5029 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5030 configuring. Tests 2 and 5 work only with 2-byte links because they output
5031 debugging information about compiled patterns.
5032
5033 33. Internal code re-arrangements:
5034
5035 (a) Moved the debugging function for printing out a compiled regex into
5036 its own source file (printint.c) and used #include to pull it into
5037 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5038 separate copies.
5039
5040 (b) Defined the list of op-code names for debugging as a macro in
5041 internal.h so that it is next to the definition of the opcodes.
5042
5043 (c) Defined a table of op-code lengths for simpler skipping along compiled
5044 code. This is again a macro in internal.h so that it is next to the
5045 definition of the opcodes.
5046
5047 34. Added support for recursive calls to individual subpatterns, along the
5048 lines of Robin Houston's patch (but implemented somewhat differently).
5049
5050 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5051 allow it to read and process whole directories in Win32. This code was
5052 contributed by Lionel Fourquaux; it has not been tested by me.
5053
5054 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5055 used to name a group. Names consist of alphanumerics and underscores, and must
5056 be unique. Back references use the syntax (?P=name) and recursive calls use
5057 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5058 numbers. The function pcre_fullinfo() can be used after compilation to extract
5059 a name/number map. There are three relevant calls:
5060
5061 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5062 PCRE_INFO_NAMECOUNT yields the number of entries
5063 PCRE_INFO_NAMETABLE yields a pointer to the map.
5064
5065 The map is a vector of fixed-size entries. The size of each entry depends on
5066 the length of the longest name used. The first two bytes of each entry are the
5067 group number, most significant byte first. There follows the corresponding
5068 name, zero terminated. The names are in alphabetical order.
5069
5070 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5071 case instead of 255. Making it the same both with and without UTF-8 support
5072 means that the same test output works with both.
5073
5074 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5075 calling malloc() with a zero argument.
5076
5077 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5078 optimization. I've improved things by keeping a bitmap of backreferences with
5079 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5080 fact referenced, the optimization can be applied. It is unlikely that a
5081 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5082 the match to follow \n) will appear inside brackets with a number greater than
5083 31, but if it does, any back reference > 31 suppresses the optimization.
5084
5085 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5086 of disabling numbered capturing parentheses. Any opening parenthesis that is
5087 not followed by ? behaves as if it were followed by ?: but named parentheses
5088 can still be used for capturing (and they will acquire numbers in the usual
5089 way).
5090
5091 41. Redesigned the return codes from the match() function into yes/no/error so
5092 that errors can be passed back from deep inside the nested calls. A malloc
5093 failure while inside a recursive subpattern call now causes the
5094 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5095
5096 42. It is now possible to set a limit on the number of times the match()
5097 function is called in a call to pcre_exec(). This facility makes it possible to
5098 limit the amount of recursion and backtracking, though not in a directly
5099 obvious way, because the match() function is used in a number of different
5100 circumstances. The count starts from zero for each position in the subject
5101 string (for non-anchored patterns). The default limit is, for compatibility, a
5102 large number, namely 10 000 000. You can change this in two ways:
5103
5104 (a) When configuring PCRE before making, you can use --with-match-limit=n
5105 to set a default value for the compiled library.
5106
5107 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5108 a different value is set. See 45 below.
5109
5110 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5111
5112 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5113 of things that can be changed at compile time. The first argument specifies
5114 what is wanted and the second points to where the information is to be placed.
5115 The current list of available information is:
5116
5117 PCRE_CONFIG_UTF8
5118
5119 The output is an integer that is set to one if UTF-8 support is available;
5120 otherwise it is set to zero.
5121
5122 PCRE_CONFIG_NEWLINE
5123
5124 The output is an integer that it set to the value of the code that is used for
5125 newline. It is either LF (10) or CR (13).
5126
5127 PCRE_CONFIG_LINK_SIZE
5128
5129 The output is an integer that contains the number of bytes used for internal
5130 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5131
5132 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5133
5134 The output is an integer that contains the threshold above which the POSIX
5135 interface uses malloc() for output vectors. See item 31 above.
5136
5137 PCRE_CONFIG_MATCH_LIMIT
5138
5139 The output is an unsigned integer that contains the default limit of the number
5140 of match() calls in a pcre_exec() execution. See 42 above.
5141
5142 44. pcretest has been upgraded by the addition of the -C option. This causes it
5143 to extract all the available output from the new pcre_config() function, and to
5144 output it. The program then exits immediately.
5145
5146 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5147 order to support additional features. One way would have been to define
5148 pcre_exec2() (for example) with extra arguments, but this would not have been
5149 extensible, and would also have required all calls to the original function to
5150 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5151 is used for passing in "extra" data from pcre_study().
5152
5153 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5154 contains the following fields:
5155
5156 flags a bitmap indicating which of the following fields are set
5157 study_data opaque data from pcre_study()
5158 match_limit a way of specifying a limit on match() calls for a specific
5159 call to pcre_exec()
5160 callout_data data for callouts (see 49 below)
5161
5162 The flag bits are also defined in pcre.h, and are
5163
5164 PCRE_EXTRA_STUDY_DATA
5165 PCRE_EXTRA_MATCH_LIMIT
5166 PCRE_EXTRA_CALLOUT_DATA
5167
5168 The pcre_study() function now returns one of these new pcre_extra blocks, with
5169 the actual study data pointed to by the study_data field, and the
5170 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5171 before. That is, this change is entirely upwards-compatible and requires no
5172 change to existing code.
5173
5174 If you want to pass in additional data to pcre_exec(), you can either place it
5175 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5176 block.
5177
5178 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5179 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5180 times with different match limits, until it finds the minimum value needed for
5181 pcre_exec() to complete. The value is then output. This can be instructive; for
5182 most simple matches the number is quite small, but for pathological cases it
5183 gets very large very quickly.
5184
5185 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5186 returns the size of the data block pointed to by the study_data field in a
5187 pcre_extra block, that is, the value that was passed as the argument to
5188 pcre_malloc() when PCRE was getting memory in which to place the information
5189 created by pcre_study(). The fourth argument should point to a size_t variable.
5190 pcretest has been extended so that this information is shown after a successful
5191 pcre_study() call when information about the compiled regex is being displayed.
5192
5193 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5194 because what follows is always an absolute path. (Later: it turns out that this
5195 is more than cosmetic for MinGW, because it doesn't like empty path
5196 components.)
5197
5198 49. Some changes have been made to the callout feature (see 28 above):
5199
5200 (i) A callout function now has three choices for what it returns:
5201
5202 0 => success, carry on matching
5203 > 0 => failure at this point, but backtrack if possible
5204 < 0 => serious error, return this value from pcre_exec()
5205
5206 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5207 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5208 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5209 use by callout functions. It will never be used by PCRE itself.
5210
5211 (ii) The pcre_extra structure (see 45 above) has a void * field called
5212 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5213 pcre_callout_block structure has a field of the same name. The contents of
5214 the field passed in the pcre_extra structure are passed to the callout
5215 function in the corresponding field in the callout block. This makes it
5216 easier to use the same callout-containing regex from multiple threads. For
5217 testing, the pcretest program has a new data escape
5218
5219 \C*n pass the number n (may be negative) as callout_data
5220
5221 If the callout function in pcretest receives a non-zero value as
5222 callout_data, it returns that value.
5223
5224 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5225 there were some redundant $(CFLAGS) in commands that are now specified as
5226 $(LINK), which already includes $(CFLAGS).
5227
5228 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5229 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5230 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5231 one-byte characters throughout. Note that case-insensitive matching applies
5232 only to characters whose values are less than 256. PCRE doesn't support the
5233 notion of cases for higher-valued characters.
5234
5235 (i) A character class whose characters are all within 0-255 is handled as
5236 a bit map, and the map is inverted for negative classes. Previously, a
5237 character > 255 always failed to match such a class; however it should
5238 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5239
5240 (ii) A negated character class with a single character < 255 is coded as
5241 "not this character" (OP_NOT). This wasn't working properly when the test
5242 character was multibyte, either singly or repeated.
5243
5244 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5245 mode, for example: \x{100}{2,3}.
5246
5247 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5248 singly or repeated) now correctly test multibyte characters. However,
5249 PCRE doesn't recognize any characters with values greater than 255 as
5250 digits, spaces, or word characters. Such characters always match \D, \S,
5251 and \W, and never match \d, \s, or \w.
5252
5253 (v) Classes may now contain characters and character ranges with values
5254 greater than 255. For example: [ab\x{100}-\x{400}].
5255
5256 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5257 PCRE in UTF-8 mode.
5258
5259 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5260 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5261 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5262 value.)
5263
5264 53. The single man page has become too large. I have therefore split it up into
5265 a number of separate man pages. These also give rise to individual HTML pages;
5266 these are now put in a separate directory, and there is an index.html page that
5267 lists them all. Some hyperlinking between the pages has been installed.
5268
5269 54. Added convenience functions for handling named capturing parentheses.
5270
5271 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5272 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5273 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5274 are faulted.
5275
5276 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5277 calling configure. These values are used when compiling the dftables.c program
5278 which is run to generate the source of the default character tables. They
5279 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5280 you will need to set these values.
5281
5282 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5283
5284
5285 Version 3.9 02-Jan-02
5286 ---------------------
5287
5288 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5289
5290 2. If --disable-static was given, the building process failed when trying to
5291 build pcretest and pcregrep. (For some reason it was using libtool to compile
5292 them, which is not right, as they aren't part of the library.)
5293
5294
5295 Version 3.8 18-Dec-01
5296 ---------------------
5297
5298 1. The experimental UTF-8 code was completely screwed up. It was packing the
5299 bytes in the wrong order. How dumb can you get?
5300
5301
5302 Version 3.7 29-Oct-01
5303 ---------------------
5304
5305 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5306 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5307 this didn't happen under Solaris 8, where I normally test things.
5308
5309 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5310 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5311 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5312 this fix an hour or so after the initial 3.7 release.)
5313
5314
5315 Version 3.6 23-Oct-01
5316 ---------------------
5317
5318 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5319 offsets passed as NULL with zero offset count.
5320
5321 2. The config.guess and config.sub files had not been updated when I moved to
5322 the latest autoconf.
5323
5324
5325 Version 3.5 15-Aug-01
5326 ---------------------
5327
5328 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5329 had been forgotten.
5330
5331 2. By using declared but undefined structures, we can avoid using "void"
5332 definitions in pcre.h while keeping the internal definitions of the structures
5333 private.
5334
5335 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5336 user point of view, this means that both static and shared libraries are built
5337 by default, but this can be individually controlled. More of the work of
5338 handling this static/shared cases is now inside libtool instead of PCRE's make
5339 file.
5340
5341 4. The pcretest utility is now installed along with pcregrep because it is
5342 useful for users (to test regexs) and by doing this, it automatically gets
5343 relinked by libtool. The documentation has been turned into a man page, so
5344 there are now .1, .txt, and .html versions in /doc.
5345
5346 5. Upgrades to pcregrep:
5347 (i) Added long-form option names like gnu grep.
5348 (ii) Added --help to list all options with an explanatory phrase.
5349 (iii) Added -r, --recursive to recurse into sub-directories.
5350 (iv) Added -f, --file to read patterns from a file.
5351
5352 6. pcre_exec() was referring to its "code" argument before testing that
5353 argument for NULL (and giving an error if it was NULL).
5354
5355 7. Upgraded Makefile.in to allow for compiling in a different directory from
5356 the source directory.
5357
5358 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5359 options bits, the pointer it was passed was to an int instead of to an unsigned
5360 long int. This mattered only on 64-bit systems.
5361
5362 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5363 generated) instead of pcre.in, which it its source. Also made the same change
5364 in several of the .c files.
5365
5366 10. A new release of gcc defines printf() as a macro, which broke pcretest
5367 because it had an ifdef in the middle of a string argument for printf(). Fixed
5368 by using separate calls to printf().
5369
5370 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5371 script, to force use of CR or LF instead of \n in the source. On non-Unix
5372 systems, the value can be set in config.h.
5373
5374 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5375 absolute limit. Changed the text of the error message to make this clear, and
5376 likewise updated the man page.
5377
5378 13. The limit of 99 on the number of capturing subpatterns has been removed.
5379 The new limit is 65535, which I hope will not be a "real" limit.
5380
5381
5382 Version 3.4 22-Aug-00
5383 ---------------------
5384
5385 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5386
5387 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5388
5389
5390 Version 3.3 01-Aug-00
5391 ---------------------
5392
5393 1. If an octal character was given, but the value was greater than \377, it
5394 was not getting masked to the least significant bits, as documented. This could
5395 lead to crashes in some systems.
5396
5397 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5398 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5399
5400 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5401 These just pass their arguments on to (pcre_free)(), but they are provided
5402 because some uses of PCRE bind it to non-C systems that can call its functions,
5403 but cannot call free() or pcre_free() directly.
5404
5405 4. Add "make test" as a synonym for "make check". Corrected some comments in
5406 the Makefile.
5407
5408 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5409 Makefile.
5410
5411 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5412 command called pgrep for grepping around the active processes.
5413
5414 7. Added the beginnings of support for UTF-8 character strings.
5415
5416 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5417 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5418 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5419 out for the ar command.)
5420
5421
5422 Version 3.2 12-May-00
5423 ---------------------
5424
5425 This is purely a bug fixing release.
5426
5427 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5428 of ZA. This was just one example of several cases that could provoke this bug,
5429 which was introduced by change 9 of version 2.00. The code for breaking
5430 infinite loops after an iteration that matches an empty string was't working
5431 correctly.
5432
5433 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5434 when matched against abbab (for example). After matching an empty string, it
5435 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5436 caused it to match further down the string than it should.
5437
5438 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5439 was there because it doesn't seem to be needed, and it causes trouble on some
5440 systems, as it is not a Standard C header. It has been removed.
5441
5442 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5443 were reported on the Macintosh. The changes were from
5444
5445 while ((c = *(++ptr)) != 0 && c != '\n');
5446 to
5447 while ((c = *(++ptr)) != 0 && c != '\n') ;
5448
5449 Totally extraordinary, but if that's what it takes...
5450
5451 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5452 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5453 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5454 assumes the way PCRE uses memmove() (always moving upwards).
5455
5456 6. PCRE is being used in one environment where strchr() is not available. There
5457 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5458 faster code anyway.
5459
5460
5461 Version 3.1 09-Feb-00
5462 ---------------------
5463
5464 The only change in this release is the fixing of some bugs in Makefile.in for
5465 the "install" target:
5466
5467 (1) It was failing to install pcreposix.h.
5468
5469 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5470
5471
5472 Version 3.0 01-Feb-00
5473 ---------------------
5474
5475 1. Add support for the /+ modifier to perltest (to output $` like it does in
5476 pcretest).
5477
5478 2. Add support for the /g modifier to perltest.
5479
5480 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5481 matches null strings.
5482
5483 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5484 pattern. Perl treats empty patterns specially - it reuses the most recent
5485 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5486 effect.
5487
5488 5. The POSIX interface was broken in that it was just handing over the POSIX
5489 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5490 required a bigger vector, with some working space on the end. This means that
5491 the POSIX wrapper now has to get and free some memory, and copy the results.
5492
5493 6. Added some simple autoconf support, placing the test data and the
5494 documentation in separate directories, re-organizing some of the
5495 information files, and making it build pcre-config (a GNU standard). Also added
5496 libtool support for building PCRE as a shared library, which is now the
5497 default.
5498
5499 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5500 09 are not valid octal constants. Single digits will be used for minor values
5501 less than 10.
5502
5503 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5504 existing programs that set these in the POSIX interface can use PCRE without
5505 modification.
5506
5507 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5508 return all that pcre_info() returns, plus additional data. The pcre_info()
5509 function is retained for compatibility, but is considered to be obsolete.
5510
5511 10. Added experimental recursion feature (?R) to handle one common case that
5512 Perl 5.6 will be able to do with (?p{...}).
5513
5514 11. Added support for POSIX character classes like [:alpha:], which Perl is
5515 adopting.
5516
5517
5518 Version 2.08 31-Aug-99
5519 ----------------------
5520
5521 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5522 trying to match at the startoffset position, but instead was moving forward to
5523 the next newline as if a previous match had failed.
5524
5525 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5526 and could get into a loop if a null string was matched other than at the start
5527 of the subject.
5528
5529 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5530 be distinguished at compile time, and for completeness also added PCRE_DATE.
5531
5532 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5533 in GnuWin32 environments.
5534
5535
5536 Version 2.07 29-Jul-99
5537 ----------------------
5538
5539 1. The documentation is now supplied in plain text form and HTML as well as in
5540 the form of man page sources.
5541
5542 2. C++ compilers don't like assigning (void *) values to other pointer types.
5543 In particular this affects malloc(). Although there is no problem in Standard
5544 C, I've put in casts to keep C++ compilers happy.
5545
5546 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5547 should be (const char *).
5548
5549 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5550 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5551 However, I haven't made this a standard facility. The documentation doesn't
5552 mention it, and the Makefile doesn't support it.
5553
5554 5. The Makefile now contains an "install" target, with editable destinations at
5555 the top of the file. The pcretest program is not installed.
5556
5557 6. pgrep -V now gives the PCRE version number and date.
5558
5559 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5560 causing the entire string to be ignored, instead of just the last character.
5561
5562 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5563 non-matching string, it can take a very, very long time, even for strings of
5564 quite modest length, because of the nested recursion. PCRE now does better in
5565 some of these cases. It does this by remembering the last required literal
5566 character in the pattern, and pre-searching the subject to ensure it is present
5567 before running the real match. In other words, it applies a heuristic to detect
5568 some types of certain failure quickly, and in the above example, if presented
5569 with a string that has no trailing " it gives "no match" very quickly.
5570
5571 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5572 other alternatives are tried instead.
5573
5574
5575 Version 2.06 09-Jun-99
5576 ----------------------
5577
5578 1. Change pcretest's output for amount of store used to show just the code
5579 space, because the remainder (the data block) varies in size between 32-bit and
5580 64-bit systems.
5581
5582 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5583 start matching at. This allows lookbehinds to work when searching for multiple
5584 occurrences in a string.
5585
5586 3. Added additional options to pcretest for testing multiple occurrences:
5587
5588 /+ outputs the rest of the string that follows a match
5589 /g loops for multiple occurrences, using the new startoffset argument
5590 /G loops for multiple occurrences by passing an incremented pointer
5591
5592 4. PCRE wasn't doing the "first character" optimization for patterns starting
5593 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5594 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5595 the letter 'x'. On long subject strings, this gives a significant speed-up.
5596
5597
5598 Version 2.05 21-Apr-99
5599 ----------------------
5600
5601 1. Changed the type of magic_number from int to long int so that it works
5602 properly on 16-bit systems.
5603
5604 2. Fixed a bug which caused patterns starting with .* not to work correctly
5605 when the subject string contained newline characters. PCRE was assuming
5606 anchoring for such patterns in all cases, which is not correct because .* will
5607 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5608 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5609 must be retried after every newline in the subject.
5610
5611
5612 Version 2.04 18-Feb-99
5613 ----------------------
5614
5615 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5616 computation of the store needed to hold the pattern was incorrect (too large).
5617 If such patterns were nested a few deep, this could multiply and become a real
5618 problem.
5619
5620 2. Added /M option to pcretest to show the memory requirement of a specific
5621 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5622
5623 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5624 compiled in such a way that the backtracking after subsequent failure was
5625 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5626 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5627
5628
5629 Version 2.03 02-Feb-99
5630 ----------------------
5631
5632 1. Fixed typo and small mistake in man page.
5633
5634 2. Added 4th condition (GPL supersedes if conflict) and created separate
5635 LICENCE file containing the conditions.
5636
5637 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5638 Perl, that is the internal \ allows the delimiter to be included in the
5639 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5640 the final delimiter, add \ to the end of the pattern (to test the error).
5641
5642 4. Added the convenience functions for extracting substrings after a successful
5643 match. Updated pcretest to make it able to test these functions.
5644
5645
5646 Version 2.02 14-Jan-99
5647 ----------------------
5648
5649 1. Initialized the working variables associated with each extraction so that
5650 their saving and restoring doesn't refer to uninitialized store.
5651
5652 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5653 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5654 fix the problem.
5655
5656 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5657 calls, and wasn't printing the correct value for compiling calls. Increased the
5658 default value of LOOPREPEAT, and the number of significant figures in the
5659 times.
5660
5661 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5662
5663 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5664 a building problem on Windows NT with a FAT file system.
5665
5666
5667 Version 2.01 21-Oct-98
5668 ----------------------
5669
5670 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5671 to character tables built by pcre_maketables() in the current locale. If NULL
5672 is passed, the default tables are used.
5673
5674
5675 Version 2.00 24-Sep-98
5676 ----------------------
5677
5678 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5679 it any more.
5680
5681 2. Allow quantification of (?>) groups, and make it work correctly.
5682
5683 3. The first character computation wasn't working for (?>) groups.
5684
5685 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5686 end of the subject) and add 5.005's \z, which really does match only at the
5687 very end of the subject.
5688
5689 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5690
5691 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5692 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5693 localized options. All options to pcre_study() were also removed.
5694
5695 7. Add other new features from 5.005:
5696
5697 $(?<= positive lookbehind
5698 $(?<! negative lookbehind
5699 (?imsx-imsx) added the unsetting capability
5700 such a setting is global if at outer level; local otherwise
5701 (?imsx-imsx:) non-capturing groups with option setting
5702 (?(cond)re|re) conditional pattern matching
5703
5704 A backreference to itself in a repeated group matches the previous
5705 captured string.
5706
5707 8. General tidying up of studying (both automatic and via "study")
5708 consequential on the addition of new assertions.
5709
5710 9. As in 5.005, unlimited repeated groups that could match an empty substring
5711 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5712 runtime if any iteration does actually match an empty substring.
5713
5714 10. Include the RunTest script in the distribution.
5715
5716 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5717 discrepancies, some of which were old and were also with respect to 5.004. They
5718 have now been fixed.
5719
5720
5721 Version 1.09 28-Apr-98
5722 ----------------------
5723
5724 1. A negated single character class followed by a quantifier with a minimum
5725 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5726 program crashes, or just wrong answers. This did not apply to negated classes
5727 containing more than one character, or to minima other than one.
5728
5729
5730 Version 1.08 27-Mar-98
5731 ----------------------
5732
5733 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5734
5735 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5736 latter must appear before anything that relies on it in the pattern.
5737
5738
5739 Version 1.07 16-Feb-98
5740 ----------------------
5741
5742 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5743 repeat of a potentially empty string).
5744
5745
5746 Version 1.06 23-Jan-98
5747 ----------------------
5748
5749 1. Added Markus Oberhumer's little patches for C++.
5750
5751 2. Literal strings longer than 255 characters were broken.
5752
5753
5754 Version 1.05 23-Dec-97
5755 ----------------------
5756
5757 1. Negated character classes containing more than one character were failing if
5758 PCRE_CASELESS was set at run time.
5759
5760
5761 Version 1.04 19-Dec-97
5762 ----------------------
5763
5764 1. Corrected the man page, where some "const" qualifiers had been omitted.
5765
5766 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5767 input syntax.
5768
5769 3. Fixed memory leak which occurred when a regex with back references was
5770 matched with an offsets vector that wasn't big enough. The temporary memory
5771 that is used in this case wasn't being freed if the match failed.
5772
5773 4. Tidied pcretest to ensure it frees memory that it gets.
5774
5775 5. Temporary memory was being obtained in the case where the passed offsets
5776 vector was exactly big enough.
5777
5778 6. Corrected definition of offsetof() from change 5 below.
5779
5780 7. I had screwed up change 6 below and broken the rules for the use of
5781 setjmp(). Now fixed.
5782
5783
5784 Version 1.03 18-Dec-97
5785 ----------------------
5786
5787 1. A erroneous regex with a missing opening parenthesis was correctly
5788 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5789 on some systems.
5790
5791 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5792 it was reported that one broken compiler failed on the former because "code" is
5793 also an independent variable.
5794
5795 3. The erroneous regex a[]b caused an array overrun reference.
5796
5797 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5798 fail on data ending with that character. (It was going on too far, and checking
5799 the next character, typically a binary zero.) This was specific to the
5800 optimized code for single-character negative classes.
5801
5802 5. Added a contributed patch from the TIN world which does the following:
5803
5804 + Add an undef for memmove, in case the the system defines a macro for it.
5805
5806 + Add a definition of offsetof(), in case there isn't one. (I don't know
5807 the reason behind this - offsetof() is part of the ANSI standard - but
5808 it does no harm).
5809
5810 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5811 most of the places where whitespace preceded '#'. I have given up and
5812 allowed the remaining 2 cases to be at the margin.
5813
5814 + Rename some variables in pcre to eliminate shadowing. This seems very
5815 pedantic, but does no harm, of course.
5816
5817 6. Moved the call to setjmp() into its own function, to get rid of warnings
5818 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5819
5820 7. Constructs such as \d{8,} were compiling into the equivalent of
5821 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5822 outcome, but in this particular case used more store than had been allocated,
5823 which caused the bug to be discovered because it threw up an internal error.
5824
5825 8. The debugging code in both pcre and pcretest for outputting the compiled
5826 form of a regex was going wrong in the case of back references followed by
5827 curly-bracketed repeats.
5828
5829
5830 Version 1.02 12-Dec-97
5831 ----------------------
5832
5833 1. Typos in pcre.3 and comments in the source fixed.
5834
5835 2. Applied a contributed patch to get rid of places where it used to remove
5836 'const' from variables, and fixed some signed/unsigned and uninitialized
5837 variable warnings.
5838
5839 3. Added the "runtest" target to Makefile.
5840
5841 4. Set default compiler flag to -O2 rather than just -O.
5842
5843
5844 Version 1.01 19-Nov-97
5845 ----------------------
5846
5847 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5848 like /([ab]*)*/, that is, for classes with more than one character in them.
5849
5850 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5851 as /((?>a*))*/ (a PCRE_EXTRA facility).
5852
5853
5854 Version 1.00 18-Nov-97
5855 ----------------------
5856
5857 1. Added compile-time macros to support systems such as SunOS4 which don't have
5858 memmove() or strerror() but have other things that can be used instead.
5859
5860 2. Arranged that "make clean" removes the executables.
5861
5862
5863 Version 0.99 27-Oct-97
5864 ----------------------
5865
5866 1. Fixed bug in code for optimizing classes with only one character. It was
5867 initializing a 32-byte map regardless, which could cause it to run off the end
5868 of the memory it had got.
5869
5870 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5871
5872
5873 Version 0.98 22-Oct-97
5874 ----------------------
5875
5876 1. Fixed bug in code for handling temporary memory usage when there are more
5877 back references than supplied space in the ovector. This could cause segfaults.
5878
5879
5880 Version 0.97 21-Oct-97
5881 ----------------------
5882
5883 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5884
5885 2. Optimized negated single characters not to use a bit map.
5886
5887 3. Brought error texts together as macro definitions; clarified some of them;
5888 fixed one that was wrong - it said "range out of order" when it meant "invalid
5889 escape sequence".
5890
5891 4. Changed some char * arguments to const char *.
5892
5893 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5894
5895 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5896 pcretest.
5897
5898
5899 Version 0.96 16-Oct-97
5900 ----------------------
5901
5902 1. Added a simple "pgrep" utility to the distribution.
5903
5904 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5905 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5906 where "ddd" means "one or more decimal digits".
5907
5908 3. Fixed serious bug. If a pattern had a back reference, but the call to
5909 pcre_exec() didn't supply a large enough ovector to record the related
5910 identifying subpattern, the match always failed. PCRE now remembers the number
5911 of the largest back reference, and gets some temporary memory in which to save
5912 the offsets during matching if necessary, in order to ensure that
5913 backreferences always work.
5914
5915 4. Increased the compatibility with Perl in a number of ways:
5916
5917 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5918 to request this handling. The option can be set at compile or exec time.
5919
5920 (b) $ matches before a terminating newline by default; an option
5921 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5922 mode). The option can be set at compile or exec time.
5923
5924 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5925 the same as Perl's. If the decimal number it represents is less than 10
5926 or there aren't that many previous left capturing parentheses, an octal
5927 escape is read. Inside a character class, it's always an octal escape,
5928 even if it is a single digit.
5929
5930 (d) An escaped but undefined alphabetic character is taken as a literal,
5931 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5932 escapes.
5933
5934 (e) {0} is now permitted. (The previous item is removed from the compiled
5935 pattern).
5936
5937 5. Changed all the names of code files so that the basic parts are no longer
5938 than 10 characters, and abolished the teeny "globals.c" file.
5939
5940 6. Changed the handling of character classes; they are now done with a 32-byte
5941 bit map always.
5942
5943 7. Added the -d and /D options to pcretest to make it possible to look at the
5944 internals of compilation without having to recompile pcre.
5945
5946
5947 Version 0.95 23-Sep-97
5948 ----------------------
5949
5950 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5951 \x20 at the start of a run of normal characters. These were being treated as
5952 real characters, instead of the source characters being re-checked.
5953
5954
5955 Version 0.94 18-Sep-97
5956 ----------------------
5957
5958 1. The functions are now thread-safe, with the caveat that the global variables
5959 containing pointers to malloc() and free() or alternative functions are the
5960 same for all threads.
5961
5962 2. Get pcre_study() to generate a bitmap of initial characters for non-
5963 anchored patterns when this is possible, and use it if passed to pcre_exec().
5964
5965
5966 Version 0.93 15-Sep-97
5967 ----------------------
5968
5969 1. /(b)|(:+)/ was computing an incorrect first character.
5970
5971 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5972 but not actually doing anything yet.
5973
5974 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5975 as Perl does (e.g. [-az] or [az-]).
5976
5977 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5978 all possible positions.
5979
5980 5. Split up into different modules to avoid including unneeded functions in a
5981 compiled binary. However, compile and exec are still in one module. The "study"
5982 function is split off.
5983
5984 6. The character tables are now in a separate module whose source is generated
5985 by an auxiliary program - but can then be edited by hand if required. There are
5986 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5987 toupper() in the code.
5988
5989 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5990 make them global. Abolish the function for setting them, as the caller can now
5991 set them directly.
5992
5993
5994 Version 0.92 11-Sep-97
5995 ----------------------
5996
5997 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5998 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5999
6000 2. Caseless matching was not working in character classes if the characters in
6001 the pattern were in upper case.
6002
6003 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
6004
6005 4. Make PCRE_ANCHORED public and accept as a compile option.
6006
6007 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
6008 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
6009 pass them.
6010
6011 6. Give an error if bad option bits passed at compile or run time.
6012
6013 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
6014 pcretest to cause it to pass that flag.
6015
6016 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
6017 options, and the first character, if set.
6018
6019 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
6020
6021
6022 Version 0.91 10-Sep-97
6023 ----------------------
6024
6025 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
6026 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
6027
6028 2. PCRE was looping on encountering an indefinitely repeated back reference to
6029 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
6030 Perl does - treats the match as successful.
6031
6032 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5