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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1662 - (show annotations)
Fri Jul 15 09:57:53 2016 UTC (9 days, 4 hours ago) by zherczeg
File size: 290005 byte(s)
Error occurred while calculating annotation data.
Fix unaligned accesses on x86. Patch by Marc Mutz.
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 17-June-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 unaligned accesses on x86. Patch by Marc Mutz.
21
22
23 Version 8.39 14-June-2016
24 -------------------------
25
26 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
27 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
28 misbehaved. This bug was found by the LLVM fuzzer.
29
30 2. Similar to the above, if an isolated \E was present between an item and its
31 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
32 bug was found by the LLVM fuzzer.
33
34 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
35 working correctly in UCP mode.
36
37 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
38 was set when the pmatch argument was NULL. It now returns REG_INVARG.
39
40 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
41
42 6. An empty \Q\E sequence between an item and its qualifier caused
43 pcre_compile() to misbehave when auto callouts were enabled. This bug was
44 found by the LLVM fuzzer.
45
46 7. If a pattern that was compiled with PCRE_EXTENDED started with white
47 space or a #-type comment that was followed by (?-x), which turns off
48 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
49 pcre_compile() assumed that (?-x) applied to the whole pattern and
50 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
51
52 8. A call of pcre_copy_named_substring() for a named substring whose number
53 was greater than the space in the ovector could cause a crash.
54
55 9. Yet another buffer overflow bug involved duplicate named groups with a
56 group that reset capture numbers (compare 8.38/7 below). Once again, I have
57 just allowed for more memory, even if not needed. (A proper fix is
58 implemented in PCRE2, but it involves a lot of refactoring.)
59
60 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
61 start of the match to be earlier than the end.
62
63 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
64
65 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
66 assertion, caused pcretest to generate incorrect output, and also to read
67 uninitialized memory (detected by ASAN or valgrind).
68
69 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
70 nested set of parentheses of sufficient size caused an overflow of the
71 compiling workspace (which was diagnosed, but of course is not desirable).
72
73 14. And yet another buffer overflow bug involving duplicate named groups, this
74 time nested, with a nested back reference. Yet again, I have just allowed
75 for more memory, because anything more needs all the refactoring that has
76 been done for PCRE2. An example pattern that provoked this bug is:
77 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
78 registered as CVE-2016-1283.
79
80 15. pcretest went into a loop if global matching was requested with an ovector
81 size less than 2. It now gives an error message. This bug was found by
82 afl-fuzz.
83
84 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
85 ("assertion expected") when (?(?C) was not followed by an opening
86 parenthesis.
87
88 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
89 actually affect anything, by sheer luck.
90
91 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
92 static compilation.
93
94 19. Modified the RunTest script to incorporate a valgrind suppressions file so
95 that certain errors, provoked by the SSE2 instruction set when JIT is used,
96 are ignored.
97
98 20. A racing condition is fixed in JIT reported by Mozilla.
99
100 21. Minor code refactor to avoid "array subscript is below array bounds"
101 compiler warning.
102
103 22. Minor code refactor to avoid "left shift of negative number" warning.
104
105 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
106 UCP support.
107
108 24. Refactor to avoid compiler warnings in pcrecpp.cc.
109
110 25. Refactor to fix a typo in pcre_jit_test.c
111
112 26. Patch to support compiling pcrecpp.cc with Intel compiler.
113
114
115 Version 8.38 23-November-2015
116 -----------------------------
117
118 1. If a group that contained a recursive back reference also contained a
119 forward reference subroutine call followed by a non-forward-reference
120 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
121 compile correct code, leading to undefined behaviour or an internally
122 detected error. This bug was discovered by the LLVM fuzzer.
123
124 2. Quantification of certain items (e.g. atomic back references) could cause
125 incorrect code to be compiled when recursive forward references were
126 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
127 This bug was discovered by the LLVM fuzzer.
128
129 3. A repeated conditional group whose condition was a reference by name caused
130 a buffer overflow if there was more than one group with the given name.
131 This bug was discovered by the LLVM fuzzer.
132
133 4. A recursive back reference by name within a group that had the same name as
134 another group caused a buffer overflow. For example:
135 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
136
137 5. A forward reference by name to a group whose number is the same as the
138 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
139 a buffer overflow at compile time. This bug was discovered by the LLVM
140 fuzzer.
141
142 6. A lookbehind assertion within a set of mutually recursive subpatterns could
143 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
144
145 7. Another buffer overflow bug involved duplicate named groups with a
146 reference between their definition, with a group that reset capture
147 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
148 fixed by always allowing for more memory, even if not needed. (A proper fix
149 is implemented in PCRE2, but it involves more refactoring.)
150
151 8. There was no check for integer overflow in subroutine calls such as (?123).
152
153 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
154 being treated as a literal 'l' instead of causing an error.
155
156 10. There was a buffer overflow if pcre_exec() was called with an ovector of
157 size 1. This bug was found by american fuzzy lop.
158
159 11. If a non-capturing group containing a conditional group that could match
160 an empty string was repeated, it was not identified as matching an empty
161 string itself. For example: /^(?:(?(1)x|)+)+$()/.
162
163 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
164 \a and \e in test subject lines.
165
166 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
167 instead of the EBCDIC value.
168
169 14. The handling of \c in an EBCDIC environment has been revised so that it is
170 now compatible with the specification in Perl's perlebcdic page.
171
172 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
173 ASCII/Unicode. This has now been added to the list of characters that are
174 recognized as white space in EBCDIC.
175
176 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
177 error (correctly) when used outside a class, but did not give an error
178 within a class.
179
180 17. \h within a class was incorrectly compiled in EBCDIC environments.
181
182 18. A pattern with an unmatched closing parenthesis that contained a backward
183 assertion which itself contained a forward reference caused buffer
184 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
185
186 19. JIT should return with error when the compiled pattern requires more stack
187 space than the maximum.
188
189 20. A possessively repeated conditional group that could match an empty string,
190 for example, /(?(R))*+/, was incorrectly compiled.
191
192 21. Fix infinite recursion in the JIT compiler when certain patterns such as
193 /(?:|a|){100}x/ are analysed.
194
195 22. Some patterns with character classes involving [: and \\ were incorrectly
196 compiled and could cause reading from uninitialized memory or an incorrect
197 error diagnosis.
198
199 23. Pathological patterns containing many nested occurrences of [: caused
200 pcre_compile() to run for a very long time.
201
202 24. A conditional group with only one branch has an implicit empty alternative
203 branch and must therefore be treated as potentially matching an empty
204 string.
205
206 25. If (?R was followed by - or + incorrect behaviour happened instead of a
207 diagnostic.
208
209 26. Arrange to give up on finding the minimum matching length for overly
210 complex patterns.
211
212 27. Similar to (4) above: in a pattern with duplicated named groups and an
213 occurrence of (?| it is possible for an apparently non-recursive back
214 reference to become recursive if a later named group with the relevant
215 number is encountered. This could lead to a buffer overflow. Wen Guanxing
216 from Venustech ADLAB discovered this bug.
217
218 28. If pcregrep was given the -q option with -c or -l, or when handling a
219 binary file, it incorrectly wrote output to stdout.
220
221 29. The JIT compiler did not restore the control verb head in case of *THEN
222 control verbs. This issue was found by Karl Skomski with a custom LLVM
223 fuzzer.
224
225 30. Error messages for syntax errors following \g and \k were giving inaccurate
226 offsets in the pattern.
227
228 31. Added a check for integer overflow in conditions (?(<digits>) and
229 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
230 fuzzer.
231
232 32. Handling recursive references such as (?2) when the reference is to a group
233 later in the pattern uses code that is very hacked about and error-prone.
234 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
235 give an internal error if it is obvious that compiling has gone wrong.
236
237 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
238 This issue was found by Karl Skomski with a custom LLVM fuzzer.
239
240 34. The JIT compiler should restore the control chain for empty possessive
241 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
242
243 35. Match limit check added to JIT recursion. This issue was found by Karl
244 Skomski with a custom LLVM fuzzer.
245
246 36. Yet another case similar to 27 above has been circumvented by an
247 unconditional allocation of extra memory. This issue is fixed "properly" in
248 PCRE2 by refactoring the way references are handled. Wen Guanxing
249 from Venustech ADLAB discovered this bug.
250
251 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
252 with a custom LLVM fuzzer.
253
254 38. Fixed a corner case of range optimization in JIT.
255
256 39. An incorrect error "overran compiling workspace" was given if there were
257 exactly enough group forward references such that the last one extended
258 into the workspace safety margin. The next one would have expanded the
259 workspace. The test for overflow was not including the safety margin.
260
261 40. A match limit issue is fixed in JIT which was found by Karl Skomski
262 with a custom LLVM fuzzer.
263
264 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
265 work under Windows. (Why has it taken so long for anyone to notice?)
266
267 42. In a character class such as [\W\p{Any}] where both a negative-type escape
268 ("not a word character") and a property escape were present, the property
269 escape was being ignored.
270
271 43. Fix crash caused by very long (*MARK) or (*THEN) names.
272
273 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
274 by a single ASCII character in a class item, was incorrectly compiled in
275 UCP mode. The POSIX class got lost, but only if the single character
276 followed it.
277
278 45. [:punct:] in UCP mode was matching some characters in the range 128-255
279 that should not have been matched.
280
281 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
282 class, all characters with code points greater than 255 are in the class.
283 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
284 such as \w are turned into Unicode properties), wide characters were not
285 correctly handled, and could fail to match.
286
287
288 Version 8.37 28-April-2015
289 --------------------------
290
291 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
292 for those parentheses to be closed with whatever has been captured so far.
293 However, it was failing to mark any other groups between the hightest
294 capture so far and the currrent group as "unset". Thus, the ovector for
295 those groups contained whatever was previously there. An example is the
296 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
297
298 2. If an assertion condition was quantified with a minimum of zero (an odd
299 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
300
301 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
302 unrecognized modifier, a crash could occur.
303
304 4. An attempt to do global matching in pcretest with a zero-length ovector
305 caused a crash.
306
307 5. Fixed a memory leak during matching that could occur for a subpattern
308 subroutine call (recursive or otherwise) if the number of captured groups
309 that had to be saved was greater than ten.
310
311 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
312 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
313 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
314
315 7. A UTF pattern containing a "not" match of a non-ASCII character and a
316 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
317
318 8. When a pattern is compiled, it remembers the highest back reference so that
319 when matching, if the ovector is too small, extra memory can be obtained to
320 use instead. A conditional subpattern whose condition is a check on a
321 capture having happened, such as, for example in the pattern
322 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
323 setting the highest backreference number. This mattered only if pcre_exec()
324 was called with an ovector that was too small to hold the capture, and there
325 was no other kind of back reference (a situation which is probably quite
326 rare). The effect of the bug was that the condition was always treated as
327 FALSE when the capture could not be consulted, leading to a incorrect
328 behaviour by pcre_exec(). This bug has been fixed.
329
330 9. A reference to a duplicated named group (either a back reference or a test
331 for being set in a conditional) that occurred in a part of the pattern where
332 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
333 to be incorrectly calculated, leading to overwriting.
334
335 10. A mutually recursive set of back references such as (\2)(\1) caused a
336 segfault at study time (while trying to find the minimum matching length).
337 The infinite loop is now broken (with the minimum length unset, that is,
338 zero).
339
340 11. If an assertion that was used as a condition was quantified with a minimum
341 of zero, matching went wrong. In particular, if the whole group had
342 unlimited repetition and could match an empty string, a segfault was
343 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
344 assertions to be quantified, but not if they are being used as conditions,
345 so the above pattern is faulted by Perl. PCRE has now been changed so that
346 it also rejects such patterns.
347
348 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
349 failed to allow the zero-repeat case if pcre2_exec() was called with an
350 ovector too small to capture the group.
351
352 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
353 Red Hat Product Security:
354
355 (a) A crash if /K and /F were both set with the option to save the compiled
356 pattern.
357
358 (b) Another crash if the option to print captured substrings in a callout
359 was combined with setting a null ovector, for example \O\C+ as a subject
360 string.
361
362 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
363 forward reference repeated a large (but limited) number of times within a
364 repeated outer group that has a zero minimum quantifier, caused incorrect
365 code to be compiled, leading to the error "internal error:
366 previously-checked referenced subpattern not found" when an incorrect
367 memory address was read. This bug was reported as "heap overflow",
368 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
369 CVE-2015-2325.
370
371 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
372 call within a group that also contained a recursive back reference caused
373 incorrect code to be compiled. This bug was reported as "heap overflow",
374 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
375 number CVE-2015-2326.
376
377 24. Computing the size of the JIT read-only data in advance has been a source
378 of various issues, and new ones are still appear unfortunately. To fix
379 existing and future issues, size computation is eliminated from the code,
380 and replaced by on-demand memory allocation.
381
382 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
383 adjacent to the end of the range, and the range contained characters with
384 more than one other case, caused incorrect behaviour when compiled in UTF
385 mode. In that example, the range a-j was left out of the class.
386
387 26. Fix JIT compilation of conditional blocks, which assertion
388 is converted to (*FAIL). E.g: /(?(?!))/.
389
390 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
391 discovered by the LLVM fuzzer.
392
393 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
394 when this assertion was used as a condition, for example (?(?!)a|b). In
395 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
396 error about an unsupported item.
397
398 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
399 possessification code could take exponential time to complete. A recursion
400 depth limit of 1000 has been imposed to limit the resources used by this
401 optimization.
402
403 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
404 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
405 because \S ensures they are all in the class. The code for doing this was
406 interacting badly with the code for computing the amount of space needed to
407 compile the pattern, leading to a buffer overflow. This bug was discovered
408 by the LLVM fuzzer.
409
410 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
411 other kinds of group caused stack overflow at compile time. This bug was
412 discovered by the LLVM fuzzer.
413
414 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
415 between a subroutine call and its quantifier was incorrectly compiled,
416 leading to buffer overflow or other errors. This bug was discovered by the
417 LLVM fuzzer.
418
419 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
420 assertion after (?(. The code was failing to check the character after
421 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
422 was discovered by the LLVM fuzzer.
423
424 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
425 a fixed maximum following a group that contains a subroutine reference was
426 incorrectly compiled and could trigger buffer overflow. This bug was
427 discovered by the LLVM fuzzer.
428
429 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
430 caused a stack overflow instead of the diagnosis of a non-fixed length
431 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
432
433 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
434 (e.g. /(?<=\Ka)/) could make pcregrep loop.
435
436 37. There was a similar problem to 36 in pcretest for global matches.
437
438 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
439 and a subsequent item in the pattern caused a non-match, backtracking over
440 the repeated \X did not stop, but carried on past the start of the subject,
441 causing reference to random memory and/or a segfault. There were also some
442 other cases where backtracking after \C could crash. This set of bugs was
443 discovered by the LLVM fuzzer.
444
445 39. The function for finding the minimum length of a matching string could take
446 a very long time if mutual recursion was present many times in a pattern,
447 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
448 method has been implemented. This infelicity was discovered by the LLVM
449 fuzzer.
450
451 40. Static linking against the PCRE library using the pkg-config module was
452 failing on missing pthread symbols.
453
454
455 Version 8.36 26-September-2014
456 ------------------------------
457
458 1. Got rid of some compiler warnings in the C++ modules that were shown up by
459 -Wmissing-field-initializers and -Wunused-parameter.
460
461 2. The tests for quantifiers being too big (greater than 65535) were being
462 applied after reading the number, and stupidly assuming that integer
463 overflow would give a negative number. The tests are now applied as the
464 numbers are read.
465
466 3. Tidy code in pcre_exec.c where two branches that used to be different are
467 now the same.
468
469 4. The JIT compiler did not generate match limit checks for certain
470 bracketed expressions with quantifiers. This may lead to exponential
471 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
472 issue should be resolved now.
473
474 5. Fixed an issue, which occures when nested alternatives are optimized
475 with table jumps.
476
477 6. Inserted two casts and changed some ints to size_t in the light of some
478 reported 64-bit compiler warnings (Bugzilla 1477).
479
480 7. Fixed a bug concerned with zero-minimum possessive groups that could match
481 an empty string, which sometimes were behaving incorrectly in the
482 interpreter (though correctly in the JIT matcher). This pcretest input is
483 an example:
484
485 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
486 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
487
488 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
489 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
490 for an empty string was breaking the inner loop and carrying on at a lower
491 level, when possessive repeated groups should always return to a higher
492 level as they have no backtrack points in them. The empty string test now
493 occurs at the outer level.
494
495 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
496 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
497
498 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
499 doesn't).
500
501 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
502 to hold a pattern to be incorrectly computed (too small) when there were
503 named back references to duplicated names. This could cause "internal
504 error: code overflow" or "double free or corruption" or other memory
505 handling errors.
506
507 11. When named subpatterns had the same prefixes, back references could be
508 confused. For example, in this pattern:
509
510 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
511
512 the reference to 'Name' was incorrectly treated as a reference to a
513 duplicate name.
514
515 12. A pattern such as /^s?c/mi8 where the optional character has more than
516 one "other case" was incorrectly compiled such that it would only try to
517 match starting at "c".
518
519 13. When a pattern starting with \s was studied, VT was not included in the
520 list of possible starting characters; this should have been part of the
521 8.34/18 patch.
522
523 14. If a character class started [\Qx]... where x is any character, the class
524 was incorrectly terminated at the ].
525
526 15. If a pattern that started with a caseless match for a character with more
527 than one "other case" was studied, PCRE did not set up the starting code
528 unit bit map for the list of possible characters. Now it does. This is an
529 optimization improvement, not a bug fix.
530
531 16. The Unicode data tables have been updated to Unicode 7.0.0.
532
533 17. Fixed a number of memory leaks in pcregrep.
534
535 18. Avoid a compiler warning (from some compilers) for a function call with
536 a cast that removes "const" from an lvalue by using an intermediate
537 variable (to which the compiler does not object).
538
539 19. Incorrect code was compiled if a group that contained an internal recursive
540 back reference was optional (had quantifier with a minimum of zero). This
541 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
542 caused segmentation faults because of stack overflows at compile time.
543
544 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
545 group that is quantified with an indefinite repeat, caused a compile-time
546 loop which used up all the system stack and provoked a segmentation fault.
547 This was not the same bug as 19 above.
548
549 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
550 Patch by Mike Frysinger.
551
552
553 Version 8.35 04-April-2014
554 --------------------------
555
556 1. A new flag is set, when property checks are present in an XCLASS.
557 When this flag is not set, PCRE can perform certain optimizations
558 such as studying these XCLASS-es.
559
560 2. The auto-possessification of character sets were improved: a normal
561 and an extended character set can be compared now. Furthermore
562 the JIT compiler optimizes more character set checks.
563
564 3. Got rid of some compiler warnings for potentially uninitialized variables
565 that show up only when compiled with -O2.
566
567 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
568 of a match later then the end of the match. The pcretest program was not
569 handling the case sensibly - it was outputting from the start to the next
570 binary zero. It now reports this situation in a message, and outputs the
571 text from the end to the start.
572
573 5. Fast forward search is improved in JIT. Instead of the first three
574 characters, any three characters with fixed position can be searched.
575 Search order: first, last, middle.
576
577 6. Improve character range checks in JIT. Characters are read by an inprecise
578 function now, which returns with an unknown value if the character code is
579 above a certain threshold (e.g: 256). The only limitation is that the value
580 must be bigger than the threshold as well. This function is useful when
581 the characters above the threshold are handled in the same way.
582
583 7. The macros whose names start with RAWUCHAR are placeholders for a future
584 mode in which only the bottom 21 bits of 32-bit data items are used. To
585 make this more memorable for those maintaining the code, the names have
586 been changed to start with UCHAR21, and an extensive comment has been added
587 to their definition.
588
589 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
590 to the export list in Makefile.am (they were accidentally omitted from the
591 8.34 tarball).
592
593 9. The informational output from pcretest used the phrase "starting byte set"
594 which is inappropriate for the 16-bit and 32-bit libraries. As the output
595 for "first char" and "need char" really means "non-UTF-char", I've changed
596 "byte" to "char", and slightly reworded the output. The documentation about
597 these values has also been (I hope) clarified.
598
599 10. Another JIT related optimization: use table jumps for selecting the correct
600 backtracking path, when more than four alternatives are present inside a
601 bracket.
602
603 11. Empty match is not possible, when the minimum length is greater than zero,
604 and there is no \K in the pattern. JIT should avoid empty match checks in
605 such cases.
606
607 12. In a caseless character class with UCP support, when a character with more
608 than one alternative case was not the first character of a range, not all
609 the alternative cases were added to the class. For example, s and \x{17f}
610 are both alternative cases for S: the class [RST] was handled correctly,
611 but [R-T] was not.
612
613 13. The configure.ac file always checked for pthread support when JIT was
614 enabled. This is not used in Windows, so I have put this test inside a
615 check for the presence of windows.h (which was already tested for).
616
617 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
618 The algorithm provides a way to skip certain starting offsets, and usually
619 faster than linear prefix searches.
620
621 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
622 as for 'fr_FR' and 'french'. For some reason, however, it then used the
623 Windows-specific input and output files, which have 'french' screwed in.
624 So this could never have worked. One of the problems with locales is that
625 they aren't always the same. I have now updated RunTest so that it checks
626 the output of the locale test (test 3) against three different output
627 files, and it allows the test to pass if any one of them matches. With luck
628 this should make the test pass on some versions of Solaris where it was
629 failing. Because of the uncertainty, the script did not used to stop if
630 test 3 failed; it now does. If further versions of a French locale ever
631 come to light, they can now easily be added.
632
633 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
634 there was a message during ./configure, but it did not stop. This now
635 provokes an error. The invalid example in README has been corrected.
636 If a value less than the minimum is given, the minimum value has always
637 been used, but now a warning is given.
638
639 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
640 was a bug in the test system, which is now fixed. Also, the list of various
641 configurations that are tested for each release did not have one with both
642 16/32 bits and --enable-bar-anycrlf. It now does.
643
644 18. pcretest was missing "-C bsr" for displaying the \R default setting.
645
646 19. Little endian PowerPC systems are supported now by the JIT compiler.
647
648 20. The fast forward newline mechanism could enter to an infinite loop on
649 certain invalid UTF-8 input. Although we don't support these cases
650 this issue can be fixed by a performance optimization.
651
652 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
653 not take account if existing stack usage. There is now a new global
654 variable called pcre_stack_guard that can be set to point to an external
655 function to check stack availability. It is called at the start of
656 processing every parenthesized group.
657
658 22. A typo in the code meant that in ungreedy mode the max/min qualifier
659 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
660 not match "ab".
661
662 23. When UTF was disabled, the JIT program reported some incorrect compile
663 errors. These messages are silenced now.
664
665 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
666 compiler.
667
668 25. Change all the temporary files used in RunGrepTest to be different to those
669 used by RunTest so that the tests can be run simultaneously, for example by
670 "make -j check".
671
672
673 Version 8.34 15-December-2013
674 -----------------------------
675
676 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
677 executable memory. Patch inspired by Carsten Klein.
678
679 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
680 this macro is never tested and has no effect, because the work to support
681 coverage involves only compiling and linking options and special targets in
682 the Makefile. The comment in config.h implied that defining the macro would
683 enable coverage support, which is totally false. There was also support for
684 setting this macro in the CMake files (my fault, I just copied it from
685 configure). SUPPORT_GCOV has now been removed.
686
687 3. Make a small performance improvement in strlen16() and strlen32() in
688 pcretest.
689
690 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
691 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
692 these cases). There was also one in pcretest.c.
693
694 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
695
696 6. In UTF mode, the code for checking whether a group could match an empty
697 string (which is used for indefinitely repeated groups to allow for
698 breaking an infinite loop) was broken when the group contained a repeated
699 negated single-character class with a character that occupied more than one
700 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
701 UTF-8 mode). The effect was undefined: the group might or might not be
702 deemed as matching an empty string, or the program might have crashed.
703
704 7. The code for checking whether a group could match an empty string was not
705 recognizing that \h, \H, \v, \V, and \R must match a character.
706
707 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
708 an empty string. If it can, pcretest shows this in its information output.
709
710 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
711 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
712 matched by pcre_exec() without using JIT:
713
714 (a) If the rest of the pattern did not match after a maximal run of
715 grapheme clusters, the code for backing up to try with fewer of them
716 did not always back up over a full grapheme when characters that do not
717 have the modifier quality were involved, e.g. Hangul syllables.
718
719 (b) If the match point in a subject started with modifier character, and
720 there was no match, the code could incorrectly back up beyond the match
721 point, and potentially beyond the first character in the subject,
722 leading to a segfault or an incorrect match result.
723
724 10. A conditional group with an assertion condition could lead to PCRE
725 recording an incorrect first data item for a match if no other first data
726 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
727 first data item, and therefore matched "ca" after "c" instead of at the
728 start.
729
730 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
731 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
732
733 12. The source of pcregrep now includes z/OS-specific code so that it can be
734 compiled for z/OS as part of the special z/OS distribution.
735
736 13. Added the -T and -TM options to pcretest.
737
738 14. The code in pcre_compile.c for creating the table of named capturing groups
739 has been refactored. Instead of creating the table dynamically during the
740 actual compiling pass, the information is remembered during the pre-compile
741 pass (on the stack unless there are more than 20 named groups, in which
742 case malloc() is used) and the whole table is created before the actual
743 compile happens. This has simplified the code (it is now nearly 150 lines
744 shorter) and prepared the way for better handling of references to groups
745 with duplicate names.
746
747 15. A back reference to a named subpattern when there is more than one of the
748 same name now checks them in the order in which they appear in the pattern.
749 The first one that is set is used for the reference. Previously only the
750 first one was inspected. This change makes PCRE more compatible with Perl.
751
752 16. Unicode character properties were updated from Unicode 6.3.0.
753
754 17. The compile-time code for auto-possessification has been refactored, based
755 on a patch by Zoltan Herczeg. It now happens after instead of during
756 compilation. The code is cleaner, and more cases are handled. The option
757 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
758 options in pcretest are provided to set it. It can also be set by
759 (*NO_AUTO_POSSESS) at the start of a pattern.
760
761 18. The character VT has been added to the default ("C" locale) set of
762 characters that match \s and are generally treated as white space,
763 following this same change in Perl 5.18. There is now no difference between
764 "Perl space" and "POSIX space". Whether VT is treated as white space in
765 other locales depends on the locale.
766
767 19. The code for checking named groups as conditions, either for being set or
768 for being recursed, has been refactored (this is related to 14 and 15
769 above). Processing unduplicated named groups should now be as fast at
770 numerical groups, and processing duplicated groups should be faster than
771 before.
772
773 20. Two patches to the CMake build system, by Alexander Barkov:
774
775 (1) Replace the "source" command by "." in CMakeLists.txt because
776 "source" is a bash-ism.
777
778 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
779 without these the CMake build does not work on Solaris.
780
781 21. Perl has changed its handling of \8 and \9. If there is no previously
782 encountered capturing group of those numbers, they are treated as the
783 literal characters 8 and 9 instead of a binary zero followed by the
784 literals. PCRE now does the same.
785
786 22. Following Perl, added \o{} to specify codepoints in octal, making it
787 possible to specify values greater than 0777 and also making them
788 unambiguous.
789
790 23. Perl now gives an error for missing closing braces after \x{... instead of
791 treating the string as literal. PCRE now does the same.
792
793 24. RunTest used to grumble if an inappropriate test was selected explicitly,
794 but just skip it when running all tests. This make it awkward to run ranges
795 of tests when one of them was inappropriate. Now it just skips any
796 inappropriate tests, as it always did when running all tests.
797
798 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
799 character types such as \d or \w, too many callouts were inserted, and the
800 data that they returned was rubbish.
801
802 26. In UCP mode, \s was not matching two of the characters that Perl matches,
803 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
804 were matched by \h. The code has now been refactored so that the lists of
805 the horizontal and vertical whitespace characters used for \h and \v (which
806 are defined only in one place) are now also used for \s.
807
808 27. Add JIT support for the 64 bit TileGX architecture.
809 Patch by Jiong Wang (Tilera Corporation).
810
811 28. Possessive quantifiers for classes (both explicit and automatically
812 generated) now use special opcodes instead of wrapping in ONCE brackets.
813
814 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
815 (because it's meaningless), this was not happening when PCRE_CASELESS was
816 set. Not wrong, but inefficient.
817
818 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
819 properties for \w, \d, etc) is present in a test regex. Otherwise if the
820 test contains no characters greater than 255, Perl doesn't realise it
821 should be using Unicode semantics.
822
823 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
824 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
825 does in Unicode mode.
826
827 32. Added the "forbid" facility to pcretest so that putting tests into the
828 wrong test files can sometimes be quickly detected.
829
830 33. There is now a limit (default 250) on the depth of nesting of parentheses.
831 This limit is imposed to control the amount of system stack used at compile
832 time. It can be changed at build time by --with-parens-nest-limit=xxx or
833 the equivalent in CMake.
834
835 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
836 errors. Perl warns for these when in warning mode, but PCRE has no facility
837 for giving warnings.
838
839 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
840 However, this was not working for (?!) because it is optimized to (*FAIL),
841 for which PCRE does not allow quantifiers. The optimization is now disabled
842 when a quantifier follows (?!). I can't see any use for this, but it makes
843 things uniform.
844
845 36. Perl no longer allows group names to start with digits, so I have made this
846 change also in PCRE. It simplifies the code a bit.
847
848 37. In extended mode, Perl ignores spaces before a + that indicates a
849 possessive quantifier. PCRE allowed a space before the quantifier, but not
850 before the possessive +. It now does.
851
852 38. The use of \K (reset reported match start) within a repeated possessive
853 group such as (a\Kb)*+ was not working.
854
855 40. Document that the same character tables must be used at compile time and
856 run time, and that the facility to pass tables to pcre_exec() and
857 pcre_dfa_exec() is for use only with saved/restored patterns.
858
859 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
860 features for Builds with MSVC:
861
862 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
863 builds). The .rc files can be used to set FileDescription and many other
864 attributes.
865
866 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
867 This allows higher-level build scripts which want .pdb files to avoid
868 hard-coding the exact files needed."
869
870 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
871 mean "start of word" and "end of word", respectively, as a transition aid.
872
873 43. A minimizing repeat of a class containing codepoints greater than 255 in
874 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
875 compiled to use the heap for recursion.
876
877 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
878 is configured.
879
880
881 Version 8.33 28-May-2013
882 ------------------------
883
884 1. Added 'U' to some constants that are compared to unsigned integers, to
885 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
886 variables that are added to signed variables, to ensure the result is
887 signed and can be negated.
888
889 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
890 CMake config files.
891
892 3. Revise the creation of config.h.generic so that all boolean macros are
893 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
894 overriding via -D on the command line possible.
895
896 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
897 to unsigned int is reported to make a quite noticeable speed difference in
898 a specific Windows environment. Testing on Linux did also appear to show
899 some benefit (and it is clearly not harmful). Also fixed the definition of
900 Xop which should be unsigned.
901
902 5. Related to (4), changing the definition of the intermediate variable cc
903 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
904 improvements.
905
906 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
907 unnecessary spaces.
908
909 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
910 and later.
911
912 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
913
914 9. Optimizing fast_forward_start_bits in JIT.
915
916 10. Adding support for callouts in JIT, and fixing some issues revealed
917 during this work. Namely:
918
919 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
920
921 (b) Minimum length was not checked before the matching is started.
922
923 11. The value of capture_last that is passed to callouts was incorrect in some
924 cases when there was a capture on one path that was subsequently abandoned
925 after a backtrack. Also, the capture_last value is now reset after a
926 recursion, since all captures are also reset in this case.
927
928 12. The interpreter no longer returns the "too many substrings" error in the
929 case when an overflowing capture is in a branch that is subsequently
930 abandoned after a backtrack.
931
932 13. In the pathological case when an offset vector of size 2 is used, pcretest
933 now prints out the matched string after a yield of 0 or 1.
934
935 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
936 Only supported by the JIT compiler at the moment.
937
938 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
939
940 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
941 offset of the starting point of the matching process, provided the offsets
942 vector is large enough.
943
944 17. The \A escape now records a lookbehind value of 1, though its execution
945 does not actually inspect the previous character. This is to ensure that,
946 in partial multi-segment matching, at least one character from the old
947 segment is retained when a new segment is processed. Otherwise, if there
948 are no lookbehinds in the pattern, \A might match incorrectly at the start
949 of a new segment.
950
951 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
952
953 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
954 gives some modest performance improvement in 8-bit mode.
955
956 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
957 be expressed in certain programming languages using Universal Character
958 Names.
959
960 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
961 which points out that "non characters" are not "characters that may not
962 appear in Unicode strings" but rather "characters that are reserved for
963 internal use and have only local meaning".
964
965 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
966 there was a conditional group that depended on an assertion, if the
967 assertion was false, the callout that immediately followed the alternation
968 in the condition was skipped when pcre_exec() was used for matching.
969
970 23. Allow an explicit callout to be inserted before an assertion that is the
971 condition for a conditional group, for compatibility with automatic
972 callouts, which always insert a callout at this point.
973
974 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
975 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
976
977 25. (*PRUNE) is now supported by the JIT compiler.
978
979 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
980
981 27. Fix the case where there are two or more SKIPs with arguments that may be
982 ignored.
983
984 28. (*SKIP) is now supported by the JIT compiler.
985
986 29. (*THEN) is now supported by the JIT compiler.
987
988 30. Update RunTest with additional test selector options.
989
990 31. The way PCRE handles backtracking verbs has been changed in two ways.
991
992 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
993 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
994 backtracking. In some cases this makes it more Perl-compatible, but Perl's
995 rather obscure rules do not always do the same thing.
996
997 (2) Previously, backtracking verbs were confined within assertions. This is
998 no longer the case for positive assertions, except for (*ACCEPT). Again,
999 this sometimes improves Perl compatibility, and sometimes does not.
1000
1001 32. A number of tests that were in test 2 because Perl did things differently
1002 have been moved to test 1, because either Perl or PCRE has changed, and
1003 these tests are now compatible.
1004
1005 32. Backtracking control verbs are now handled in the same way in JIT and
1006 interpreter.
1007
1008 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1009 contained a forward subroutine reference caused a compile error.
1010
1011 34. Auto-detect and optimize limited repetitions in JIT.
1012
1013 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1014 blocking (*UTF) etc.
1015
1016 36. In the interpreter, maximizing pattern repetitions for characters and
1017 character types now use tail recursion, which reduces stack usage.
1018
1019 37. The value of the max lookbehind was not correctly preserved if a compiled
1020 and saved regex was reloaded on a host of different endianness.
1021
1022 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1023 of the compiled pattern block, expand the flags field from 16 to 32 bits
1024 because it was almost full.
1025
1026 39. Try madvise first before posix_madvise.
1027
1028 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1029 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1030 original change locked out all matches of empty strings. This has been
1031 changed so that one match of an empty string per line is recognized.
1032 Subsequent searches on the same line (for colouring or for --only-matching,
1033 for example) do not recognize empty strings.
1034
1035 41. Applied a user patch to fix a number of spelling mistakes in comments.
1036
1037 42. Data lines longer than 65536 caused pcretest to crash.
1038
1039 43. Clarified the data type for length and startoffset arguments for pcre_exec
1040 and pcre_dfa_exec in the function-specific man pages, where they were
1041 explicitly stated to be in bytes, never having been updated. I also added
1042 some clarification to the pcreapi man page.
1043
1044 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1045 a segmentation fault.
1046
1047
1048 Version 8.32 30-November-2012
1049 -----------------------------
1050
1051 1. Improved JIT compiler optimizations for first character search and single
1052 character iterators.
1053
1054 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1055 Patch by Daniel Richard G.
1056
1057 3. Single character iterator optimizations in the JIT compiler.
1058
1059 4. Improved JIT compiler optimizations for character ranges.
1060
1061 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1062 Reported by Giuseppe D'Angelo.
1063
1064 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1065 of a line, was being set incorrectly in cases where .* appeared inside
1066 atomic brackets at the start of a pattern, or where there was a subsequent
1067 *PRUNE or *SKIP.
1068
1069 7. Improved instruction cache flush for POWER/PowerPC.
1070 Patch by Daniel Richard G.
1071
1072 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1073 grep:
1074
1075 (a) There is now no limit to the number of patterns to be matched.
1076
1077 (b) An error is given if a pattern is too long.
1078
1079 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1080 are now supported.
1081
1082 (d) --exclude-from and --include-from (multiple use) have been added.
1083
1084 (e) Exclusions and inclusions now apply to all files and directories, not
1085 just to those obtained from scanning a directory recursively.
1086
1087 (f) Multiple uses of -f and --file-list are now supported.
1088
1089 (g) In a Windows environment, the default for -d has been changed from
1090 "read" (the GNU grep default) to "skip", because otherwise the presence
1091 of a directory in the file list provokes an error.
1092
1093 (h) The documentation has been revised and clarified in places.
1094
1095 9. Improve the matching speed of capturing brackets.
1096
1097 10. Changed the meaning of \X so that it now matches a Unicode extended
1098 grapheme cluster.
1099
1100 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1101 out POSIX threads when JIT support is configured.
1102
1103 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1104
1105 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1106 structure could go wrong in environments where size_t is not the same size
1107 as int.
1108
1109 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1110 set.
1111
1112 15. The EBCDIC support had decayed; later updates to the code had included
1113 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1114 general tidy up of EBCDIC-related issues, and the documentation was also
1115 not quite right. There is now a test that can be run on ASCII systems to
1116 check some of the EBCDIC-related things (but is it not a full test).
1117
1118 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1119 in a small tidy to the code.
1120
1121 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1122
1123 18. If the --only-matching (-o) option in pcregrep is specified multiple
1124 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1125 substrings matched by the 1st and 2nd capturing parentheses. A separating
1126 string can be specified by --om-separator (default empty).
1127
1128 19. Improving the first n character searches.
1129
1130 20. Turn case lists for horizontal and vertical white space into macros so that
1131 they are defined only once.
1132
1133 21. This set of changes together give more compatible Unicode case-folding
1134 behaviour for characters that have more than one other case when UCP
1135 support is available.
1136
1137 (a) The Unicode property table now has offsets into a new table of sets of
1138 three or more characters that are case-equivalent. The MultiStage2.py
1139 script that generates these tables (the pcre_ucd.c file) now scans
1140 CaseFolding.txt instead of UnicodeData.txt for character case
1141 information.
1142
1143 (b) The code for adding characters or ranges of characters to a character
1144 class has been abstracted into a generalized function that also handles
1145 case-independence. In UTF-mode with UCP support, this uses the new data
1146 to handle characters with more than one other case.
1147
1148 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1149 whose other case is greater than 256 are now correctly matched
1150 caselessly. Previously, the high codepoint matched the low one, but not
1151 vice versa.
1152
1153 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1154 of the new class addition function, using character lists defined as
1155 macros alongside the case definitions of 20 above.
1156
1157 (e) Caseless back references now work with characters that have more than
1158 one other case.
1159
1160 (f) General caseless matching of characters with more than one other case
1161 is supported.
1162
1163 22. Unicode character properties were updated from Unicode 6.2.0
1164
1165 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1166
1167 24. Add support for 32-bit character strings, and UTF-32
1168
1169 25. Major JIT compiler update (code refactoring and bugfixing).
1170 Experimental Sparc 32 support is added.
1171
1172 26. Applied a modified version of Daniel Richard G's patch to create
1173 pcre.h.generic and config.h.generic by "make" instead of in the
1174 PrepareRelease script.
1175
1176 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1177 pcre_compile.c when checking for a zero character.
1178
1179 28. Introducing a native interface for JIT. Through this interface, the compiled
1180 machine code can be directly executed. The purpose of this interface is to
1181 provide fast pattern matching, so several sanity checks are not performed.
1182 However, feature tests are still performed. The new interface provides
1183 1.4x speedup compared to the old one.
1184
1185 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1186 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1187 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1188
1189 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1190 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1191 the "old" RFC 2279). Instead, it ended up passing a negative length to
1192 pcre_exec().
1193
1194 31. Add support for GCC's visibility feature to hide internal functions.
1195
1196 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1197 "unknown -C option" after outputting 0 or 1.
1198
1199 33. There is now support for generating a code coverage report for the test
1200 suite in environments where gcc is the compiler and lcov is installed. This
1201 is mainly for the benefit of the developers.
1202
1203 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1204 unaddressable using valgrind annotations, allowing valgrind to detect
1205 invalid memory accesses. This is mainly for the benefit of the developers.
1206
1207 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1208
1209 26. Give configure error if --enable-cpp but no C++ compiler found.
1210
1211
1212 Version 8.31 06-July-2012
1213 -------------------------
1214
1215 1. Fixing a wrong JIT test case and some compiler warnings.
1216
1217 2. Removed a bashism from the RunTest script.
1218
1219 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1220 to unsigned type, result still unsigned" that was given by an MS compiler
1221 on encountering the code "-sizeof(xxx)".
1222
1223 4. Partial matching support is added to the JIT compiler.
1224
1225 5. Fixed several bugs concerned with partial matching of items that consist
1226 of more than one character:
1227
1228 (a) /^(..)\1/ did not partially match "aba" because checking references was
1229 done on an "all or nothing" basis. This also applied to repeated
1230 references.
1231
1232 (b) \R did not give a hard partial match if \r was found at the end of the
1233 subject.
1234
1235 (c) \X did not give a hard partial match after matching one or more
1236 characters at the end of the subject.
1237
1238 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1239 a partial match for the string "\r".
1240
1241 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1242 a partial match for a CR character at the end of the subject string.
1243
1244 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1245 running pcretest, the text "(JIT)" added to the output whenever JIT is
1246 actually used to run the match.
1247
1248 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1249 or /S+[+] with a digit between 1 and 7.
1250
1251 8. OP_NOT now supports any UTF character not just single-byte ones.
1252
1253 9. (*MARK) control verb is now supported by the JIT compiler.
1254
1255 10. The command "./RunTest list" lists the available tests without actually
1256 running any of them. (Because I keep forgetting what they all are.)
1257
1258 11. Add PCRE_INFO_MAXLOOKBEHIND.
1259
1260 12. Applied a (slightly modified) user-supplied patch that improves performance
1261 when the heap is used for recursion (compiled with --disable-stack-for-
1262 recursion). Instead of malloc and free for each heap frame each time a
1263 logical recursion happens, frames are retained on a chain and re-used where
1264 possible. This sometimes gives as much as 30% improvement.
1265
1266 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1267 call.
1268
1269 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1270
1271 15. It is now possible to link pcretest with libedit as an alternative to
1272 libreadline.
1273
1274 16. (*COMMIT) control verb is now supported by the JIT compiler.
1275
1276 17. The Unicode data tables have been updated to Unicode 6.1.0.
1277
1278 18. Added --file-list option to pcregrep.
1279
1280 19. Added binary file support to pcregrep, including the -a, --binary-files,
1281 -I, and --text options.
1282
1283 20. The madvise function is renamed for posix_madvise for QNX compatibility
1284 reasons. Fixed by Giuseppe D'Angelo.
1285
1286 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1287
1288 22. Changed the option for creating symbolic links for 16-bit man pages from
1289 -s to -sf so that re-installing does not cause issues.
1290
1291 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1292
1293 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1294 matches in certain environments (the workspace was not being correctly
1295 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1296 some of the workspace data at the beginning of a restart.
1297
1298 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1299 was not doing so when it should - probably a typo introduced by SVN 528
1300 (change 8.10/14).
1301
1302 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1303 \w+ when the character tables indicated that \x{c4} was a word character.
1304 There were several related cases, all because the tests for doing a table
1305 lookup were testing for characters less than 127 instead of 255.
1306
1307 27. If a pattern contains capturing parentheses that are not used in a match,
1308 their slots in the ovector are set to -1. For those that are higher than
1309 any matched groups, this happens at the end of processing. In the case when
1310 there were back references that the ovector was too small to contain
1311 (causing temporary malloc'd memory to be used during matching), and the
1312 highest capturing number was not used, memory off the end of the ovector
1313 was incorrectly being set to -1. (It was using the size of the temporary
1314 memory instead of the true size.)
1315
1316 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1317 ovector size, it uses memory at the end of the block that it has got.
1318
1319 29. Check for an overlong MARK name and give an error at compile time. The
1320 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1321
1322 30. JIT compiler update.
1323
1324 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1325 Rill for the patch.
1326
1327 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1328
1329 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1330
1331 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1332 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1333 was enabled.
1334
1335 35. Improve JIT code generation for greedy plus quantifier.
1336
1337 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1338 1 to "aa" instead of to an empty string. The bug affected repeated groups
1339 that could potentially match an empty string.
1340
1341 37. Optimizing single character iterators in JIT.
1342
1343 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1344 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1345 codepoints that are too big for the mode are faulted, and in a UTF mode,
1346 disallowed codepoints are also faulted.
1347
1348 39. If PCRE was compiled with UTF support, in three places in the DFA
1349 matcher there was code that should only have been obeyed in UTF mode, but
1350 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1351 processing when bytes with values greater than 127 were present. In 16-bit
1352 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1353 both cases the values are those that cannot be the first data item in a UTF
1354 character. The three items that might have provoked this were recursions,
1355 possessively repeated groups, and atomic groups.
1356
1357 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1358 and pcregrep, because some OS require shared objects to be explicitly
1359 passed to ld, causing the link step to fail if they are not.
1360
1361 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1362 mode, patterns that started with \h* or \R* might be incorrectly matched.
1363
1364
1365 Version 8.30 04-February-2012
1366 -----------------------------
1367
1368 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1369 name is defined in ctype.h.
1370
1371 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1372 only in quite long subpatterns.
1373
1374 3. Removed the function pcre_info(), which has been obsolete and deprecated
1375 since it was replaced by pcre_fullinfo() in February 2000.
1376
1377 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1378 match a (*MARK), and the match failed at the start of the subject, a
1379 reference to memory before the start of the subject could occur. This bug
1380 was introduced by fix 17 of release 8.21.
1381
1382 5. A reference to an unset group with zero minimum repetition was giving
1383 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1384 /(another)?(\1?)test/ matched against "hello world test". This bug was
1385 introduced in release 8.13.
1386
1387 6. Add support for 16-bit character strings (a large amount of work involving
1388 many changes and refactorings).
1389
1390 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1391 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1392 from a file.
1393
1394 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1395 rounding is not applied in this particular case).
1396
1397 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1398 if they appear, or are escaped, in patterns.
1399
1400 10. Get rid of a number of -Wunused-but-set-variable warnings.
1401
1402 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1403 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1404 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1405 also returns the mark "x". This bug applied to capturing parentheses,
1406 non-capturing parentheses, and atomic parentheses. It also applied to some
1407 assertions.
1408
1409 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1410 information out of configure.ac instead of relying on pcre.h.generic, which
1411 is not stored in the repository.
1412
1413 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1414 -lreadline.
1415
1416 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1417
1418 15. Applied Graycode's patch to put the top-level frame on the stack rather
1419 than the heap when not using the stack for recursion. This gives a
1420 performance improvement in many cases when recursion is not deep.
1421
1422 16. Experimental code added to "pcretest -C" to output the stack frame size.
1423
1424
1425 Version 8.21 12-Dec-2011
1426 ------------------------
1427
1428 1. Updating the JIT compiler.
1429
1430 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1431 are added as well.
1432
1433 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1434 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1435 calling _pcre_jit_exec. Some extra comments are added.
1436
1437 4. (*MARK) settings inside atomic groups that do not contain any capturing
1438 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1439 was introduced by change 18 for 8.20.
1440
1441 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1442 ECMA-262 standard.
1443
1444 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1445 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1446 This bug was probably introduced by change 9 of 8.13.
1447
1448 7. While fixing 6 above, I noticed that a number of other items were being
1449 incorrectly rejected as "not fixed length". This arose partly because newer
1450 opcodes had not been added to the fixed-length checking code. I have (a)
1451 corrected the bug and added tests for these items, and (b) arranged for an
1452 error to occur if an unknown opcode is encountered while checking for fixed
1453 length instead of just assuming "not fixed length". The items that were
1454 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1455 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1456 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1457
1458 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1459 being incorrectly compiled and would have given unpredicatble results.
1460
1461 9. A possessively repeated subpattern with minimum repeat count greater than
1462 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1463 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1464 the first (A) could occur when it should not.
1465
1466 10. Add a cast and remove a redundant test from the code.
1467
1468 11. JIT should use pcre_malloc/pcre_free for allocation.
1469
1470 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1471 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1472 is anything other than /usr, -L is still shown).
1473
1474 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1475
1476 14. Perl does not support \N without a following name in a [] class; PCRE now
1477 also gives an error.
1478
1479 15. If a forward reference was repeated with an upper limit of around 2000,
1480 it caused the error "internal error: overran compiling workspace". The
1481 maximum number of forward references (including repeats) was limited by the
1482 internal workspace, and dependent on the LINK_SIZE. The code has been
1483 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1484 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1485 of around 200,000 forward references. While doing this, I also speeded up
1486 the filling in of repeated forward references.
1487
1488 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1489 incorrectly expecting the subject to contain another "a" after the start.
1490
1491 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1492 in the match, the SKIP should be ignored. This was not happening; instead
1493 the SKIP was being treated as NOMATCH. For patterns such as
1494 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1495 tested.
1496
1497 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1498 now much more compatible with Perl, in particular in cases where the result
1499 is a non-match for a non-anchored pattern. For example, if
1500 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1501 "m", where previously it did not return a name. A side effect of this
1502 change is that for partial matches, the last encountered mark name is
1503 returned, as for non matches. A number of tests that were previously not
1504 Perl-compatible have been moved into the Perl-compatible test files. The
1505 refactoring has had the pleasing side effect of removing one argument from
1506 the match() function, thus reducing its stack requirements.
1507
1508 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1509 subsequent uses of /S (without +) incorrectly behaved like /S+.
1510
1511 21. Retrieve executable code size support for the JIT compiler and fixing
1512 some warnings.
1513
1514 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1515 not work when the shorter character appeared right at the end of the
1516 subject string.
1517
1518 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1519 systems.
1520
1521 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1522 output it when the /M option is used in pcretest.
1523
1524 25. The CheckMan script was not being included in the distribution. Also, added
1525 an explicit "perl" to run Perl scripts from the PrepareRelease script
1526 because this is reportedly needed in Windows.
1527
1528 26. If study data was being save in a file and studying had not found a set of
1529 "starts with" bytes for the pattern, the data written to the file (though
1530 never used) was taken from uninitialized memory and so caused valgrind to
1531 complain.
1532
1533 27. Updated RunTest.bat as provided by Sheri Pierce.
1534
1535 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1536
1537 29. Computation of memory usage for the table of capturing group names was
1538 giving an unnecessarily large value.
1539
1540
1541 Version 8.20 21-Oct-2011
1542 ------------------------
1543
1544 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1545 a POSIX class. After further experiments with Perl, which convinced me that
1546 Perl has bugs and confusions, a closing square bracket is no longer allowed
1547 in a POSIX name. This bug also affected patterns with classes that started
1548 with full stops.
1549
1550 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1551 captured substring, but while checking the failing first alternative,
1552 substring 1 is temporarily captured. If the output vector supplied to
1553 pcre_exec() was not big enough for this capture, the yield of the function
1554 was still zero ("insufficient space for captured substrings"). This cannot
1555 be totally fixed without adding another stack variable, which seems a lot
1556 of expense for a edge case. However, I have improved the situation in cases
1557 such as /(a)(b)x|abc/ matched against "abc", where the return code
1558 indicates that fewer than the maximum number of slots in the ovector have
1559 been set.
1560
1561 3. Related to (2) above: when there are more back references in a pattern than
1562 slots in the output vector, pcre_exec() uses temporary memory during
1563 matching, and copies in the captures as far as possible afterwards. It was
1564 using the entire output vector, but this conflicts with the specification
1565 that only 2/3 is used for passing back captured substrings. Now it uses
1566 only the first 2/3, for compatibility. This is, of course, another edge
1567 case.
1568
1569 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1570 main code base, and can be used by building with --enable-jit. When this is
1571 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1572 runtime --no-jit option is given.
1573
1574 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1575 ovector, the return from the function was zero, implying that there were
1576 other matches that did not fit. The correct "exactly full" value is now
1577 returned.
1578
1579 6. If a subpattern that was called recursively or as a subroutine contained
1580 (*PRUNE) or any other control that caused it to give a non-standard return,
1581 invalid errors such as "Error -26 (nested recursion at the same subject
1582 position)" or even infinite loops could occur.
1583
1584 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1585 computing the minimum length on reaching *ACCEPT, and so ended up with the
1586 wrong value of 1 rather than 0. Further investigation indicates that
1587 computing a minimum subject length in the presence of *ACCEPT is difficult
1588 (think back references, subroutine calls), and so I have changed the code
1589 so that no minimum is registered for a pattern that contains *ACCEPT.
1590
1591 8. If (*THEN) was present in the first (true) branch of a conditional group,
1592 it was not handled as intended. [But see 16 below.]
1593
1594 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1595 Sheri Pierce.
1596
1597 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1598 the first byte in a match must be "a".
1599
1600 11. Change 17 for 8.13 increased the recursion depth for patterns like
1601 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1602 pattern contains any instances of (*THEN). If it does not, the old
1603 optimizations are restored. It would be nice to do this on a per-group
1604 basis, but at the moment that is not feasible.
1605
1606 12. In some environments, the output of pcretest -C is CRLF terminated. This
1607 broke RunTest's code that checks for the link size. A single white space
1608 character after the value is now allowed for.
1609
1610 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1611 For "fr", it uses the Windows-specific input and output files.
1612
1613 14. If (*THEN) appeared in a group that was called recursively or as a
1614 subroutine, it did not work as intended. [But see next item.]
1615
1616 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1617 pattern fragments (but not containing any | characters). If A and B are
1618 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1619 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1620 D. In other words, Perl considers parentheses that do not contain any |
1621 characters to be part of a surrounding alternative, whereas PCRE was
1622 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1623 differently. PCRE now behaves in the same way as Perl, except in the case
1624 of subroutine/recursion calls such as (?1) which have in any case always
1625 been different (but PCRE had them first :-).
1626
1627 16. Related to 15 above: Perl does not treat the | in a conditional group as
1628 creating alternatives. Such a group is treated in the same way as an
1629 ordinary group without any | characters when processing (*THEN). PCRE has
1630 been changed to match Perl's behaviour.
1631
1632 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1633 RunGrepTest script failed.
1634
1635 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1636 inevitable for groups that contain captures, but it can lead to a lot of
1637 stack use in large patterns. The old behaviour has been restored for atomic
1638 groups that do not contain any capturing parentheses.
1639
1640 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1641 suppress the check for a minimum subject length at run time. (If it was
1642 given to pcre_exec() or pcre_dfa_exec() it did work.)
1643
1644 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1645 fail to work when decoding hex characters in data strings in EBCDIC
1646 environments.
1647
1648 21. It appears that in at least one Mac OS environment, the isxdigit() function
1649 is implemented as a macro that evaluates to its argument more than once,
1650 contravening the C 90 Standard (I haven't checked a later standard). There
1651 was an instance in pcretest which caused it to go wrong when processing
1652 \x{...} escapes in subject strings. The has been rewritten to avoid using
1653 things like p++ in the argument of isxdigit().
1654
1655
1656 Version 8.13 16-Aug-2011
1657 ------------------------
1658
1659 1. The Unicode data tables have been updated to Unicode 6.0.0.
1660
1661 2. Two minor typos in pcre_internal.h have been fixed.
1662
1663 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1664 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1665 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1666
1667 4. There were a number of related bugs in the code for matching backrefences
1668 caselessly in UTF-8 mode when codes for the characters concerned were
1669 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1670 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1671 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1672 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1673 2-byte code at the end of the subject (it thought there wasn't enough data
1674 left).
1675
1676 5. Comprehensive information about what went wrong is now returned by
1677 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1678 as the output vector has at least 2 elements. The offset of the start of
1679 the failing character and a reason code are placed in the vector.
1680
1681 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1682 now returned is for the first byte of the failing character, instead of the
1683 last byte inspected. This is an incompatible change, but I hope it is small
1684 enough not to be a problem. It makes the returned offset consistent with
1685 pcre_exec() and pcre_dfa_exec().
1686
1687 7. pcretest now gives a text phrase as well as the error number when
1688 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1689 failure, the offset and reason code are output.
1690
1691 8. When \R was used with a maximizing quantifier it failed to skip backwards
1692 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1693 back over a single character (\n). This seems wrong (because it treated the
1694 two characters as a single entity when going forwards), conflicts with the
1695 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1696 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1697 has been changed.
1698
1699 9. Some internal refactoring has changed the processing so that the handling
1700 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1701 time (the PCRE_DOTALL option was changed this way some time ago: version
1702 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1703 which was always a bit of a fudge. It also means that there is one less
1704 argument for the match() function, which reduces its stack requirements
1705 slightly. This change also fixes an incompatibility with Perl: the pattern
1706 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1707
1708 10. More internal refactoring has drastically reduced the number of recursive
1709 calls to match() for possessively repeated groups such as (abc)++ when
1710 using pcre_exec().
1711
1712 11. While implementing 10, a number of bugs in the handling of groups were
1713 discovered and fixed:
1714
1715 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1716 (a|)*(?1) gave a compile-time internal error.
1717 ((a|)+)+ did not notice that the outer group could match an empty string.
1718 (^a|^)+ was not marked as anchored.
1719 (.*a|.*)+ was not marked as matching at start or after a newline.
1720
1721 12. Yet more internal refactoring has removed another argument from the match()
1722 function. Special calls to this function are now indicated by setting a
1723 value in a variable in the "match data" data block.
1724
1725 13. Be more explicit in pcre_study() instead of relying on "default" for
1726 opcodes that mean there is no starting character; this means that when new
1727 ones are added and accidentally left out of pcre_study(), testing should
1728 pick them up.
1729
1730 14. The -s option of pcretest has been documented for ages as being an old
1731 synonym of -m (show memory usage). I have changed it to mean "force study
1732 for every regex", that is, assume /S for every regex. This is similar to -i
1733 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1734 using it. It makes it easier to run collections of tests with and without
1735 study enabled, and thereby test pcre_study() more easily. All the standard
1736 tests are now run with and without -s (but some patterns can be marked as
1737 "never study" - see 20 below).
1738
1739 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1740 restoration of the capturing data to the outer values was not happening
1741 correctly.
1742
1743 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1744 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1745 pattern had matched an empty string, and so incorrectly returned a no
1746 match.
1747
1748 17. There was optimizing code for the last branch of non-capturing parentheses,
1749 and also for the obeyed branch of a conditional subexpression, which used
1750 tail recursion to cut down on stack usage. Unfortunately, now that there is
1751 the possibility of (*THEN) occurring in these branches, tail recursion is
1752 no longer possible because the return has to be checked for (*THEN). These
1753 two optimizations have therefore been removed. [But see 8.20/11 above.]
1754
1755 18. If a pattern containing \R was studied, it was assumed that \R always
1756 matched two bytes, thus causing the minimum subject length to be
1757 incorrectly computed because \R can also match just one byte.
1758
1759 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1760 was incorrectly computed.
1761
1762 20. If /S is present twice on a test pattern in pcretest input, it now
1763 *disables* studying, thereby overriding the use of -s on the command line
1764 (see 14 above). This is necessary for one or two tests to keep the output
1765 identical in both cases.
1766
1767 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1768 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1769
1770 22. When an atomic group that contained a capturing parenthesis was
1771 successfully matched, but the branch in which it appeared failed, the
1772 capturing was not being forgotten if a higher numbered group was later
1773 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1774 group 1 to "a", when in fact it should be unset. This applied to multi-
1775 branched capturing and non-capturing groups, repeated or not, and also to
1776 positive assertions (capturing in negative assertions does not happen
1777 in PCRE) and also to nested atomic groups.
1778
1779 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1780 subject after a captured substring, to make it easier to tell which of a
1781 number of identical substrings has been captured.
1782
1783 24. The way atomic groups are processed by pcre_exec() has been changed so that
1784 if they are repeated, backtracking one repetition now resets captured
1785 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1786 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1787 "aaa". Previously, it would have been "a". As part of this code
1788 refactoring, the way recursive calls are handled has also been changed.
1789
1790 25. If an assertion condition captured any substrings, they were not passed
1791 back unless some other capturing happened later. For example, if
1792 (?(?=(a))a) was matched against "a", no capturing was returned.
1793
1794 26. When studying a pattern that contained subroutine calls or assertions,
1795 the code for finding the minimum length of a possible match was handling
1796 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1797 group 1 called group 2 while simultaneously a separate group 2 called group
1798 1). A stack overflow occurred in this case. I have fixed this by limiting
1799 the recursion depth to 10.
1800
1801 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1802 Fortmann. This supports explicit test numbers on the command line, and has
1803 argument validation and error reporting.
1804
1805 28. An instance of \X with an unlimited repeat could fail if at any point the
1806 first character it looked at was a mark character.
1807
1808 29. Some minor code refactoring concerning Unicode properties and scripts
1809 should reduce the stack requirement of match() slightly.
1810
1811 30. Added the '=' option to pcretest to check the setting of unused capturing
1812 slots at the end of the pattern, which are documented as being -1, but are
1813 not included in the return count.
1814
1815 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1816 compiled something random. Now it gives a compile-time error (as does
1817 Perl).
1818
1819 32. A *MARK encountered during the processing of a positive assertion is now
1820 recorded and passed back (compatible with Perl).
1821
1822 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1823 had alternative anchored branches, the search for a second match in a line
1824 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1825 matched the line "0102" twice. The same bug affected patterns that started
1826 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1827 twice.
1828
1829 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1830 does, and because of capturing effects, quantifying parenthesized
1831 assertions may at times be useful. Quantifiers are now allowed for
1832 parenthesized assertions.
1833
1834 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1835
1836 36. \g was being checked for fancy things in a character class, when it should
1837 just be a literal "g".
1838
1839 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1840 appearance of a nested POSIX class supersedes an apparent external class.
1841 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1842 unescaped square brackets may also appear as part of class names. For
1843 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1844 more like Perl. (But see 8.20/1 above.)
1845
1846 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1847 was because it thought it was \N{name}, which is not supported).
1848
1849 39. Add minix to OS list not supporting the -S option in pcretest.
1850
1851 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1852 cannot analyze patterns in sufficient detail to catch mutual recursions
1853 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1854 subgroup is called recursively as a subpattern for a second time at the
1855 same position in the subject string. In previous releases this might have
1856 been caught by the recursion limit, or it might have run out of stack.
1857
1858 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1859 happen only once. PCRE was, however incorrectly giving a compile time error
1860 "recursive call could loop indefinitely" because it cannot analyze the
1861 pattern in sufficient detail. The compile time test no longer happens when
1862 PCRE is compiling a conditional subpattern, but actual runaway loops are
1863 now caught at runtime (see 40 above).
1864
1865 42. It seems that Perl allows any characters other than a closing parenthesis
1866 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1867 has been changed to be the same.
1868
1869 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1870 as not to get warnings when autogen.sh is called. Also changed
1871 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1872
1873 44. To help people who use pcregrep to scan files containing exceedingly long
1874 lines, the following changes have been made:
1875
1876 (a) The default value of the buffer size parameter has been increased from
1877 8K to 20K. (The actual buffer used is three times this size.)
1878
1879 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1880 PCRE is built.
1881
1882 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1883 to be set at run time.
1884
1885 (d) Numerical values in pcregrep options can be followed by K or M, for
1886 example --buffer-size=50K.
1887
1888 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1889 given and the return code is set to 2.
1890
1891 45. Add a pointer to the latest mark to the callout data block.
1892
1893 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1894 partial match of an empty string instead of no match. This was specific to
1895 the use of ".".
1896
1897 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1898 complete match instead of a partial match. This bug was dependent on both
1899 the PCRE_UTF8 and PCRE_DOTALL options being set.
1900
1901 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1902 starting byte set, because \b was not being ignored.
1903
1904
1905 Version 8.12 15-Jan-2011
1906 ------------------------
1907
1908 1. Fixed some typos in the markup of the man pages, and wrote a script that
1909 checks for such things as part of the documentation building process.
1910
1911 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1912 --match-limit and --recursion-limit options (added for 8.11). In
1913 particular, this made one of the standard tests fail. (The integer value
1914 went into the wrong half of a long int.)
1915
1916 3. If the --colour option was given to pcregrep with -v (invert match), it
1917 did strange things, either producing crazy output, or crashing. It should,
1918 of course, ignore a request for colour when reporting lines that do not
1919 match.
1920
1921 4. Another pcregrep bug caused similar problems if --colour was specified with
1922 -M (multiline) and the pattern match finished with a line ending.
1923
1924 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1925 matched in multiline mode, the following line was shown as part of the
1926 match. This seems wrong, so I have changed it.
1927
1928 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1929 the check for further matches in the same line (so they could be coloured)
1930 to overrun the end of the current line. If another match was found, it was
1931 incorrectly shown (and then shown again when found in the next line).
1932
1933 7. If pcregrep was compiled under Windows, there was a reference to the
1934 function pcregrep_exit() before it was defined. I am assuming this was
1935 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1936 reported by a user. I've moved the definition above the reference.
1937
1938
1939 Version 8.11 10-Dec-2010
1940 ------------------------
1941
1942 1. (*THEN) was not working properly if there were untried alternatives prior
1943 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1944 backtracked to try for "b" instead of moving to the next alternative branch
1945 at the same level (in this case, to look for "c"). The Perl documentation
1946 is clear that when (*THEN) is backtracked onto, it goes to the "next
1947 alternative in the innermost enclosing group".
1948
1949 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1950 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1951 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1952 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1953 (*THEN).
1954
1955 3. If \s appeared in a character class, it removed the VT character from
1956 the class, even if it had been included by some previous item, for example
1957 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1958 of \s, but is part of the POSIX "space" class.)
1959
1960 4. A partial match never returns an empty string (because you can always
1961 match an empty string at the end of the subject); however the checking for
1962 an empty string was starting at the "start of match" point. This has been
1963 changed to the "earliest inspected character" point, because the returned
1964 data for a partial match starts at this character. This means that, for
1965 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1966 (previously it gave "no match").
1967
1968 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1969 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1970 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1971 has an implication that the given string is incomplete (because a partial
1972 match is preferred over a full match). For this reason, these items now
1973 give a partial match in this situation. [Aside: previously, the one case
1974 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1975 match rather than a full match, which was wrong by the old rules, but is
1976 now correct.]
1977
1978 6. There was a bug in the handling of #-introduced comments, recognized when
1979 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1980 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1981 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1982 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1983 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1984 places in pcre_compile().
1985
1986 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1987 comments when looking ahead for named forward references to subpatterns,
1988 the only newline sequence it recognized was NL. It now handles newlines
1989 according to the set newline convention.
1990
1991 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1992 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1993 cater for a lack of strerror(). These oversights have been fixed.
1994
1995 9. Added --match-limit and --recursion-limit to pcregrep.
1996
1997 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1998
1999 11. When the -o option was used, pcregrep was setting a return code of 1, even
2000 when matches were found, and --line-buffered was not being honoured.
2001
2002 12. Added an optional parentheses number to the -o and --only-matching options
2003 of pcregrep.
2004
2005 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2006 can match an empty string. The code to do it in pcretest and pcredemo
2007 needed fixing:
2008
2009 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2010 only one byte after an empty string match just before CRLF (this case
2011 just got forgotten; "any" and "anycrlf" were OK).
2012
2013 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2014 mode when an empty string match preceded an ASCII character followed by
2015 a non-ASCII character. (The code for advancing by one character rather
2016 than one byte was nonsense.)
2017
2018 (c) The pcredemo.c sample program did not have any code at all to handle
2019 the cases when CRLF is a valid newline sequence.
2020
2021 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2022 as a starting offset was within the subject string. There is now a new
2023 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2024 negative or greater than the length of the string. In order to test this,
2025 pcretest is extended to allow the setting of negative starting offsets.
2026
2027 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2028 starting offset points to the beginning of a UTF-8 character was
2029 unnecessarily clumsy. I tidied it up.
2030
2031 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2032 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2033
2034 17. Nobody had reported that the --include_dir option, which was added in
2035 release 7.7 should have been called --include-dir (hyphen, not underscore)
2036 for compatibility with GNU grep. I have changed it to --include-dir, but
2037 left --include_dir as an undocumented synonym, and the same for
2038 --exclude-dir, though that is not available in GNU grep, at least as of
2039 release 2.5.4.
2040
2041 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2042 characters from a string of bytes) have been redefined so as not to use
2043 loops, in order to improve performance in some environments. At the same
2044 time, I abstracted some of the common code into auxiliary macros to save
2045 repetition (this should not affect the compiled code).
2046
2047 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2048 compile-time error is now given if \c is not followed by an ASCII
2049 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2050 different, and any byte value is allowed.)
2051
2052 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2053 START_OPTIMIZE option, which is now allowed at compile time - but just
2054 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2055 to pcregrep and other applications that have no direct access to PCRE
2056 options. The new /Y option in pcretest sets this option when calling
2057 pcre_compile().
2058
2059 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2060 back references. Groups containing recursive back references were forced to
2061 be atomic by that change, but in the case of named groups, the amount of
2062 memory required was incorrectly computed, leading to "Failed: internal
2063 error: code overflow". This has been fixed.
2064
2065 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2066 pcretest.c, to avoid build problems in some Borland environments.
2067
2068
2069 Version 8.10 25-Jun-2010
2070 ------------------------
2071
2072 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2073 THEN.
2074
2075 2. (*ACCEPT) was not working when inside an atomic group.
2076
2077 3. Inside a character class, \B is treated as a literal by default, but
2078 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2079 causes the error). The code is unchanged, but I tidied the documentation.
2080
2081 4. Inside a character class, PCRE always treated \R and \X as literals,
2082 whereas Perl faults them if its -w option is set. I have changed PCRE so
2083 that it faults them when PCRE_EXTRA is set.
2084
2085 5. Added support for \N, which always matches any character other than
2086 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2087
2088 6. When compiling pcregrep with newer versions of gcc which may have
2089 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2090 declared with attribute warn_unused_result" were given. Just casting the
2091 result to (void) does not stop the warnings; a more elaborate fudge is
2092 needed. I've used a macro to implement this.
2093
2094 7. Minor change to pcretest.c to avoid a compiler warning.
2095
2096 8. Added four artifical Unicode properties to help with an option to make
2097 \s etc use properties (see next item). The new properties are: Xan
2098 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2099
2100 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2101 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2102 this option. Modified pcretest to add /W to test this facility. Added
2103 REG_UCP to make it available via the POSIX interface.
2104
2105 10. Added --line-buffered to pcregrep.
2106
2107 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2108 studied, and the match started with a letter with a code point greater than
2109 127 whose first byte was different to the first byte of the other case of
2110 the letter, the other case of this starting letter was not recognized
2111 (#976).
2112
2113 12. If a pattern that was studied started with a repeated Unicode property
2114 test, for example, \p{Nd}+, there was the theoretical possibility of
2115 setting up an incorrect bitmap of starting bytes, but fortunately it could
2116 not have actually happened in practice until change 8 above was made (it
2117 added property types that matched character-matching opcodes).
2118
2119 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2120 possible starting bytes for non-anchored patterns.
2121
2122 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2123 \R, and also a number of cases that involve Unicode properties, both
2124 explicit and implicit when PCRE_UCP is set.
2125
2126 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2127 input, it could crash or give wrong results if characters with values
2128 greater than 0xc0 were present in the subject string. (Detail: it assumed
2129 UTF-8 input when processing these items.)
2130
2131 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2132 size_t is 64-bit (#991).
2133
2134 17. Added a check for running out of memory when PCRE is compiled with
2135 --disable-stack-for-recursion (#990).
2136
2137 18. If the last data line in a file for pcretest does not have a newline on
2138 the end, a newline was missing in the output.
2139
2140 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2141 less than 128) in its various bitmaps. However, there is a facility for
2142 generating tables according to the current locale when PCRE is compiled. It
2143 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2144 characters, are recognized by isspace() and therefore were getting set in
2145 these tables, and indeed these tables seem to approximate to ISO 8859. This
2146 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2147 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2148 which of course cannot start UTF-8 characters. I have changed the code so
2149 that only real ASCII characters (less than 128) and the correct starting
2150 bytes for UTF-8 encodings are set for characters greater than 127 when in
2151 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2152 altogether.)
2153
2154 20. Added the /T option to pcretest so as to be able to run tests with non-
2155 standard character tables, thus making it possible to include the tests
2156 used for 19 above in the standard set of tests.
2157
2158 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2159 reference to a subpattern the other side of a comment that contains an
2160 opening parenthesis caused either an internal compiling error, or a
2161 reference to the wrong subpattern.
2162
2163
2164 Version 8.02 19-Mar-2010
2165 ------------------------
2166
2167 1. The Unicode data tables have been updated to Unicode 5.2.0.
2168
2169 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2170 configured.
2171
2172 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2173 original author of that file, following a query about its status.
2174
2175 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2176 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2177
2178 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2179 quantifier applied to a forward-referencing subroutine call, could compile
2180 incorrect code or give the error "internal error: previously-checked
2181 referenced subpattern not found".
2182
2183 6. Both MS Visual Studio and Symbian OS have problems with initializing
2184 variables to point to external functions. For these systems, therefore,
2185 pcre_malloc etc. are now initialized to local functions that call the
2186 relevant global functions.
2187
2188 7. There were two entries missing in the vectors called coptable and poptable
2189 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2190 I've fixed the data, and added a kludgy way of testing at compile time that
2191 the lengths are correct (equal to the number of opcodes).
2192
2193 8. Following on from 7, I added a similar kludge to check the length of the
2194 eint vector in pcreposix.c.
2195
2196 9. Error texts for pcre_compile() are held as one long string to avoid too
2197 much relocation at load time. To find a text, the string is searched,
2198 counting zeros. There was no check for running off the end of the string,
2199 which could happen if a new error number was added without updating the
2200 string.
2201
2202 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2203
2204 11. \K was not working if it appeared in an atomic group or in a group that
2205 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2206 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2207 the assertion is positive, but not if it is negative.
2208
2209 12. Change 11 fortuitously reduced the size of the stack frame used in the
2210 "match()" function of pcre_exec.c by one pointer. Forthcoming
2211 implementation of support for (*MARK) will need an extra pointer on the
2212 stack; I have reserved it now, so that the stack frame size does not
2213 decrease.
2214
2215 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2216 item in branch that calls a recursion is a subroutine call - as in the
2217 second branch in the above example - was incorrectly given the compile-
2218 time error "recursive call could loop indefinitely" because pcre_compile()
2219 was not correctly checking the subroutine for matching a non-empty string.
2220
2221 14. The checks for overrunning compiling workspace could trigger after an
2222 overrun had occurred. This is a "should never occur" error, but it can be
2223 triggered by pathological patterns such as hundreds of nested parentheses.
2224 The checks now trigger 100 bytes before the end of the workspace.
2225
2226 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2227
2228
2229 Version 8.01 19-Jan-2010
2230 ------------------------
2231
2232 1. If a pattern contained a conditional subpattern with only one branch (in
2233 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2234 computed the wrong minimum data length (which is of course zero for such
2235 subpatterns). This could cause incorrect "no match" results.
2236
2237 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2238 the pattern is reset in the first branch, pcre_compile() failed with
2239 "internal error: code overflow at offset...". This happened only when
2240 the reset was to the original external option setting. (An optimization
2241 abstracts leading options settings into an external setting, which was the
2242 cause of this.)
2243
2244 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2245 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2246 assertion pattern did not match (meaning that the assertion was true), it
2247 was incorrectly treated as false if the SKIP had been reached during the
2248 matching. This also applied to assertions used as conditions.
2249
2250 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2251 assertion subpattern, including such a pattern used as a condition,
2252 unpredictable results occurred, instead of the error return
2253 PCRE_ERROR_DFA_UITEM.
2254
2255 5. The C++ GlobalReplace function was not working like Perl for the special
2256 situation when an empty string is matched. It now does the fancy magic
2257 stuff that is necessary.
2258
2259 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2260 removed. (These were left over from very, very early versions of PCRE.)
2261
2262 7. Some cosmetic changes to the code to make life easier when compiling it
2263 as part of something else:
2264
2265 (a) Change DEBUG to PCRE_DEBUG.
2266
2267 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2268 called "current" as "current_branch", to prevent a collision with the
2269 Linux macro when compiled as a kernel module.
2270
2271 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2272 prevent a collision with the Linux macro when compiled as a kernel
2273 module.
2274
2275 8. In pcre_compile() there are some checks for integer overflows that used to
2276 cast potentially large values to (double). This has been changed to that
2277 when building, a check for int64_t is made, and if it is found, it is used
2278 instead, thus avoiding the use of floating point arithmetic. (There is no
2279 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2280 double.
2281
2282 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2283 2005 (difference between two addresses compared to an unsigned value).
2284
2285 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2286 custom one, because of the following reported problem in Windows:
2287
2288 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2289 under Win32.
2290 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2291 therefore missing the function definition.
2292 - The compiler thus generates a "C" signature for the test function.
2293 - The linker fails to find the "C" function.
2294 - PCRE fails to configure if asked to do so against libbz2.
2295
2296 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2297 messages were output:
2298
2299 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2300 rerunning libtoolize, to keep the correct libtool macros in-tree.
2301 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2302
2303 I have done both of these things.
2304
2305 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2306 most of the time, it *can* run out if it is given a pattern that contains a
2307 runaway infinite recursion. I updated the discussion in the pcrestack man
2308 page.
2309
2310 13. Now that we have gone to the x.xx style of version numbers, the minor
2311 version may start with zero. Using 08 or 09 is a bad idea because users
2312 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2313 interpreted as invalid octal numbers. I've updated the previous comment in
2314 configure.ac, and also added a check that gives an error if 08 or 09 are
2315 used.
2316
2317 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2318 causing partial matching to fail when the end of the subject matched \W
2319 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2320
2321 15. There were some discrepancies between the declarations in pcre_internal.h
2322 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2323 their definitions. The declarations used "const uschar *" and the
2324 definitions used USPTR. Even though USPTR is normally defined as "const
2325 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2326 reported that: "This difference in casting confuses some C++ compilers, for
2327 example, SunCC recognizes above declarations as different functions and
2328 generates broken code for hbpcre." I have changed the declarations to use
2329 USPTR.
2330
2331 16. GNU libtool is named differently on some systems. The autogen.sh script now
2332 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2333 (FreeBSD).
2334
2335 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2336 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2337 comment: "Figure out how to create a longlong from a string: strtoll and
2338 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2339 instance, but it only takes 2 args instead of 3!"
2340
2341 18. A subtle bug concerned with back references has been fixed by a change of
2342 specification, with a corresponding code fix. A pattern such as
2343 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2344 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2345 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2346 same bug. Such groups have to be quantified to be useful, or contained
2347 inside another quantified group. (If there's no repetition, the reference
2348 can never match.) The problem arises because, having left the group and
2349 moved on to the rest of the pattern, a later failure that backtracks into
2350 the group uses the captured value from the final iteration of the group
2351 rather than the correct earlier one. I have fixed this in PCRE by forcing
2352 any group that contains a reference to itself to be an atomic group; that
2353 is, there cannot be any backtracking into it once it has completed. This is
2354 similar to recursive and subroutine calls.
2355
2356
2357 Version 8.00 19-Oct-09
2358 ----------------------
2359
2360 1. The table for translating pcre_compile() error codes into POSIX error codes
2361 was out-of-date, and there was no check on the pcre_compile() error code
2362 being within the table. This could lead to an OK return being given in
2363 error.
2364
2365 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2366 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2367 in a Windows environment.
2368
2369 3. The pcregrep --count option prints the count for each file even when it is
2370 zero, as does GNU grep. However, pcregrep was also printing all files when
2371 --files-with-matches was added. Now, when both options are given, it prints
2372 counts only for those files that have at least one match. (GNU grep just
2373 prints the file name in this circumstance, but including the count seems
2374 more useful - otherwise, why use --count?) Also ensured that the
2375 combination -clh just lists non-zero counts, with no names.
2376
2377 4. The long form of the pcregrep -F option was incorrectly implemented as
2378 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2379 but it seems right to fix it, and I didn't think it was worth preserving
2380 the old behaviour.
2381
2382 5. The command line items --regex=pattern and --regexp=pattern were not
2383 recognized by pcregrep, which required --regex pattern or --regexp pattern
2384 (with a space rather than an '='). The man page documented the '=' forms,
2385 which are compatible with GNU grep; these now work.
2386
2387 6. No libpcreposix.pc file was created for pkg-config; there was just
2388 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2389
2390 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2391 when UCP support is not needed, by modifying the Python script that
2392 generates it from Unicode data files. This should not matter if the module
2393 is correctly used as a library, but I received one complaint about 50K of
2394 unwanted data. My guess is that the person linked everything into his
2395 program rather than using a library. Anyway, it does no harm.
2396
2397 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2398 was a minimum greater than 1 for a wide character in a possessive
2399 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2400 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2401 character. Chaos in the form of incorrect output or a compiling loop could
2402 result.
2403
2404 9. The restrictions on what a pattern can contain when partial matching is
2405 requested for pcre_exec() have been removed. All patterns can now be
2406 partially matched by this function. In addition, if there are at least two
2407 slots in the offset vector, the offset of the earliest inspected character
2408 for the match and the offset of the end of the subject are set in them when
2409 PCRE_ERROR_PARTIAL is returned.
2410
2411 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2412 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2413 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2414 and may be more useful for multi-segment matching.
2415
2416 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2417 used to be given if ever the end of the subject was reached; now it is
2418 given only if matching could not proceed because another character was
2419 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2420 string "Z", which now yields "no match" instead of "partial match". In the
2421 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2422 final character ended with (*FAIL).
2423
2424 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2425 if the pattern had a "must contain" character that was already found in the
2426 earlier partial match, unless partial matching was again requested. For
2427 example, with the pattern /dog.(body)?/, the "must contain" character is
2428 "g". If the first part-match was for the string "dog", restarting with
2429 "sbody" failed. This bug has been fixed.
2430
2431 13. The string returned by pcre_dfa_exec() after a partial match has been
2432 changed so that it starts at the first inspected character rather than the
2433 first character of the match. This makes a difference only if the pattern
2434 starts with a lookbehind assertion or \b or \B (\K is not supported by
2435 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2436 matching functions compatible, and I think it's the right thing to do.
2437
2438 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2439 so that the demonstration program is easily available in environments where
2440 PCRE has not been installed from source.
2441
2442 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2443 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2444 library.
2445
2446 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2447 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2448 is not the first non-POSIX option to be added. Clearly some people find
2449 these options useful.
2450
2451 17. If a caller to the POSIX matching function regexec() passes a non-zero
2452 value for nmatch with a NULL value for pmatch, the value of
2453 nmatch is forced to zero.
2454
2455 18. RunGrepTest did not have a test for the availability of the -u option of
2456 the diff command, as RunTest does. It now checks in the same way as
2457 RunTest, and also checks for the -b option.
2458
2459 19. If an odd number of negated classes containing just a single character
2460 interposed, within parentheses, between a forward reference to a named
2461 subpattern and the definition of the subpattern, compilation crashed with
2462 an internal error, complaining that it could not find the referenced
2463 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2464 [The bug was that it was starting one character too far in when skipping
2465 over the character class, thus treating the ] as data rather than
2466 terminating the class. This meant it could skip too much.]
2467
2468 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2469 /g option in pcretest when the pattern contains \K, which makes it possible
2470 to have an empty string match not at the start, even when the pattern is
2471 anchored. Updated pcretest and pcredemo to use this option.
2472
2473 21. If the maximum number of capturing subpatterns in a recursion was greater
2474 than the maximum at the outer level, the higher number was returned, but
2475 with unset values at the outer level. The correct (outer level) value is
2476 now given.
2477
2478 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2479 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2480 make it do so. The string so far is captured, making this feature
2481 compatible with Perl.
2482
2483 23. The tests have been re-organized, adding tests 11 and 12, to make it
2484 possible to check the Perl 5.10 features against Perl 5.10.
2485
2486 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2487 pattern matches a fixed length string. PCRE did not allow this; now it
2488 does. Neither allows recursion.
2489
2490 25. I finally figured out how to implement a request to provide the minimum
2491 length of subject string that was needed in order to match a given pattern.
2492 (It was back references and recursion that I had previously got hung up
2493 on.) This code has now been added to pcre_study(); it finds a lower bound
2494 to the length of subject needed. It is not necessarily the greatest lower
2495 bound, but using it to avoid searching strings that are too short does give
2496 some useful speed-ups. The value is available to calling programs via
2497 pcre_fullinfo().
2498
2499 26. While implementing 25, I discovered to my embarrassment that pcretest had
2500 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2501 study optimizations had never been tested with that matching function.
2502 Oops. What is worse, even when it was passed study data, there was a bug in
2503 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2504 were also very few tests of studied patterns with pcre_dfa_exec().
2505
2506 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2507 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2508 on the other side of the coin, they are no longer allowed to have different
2509 names, because these cannot be distinguished in PCRE, and this has caused
2510 confusion. (This is a difference from Perl.)
2511
2512 28. When duplicate subpattern names are present (necessarily with different
2513 numbers, as required by 27 above), and a test is made by name in a
2514 conditional pattern, either for a subpattern having been matched, or for
2515 recursion in such a pattern, all the associated numbered subpatterns are
2516 tested, and the overall condition is true if the condition is true for any
2517 one of them. This is the way Perl works, and is also more like the way
2518 testing by number works.
2519
2520
2521 Version 7.9 11-Apr-09
2522 ---------------------
2523
2524 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2525 (pcretest), all targets were linked against these libraries. This included
2526 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2527 libraries. This caused unwanted dependencies to be created. This problem
2528 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2529 pcretest is linked with readline.
2530
2531 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2532 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2533 moved outside it again, because FALSE and TRUE are already defined in AIX,
2534 but BOOL is not.
2535
2536 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2537 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2538
2539 4. The pcregrep documentation said spaces were inserted as well as colons (or
2540 hyphens) following file names and line numbers when outputting matching
2541 lines. This is not true; no spaces are inserted. I have also clarified the
2542 wording for the --colour (or --color) option.
2543
2544 5. In pcregrep, when --colour was used with -o, the list of matching strings
2545 was not coloured; this is different to GNU grep, so I have changed it to be
2546 the same.
2547
2548 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2549 each matching line was coloured. Now it goes on to look for further matches
2550 of any of the test patterns, which is the same behaviour as GNU grep.
2551
2552 7. A pattern that could match an empty string could cause pcregrep to loop; it
2553 doesn't make sense to accept an empty string match in pcregrep, so I have
2554 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2555 seems to be how GNU grep behaves. [But see later change 40 for release
2556 8.33.]
2557
2558 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2559 start or after a newline", because the conditional assertion was not being
2560 correctly handled. The rule now is that both the assertion and what follows
2561 in the first alternative must satisfy the test.
2562
2563 9. If auto-callout was enabled in a pattern with a conditional group whose
2564 condition was an assertion, PCRE could crash during matching, both with
2565 pcre_exec() and pcre_dfa_exec().
2566
2567 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2568 used for matching.
2569
2570 11. Unicode property support in character classes was not working for
2571 characters (bytes) greater than 127 when not in UTF-8 mode.
2572
2573 12. Added the -M command line option to pcretest.
2574
2575 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2576
2577 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2578
2579 16. Added comments and documentation about mis-use of no_arg in the C++
2580 wrapper.
2581
2582 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2583 from Martin Jerabek that uses macro names for all relevant character and
2584 string constants.
2585
2586 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2587 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2588 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2589 these, but not everybody uses configure.
2590
2591 19. A conditional group that had only one branch was not being correctly
2592 recognized as an item that could match an empty string. This meant that an
2593 enclosing group might also not be so recognized, causing infinite looping
2594 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2595 with the subject "ab", where knowledge that the repeated group can match
2596 nothing is needed in order to break the loop.
2597
2598 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2599 exec(), but without supplying a callout function, matching went wrong.
2600
2601 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2602 leak if the size of the offset vector was greater than 30. When the vector
2603 is smaller, the saved offsets during recursion go onto a local stack
2604 vector, but for larger vectors malloc() is used. It was failing to free
2605 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2606 error, in fact).
2607
2608 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2609 heapframe that is used only when UTF-8 support is enabled. This caused no
2610 problem, but was untidy.
2611
2612 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2613 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2614 included within another project.
2615
2616 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2617 slightly modified by me:
2618
2619 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2620 not building pcregrep.
2621
2622 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2623 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2624
2625 25. Forward references, both numeric and by name, in patterns that made use of
2626 duplicate group numbers, could behave incorrectly or give incorrect errors,
2627 because when scanning forward to find the reference group, PCRE was not
2628 taking into account the duplicate group numbers. A pattern such as
2629 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2630
2631 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2632 the feature of a custom pointer more persuasive (as requested by a user).
2633
2634 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2635 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2636 pre-defined.
2637
2638 28. Added support for (*UTF8) at the start of a pattern.
2639
2640 29. Arrange for flags added by the "release type" setting in CMake to be shown
2641 in the configuration summary.
2642
2643
2644 Version 7.8 05-Sep-08
2645 ---------------------
2646
2647 1. Replaced UCP searching code with optimized version as implemented for Ad
2648 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2649 stage table and inline lookup instead of a function, giving speed ups of 2
2650 to 5 times on some simple patterns that I tested. Permission was given to
2651 distribute the MultiStage2.py script that generates the tables (it's not in
2652 the tarball, but is in the Subversion repository).
2653
2654 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2655 scripts.
2656
2657 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2658 a group with a zero qualifier. The result of the study could be incorrect,
2659 or the function might crash, depending on the pattern.
2660
2661 4. Caseless matching was not working for non-ASCII characters in back
2662 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2663 It now works when Unicode Property Support is available.
2664
2665 5. In pcretest, an escape such as \x{de} in the data was always generating
2666 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2667 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2668 truncation.
2669
2670 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2671
2672 7. Added two (int) casts to pcregrep when printing the difference of two
2673 pointers, in case they are 64-bit values.
2674
2675 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2676 test 2 if it fails.
2677
2678 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2679 and a #define of that name to empty if it is not externally set. This is to
2680 allow users of MSVC to set it if necessary.
2681
2682 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2683 the convenience functions in the pcre_get.c source file.
2684
2685 11. An option change at the start of a pattern that had top-level alternatives
2686 could cause overwriting and/or a crash. This command provoked a crash in
2687 some environments:
2688
2689 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2690
2691 This potential security problem was recorded as CVE-2008-2371.
2692
2693 12. For a pattern where the match had to start at the beginning or immediately
2694 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2695 pcre_dfa_exec() could read past the end of the passed subject if there was
2696 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2697 pcretest so that it places the subject at the end of its malloc-ed buffer.
2698
2699 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2700 exec() might read past the end of the data buffer in UTF-8 mode.
2701
2702 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2703 the data contained the byte 0x85 as part of a UTF-8 character within its
2704 first line. This applied both to normal and DFA matching.
2705
2706 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2707 /^[^d]*?$/8 failed to match "abc".
2708
2709 16. Added a missing copyright notice to pcrecpp_internal.h.
2710
2711 17. Make it more clear in the documentation that values returned from
2712 pcre_exec() in ovector are byte offsets, not character counts.
2713
2714 18. Tidied a few places to stop certain compilers from issuing warnings.
2715
2716 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2717 supplied by Stefan Weber. I made a further small update for 7.8 because
2718 there is a change of source arrangements: the pcre_searchfuncs.c module is
2719 replaced by pcre_ucd.c.
2720
2721
2722 Version 7.7 07-May-08
2723 ---------------------
2724
2725 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2726 a string to a long long, pretend we don't even have a long long." This is
2727 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2728
2729 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2730 pre-7.6 versions, which defined a global no_arg variable instead of putting
2731 it in the RE class. (See also #8 below.)
2732
2733 3. Remove a line of dead code, identified by coverity and reported by Nuno
2734 Lopes.
2735
2736 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2737
2738 (1) The include/exclude patterns were being applied to the whole pathnames
2739 of files, instead of just to the final components.
2740
2741 (2) If there was more than one level of directory, the subdirectories were
2742 skipped unless they satisfied the include/exclude conditions. This is
2743 inconsistent with GNU grep (and could even be seen as contrary to the
2744 pcregrep specification - which I improved to make it absolutely clear).
2745 The action now is always to scan all levels of directory, and just
2746 apply the include/exclude patterns to regular files.
2747
2748 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2749 --exclude_dir in the tests to avoid scanning .svn directories.
2750
2751 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2752 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2753 doesn't support NULs in patterns.
2754
2755 7. Added some missing "const"s to declarations of static tables in
2756 pcre_compile.c and pcre_dfa_exec.c.
2757
2758 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2759 caused by fix #2 above. (Subsequently also a second patch to fix the
2760 first patch. And a third patch - this was a messy problem.)
2761
2762 9. Applied Craig's patch to remove the use of push_back().
2763
2764 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2765 matching function regexec().
2766
2767 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2768 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2769 references. PCRE supports relative numbers with this syntax (I don't think
2770 Oniguruma does).
2771
2772 12. Previously, a group with a zero repeat such as (...){0} was completely
2773 omitted from the compiled regex. However, this means that if the group
2774 was called as a subroutine from elsewhere in the pattern, things went wrong
2775 (an internal error was given). Such groups are now left in the compiled
2776 pattern, with a new opcode that causes them to be skipped at execution
2777 time.
2778
2779 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2780 to the way PCRE behaves:
2781
2782 (a) A lone ] character is dis-allowed (Perl treats it as data).
2783
2784 (b) A back reference to an unmatched subpattern matches an empty string
2785 (Perl fails the current match path).
2786
2787 (c) A data ] in a character class must be notated as \] because if the
2788 first data character in a class is ], it defines an empty class. (In
2789 Perl it is not possible to have an empty class.) The empty class []
2790 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2791 The negative empty class [^] matches any one character, independently
2792 of the DOTALL setting.
2793
2794 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2795 non-existent subpattern following a character class starting with ']' and
2796 containing () gave an internal compiling error instead of "reference to
2797 non-existent subpattern". Fortunately, when the pattern did exist, the
2798 compiled code was correct. (When scanning forwards to check for the
2799 existence of the subpattern, it was treating the data ']' as terminating
2800 the class, so got the count wrong. When actually compiling, the reference
2801 was subsequently set up correctly.)
2802
2803 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2804 it was being rejected as not supported by pcre_dfa_exec(), even though
2805 other assertions are supported. I have made pcre_dfa_exec() support
2806 (*FAIL).
2807
2808 16. The implementation of 13c above involved the invention of a new opcode,
2809 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2810 cannot be changed at match time, I realized I could make a small
2811 improvement to matching performance by compiling OP_ALLANY instead of
2812 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2813 on the OP_ANY path.
2814
2815 17. Compiling pcretest on Windows with readline support failed without the
2816 following two fixes: (1) Make the unistd.h include conditional on
2817 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2818
2819 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2820 ncurses library to be included for pcretest when ReadLine support is
2821 requested, but also to allow for it to be overridden. This patch came from
2822 Daniel Bergström.
2823
2824 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2825 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2826 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2827 spotting this.
2828
2829
2830 Version 7.6 28-Jan-08
2831 ---------------------
2832
2833 1. A character class containing a very large number of characters with
2834 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2835 overflow.
2836
2837 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2838 HAVE_LONG_LONG is not defined.
2839
2840 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2841 bring them up to date and include new features. This patch includes:
2842
2843 - Fixed PH's badly added libz and libbz2 support.
2844 - Fixed a problem with static linking.
2845 - Added pcredemo. [But later removed - see 7 below.]
2846 - Fixed dftables problem and added an option.
2847 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2848 HAVE_LONG_LONG.
2849 - Added readline support for pcretest.
2850 - Added an listing of the option settings after cmake has run.
2851
2852 4. A user submitted a patch to Makefile that makes it easy to create
2853 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2854 Makefile.am that cause it to include this special target, without
2855 affecting anything else. Note that the same mingw target plus all
2856 the other distribution libraries and programs are now supported
2857 when configuring with CMake (see 6 below) instead of with
2858 Configure/Make.
2859
2860 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2861 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2862 exported in the Windows port". It has not yet been confirmed that the patch
2863 solves the problem, but it does no harm.
2864
2865 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2866 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2867 with CMake, and also correct the comment about stack recursion.
2868
2869 7. Remove the automatic building of pcredemo from the ./configure system and
2870 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2871 of a program that users should build themselves after PCRE is installed, so
2872 building it automatically is not really right. What is more, it gave
2873 trouble in some build environments.
2874
2875 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2876
2877
2878 Version 7.5 10-Jan-08
2879 ---------------------
2880
2881 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2882 values in parens when parsing an RE using the C++ wrapper."
2883
2884 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2885 Characters greater than 255 were excluded from the class instead of being
2886 included.
2887
2888 3. The same bug as (2) above applied to negated POSIX classes such as
2889 [:^space:].
2890
2891 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2892 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2893 I have changed it.
2894
2895 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2896 first named subpattern) and a construct such as (?&a) would reference the
2897 first named subpattern whose name started with "a" (in other words, the
2898 length check was missing). Both these problems are fixed. "Subpattern name
2899 expected" is now given for (?&) (a zero-length name), and this patch also
2900 makes it give the same error for \k'' (previously it complained that that
2901 was a reference to a non-existent subpattern).
2902
2903 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2904 this is right because (?- can be followed by option settings as well as by
2905 digits. I have, however, made the messages clearer.
2906
2907 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2908 than the number used in the conditional) now cause a compile-time error.
2909 This is actually not compatible with Perl, which accepts such patterns, but
2910 treats the conditional as always being FALSE (as PCRE used to), but it
2911 seems to me that giving a diagnostic is better.
2912
2913 8. Change "alphameric" to the more common word "alphanumeric" in comments
2914 and messages.
2915
2916 9. Fix two occurrences of "backslash" in comments that should have been
2917 "backspace".
2918
2919 10. Remove two redundant lines of code that can never be obeyed (their function
2920 was moved elsewhere).
2921
2922 11. The program that makes PCRE's Unicode character property table had a bug
2923 which caused it to generate incorrect table entries for sequences of
2924 characters that have the same character type, but are in different scripts.
2925 It amalgamated them into a single range, with the script of the first of
2926 them. In other words, some characters were in the wrong script. There were
2927 thirteen such cases, affecting characters in the following ranges:
2928
2929 U+002b0 - U+002c1
2930 U+0060c - U+0060d
2931 U+0061e - U+00612
2932 U+0064b - U+0065e
2933 U+0074d - U+0076d
2934 U+01800 - U+01805
2935 U+01d00 - U+01d77
2936 U+01d9b - U+01dbf
2937 U+0200b - U+0200f
2938 U+030fc - U+030fe
2939 U+03260 - U+0327f
2940 U+0fb46 - U+0fbb1
2941 U+10450 - U+1049d
2942
2943 12. The -o option (show only the matching part of a line) for pcregrep was not
2944 compatible with GNU grep in that, if there was more than one match in a
2945 line, it showed only the first of them. It now behaves in the same way as
2946 GNU grep.
2947
2948 13. If the -o and -v options were combined for pcregrep, it printed a blank
2949 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2950 does the same. The return code can be used to tell if there were any
2951 non-matching lines.
2952
2953 14. Added --file-offsets and --line-offsets to pcregrep.
2954
2955 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2956 infinitely looping recursion. The bug was that positive lookaheads were not
2957 being skipped when checking for a possible empty match (negative lookaheads
2958 and both kinds of lookbehind were skipped).
2959
2960 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2961 inclusion of <windows.h> to before rather than after the definition of
2962 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2963
2964 17. Specifying a possessive quantifier with a specific limit for a Unicode
2965 character property caused pcre_compile() to compile bad code, which led at
2966 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2967 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2968 caused the error; without that there was no problem.
2969
2970 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2971
2972 19. Added --enable-pcretest-libreadline.
2973
2974 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2975 RE::GlobalReplace(). As a result, the number of replacements returned was
2976 double what it should be. I removed one of the increments, but Craig sent a
2977 later patch that removed the other one (the right fix) and added unit tests
2978 that check the return values (which was not done before).
2979
2980 21. Several CMake things:
2981
2982 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2983 the names libpcre and libpcreposix, not just pcre and pcreposix.
2984
2985 (2) The above change means that pcretest and pcregrep are now correctly
2986 linked with the newly-built libraries, not previously installed ones.
2987
2988 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2989
2990 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2991 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2992 UTF-8 newline character). The key issue is that the pattern starts .*;
2993 this means that the match must be either at the beginning, or after a
2994 newline. The bug was in the code for advancing after a failed match and
2995 checking that the new position followed a newline. It was not taking
2996 account of UTF-8 characters correctly.
2997
2998 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2999 character classes. PCRE was not treating the sequence [:...:] as a
3000 character class unless the ... were all letters. Perl, however, seems to
3001 allow any characters between [: and :], though of course it rejects as
3002 unknown any "names" that contain non-letters, because all the known class
3003 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3004 for example, whereas PCRE did not - it did not recognize a POSIX character
3005 class. This seemed a bit dangerous, so the code has been changed to be
3006 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3007 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3008 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3009 Perl does, and where it didn't before.
3010
3011 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3012 Windows environments %n is disabled by default.
3013
3014
3015 Version 7.4 21-Sep-07
3016 ---------------------
3017
3018 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3019 means that a class such as [\s] counted as "explicit reference to CR or
3020 LF". That isn't really right - the whole point of the change was to try to
3021 help when there was an actual mention of one of the two characters. So now
3022 the change happens only if \r or \n (or a literal CR or LF) character is
3023 encountered.
3024
3025 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3026 of both had grown to the point where there were only 3 bits left.
3027 Fortunately, there was spare space in the data structure, and so I have
3028 moved the internal flags into a new 16-bit field to free up more option
3029 bits.
3030
3031 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3032 but did not set the internal JCHANGED flag - either of these is enough to
3033 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3034 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3035 start sets both bits.
3036
3037 4. Added options (at build time, compile time, exec time) to change \R from
3038 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3039
3040 5. doc/pcresyntax.html was missing from the distribution.
3041
3042 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3043 compatibility, even though it is no longer used.
3044
3045 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3046 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3047 windows.h file is present (where different names are used). [This was
3048 reversed later after testing - see 16 below.]
3049
3050 8. Changed all #include <config.h> to #include "config.h". There were also
3051 some further <pcre.h> cases that I changed to "pcre.h".
3052
3053 9. When pcregrep was used with the --colour option, it missed the line ending
3054 sequence off the lines that it output.
3055
3056 10. It was pointed out to me that arrays of string pointers cause lots of
3057 relocations when a shared library is dynamically loaded. A technique of
3058 using a single long string with a table of offsets can drastically reduce
3059 these. I have refactored PCRE in four places to do this. The result is
3060 dramatic:
3061
3062 Originally: 290
3063 After changing UCP table: 187
3064 After changing error message table: 43
3065 After changing table of "verbs" 36
3066 After changing table of Posix names 22
3067
3068 Thanks to the folks working on Gregex for glib for this insight.
3069
3070 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3071 unicode-properties was also set.
3072
3073 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3074
3075 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3076 checked only for CRLF.
3077
3078 14. Added casts to pcretest.c to avoid compiler warnings.
3079
3080 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3081
3082 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3083 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3084 entirely. This removes changes made in 7 above.
3085
3086 17. The CMake files have been updated, and there is now more information about
3087 building with CMake in the NON-UNIX-USE document.
3088
3089
3090 Version 7.3 28-Aug-07
3091 ---------------------
3092
3093 1. In the rejigging of the build system that eventually resulted in 7.1, the
3094 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3095 brackets there is not right, since it causes compilers to look for an
3096 installed pcre.h, not the version that is in the source that is being
3097 compiled (which of course may be different). I have changed it back to:
3098
3099 #include "pcre.h"
3100
3101 I have a vague recollection that the change was concerned with compiling in
3102 different directories, but in the new build system, that is taken care of
3103 by the VPATH setting the Makefile.
3104
3105 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3106 when the subject happened to end in the byte 0x85 (e.g. if the last
3107 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3108 characters but of course it shouldn't be taken as a newline when it is part
3109 of another character. The bug was that, for an unlimited repeat of . in
3110 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3111 characters when looking for a newline.
3112
3113 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3114
3115 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3116 in debug output.
3117
3118 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3119 long printing in the pcrecpp unittest when running under MinGW.
3120
3121 6. ESC_K was left out of the EBCDIC table.
3122
3123 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3124 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3125 limit also applies to "virtual nesting" when a pattern is recursive, and in
3126 this case 1000 isn't so big. I have been able to remove this limit at the
3127 expense of backing off one optimization in certain circumstances. Normally,
3128 when pcre_exec() would call its internal match() function recursively and
3129 immediately return the result unconditionally, it uses a "tail recursion"
3130 feature to save stack. However, when a subpattern that can match an empty
3131 string has an unlimited repetition quantifier, it no longer makes this
3132 optimization. That gives it a stack frame in which to save the data for
3133 checking that an empty string has been matched. Previously this was taken
3134 from the 1000-entry workspace that had been reserved. So now there is no
3135 explicit limit, but more stack is used.
3136
3137 8. Applied Daniel's patches to solve problems with the import/export magic
3138 syntax that is required for Windows, and which was going wrong for the
3139 pcreposix and pcrecpp parts of the library. These were overlooked when this
3140 problem was solved for the main library.
3141
3142 9. There were some crude static tests to avoid integer overflow when computing
3143 the size of patterns that contain repeated groups with explicit upper
3144 limits. As the maximum quantifier is 65535, the maximum group length was
3145 set at 30,000 so that the product of these two numbers did not overflow a
3146 32-bit integer. However, it turns out that people want to use groups that
3147 are longer than 30,000 bytes (though not repeat them that many times).
3148 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3149 made it possible to implement the integer overflow checks in a much more
3150 dynamic way, which I have now done. The artificial limitation on group
3151 length has been removed - we now have only the limit on the total length of
3152 the compiled pattern, which depends on the LINK_SIZE setting.
3153
3154 10. Fixed a bug in the documentation for get/copy named substring when
3155 duplicate names are permitted. If none of the named substrings are set, the
3156 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3157 empty string.
3158
3159 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3160 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3161 because the ] is interpreted as the first data character and the
3162 terminating ] is not found. PCRE has been made compatible with Perl in this
3163 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3164 cause memory overwriting.
3165
3166 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3167 string has been matched (to stop an infinite loop). It was not recognizing
3168 a conditional subpattern that could match an empty string if that
3169 subpattern was within another subpattern. For example, it looped when
3170 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3171 condition was not nested. This bug has been fixed.
3172
3173 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3174 past the start of the subject in the presence of bytes with the top bit
3175 set, for example "\x8aBCD".
3176
3177 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3178 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3179
3180 14. Optimized (?!) to (*FAIL).
3181
3182 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3183 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3184 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3185 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3186 does: it's just the validity check that is more restrictive.
3187
3188 16. Inserted checks for integer overflows during escape sequence (backslash)
3189 processing, and also fixed erroneous offset values for syntax errors during
3190 backslash processing.
3191
3192 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3193 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3194
3195 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3196 caused an overrun.
3197
3198 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3199 something other than just ASCII characters) inside a group that had an
3200 unlimited repeat caused a loop at compile time (while checking to see
3201 whether the group could match an empty string).
3202
3203 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3204 [\P{Any}] did so. (Error in the code for printing property names.)
3205
3206 21. An orphan \E inside a character class could cause a crash.
3207
3208 22. A repeated capturing bracket such as (A)? could cause a wild memory
3209 reference during compilation.
3210
3211 23. There are several functions in pcre_compile() that scan along a compiled
3212 expression for various reasons (e.g. to see if it's fixed length for look
3213 behind). There were bugs in these functions when a repeated \p or \P was
3214 present in the pattern. These operators have additional parameters compared
3215 with \d, etc, and these were not being taken into account when moving along
3216 the compiled data. Specifically:
3217
3218 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3219 length.
3220
3221 (b) An item such as \pL+ within a repeated group could cause crashes or
3222 loops.
3223
3224 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3225 "reference to non-existent subpattern" error.
3226
3227 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3228
3229 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3230 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3231
3232 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3233
3234 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3235 character were causing crashes (broken optimization).
3236
3237 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3238 \p or \P) caused a compile-time loop.
3239
3240 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3241 break. For example, the unstudied pattern [\r\n]A does not match the string
3242 "\r\nA" because change 7.0/46 below moves the current point on by two
3243 characters after failing to match at the start. However, the pattern \nA
3244 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3245 the same is true. There doesn't seem any very clean way out of this, but
3246 what I have chosen to do makes the common cases work: PCRE now takes note
3247 of whether there can be an explicit match for \r or \n anywhere in the
3248 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3249 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3250 pattern has explicit CR or LF references.
3251
3252 29. Added (*CR) etc for changing newline setting at start of pattern.
3253
3254
3255 Version 7.2 19-Jun-07
3256 ---------------------
3257
3258 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3259 which is apparently normally available under Windows.
3260
3261 2. Re-jig the pcregrep tests with different newline settings in an attempt
3262 to make them independent of the local environment's newline setting.
3263
3264 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3265
3266 4. Some of the "internals" tests were previously cut out when the link size
3267 was not 2, because the output contained actual offsets. The recent new
3268 "Z" feature of pcretest means that these can be cut out, making the tests
3269 usable with all link sizes.
3270
3271 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3272 stack recursion. This gives a massive performance boost under BSD, but just
3273 a small improvement under Linux. However, it saves one field in the frame
3274 in all cases.
3275
3276 6. Added more features from the forthcoming Perl 5.10:
3277
3278 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3279 recursion call. It refers to the nth most recently opened parentheses.
3280
3281 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3282 to be opened parentheses.
3283
3284 (c) Conditions that refer to capturing parentheses can be specified
3285 relatively, for example, (?(-2)... or (?(+3)...
3286
3287 (d) \K resets the start of the current match so that everything before
3288 is not part of it.
3289
3290 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3291
3292 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3293 reference syntax.
3294
3295 (g) (?| introduces a group in which the numbering of parentheses in each
3296 alternative starts with the same number.
3297
3298 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3299
3300 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3301 PCRE_INFO_JCHANGED.
3302
3303 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3304 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3305 for detecting groups that can match an empty string.
3306
3307 9. A pattern with a very large number of alternatives (more than several
3308 hundred) was running out of internal workspace during the pre-compile
3309 phase, where pcre_compile() figures out how much memory will be needed. A
3310 bit of new cunning has reduced the workspace needed for groups with
3311 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3312 workspace instead of running out of the 4096 that are available.
3313
3314 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3315
3316 11. Applied patch from Google to remove an optimization that didn't quite work.
3317 The report of the bug said:
3318
3319 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3320 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3321 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3322
3323 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3324 it matched the wrong number of bytes.
3325
3326
3327 Version 7.1 24-Apr-07
3328 ---------------------
3329
3330 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3331 that is more "standard", making use of automake and other Autotools. There
3332 is some re-arrangement of the files and adjustment of comments consequent
3333 on this.
3334
3335 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3336 for recursive directory scanning broke on some systems because the files
3337 are not scanned in any specific order and on different systems the order
3338 was different. A call to "sort" has been inserted into RunGrepTest for the
3339 approprate test as a short-term fix. In the longer term there may be an
3340 alternative.
3341
3342 3. I had an email from Eric Raymond about problems translating some of PCRE's
3343 man pages to HTML (despite the fact that I distribute HTML pages, some
3344 people do their own conversions for various reasons). The problems
3345 concerned the use of low-level troff macros .br and .in. I have therefore
3346 removed all such uses from the man pages (some were redundant, some could
3347 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3348 HTML has been updated to handle .nf/.fi and to complain if it encounters
3349 .br or .in.
3350
3351 4. Updated comments in configure.ac that get placed in config.h.in and also
3352 arranged for config.h to be included in the distribution, with the name
3353 config.h.generic, for the benefit of those who have to compile without
3354 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3355
3356 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3357 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3358 makevp.bat for the current PCRE, using the additional files
3359 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3360
3361 6. A Windows user reported a minor discrepancy with test 2, which turned out
3362 to be caused by a trailing space on an input line that had got lost in his
3363 copy. The trailing space was an accident, so I've just removed it.
3364
3365 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3366 that is needed.
3367
3368 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3369 as "const" (a) because they are and (b) because it helps the PHP
3370 maintainers who have recently made a script to detect big data structures
3371 in the php code that should be moved to the .rodata section. I remembered
3372 to update Builducptable as well, so it won't revert if ucptable.h is ever
3373 re-created.
3374
3375 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3376 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3377 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3378 support is not required. This saves 1.5-2K of code, which is important in
3379 some applications.
3380
3381 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3382 so as not to refer to the tables, even though these functions will never be
3383 called when UTF-8 support is disabled. Otherwise there are problems with a
3384 shared library.
3385
3386 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3387
3388 (a) It was defining its arguments as char * instead of void *.
3389
3390 (b) It was assuming that all moves were upwards in memory; this was true
3391 a long time ago when I wrote it, but is no longer the case.
3392
3393 The emulated memove() is provided for those environments that have neither
3394 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3395 is clearly not the case, as these two bugs were recently reported.
3396
3397 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3398 and Detrail to create the HTML documentation, the .txt form of the man
3399 pages, and it removes trailing spaces from listed files. It also creates
3400 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3401 case, it wraps all the #defines with #ifndefs. This script should be run
3402 before "make dist".
3403
3404 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3405 with Unicode property support.
3406
3407 (a) For a maximizing quantifier, if the two different cases of the
3408 character were of different lengths in their UTF-8 codings (there are
3409 some cases like this - I found 11), and the matching function had to
3410 back up over a mixture of the two cases, it incorrectly assumed they
3411 were both the same length.
3412
3413 (b) When PCRE was configured to use the heap rather than the stack for
3414 recursion during matching, it was not correctly preserving the data for
3415 the other case of a UTF-8 character when checking ahead for a match
3416 while processing a minimizing repeat. If the check also involved
3417 matching a wide character, but failed, corruption could cause an
3418 erroneous result when trying to check for a repeat of the original
3419 character.
3420
3421 13. Some tidying changes to the testing mechanism:
3422
3423 (a) The RunTest script now detects the internal link size and whether there
3424 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3425 values substituted by "configure". (The RunGrepTest script already did
3426 this for UTF-8.) The configure.ac script no longer substitutes the
3427 relevant variables.
3428
3429 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3430 with length and offset values. This means that the output is different
3431 for different internal link sizes. Test 2 is skipped for link sizes
3432 other than 2 because of this, bypassing the problem. Unfortunately,
3433 there was also a test in test 3 (the locale tests) that used /B and
3434 failed for link sizes other than 2. Rather than cut the whole test out,
3435 I have added a new /Z option to pcretest that replaces the length and
3436 offset values with spaces. This is now used to make test 3 independent
3437 of link size. (Test 2 will be tidied up later.)
3438
3439 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3440 segmentation fault instead of returning the appropriate error message.
3441
3442 15. In multiline mode when the newline sequence was set to "any", the pattern
3443 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3444 This doesn't seem right; it now treats the CRLF combination as the line
3445 ending, and so does not match in that case. It's only a pattern such as ^$
3446 that would hit this one: something like ^ABC$ would have failed after \r
3447 and then tried again after \r\n.
3448
3449 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3450 in an attempt to make files that differ only in their line terminators
3451 compare equal. This works on Linux.
3452
3453 17. Under certain error circumstances pcregrep might try to free random memory
3454 as it exited. This is now fixed, thanks to valgrind.
3455
3456 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3457 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3458 was because its rules for how to advance for /g after matching an empty
3459 string at the end of a line did not allow for this case. They now check for
3460 it specially.
3461
3462 20. pcretest is supposed to handle patterns and data of any length, by
3463 extending its buffers when necessary. It was getting this wrong when the
3464 buffer for a data line had to be extended.
3465
3466 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3467 CRLF as a newline sequence.
3468
3469 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3470 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3471 I have nevertheless tidied it up.
3472
3473 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3474
3475 24. Added a man page for pcre-config.
3476
3477
3478 Version 7.0 19-Dec-06
3479 ---------------------
3480
3481 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3482 moving to gcc 4.1.1.
3483
3484 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3485 sys/time.h, which is documented as needed for this function. It doesn't
3486 seem to matter on Linux, but it showed up on some releases of OS X.
3487
3488 3. It seems that there are systems where bytes whose values are greater than
3489 127 match isprint() in the "C" locale. The "C" locale should be the
3490 default when a C program starts up. In most systems, only ASCII printing
3491 characters match isprint(). This difference caused the output from pcretest
3492 to vary, making some of the tests fail. I have changed pcretest so that:
3493
3494 (a) When it is outputting text in the compiled version of a pattern, bytes
3495 other than 32-126 are always shown as hex escapes.
3496
3497 (b) When it is outputting text that is a matched part of a subject string,
3498 it does the same, unless a different locale has been set for the match
3499 (using the /L modifier). In this case, it uses isprint() to decide.
3500
3501 4. Fixed a major bug that caused incorrect computation of the amount of memory
3502 required for a compiled pattern when options that changed within the
3503 pattern affected the logic of the preliminary scan that determines the
3504 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3505 that the computed length was too small. The symptoms of this bug were
3506 either the PCRE error "internal error: code overflow" from pcre_compile(),
3507 or a glibc crash with a message such as "pcretest: free(): invalid next
3508 size (fast)". Examples of patterns that provoked this bug (shown in
3509 pcretest format) are:
3510
3511 /(?-x: )/x
3512 /(?x)(?-x: \s*#\s*)/
3513 /((?i)[\x{c0}])/8
3514 /(?i:[\x{c0}])/8
3515
3516 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3517 is now done differently.
3518
3519 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3520 wrapper classes; (b) implement a new function in the C++ scanner that is
3521 more efficient than the old way of doing things because it avoids levels of
3522 recursion in the regex matching; (c) add a paragraph to the documentation
3523 for the FullMatch() function.
3524
3525 6. The escape sequence \n was being treated as whatever was defined as
3526 "newline". Not only was this contrary to the documentation, which states
3527 that \n is character 10 (hex 0A), but it also went horribly wrong when
3528 "newline" was defined as CRLF. This has been fixed.
3529
3530 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3531 was being set to -1 for the "end of line" case (supposedly a value that no
3532 character can have). Though this value is never used (the check for end of
3533 line is "zero bytes in current character"), it caused compiler complaints.
3534 I've changed it to 0xffffffff.
3535
3536 8. In pcre_version.c, the version string was being built by a sequence of
3537 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3538 string (as it is for production releases) called a macro with an empty
3539 argument. The C standard says the result of this is undefined. The gcc
3540 compiler treats it as an empty string (which was what was wanted) but it is
3541 reported that Visual C gives an error. The source has been hacked around to
3542 avoid this problem.
3543
3544 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3545 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3546 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3547 of them did).
3548
3549 10. Originally, pcretest opened its input and output without "b"; then I was
3550 told that "b" was needed in some environments, so it was added for release
3551 5.0 to both the input and output. (It makes no difference on Unix-like
3552 systems.) Later I was told that it is wrong for the input on Windows. I've
3553 now abstracted the modes into two macros, to make it easier to fiddle with
3554 them, and removed "b" from the input mode under Windows.
3555
3556 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3557
3558 12. Added -help and --help to pcretest as an official way of being reminded
3559 of the options.
3560
3561 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3562 and pcrecpp.cc because they annoy compilers at high warning levels.
3563
3564 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3565
3566 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3567 scripts are not C programs :-) and which was not noticed because it works
3568 on Linux.
3569
3570 16. pcretest is supposed to handle any length of pattern and data line (as one
3571 line or as a continued sequence of lines) by extending its input buffer if
3572 necessary. This feature was broken for very long pattern lines, leading to
3573 a string of junk being passed to pcre_compile() if the pattern was longer
3574 than about 50K.
3575
3576 17. I have done a major re-factoring of the way pcre_compile() computes the
3577 amount of memory needed for a compiled pattern. Previously, there was code
3578 that made a preliminary scan of the pattern in order to do this. That was
3579 OK when PCRE was new, but as the facilities have expanded, it has become
3580 harder and harder to keep it in step with the real compile phase, and there
3581 have been a number of bugs (see for example, 4 above). I have now found a
3582 cunning way of running the real compile function in a "fake" mode that
3583 enables it to compute how much memory it would need, while actually only
3584 ever using a few hundred bytes of working memory and without too many
3585 tests of the mode. This should make future maintenance and development
3586 easier. A side effect of this work is that the limit of 200 on the nesting
3587 depth of parentheses has been removed (though this was never a serious
3588 limitation, I suspect). However, there is a downside: pcre_compile() now
3589 runs more slowly than before (30% or more, depending on the pattern). I
3590 hope this isn't a big issue. There is no effect on runtime performance.
3591
3592 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3593 newline (only possible for the last line of a file) and it was a
3594 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3595
3596 19. Added additional timing features to pcretest. (1) The -tm option now times
3597 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3598 separate command line item, by a number that specifies the number of
3599 repeats to use when timing. The default is 50000; this gives better
3600 precision, but takes uncomfortably long for very large patterns.
3601
3602 20. Extended pcre_study() to be more clever in cases where a branch of a
3603 subpattern has no definite first character. For example, (a*|b*)[cd] would
3604 previously give no result from pcre_study(). Now it recognizes that the
3605 first character must be a, b, c, or d.
3606
3607 21. There was an incorrect error "recursive call could loop indefinitely" if
3608 a subpattern (or the entire pattern) that was being tested for matching an
3609 empty string contained only one non-empty item after a nested subpattern.
3610 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3611 incorrectly, because the \d was being skipped in the check.
3612
3613 22. The pcretest program now has a new pattern option /B and a command line
3614 option -b, which is equivalent to adding /B to every pattern. This causes
3615 it to show the compiled bytecode, without the additional information that
3616 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3617 is the same as /B/I).
3618
3619 23. A new optimization is now able automatically to treat some sequences such
3620 as a*b as a*+b. More specifically, if something simple (such as a character
3621 or a simple class like \d) has an unlimited quantifier, and is followed by
3622 something that cannot possibly match the quantified thing, the quantifier
3623 is automatically "possessified".
3624
3625 24. A recursive reference to a subpattern whose number was greater than 39
3626 went wrong under certain circumstances in UTF-8 mode. This bug could also
3627 have affected the operation of pcre_study().
3628
3629 25. Realized that a little bit of performance could be had by replacing
3630 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3631
3632 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3633
3634 27. Possessive quantifiers such as a++ were previously implemented by turning
3635 them into atomic groups such as ($>a+). Now they have their own opcodes,
3636 which improves performance. This includes the automatically created ones
3637 from 23 above.
3638
3639 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3640 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3641 the first matched character to be a colon. This applied both to named and
3642 numbered groups.
3643
3644 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3645
3646 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3647 building PCRE on Windows easier, so I have included it in the distribution.
3648
3649 31. There is now a check in pcretest against a ridiculously large number being
3650 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3651 loop, the loop is abandoned.
3652
3653 32. Forward references to subpatterns in conditions such as (?(2)...) where
3654 subpattern 2 is defined later cause pcre_compile() to search forwards in
3655 the pattern for the relevant set of parentheses. This search went wrong
3656 when there were unescaped parentheses in a character class, parentheses
3657 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3658
3659 33. "Subroutine" calls and backreferences were previously restricted to
3660 referencing subpatterns earlier in the regex. This restriction has now
3661 been removed.
3662
3663 34. Added a number of extra features that are going to be in Perl 5.10. On the
3664 whole, these are just syntactic alternatives for features that PCRE had
3665 previously implemented using the Python syntax or my own invention. The
3666 other formats are all retained for compatibility.
3667
3668 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3669 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3670 also .NET compatible.
3671
3672 (b) A recursion or subroutine call to a named group can now be defined as
3673 (?&name) as well as (?P>name).
3674
3675 (c) A backreference to a named group can now be defined as \k<name> or
3676 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3677 5.10, are also .NET compatible.
3678
3679 (d) A conditional reference to a named group can now use the syntax
3680 (?(<name>) or (?('name') as well as (?(name).
3681
3682 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3683 groups (named and numbered) that are never evaluated inline, but can be
3684 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3685 is always false. There may be only one alternative in such a group.
3686
3687 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3688 as the simple (?(R). The condition is true only if the most recent
3689 recursion is that of the given number or name. It does not search out
3690 through the entire recursion stack.
3691
3692 (g) The escape \gN or \g{N} has been added, where N is a positive or
3693 negative number, specifying an absolute or relative reference.
3694
3695 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3696 some "unreachable code" warnings.
3697
3698 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3699 things, this adds five new scripts.
3700
3701 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3702 There were also incompatibilities regarding the handling of \Q..\E inside
3703 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3704 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3705
3706 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3707 matches an empty string, and forcibly breaks the loop. There were bugs in
3708 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3709 against aaaa the result was just "a" rather than "aaaa", for example. Two
3710 separate and independent bugs (that affected different cases) have been
3711 fixed.
3712
3713 39. Refactored the code to abolish the use of different opcodes for small
3714 capturing bracket numbers. This is a tidy that I avoided doing when I
3715 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3716 The new approach is not only tidier, it makes it possible to reduce the
3717 memory needed to fix the previous bug (38).
3718
3719 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3720 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3721 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3722 mode.
3723
3724 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3725 report.
3726
3727 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3728 copy construction and assignment in the C++ wrapper.
3729
3730 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3731 couple of bugs that could have given wrong results in the "--newline=crlf"
3732 case.
3733
3734 44. Added a number of casts and did some reorganization of signed/unsigned int
3735 variables following suggestions from Dair Grant. Also renamed the variable
3736 "this" as "item" because it is a C++ keyword.
3737
3738 45. Arranged for dftables to add
3739
3740 #include "pcre_internal.h"
3741
3742 to pcre_chartables.c because without it, gcc 4.x may remove the array
3743 definition from the final binary if PCRE is built into a static library and
3744 dead code stripping is activated.
3745
3746 46. For an unanchored pattern, if a match attempt fails at the start of a
3747 newline sequence, and the newline setting is CRLF or ANY, and the next two
3748 characters are CRLF, advance by two characters instead of one.
3749
3750
3751 Version 6.7 04-Jul-06
3752 ---------------------
3753
3754 1. In order to handle tests when input lines are enormously long, pcretest has
3755 been re-factored so that it automatically extends its buffers when
3756 necessary. The code is crude, but this _is_ just a test program. The
3757 default size has been increased from 32K to 50K.
3758
3759 2. The code in pcre_study() was using the value of the re argument before
3760 testing it for NULL. (Of course, in any sensible call of the function, it
3761 won't be NULL.)
3762
3763 3. The memmove() emulation function in pcre_internal.h, which is used on
3764 systems that lack both memmove() and bcopy() - that is, hardly ever -
3765 was missing a "static" storage class specifier.
3766
3767 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3768 containing an extended class (one that cannot be represented by a bitmap
3769 because it contains high-valued characters or Unicode property items, e.g.
3770 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3771 pattern, but PCRE should not loop if you do not (it no longer does).
3772 [Detail: two cases were found: (a) a repeated subpattern containing an
3773 extended class; (b) a recursive reference to a subpattern that followed a
3774 previous extended class. It wasn't skipping over the extended class
3775 correctly when UTF-8 mode was not set.]
3776
3777 5. A negated single-character class was not being recognized as fixed-length
3778 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3779 compile error "lookbehind assertion is not fixed length".
3780
3781 6. The RunPerlTest auxiliary script was showing an unexpected difference
3782 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3783 write a Perl script that can interpret lines of an input file either as
3784 byte characters or as UTF-8, which is what "perltest" was being required to
3785 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3786 can't do is switch easily at run time between having the "use utf8;" pragma
3787 or not. In the end, I fudged it by using the RunPerlTest script to insert
3788 "use utf8;" explicitly for the UTF-8 tests.
3789
3790 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3791 the end of the subject string, contrary to the documentation and to what
3792 Perl does. This was true of both matching functions. Now it matches only at
3793 the start of the subject and immediately after *internal* newlines.
3794
3795 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3796 a pointer to an int instead of a pointer to an unsigned long int. This
3797 caused problems on 64-bit systems.
3798
3799 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3800 instance of the 'standard' template library not being so standard".
3801
3802 10. There was no check on the number of named subpatterns nor the maximum
3803 length of a subpattern name. The product of these values is used to compute
3804 the size of the memory block for a compiled pattern. By supplying a very
3805 long subpattern name and a large number of named subpatterns, the size
3806 computation could be caused to overflow. This is now prevented by limiting
3807 the length of names to 32 characters, and the number of named subpatterns
3808 to 10,000.
3809
3810 11. Subpatterns that are repeated with specific counts have to be replicated in
3811 the compiled pattern. The size of memory for this was computed from the
3812 length of the subpattern and the repeat count. The latter is limited to
3813 65535, but there was no limit on the former, meaning that integer overflow
3814 could in principle occur. The compiled length of a repeated subpattern is
3815 now limited to 30,000 bytes in order to prevent this.
3816
3817 12. Added the optional facility to have named substrings with the same name.
3818
3819 13. Added the ability to use a named substring as a condition, using the
3820 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3821 are numbers (not recommended). Forward references are permitted.
3822
3823 14. Added forward references in named backreferences (if you see what I mean).
3824
3825 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3826 pattern could run off the end of the subject. For example, the pattern
3827 "(?s)(.{1,5})"8 did this with the subject "ab".
3828
3829 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3830 PCRE_CASELESS was set when matching characters that were quantified with ?
3831 or *.
3832
3833 17. A character class other than a single negated character that had a minimum
3834 but no maximum quantifier - for example [ab]{6,} - was not handled
3835 correctly by pce_dfa_exec(). It would match only one character.
3836
3837 18. A valid (though odd) pattern that looked like a POSIX character
3838 class but used an invalid character after [ (for example [[,abc,]]) caused
3839 pcre_compile() to give the error "Failed: internal error: code overflow" or
3840 in some cases to crash with a glibc free() error. This could even happen if
3841 the pattern terminated after [[ but there just happened to be a sequence of
3842 letters, a binary zero, and a closing ] in the memory that followed.
3843
3844 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3845 over the years. Originally (before any Unicode support), just the bottom 8
3846 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3847 output from "man perlunicode" includes this:
3848
3849 The regular expression compiler produces polymorphic opcodes. That
3850 is, the pattern adapts to the data and automatically switches to
3851 the Unicode character scheme when presented with Unicode data--or
3852 instead uses a traditional byte scheme when presented with byte
3853 data.
3854
3855 Sadly, a wide octal escape does not cause a switch, and in a string with
3856 no other multibyte characters, these octal escapes are treated as before.
3857 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3858 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3859 Unicode string.
3860
3861 I have not perpetrated such confusion in PCRE. Up till now, it took just
3862 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3863 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3864 translate to the appropriate multibyte character.
3865
3866 29. Applied some refactoring to reduce the number of warnings from Microsoft
3867 and Borland compilers. This has included removing the fudge introduced
3868 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3869 a warning about an unused variable.
3870
3871 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3872 characters since release 4.0, because Perl (from release 5.004) does not.
3873 [Or at least, is documented not to: some releases seem to be in conflict
3874 with the documentation.] However, when a pattern was studied with
3875 pcre_study() and all its branches started with \s, PCRE still included VT
3876 as a possible starting character. Of course, this did no harm; it just
3877 caused an unnecessary match attempt.
3878
3879 22. Removed a now-redundant internal flag bit that recorded the fact that case
3880 dependency changed within the pattern. This was once needed for "required
3881 byte" processing, but is no longer used. This recovers a now-scarce options
3882 bit. Also moved the least significant internal flag bit to the most-
3883 significant bit of the word, which was not previously used (hangover from
3884 the days when it was an int rather than a uint) to free up another bit for
3885 the future.
3886
3887 23. Added support for CRLF line endings as well as CR and LF. As well as the
3888 default being selectable at build time, it can now be changed at runtime
3889 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3890 specify that it is scanning data with non-default line endings.
3891
3892 24. Changed the definition of CXXLINK to make it agree with the definition of
3893 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3894
3895 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3896 recursions. This makes a big different to stack usage for some patterns.
3897
3898 26. If a subpattern containing a named recursion or subroutine reference such
3899 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3900 the space required for the compiled pattern went wrong and gave too small a
3901 value. Depending on the environment, this could lead to "Failed: internal
3902 error: code overflow at offset 49" or "glibc detected double free or
3903 corruption" errors.
3904
3905 27. Applied patches from Google (a) to support the new newline modes and (b) to
3906 advance over multibyte UTF-8 characters in GlobalReplace.
3907
3908 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3909 difference for some implementation of PCRE in some Windows version.
3910
3911 29. Added some extra testing facilities to pcretest:
3912
3913 \q<number> in a data line sets the "match limit" value
3914 \Q<number> in a data line sets the "match recursion limt" value
3915 -S <number> sets the stack size, where <number> is in megabytes
3916
3917 The -S option isn't available for Windows.
3918
3919
3920 Version 6.6 06-Feb-06
3921 ---------------------
3922
3923 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3924 in pcreposix.h. I have copied the definition from pcre.h.
3925
3926 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3927 because pcre.h is no longer a built file.
3928
3929 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3930 not normally included in the compiled code.
3931
3932
3933 Version 6.5 01-Feb-06
3934 ---------------------
3935
3936 1. When using the partial match feature with pcre_dfa_exec(), it was not
3937 anchoring the second and subsequent partial matches at the new starting
3938 point. This could lead to incorrect results. For example, with the pattern
3939 /1234/, partially matching against "123" and then "a4" gave a match.
3940
3941 2. Changes to pcregrep:
3942
3943 (a) All non-match returns from pcre_exec() were being treated as failures
3944 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3945 error message is output. Some extra information is given for the
3946 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3947 probably the only errors that are likely to be caused by users (by
3948 specifying a regex that has nested indefinite repeats, for instance).
3949 If there are more than 20 of these errors, pcregrep is abandoned.
3950
3951 (b) A binary zero was treated as data while matching, but terminated the
3952 output line if it was written out. This has been fixed: binary zeroes
3953 are now no different to any other data bytes.
3954
3955 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3956 used to set a locale for matching. The --locale=xxxx long option has
3957 been added (no short equivalent) to specify a locale explicitly on the
3958 pcregrep command, overriding the environment variables.
3959
3960 (d) When -B was used with -n, some line numbers in the output were one less
3961 than they should have been.
3962
3963 (e) Added the -o (--only-matching) option.
3964
3965 (f) If -A or -C was used with -c (count only), some lines of context were
3966 accidentally printed for the final match.
3967
3968 (g) Added the -H (--with-filename) option.
3969
3970 (h) The combination of options -rh failed to suppress file names for files
3971 that were found from directory arguments.
3972
3973 (i) Added the -D (--devices) and -d (--directories) options.
3974
3975 (j) Added the -F (--fixed-strings) option.
3976
3977 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3978
3979 (l) Added the --colo(u)r option.
3980
3981 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3982 is not present by default.
3983
3984 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3985 items such as (?R) or (?1), when the recursion could match a number of
3986 alternatives. If it matched one of the alternatives, but subsequently,
3987 outside the recursion, there was a failure, the code tried to back up into
3988 the recursion. However, because of the way PCRE is implemented, this is not
3989 possible, and the result was an incorrect result from the match.
3990
3991 In order to prevent this happening, the specification of recursion has
3992 been changed so that all such subpatterns are automatically treated as
3993 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3994
3995 4. I had overlooked the fact that, in some locales, there are characters for
3996 which isalpha() is true but neither isupper() nor islower() are true. In
3997 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3998 and ordfeminine) are like this. This affected the treatment of \w and \W
3999 when they appeared in character classes, but not when they appeared outside
4000 a character class. The bit map for "word" characters is now created
4001 separately from the results of isalnum() instead of just taking it from the
4002 upper, lower, and digit maps. (Plus the underscore character, of course.)
4003
4004 5. The above bug also affected the handling of POSIX character classes such as
4005 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4006 permanent tables. Instead, the bit maps for such a class were previously
4007 created as the appropriate unions of the upper, lower, and digit bitmaps.
4008 Now they are created by subtraction from the [[:word:]] class, which has
4009 its own bitmap.
4010
4011 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4012 It is created by subtracting the vertical space characters (\x09, \x0a,
4013 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4014 subtraction was done in the overall bitmap for a character class, meaning
4015 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4016 be recognized. This bug has been fixed.
4017
4018 7. Patches from the folks at Google:
4019
4020 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4021 real life, but is still worth protecting against".
4022
4023 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4024 regular expressions".
4025
4026 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4027 have it.
4028
4029 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4030 "configure" and the latter not, in order to fix a problem somebody had
4031 with compiling the Arg class on HP-UX.
4032
4033 (e) Improve the error-handling of the C++ wrapper a little bit.
4034
4035 (f) New tests for checking recursion limiting.
4036
4037 8. The pcre_memmove() function, which is used only if the environment does not
4038 have a standard memmove() function (and is therefore rarely compiled),
4039 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4040 returning a result (though PCRE never actually uses the result).
4041
4042 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4043 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4044 returned instead of calling malloc() with an overflowing number that would
4045 most likely cause subsequent chaos.
4046
4047 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4048
4049 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4050 with this option is matched, the nmatch and pmatch options of regexec() are
4051 ignored.
4052
4053 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4054 provided in case anyone wants to the the POSIX interface with UTF-8
4055 strings.
4056
4057 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4058 C++ linking (needed for some HP-UX environments).
4059
4060 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4061 (unused parameter) and in the pcre_printint() function (omitted "default"
4062 switch label when the default is to do nothing).
4063
4064 15. Added some code to make it possible, when PCRE is compiled as a C++
4065 library, to replace subject pointers for pcre_exec() with a smart pointer
4066 class, thus making it possible to process discontinuous strings.
4067
4068 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4069 much the same function. They were added by different people who were trying
4070 to make PCRE easy to compile on non-Unix systems. It has been suggested
4071 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4072 for compiling on Windows systems. I have therefore replaced it with
4073 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4074 defaults to "extern" for C or "extern C" for C++, which works fine on
4075 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4076 SCOPE with something explicit in config.h. In addition:
4077
4078 (a) pcreposix.h still had just "extern" instead of either of these macros;
4079 I have replaced it with PCRE_DATA_SCOPE.
4080
4081 (b) Functions such as _pcre_xclass(), which are internal to the library,
4082 but external in the C sense, all had PCRE_EXPORT in their definitions.
4083 This is apparently wrong for the Windows case, so I have removed it.
4084 (It makes no difference on Unix-like systems.)
4085
4086 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4087 of recursive calls to match(). This is different to MATCH_LIMIT because
4088 that limits the total number of calls to match(), not all of which increase
4089 the depth of recursion. Limiting the recursion depth limits the amount of
4090 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4091 when PCRE is compiled, and changed at run time. A patch from Google adds
4092 this functionality to the C++ interface.
4093
4094 18. Changes to the handling of Unicode character properties:
4095
4096 (a) Updated the table to Unicode 4.1.0.
4097
4098 (b) Recognize characters that are not in the table as "Cn" (undefined).
4099
4100 (c) I revised the way the table is implemented to a much improved format
4101 which includes recognition of ranges. It now supports the ranges that
4102 are defined in UnicodeData.txt, and it also amalgamates other
4103 characters into ranges. This has reduced the number of entries in the
4104 table from around 16,000 to around 3,000, thus reducing its size
4105 considerably. I realized I did not need to use a tree structure after
4106 all - a binary chop search is just as efficient. Having reduced the
4107 number of entries, I extended their size from 6 bytes to 8 bytes to
4108 allow for more data.
4109
4110 (d) Added support for Unicode script names via properties such as \p{Han}.
4111
4112 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4113 matching that character.
4114
4115 20. When matching a repeated Unicode property with a minimum greater than zero,
4116 (for example \pL{2,}), PCRE could look past the end of the subject if it
4117 reached it while seeking the minimum number of characters. This could
4118 happen only if some of the characters were more than one byte long, because
4119 there is a check for at least the minimum number of bytes.
4120
4121 21. Refactored the implementation of \p and \P so as to be more general, to
4122 allow for more different types of property in future. This has changed the
4123 compiled form incompatibly. Anybody with saved compiled patterns that use
4124 \p or \P will have to recompile them.
4125
4126 22. Added "Any" and "L&" to the supported property types.
4127
4128 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4129 but give a compile time error if the value is greater than 0xff.
4130
4131 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4132 accidentally not being installed or uninstalled.
4133
4134 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4135 made were to insert the current release number. This seemed silly, because
4136 it made things harder for people building PCRE on systems that don't run
4137 "configure". I have turned pcre.h into a distributed file, no longer built
4138 by "configure", with the version identification directly included. There is
4139 no longer a pcre.h.in file.
4140
4141 However, this change necessitated a change to the pcre-config script as
4142 well. It is built from pcre-config.in, and one of the substitutions was the
4143 release number. I have updated configure.ac so that ./configure now finds
4144 the release number by grepping pcre.h.
4145
4146 26. Added the ability to run the tests under valgrind.
4147
4148
4149 Version 6.4 05-Sep-05
4150 ---------------------
4151
4152 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4153 "--" to be printed when multiple files were scanned, even when none of the
4154 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4155 consider it to be a bug, and have restored the previous behaviour.
4156
4157 2. A couple of code tidies to get rid of compiler warnings.
4158
4159 3. The pcretest program used to cheat by referring to symbols in the library
4160 whose names begin with _pcre_. These are internal symbols that are not
4161 really supposed to be visible externally, and in some environments it is
4162 possible to suppress them. The cheating is now confined to including
4163 certain files from the library's source, which is a bit cleaner.
4164
4165 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4166 file's purpose clearer.
4167
4168 5. Reorganized pcre_ucp_findchar().
4169
4170
4171 Version 6.3 15-Aug-05
4172 ---------------------
4173
4174 1. The file libpcre.pc.in did not have general read permission in the tarball.
4175
4176 2. There were some problems when building without C++ support:
4177
4178 (a) If C++ support was not built, "make install" and "make test" still
4179 tried to test it.
4180
4181 (b) There were problems when the value of CXX was explicitly set. Some
4182 changes have been made to try to fix these, and ...
4183
4184 (c) --disable-cpp can now be used to explicitly disable C++ support.
4185
4186 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4187 backslash in a target when C++ was disabled. This confuses some
4188 versions of "make", apparently. Using an intermediate variable solves
4189 this. (Same for CPP_LOBJ.)
4190
4191 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4192 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4193 necessary on certain architectures.
4194
4195 4. Added a setting of -export-symbols-regex to the link command to remove
4196 those symbols that are exported in the C sense, but actually are local
4197 within the library, and not documented. Their names all begin with
4198 "_pcre_". This is not a perfect job, because (a) we have to except some
4199 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4200 available (and never for static libraries). I have made a note to try to
4201 find a way round (a) in the future.
4202
4203
4204 Version 6.2 01-Aug-05
4205 ---------------------
4206
4207 1. There was no test for integer overflow of quantifier values. A construction
4208 such as {1111111111111111} would give undefined results. What is worse, if
4209 a minimum quantifier for a parenthesized subpattern overflowed and became
4210 negative, the calculation of the memory size went wrong. This could have
4211 led to memory overwriting.
4212
4213 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4214
4215 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4216 operating environments where this matters.
4217
4218 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4219 PCRE options from within the C++ wrapper.
4220
4221 5. Named capturing subpatterns were not being correctly counted when a pattern
4222 was compiled. This caused two problems: (a) If there were more than 100
4223 such subpatterns, the calculation of the memory needed for the whole
4224 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4225 back references of the form \12, where the number was greater than 9, were
4226 not recognized as back references, even though there were sufficient
4227 previous subpatterns.
4228
4229 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4230 versions of gcc, e.g. 2.95.4.
4231
4232
4233 Version 6.1 21-Jun-05
4234 ---------------------
4235
4236 1. There was one reference to the variable "posix" in pcretest.c that was not
4237 surrounded by "#if !defined NOPOSIX".
4238
4239 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4240 the cross-check on the old pcre_info() function, for the benefit of the
4241 cut-down version of PCRE that is currently imported into Exim.
4242
4243 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4244 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4245 patterns that start (?i) but I don't think that matters. The use of (?i) is
4246 just an example; this all applies to the other options as well.
4247
4248 4. Since libtool seems to echo the compile commands it is issuing, the output
4249 from "make" can be reduced a bit by putting "@" in front of each libtool
4250 compile command.
4251
4252 5. Patch from the folks at Google for configure.in to be a bit more thorough
4253 in checking for a suitable C++ installation before trying to compile the
4254 C++ stuff. This should fix a reported problem when a compiler was present,
4255 but no suitable headers.
4256
4257 6. The man pages all had just "PCRE" as their title. I have changed them to
4258 be the relevant file name. I have also arranged that these names are
4259 retained in the file doc/pcre.txt, which is a concatenation in text format
4260 of all the man pages except the little individual ones for each function.
4261
4262 7. The NON-UNIX-USE file had not been updated for the different set of source
4263 files that come with release 6. I also added a few comments about the C++
4264 wrapper.
4265
4266
4267 Version 6.0 07-Jun-05
4268 ---------------------
4269
4270 1. Some minor internal re-organization to help with my DFA experiments.
4271
4272 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4273 didn't matter for the library itself when fully configured, but did matter
4274 when compiling without UCP support, or within Exim, where the ucp files are
4275 not imported.
4276
4277 3. Refactoring of the library code to split up the various functions into
4278 different source modules. The addition of the new DFA matching code (see
4279 below) to a single monolithic source would have made it really too
4280 unwieldy, quite apart from causing all the code to be include in a
4281 statically linked application, when only some functions are used. This is
4282 relevant even without the DFA addition now that patterns can be compiled in
4283 one application and matched in another.
4284
4285 The downside of splitting up is that there have to be some external
4286 functions and data tables that are used internally in different modules of
4287 the library but which are not part of the API. These have all had their
4288 names changed to start with "_pcre_" so that they are unlikely to clash
4289 with other external names.
4290
4291 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4292 a different (DFA) algorithm. Although it is slower than the original
4293 function, it does have some advantages for certain types of matching
4294 problem.
4295
4296 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4297 including restarting after a partial match.
4298
4299 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4300 defined when compiling for Windows was sent to me. I have put it into the
4301 code, though I have no means of testing or verifying it.
4302
4303 7. Added the pcre_refcount() auxiliary function.
4304
4305 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4306 match before or at the first newline in the subject string. In pcretest,
4307 the /f option on a pattern can be used to set this.
4308
4309 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4310 would behave wrongly. This has been present in PCRE since release 4.0.
4311
4312 10. A number of changes to the pcregrep command:
4313
4314 (a) Refactored how -x works; insert ^(...)$ instead of setting
4315 PCRE_ANCHORED and checking the length, in preparation for adding
4316 something similar for -w.
4317
4318 (b) Added the -w (match as a word) option.
4319
4320 (c) Refactored the way lines are read and buffered so as to have more
4321 than one at a time available.
4322
4323 (d) Implemented a pcregrep test script.
4324
4325 (e) Added the -M (multiline match) option. This allows patterns to match
4326 over several lines of the subject. The buffering ensures that at least
4327 8K, or the rest of the document (whichever is the shorter) is available
4328 for matching (and similarly the previous 8K for lookbehind assertions).
4329
4330 (f) Changed the --help output so that it now says
4331
4332 -w, --word-regex(p)
4333
4334 instead of two lines, one with "regex" and the other with "regexp"
4335 because that confused at least one person since the short forms are the
4336 same. (This required a bit of code, as the output is generated
4337 automatically from a table. It wasn't just a text change.)
4338
4339 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4340 option but starts with a hyphen. Could be a pattern or a path name
4341 starting with a hyphen, for instance.
4342
4343 (h) "-" can be given as a file name to represent stdin.
4344
4345 (i) When file names are being printed, "(standard input)" is used for
4346 the standard input, for compatibility with GNU grep. Previously
4347 "<stdin>" was used.
4348
4349 (j) The option --label=xxx can be used to supply a name to be used for
4350 stdin when file names are being printed. There is no short form.
4351
4352 (k) Re-factored the options decoding logic because we are going to add
4353 two more options that take data. Such options can now be given in four
4354 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4355
4356 (l) Added the -A, -B, and -C options for requesting that lines of context
4357 around matches be printed.
4358
4359 (m) Added the -L option to print the names of files that do not contain
4360 any matching lines, that is, the complement of -l.
4361
4362 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4363 continue to scan other files.
4364
4365 (o) The -s option was incorrectly implemented. For compatibility with other
4366 greps, it now suppresses the error message for a non-existent or non-
4367 accessible file (but not the return code). There is a new option called
4368 -q that suppresses the output of matching lines, which was what -s was
4369 previously doing.
4370
4371 (p) Added --include and --exclude options to specify files for inclusion
4372 and exclusion when recursing.
4373
4374 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4375 Hopefully, it now does.
4376
4377 12. Missing cast in pcre_study().
4378
4379 13. Added an "uninstall" target to the makefile.
4380
4381 14. Replaced "extern" in the function prototypes in Makefile.in with
4382 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4383 world, but is set differently for Windows.
4384
4385 15. Added a second compiling function called pcre_compile2(). The only
4386 difference is that it has an extra argument, which is a pointer to an
4387 integer error code. When there is a compile-time failure, this is set
4388 non-zero, in addition to the error test pointer being set to point to an
4389 error message. The new argument may be NULL if no error number is required
4390 (but then you may as well call pcre_compile(), which is now just a
4391 wrapper). This facility is provided because some applications need a
4392 numeric error indication, but it has also enabled me to tidy up the way
4393 compile-time errors are handled in the POSIX wrapper.
4394
4395 16. Added VPATH=.libs to the makefile; this should help when building with one
4396 prefix path and installing with another. (Or so I'm told by someone who
4397 knows more about this stuff than I do.)
4398
4399 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4400 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4401 match everything, including newlines. This is not POSIX-compatible, but
4402 somebody wanted the feature. From pcretest it can be activated by using
4403 both the P and the s flags.
4404
4405 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4406
4407 19. libpcre.pc was being incorrectly installed as executable.
4408
4409 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4410 it now also looks for '\r' so that it will work unmodified on Windows.
4411
4412 21. Added Google's contributed C++ wrapper to the distribution.
4413
4414 22. Added some untidy missing memory free() calls in pcretest, to keep
4415 Electric Fence happy when testing.
4416
4417
4418
4419 Version 5.0 13-Sep-04
4420 ---------------------
4421
4422 1. Internal change: literal characters are no longer packed up into items
4423 containing multiple characters in a single byte-string. Each character
4424 is now matched using a separate opcode. However, there may be more than one
4425 byte in the character in UTF-8 mode.
4426
4427 2. The pcre_callout_block structure has two new fields: pattern_position and
4428 next_item_length. These contain the offset in the pattern to the next match
4429 item, and its length, respectively.
4430
4431 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4432 insertion of callouts before each pattern item. Added the /C option to
4433 pcretest to make use of this.
4434
4435 4. On the advice of a Windows user, the lines
4436
4437 #if defined(_WIN32) || defined(WIN32)
4438 _setmode( _fileno( stdout ), 0x8000 );
4439 #endif /* defined(_WIN32) || defined(WIN32) */
4440
4441 have been added to the source of pcretest. This apparently does useful
4442 magic in relation to line terminators.
4443
4444 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4445 for the benefit of those environments where the "b" makes a difference.
4446
4447 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4448 to know about it. I have put a hack into configure.in that adds in code
4449 to set GCC=yes if CC=icc. This seems to end up at a point in the
4450 generated configure script that is early enough to affect the setting of
4451 compiler options, which is what is needed, but I have no means of testing
4452 whether it really works. (The user who reported this had patched the
4453 generated configure script, which of course I cannot do.)
4454
4455 LATER: After change 22 below (new libtool files), the configure script
4456 seems to know about icc (and also ecc). Therefore, I have commented out
4457 this hack in configure.in.
4458
4459 7. Added support for pkg-config (2 patches were sent in).
4460
4461 8. Negated POSIX character classes that used a combination of internal tables
4462 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4463 [[:^ascii]]. Typically, they would match almost any characters. The other
4464 POSIX classes were not broken in this way.
4465
4466 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4467 to find the match, as PCRE was deluded into thinking that the match had to
4468 start at the start point or following a newline. The same bug applied to
4469 patterns with negative forward assertions or any backward assertions
4470 preceding ".*" at the start, unless the pattern required a fixed first
4471 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4472
4473 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4474 starting at the last subject character, bytes beyond the end of the subject
4475 string were read.
4476
4477 11. Renamed the variable "class" as "classbits" to make life easier for C++
4478 users. (Previously there was a macro definition, but it apparently wasn't
4479 enough.)
4480
4481 12. Added the new field "tables" to the extra data so that tables can be passed
4482 in at exec time, or the internal tables can be re-selected. This allows
4483 a compiled regex to be saved and re-used at a later time by a different
4484 program that might have everything at different addresses.
4485
4486 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4487 -R library as well as a -L library.
4488
4489 14. The debugging options of pcretest (-d on the command line or D on a
4490 pattern) showed incorrect output for anything following an extended class
4491 that contained multibyte characters and which was followed by a quantifier.
4492
4493 15. Added optional support for general category Unicode character properties
4494 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4495 support. It adds about 90K to the size of the library. The meanings of the
4496 inbuilt class escapes such as \d and \s have NOT been changed.
4497
4498 16. Updated pcredemo.c to include calls to free() to release the memory for the
4499 compiled pattern.
4500
4501 17. The generated file chartables.c was being created in the source directory
4502 instead of in the building directory. This caused the build to fail if the
4503 source directory was different from the building directory, and was
4504 read-only.
4505
4506 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4507 file. No doubt somebody will tell me if they don't make sense... Also added
4508 Dan Mooney's comments about building on OpenVMS.
4509
4510 19. Added support for partial matching via the PCRE_PARTIAL option for
4511 pcre_exec() and the \P data escape in pcretest.
4512
4513 20. Extended pcretest with 3 new pattern features:
4514
4515 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4516 write the compiled pattern to the file whose name is "rest-of-line".
4517 This is a straight binary dump of the data, with the saved pointer to
4518 the character tables forced to be NULL. The study data, if any, is
4519 written too. After writing, pcretest reads a new pattern.
4520
4521 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4522 compiled pattern from the given file. There must not be any
4523 occurrences of "<" in the file name (pretty unlikely); if there are,
4524 pcretest will instead treat the initial "<" as a pattern delimiter.
4525 After reading in the pattern, pcretest goes on to read data lines as
4526 usual.
4527
4528 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4529 and 16-bit fields in a compiled pattern, to simulate a pattern that
4530 was compiled on a host of opposite endianness.
4531
4532 21. The pcre-exec() function can now cope with patterns that were compiled on
4533 hosts of opposite endianness, with this restriction:
4534
4535 As for any compiled expression that is saved and used later, the tables
4536 pointer field cannot be preserved; the extra_data field in the arguments
4537 to pcre_exec() should be used to pass in a tables address if a value
4538 other than the default internal tables were used at compile time.
4539
4540 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4541 now diagnosed as an error. Previously, most of the time, a negative number
4542 would have been treated as zero, but if in addition "ovector" was passed as
4543 NULL, a crash could occur.
4544
4545 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4546 new versions from the libtool 1.5 distribution (the last one is a copy of
4547 a file called libtool.m4). This seems to have fixed the need to patch
4548 "configure" to support Darwin 1.3 (which I used to do). However, I still
4549 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4550 workstation).
4551
4552 24. Changed the PCRE licence to be the more standard "BSD" licence.
4553
4554
4555 Version 4.5 01-Dec-03
4556 ---------------------
4557
4558 1. There has been some re-arrangement of the code for the match() function so
4559 that it can be compiled in a version that does not call itself recursively.
4560 Instead, it keeps those local variables that need separate instances for
4561 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4562 needs to "recurse". Keeping track of where control must go is done by means
4563 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4564 hide most of the details from the main code, and operates only if
4565 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4566 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4567 operating.
4568
4569 To make it easier for callers to provide specially tailored get/free
4570 functions for this usage, two new functions, pcre_stack_malloc, and
4571 pcre_stack_free, are used. They are always called in strict stacking order,
4572 and the size of block requested is always the same.
4573
4574 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4575 PCRE has been compiled to use the stack or the heap for recursion. The
4576 -C option of pcretest uses this to show which version is compiled.
4577
4578 A new data escape \S, is added to pcretest; it causes the amounts of store
4579 obtained and freed by both kinds of malloc/free at match time to be added
4580 to the output.
4581
4582 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4583 what's available on my current Linux desktop machine.
4584
4585 3. When matching a UTF-8 string, the test for a valid string at the start has
4586 been extended. If start_offset is not zero, PCRE now checks that it points
4587 to a byte that is the start of a UTF-8 character. If not, it returns
4588 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4589 this is necessary because there may be backward assertions in the pattern.
4590 When matching the same subject several times, it may save resources to use
4591 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4592
4593 4. The code for checking the validity of UTF-8 strings has been tightened so
4594 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4595 containing "overlong sequences".
4596
4597 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4598 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4599 should have been "&", but it just so happened that all the cases this let
4600 through by mistake were picked up later in the function.
4601
4602 6. I had used a variable called "isblank" - this is a C99 function, causing
4603 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4604
4605 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4606 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4607 so the version is shown; (c) stop "make test" echoing "./RunTest".
4608
4609 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4610
4611 9. The prototype for memmove() for systems that don't have it was using
4612 size_t, but the inclusion of the header that defines size_t was later. I've
4613 moved the #includes for the C headers earlier to avoid this.
4614
4615 10. Added some adjustments to the code to make it easier to compiler on certain
4616 special systems:
4617
4618 (a) Some "const" qualifiers were missing.
4619 (b) Added the macro EXPORT before all exported functions; by default this
4620 is defined to be empty.
4621 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4622 that it reads its output file name as an argument instead of writing
4623 to the standard output and assuming this can be redirected.
4624
4625 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4626 class containing characters with values greater than 255, PCRE compilation
4627 went into a loop.
4628
4629 12. A recursive reference to a subpattern that was within another subpattern
4630 that had a minimum quantifier of zero caused PCRE to crash. For example,
4631 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4632 recursion. If the recursively-called subpattern itself had a zero repeat,
4633 that was OK.
4634
4635 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4636 buffer into which it was copied (for escape processing) was still set at
4637 1024, so long lines caused crashes.
4638
4639 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4640 "internal error: code overflow...". This applied to any character class
4641 that was followed by a possessive quantifier.
4642
4643 15. Modified the Makefile to add libpcre.la as a prerequisite for
4644 libpcreposix.la because I was told this is needed for a parallel build to
4645 work.
4646
4647 16. If a pattern that contained .* following optional items at the start was
4648 studied, the wrong optimizing data was generated, leading to matching
4649 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4650 matching string must start with a or b or c. The correct conclusion for
4651 this pattern is that a match can start with any character.
4652
4653
4654 Version 4.4 13-Aug-03
4655 ---------------------
4656
4657 1. In UTF-8 mode, a character class containing characters with values between
4658 127 and 255 was not handled correctly if the compiled pattern was studied.
4659 In fixing this, I have also improved the studying algorithm for such
4660 classes (slightly).
4661
4662 2. Three internal functions had redundant arguments passed to them. Removal
4663 might give a very teeny performance improvement.
4664
4665 3. Documentation bug: the value of the capture_top field in a callout is *one
4666 more than* the number of the hightest numbered captured substring.
4667
4668 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4669 in incorrectly linking with a previously installed version. They now link
4670 explicitly with libpcre.la.
4671
4672 5. configure.in no longer needs to recognize Cygwin specially.
4673
4674 6. A problem in pcre.in for Windows platforms is fixed.
4675
4676 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4677 pcretest, it used to include the size of the study block as part of its
4678 output. Unfortunately, the structure contains a field that has a different
4679 size on different hardware architectures. This meant that the tests that
4680 showed this size failed. As the block is currently always of a fixed size,
4681 this information isn't actually particularly useful in pcretest output, so
4682 I have just removed it.
4683
4684 8. Three pre-processor statements accidentally did not start in column 1.
4685 Sadly, there are *still* compilers around that complain, even though
4686 standard C has not required this for well over a decade. Sigh.
4687
4688 9. In pcretest, the code for checking callouts passed small integers in the
4689 callout_data field, which is a void * field. However, some picky compilers
4690 complained about the casts involved for this on 64-bit systems. Now
4691 pcretest passes the address of the small integer instead, which should get
4692 rid of the warnings.
4693
4694 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4695 both compile and run time, and gives an error if an invalid UTF-8 sequence
4696 is found. There is a option for disabling this check in cases where the
4697 string is known to be correct and/or the maximum performance is wanted.
4698
4699 11. In response to a bug report, I changed one line in Makefile.in from
4700
4701 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4702 to
4703 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4704
4705 to look similar to other lines, but I have no way of telling whether this
4706 is the right thing to do, as I do not use Windows. No doubt I'll get told
4707 if it's wrong...
4708
4709
4710 Version 4.3 21-May-03
4711 ---------------------
4712
4713 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4714 Makefile.
4715
4716 2. Some refactoring to improve the quality of the code:
4717
4718 (i) The utf8_table... variables are now declared "const".
4719
4720 (ii) The code for \cx, which used the "case flipping" table to upper case
4721 lower case letters, now just substracts 32. This is ASCII-specific,
4722 but the whole concept of \cx is ASCII-specific, so it seems
4723 reasonable.
4724
4725 (iii) PCRE was using its character types table to recognize decimal and
4726 hexadecimal digits in the pattern. This is silly, because it handles
4727 only 0-9, a-f, and A-F, but the character types table is locale-
4728 specific, which means strange things might happen. A private
4729 table is now used for this - though it costs 256 bytes, a table is
4730 much faster than multiple explicit tests. Of course, the standard
4731 character types table is still used for matching digits in subject
4732 strings against \d.
4733
4734 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4735 ending in _t are). So I've renamed it as ESC_tee.
4736
4737 3. The first argument for regexec() in the POSIX wrapper should have been
4738 defined as "const".
4739
4740 4. Changed pcretest to use malloc() for its buffers so that they can be
4741 Electric Fenced for debugging.
4742
4743 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4744 to read one or more bytes before the start of the subject string. Often this
4745 had no effect on PCRE's behaviour, but in some circumstances it could
4746 provoke a segmentation fault.
4747
4748 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4749 to try to read one or more bytes before the start of the subject string.
4750
4751 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4752 UTF-8 support could misbehave in various ways if the subject string
4753 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4754 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4755 back over UTF-8 characters.)
4756
4757
4758 Version 4.2 14-Apr-03
4759 ---------------------
4760
4761 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4762
4763 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4764 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4765 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4766 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4767 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4768 and BUILD_EXEEXT
4769 Note: automatic setting of the BUILD variables is not yet working
4770 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4771 compile-time but not at link-time
4772 [LINK]: use for linking executables only
4773 make different versions for Windows and non-Windows
4774 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4775 libraries
4776 [LINK_FOR_BUILD]: new variable
4777 [OBJEXT]: use throughout
4778 [EXEEXT]: use throughout
4779 <winshared>: new target
4780 <wininstall>: new target
4781 <dftables.o>: use native compiler
4782 <dftables>: use native linker
4783 <install>: handle Windows platform correctly
4784 <clean>: ditto
4785 <check>: ditto
4786 copy DLL to top builddir before testing
4787
4788 As part of these changes, -no-undefined was removed again. This was reported
4789 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4790 in any case.
4791
4792 3. Some tidies to get rid of compiler warnings:
4793
4794 . In the match_data structure, match_limit was an unsigned long int, whereas
4795 match_call_count was an int. I've made them both unsigned long ints.
4796
4797 . In pcretest the fact that a const uschar * doesn't automatically cast to
4798 a void * provoked a warning.
4799
4800 . Turning on some more compiler warnings threw up some "shadow" variables
4801 and a few more missing casts.
4802
4803 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4804 option, a class that contained a single character with a value between 128
4805 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4806
4807 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4808 option, a class that contained several characters, but with at least one
4809 whose value was between 128 and 255 caused PCRE to crash.
4810
4811
4812 Version 4.1 12-Mar-03
4813 ---------------------
4814
4815 1. Compiling with gcc -pedantic found a couple of places where casts were
4816 needed, and a string in dftables.c that was longer than standard compilers are
4817 required to support.
4818
4819 2. Compiling with Sun's compiler found a few more places where the code could
4820 be tidied up in order to avoid warnings.
4821
4822 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4823 first of these names is deprecated in the latest Autoconf in favour of the name
4824 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4825 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4826 analogy I have changed it to CFLAGS_FOR_BUILD.
4827
4828 4. Added -no-undefined to the linking command in the Makefile, because this is
4829 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4830 linking step for the pcreposix library.
4831
4832 5. PCRE was failing to diagnose the case of two named groups with the same
4833 name.
4834
4835 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4836 literal character that is needed in the subject for a match, and scans along to
4837 ensure that it is present before embarking on the full matching process. This
4838 saves time in cases of nested unlimited repeats that are never going to match.
4839 Problem: the scan can take a lot of time if the subject is very long (e.g.
4840 megabytes), thus penalizing straightforward matches. It is now done only if the
4841 amount of subject to be scanned is less than 1000 bytes.
4842
4843 7. A lesser problem with the same optimization is that it was recording the
4844 first character of an anchored pattern as "needed", thus provoking a search
4845 right along the subject, even when the first match of the pattern was going to
4846 fail. The "needed" character is now not set for anchored patterns, unless it
4847 follows something in the pattern that is of non-fixed length. Thus, it still
4848 fulfils its original purpose of finding quick non-matches in cases of nested
4849 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4850
4851
4852 Version 4.0 17-Feb-03
4853 ---------------------
4854
4855 1. If a comment in an extended regex that started immediately after a meta-item
4856 extended to the end of string, PCRE compiled incorrect data. This could lead to
4857 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4858
4859 2. Moved to autoconf 2.53 and libtool 1.4.2.
4860
4861 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4862 the special perltest8 script is no longer needed - all the tests can be run
4863 from a single perltest script.
4864
4865 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4866 by \s. It has now been removed in PCRE. This means it isn't recognized as
4867 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4868 class [:space:] *does* include VT, thereby creating a mess.
4869
4870 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4871 space and tab.
4872
4873 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4874 its new features into the main test script, reducing the number of scripts.
4875
4876 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4877 were backward compatible, and made the (?i) apply to the whole pattern, as if
4878 /i were given. Now it behaves more logically, and applies the option setting
4879 only to what follows. PCRE has been changed to follow suit. However, if it
4880 finds options settings right at the start of the pattern, it extracts them into
4881 the global options, as before. Thus, they show up in the info data.
4882
4883 8. Added support for the \Q...\E escape sequence. Characters in between are
4884 treated as literals. This is slightly different from Perl in that $ and @ are
4885 also handled as literals inside the quotes. In Perl, they will cause variable
4886 interpolation. Note the following examples:
4887
4888 Pattern PCRE matches Perl matches
4889
4890 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4891 \Qabc\$xyz\E abc\$xyz abc\$xyz
4892 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4893
4894 For compatibility with Perl, \Q...\E sequences are recognized inside character
4895 classes as well as outside them.
4896
4897 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4898 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4899 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4900 signed/unsigned warnings.
4901
4902 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4903 option for pcretest, so I've replaced it by a simple function that does just
4904 that job.
4905
4906 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4907 "pcregrep -".
4908
4909 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4910 Java package. This provides some syntactic sugar for simple cases of what my
4911 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4912 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4913 item, you can use this simplified notation. Note that only makes sense with
4914 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4915 greediness, whatever the setting of the PCRE_UNGREEDY option.
4916
4917 13. A change of greediness default within a pattern was not taking effect at
4918 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4919 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4920 was abstracted outside.
4921
4922 14. PCRE now supports the \G assertion. It is true when the current matching
4923 position is at the start point of the match. This differs from \A when the
4924 starting offset is non-zero. Used with the /g option of pcretest (or similar
4925 code), it works in the same way as it does for Perl's /g option. If all
4926 alternatives of a regex begin with \G, the expression is anchored to the start
4927 match position, and the "anchored" flag is set in the compiled expression.
4928
4929 15. Some bugs concerning the handling of certain option changes within patterns
4930 have been fixed. These applied to options other than (?ims). For example,
4931 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4932 been the other way round. Some of this was related to change 7 above.
4933
4934 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4935 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4936 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4937 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4938
4939 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4940 mode. Unlike ".", it always matches newline, whatever the setting of
4941 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4942 assertions. Perl allows it, but it doesn't (in general) work because it can't
4943 calculate the length of the lookbehind. At least, that's the case for Perl
4944 5.8.0 - I've been told they are going to document that it doesn't work in
4945 future.
4946
4947 18. Added an error diagnosis for escapes that PCRE does not support: these are
4948 \L, \l, \N, \P, \p, \U, \u, and \X.
4949
4950 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4951 reading past the end of the pattern in cases such as /[abcd/.
4952
4953 20. PCRE was getting more memory than necessary for patterns with classes that
4954 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4955
4956 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4957 compiling PCRE for use with Virtual Pascal.
4958
4959 22. Small fix to the Makefile to make it work properly if the build is done
4960 outside the source tree.
4961
4962 23. Added a new extension: a condition to go with recursion. If a conditional
4963 subpattern starts with (?(R) the "true" branch is used if recursion has
4964 happened, whereas the "false" branch is used only at the top level.
4965
4966 24. When there was a very long string of literal characters (over 255 bytes
4967 without UTF support, over 250 bytes with UTF support), the computation of how
4968 much memory was required could be incorrect, leading to segfaults or other
4969 strange effects.
4970
4971 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4972 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4973 there was a subsequent back reference to those brackets. This meant that, for
4974 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4975 possible to check for precisely this case. All we can do is abandon the
4976 optimization if .* occurs inside capturing brackets when there are any back
4977 references whatsoever. (See below for a better fix that came later.)
4978
4979 26. The handling of the optimization for finding the first character of a
4980 non-anchored pattern, and for finding a character that is required later in the
4981 match were failing in some cases. This didn't break the matching; it just
4982 failed to optimize when it could. The way this is done has been re-implemented.
4983
4984 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4985
4986 28. Added a new feature that provides some of the functionality that Perl
4987 provides with (?{...}). The facility is termed a "callout". The way it is done
4988 in PCRE is for the caller to provide an optional function, by setting
4989 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4990 global variable. By default it is unset, which disables all calling out. To get
4991 the function called, the regex must include (?C) at appropriate points. This
4992 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4993 This provides a means of identifying different callout points. When PCRE
4994 reaches such a point in the regex, if pcre_callout has been set, the external
4995 function is called. It is provided with data in a structure called
4996 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4997 matching continues; if it returns a non-zero value, the match at the current
4998 point fails. However, backtracking will occur if possible. [This was changed
4999 later and other features added - see item 49 below.]
5000
5001 29. pcretest is upgraded to test the callout functionality. It provides a
5002 callout function that displays information. By default, it shows the start of
5003 the match and the current position in the text. There are some new data escapes
5004 to vary what happens:
5005
5006 \C+ in addition, show current contents of captured substrings
5007 \C- do not supply a callout function
5008 \C!n return 1 when callout number n is reached
5009 \C!n!m return 1 when callout number n is reached for the mth time
5010
5011 30. If pcregrep was called with the -l option and just a single file name, it
5012 output "<stdin>" if a match was found, instead of the file name.
5013
5014 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5015 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5016 pcre_exec(). This saves a malloc/free per call. The default value of
5017 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5018 when configuring.
5019
5020 32. The default maximum size of a compiled pattern is 64K. There have been a
5021 few cases of people hitting this limit. The code now uses macros to handle the
5022 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5023 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5024 configuring. Tests 2 and 5 work only with 2-byte links because they output
5025 debugging information about compiled patterns.
5026
5027 33. Internal code re-arrangements:
5028
5029 (a) Moved the debugging function for printing out a compiled regex into
5030 its own source file (printint.c) and used #include to pull it into
5031 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5032 separate copies.
5033
5034 (b) Defined the list of op-code names for debugging as a macro in
5035 internal.h so that it is next to the definition of the opcodes.
5036
5037 (c) Defined a table of op-code lengths for simpler skipping along compiled
5038 code. This is again a macro in internal.h so that it is next to the
5039 definition of the opcodes.
5040
5041 34. Added support for recursive calls to individual subpatterns, along the
5042 lines of Robin Houston's patch (but implemented somewhat differently).
5043
5044 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5045 allow it to read and process whole directories in Win32. This code was
5046 contributed by Lionel Fourquaux; it has not been tested by me.
5047
5048 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5049 used to name a group. Names consist of alphanumerics and underscores, and must
5050 be unique. Back references use the syntax (?P=name) and recursive calls use
5051 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5052 numbers. The function pcre_fullinfo() can be used after compilation to extract
5053 a name/number map. There are three relevant calls:
5054
5055 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5056 PCRE_INFO_NAMECOUNT yields the number of entries
5057 PCRE_INFO_NAMETABLE yields a pointer to the map.
5058
5059 The map is a vector of fixed-size entries. The size of each entry depends on
5060 the length of the longest name used. The first two bytes of each entry are the
5061 group number, most significant byte first. There follows the corresponding
5062 name, zero terminated. The names are in alphabetical order.
5063
5064 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5065 case instead of 255. Making it the same both with and without UTF-8 support
5066 means that the same test output works with both.
5067
5068 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5069 calling malloc() with a zero argument.
5070
5071 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5072 optimization. I've improved things by keeping a bitmap of backreferences with
5073 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5074 fact referenced, the optimization can be applied. It is unlikely that a
5075 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5076 the match to follow \n) will appear inside brackets with a number greater than
5077 31, but if it does, any back reference > 31 suppresses the optimization.
5078
5079 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5080 of disabling numbered capturing parentheses. Any opening parenthesis that is
5081 not followed by ? behaves as if it were followed by ?: but named parentheses
5082 can still be used for capturing (and they will acquire numbers in the usual
5083 way).
5084
5085 41. Redesigned the return codes from the match() function into yes/no/error so
5086 that errors can be passed back from deep inside the nested calls. A malloc
5087 failure while inside a recursive subpattern call now causes the
5088 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5089
5090 42. It is now possible to set a limit on the number of times the match()
5091 function is called in a call to pcre_exec(). This facility makes it possible to
5092 limit the amount of recursion and backtracking, though not in a directly
5093 obvious way, because the match() function is used in a number of different
5094 circumstances. The count starts from zero for each position in the subject
5095 string (for non-anchored patterns). The default limit is, for compatibility, a
5096 large number, namely 10 000 000. You can change this in two ways:
5097
5098 (a) When configuring PCRE before making, you can use --with-match-limit=n
5099 to set a default value for the compiled library.
5100
5101 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5102 a different value is set. See 45 below.
5103
5104 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5105
5106 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5107 of things that can be changed at compile time. The first argument specifies
5108 what is wanted and the second points to where the information is to be placed.
5109 The current list of available information is:
5110
5111 PCRE_CONFIG_UTF8
5112
5113 The output is an integer that is set to one if UTF-8 support is available;
5114 otherwise it is set to zero.
5115
5116 PCRE_CONFIG_NEWLINE
5117
5118 The output is an integer that it set to the value of the code that is used for
5119 newline. It is either LF (10) or CR (13).
5120
5121 PCRE_CONFIG_LINK_SIZE
5122
5123 The output is an integer that contains the number of bytes used for internal
5124 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5125
5126 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5127
5128 The output is an integer that contains the threshold above which the POSIX
5129 interface uses malloc() for output vectors. See item 31 above.
5130
5131 PCRE_CONFIG_MATCH_LIMIT
5132
5133 The output is an unsigned integer that contains the default limit of the number
5134 of match() calls in a pcre_exec() execution. See 42 above.
5135
5136 44. pcretest has been upgraded by the addition of the -C option. This causes it
5137 to extract all the available output from the new pcre_config() function, and to
5138 output it. The program then exits immediately.
5139
5140 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5141 order to support additional features. One way would have been to define
5142 pcre_exec2() (for example) with extra arguments, but this would not have been
5143 extensible, and would also have required all calls to the original function to
5144 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5145 is used for passing in "extra" data from pcre_study().
5146
5147 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5148 contains the following fields:
5149
5150 flags a bitmap indicating which of the following fields are set
5151 study_data opaque data from pcre_study()
5152 match_limit a way of specifying a limit on match() calls for a specific
5153 call to pcre_exec()
5154 callout_data data for callouts (see 49 below)
5155
5156 The flag bits are also defined in pcre.h, and are
5157
5158 PCRE_EXTRA_STUDY_DATA
5159 PCRE_EXTRA_MATCH_LIMIT
5160 PCRE_EXTRA_CALLOUT_DATA
5161
5162 The pcre_study() function now returns one of these new pcre_extra blocks, with
5163 the actual study data pointed to by the study_data field, and the
5164 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5165 before. That is, this change is entirely upwards-compatible and requires no
5166 change to existing code.
5167
5168 If you want to pass in additional data to pcre_exec(), you can either place it
5169 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5170 block.
5171
5172 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5173 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5174 times with different match limits, until it finds the minimum value needed for
5175 pcre_exec() to complete. The value is then output. This can be instructive; for
5176 most simple matches the number is quite small, but for pathological cases it
5177 gets very large very quickly.
5178
5179 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5180 returns the size of the data block pointed to by the study_data field in a
5181 pcre_extra block, that is, the value that was passed as the argument to
5182 pcre_malloc() when PCRE was getting memory in which to place the information
5183 created by pcre_study(). The fourth argument should point to a size_t variable.
5184 pcretest has been extended so that this information is shown after a successful
5185 pcre_study() call when information about the compiled regex is being displayed.
5186
5187 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5188 because what follows is always an absolute path. (Later: it turns out that this
5189 is more than cosmetic for MinGW, because it doesn't like empty path
5190 components.)
5191
5192 49. Some changes have been made to the callout feature (see 28 above):
5193
5194 (i) A callout function now has three choices for what it returns:
5195
5196 0 => success, carry on matching
5197 > 0 => failure at this point, but backtrack if possible
5198 < 0 => serious error, return this value from pcre_exec()
5199
5200 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5201 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5202 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5203 use by callout functions. It will never be used by PCRE itself.
5204
5205 (ii) The pcre_extra structure (see 45 above) has a void * field called
5206 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5207 pcre_callout_block structure has a field of the same name. The contents of
5208 the field passed in the pcre_extra structure are passed to the callout
5209 function in the corresponding field in the callout block. This makes it
5210 easier to use the same callout-containing regex from multiple threads. For
5211 testing, the pcretest program has a new data escape
5212
5213 \C*n pass the number n (may be negative) as callout_data
5214
5215 If the callout function in pcretest receives a non-zero value as
5216 callout_data, it returns that value.
5217
5218 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5219 there were some redundant $(CFLAGS) in commands that are now specified as
5220 $(LINK), which already includes $(CFLAGS).
5221
5222 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5223 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5224 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5225 one-byte characters throughout. Note that case-insensitive matching applies
5226 only to characters whose values are less than 256. PCRE doesn't support the
5227 notion of cases for higher-valued characters.
5228
5229 (i) A character class whose characters are all within 0-255 is handled as
5230 a bit map, and the map is inverted for negative classes. Previously, a
5231 character > 255 always failed to match such a class; however it should
5232 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5233
5234 (ii) A negated character class with a single character < 255 is coded as
5235 "not this character" (OP_NOT). This wasn't working properly when the test
5236 character was multibyte, either singly or repeated.
5237
5238 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5239 mode, for example: \x{100}{2,3}.
5240
5241 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5242 singly or repeated) now correctly test multibyte characters. However,
5243 PCRE doesn't recognize any characters with values greater than 255 as
5244 digits, spaces, or word characters. Such characters always match \D, \S,
5245 and \W, and never match \d, \s, or \w.
5246
5247 (v) Classes may now contain characters and character ranges with values
5248 greater than 255. For example: [ab\x{100}-\x{400}].
5249
5250 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5251 PCRE in UTF-8 mode.
5252
5253 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5254 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5255 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5256 value.)
5257
5258 53. The single man page has become too large. I have therefore split it up into
5259 a number of separate man pages. These also give rise to individual HTML pages;
5260 these are now put in a separate directory, and there is an index.html page that
5261 lists them all. Some hyperlinking between the pages has been installed.
5262
5263 54. Added convenience functions for handling named capturing parentheses.
5264
5265 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5266 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5267 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5268 are faulted.
5269
5270 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5271 calling configure. These values are used when compiling the dftables.c program
5272 which is run to generate the source of the default character tables. They
5273 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5274 you will need to set these values.
5275
5276 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5277
5278
5279 Version 3.9 02-Jan-02
5280 ---------------------
5281
5282 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5283
5284 2. If --disable-static was given, the building process failed when trying to
5285 build pcretest and pcregrep. (For some reason it was using libtool to compile
5286 them, which is not right, as they aren't part of the library.)
5287
5288
5289 Version 3.8 18-Dec-01
5290 ---------------------
5291
5292 1. The experimental UTF-8 code was completely screwed up. It was packing the
5293 bytes in the wrong order. How dumb can you get?
5294
5295
5296 Version 3.7 29-Oct-01
5297 ---------------------
5298
5299 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5300 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5301 this didn't happen under Solaris 8, where I normally test things.
5302
5303 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5304 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5305 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5306 this fix an hour or so after the initial 3.7 release.)
5307
5308
5309 Version 3.6 23-Oct-01
5310 ---------------------
5311
5312 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5313 offsets passed as NULL with zero offset count.
5314
5315 2. The config.guess and config.sub files had not been updated when I moved to
5316 the latest autoconf.
5317
5318
5319 Version 3.5 15-Aug-01
5320 ---------------------
5321
5322 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5323 had been forgotten.
5324
5325 2. By using declared but undefined structures, we can avoid using "void"
5326 definitions in pcre.h while keeping the internal definitions of the structures
5327 private.
5328
5329 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5330 user point of view, this means that both static and shared libraries are built
5331 by default, but this can be individually controlled. More of the work of
5332 handling this static/shared cases is now inside libtool instead of PCRE's make
5333 file.
5334
5335 4. The pcretest utility is now installed along with pcregrep because it is
5336 useful for users (to test regexs) and by doing this, it automatically gets
5337 relinked by libtool. The documentation has been turned into a man page, so
5338 there are now .1, .txt, and .html versions in /doc.
5339
5340 5. Upgrades to pcregrep:
5341 (i) Added long-form option names like gnu grep.
5342 (ii) Added --help to list all options with an explanatory phrase.
5343 (iii) Added -r, --recursive to recurse into sub-directories.
5344 (iv) Added -f, --file to read patterns from a file.
5345
5346 6. pcre_exec() was referring to its "code" argument before testing that
5347 argument for NULL (and giving an error if it was NULL).
5348
5349 7. Upgraded Makefile.in to allow for compiling in a different directory from
5350 the source directory.
5351
5352 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5353 options bits, the pointer it was passed was to an int instead of to an unsigned
5354 long int. This mattered only on 64-bit systems.
5355
5356 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5357 generated) instead of pcre.in, which it its source. Also made the same change
5358 in several of the .c files.
5359
5360 10. A new release of gcc defines printf() as a macro, which broke pcretest
5361 because it had an ifdef in the middle of a string argument for printf(). Fixed
5362 by using separate calls to printf().
5363
5364 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5365 script, to force use of CR or LF instead of \n in the source. On non-Unix
5366 systems, the value can be set in config.h.
5367
5368 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5369 absolute limit. Changed the text of the error message to make this clear, and
5370 likewise updated the man page.
5371
5372 13. The limit of 99 on the number of capturing subpatterns has been removed.
5373 The new limit is 65535, which I hope will not be a "real" limit.
5374
5375
5376 Version 3.4 22-Aug-00
5377 ---------------------
5378
5379 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5380
5381 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5382
5383
5384 Version 3.3 01-Aug-00
5385 ---------------------
5386
5387 1. If an octal character was given, but the value was greater than \377, it
5388 was not getting masked to the least significant bits, as documented. This could
5389 lead to crashes in some systems.
5390
5391 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5392 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5393
5394 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5395 These just pass their arguments on to (pcre_free)(), but they are provided
5396 because some uses of PCRE bind it to non-C systems that can call its functions,
5397 but cannot call free() or pcre_free() directly.
5398
5399 4. Add "make test" as a synonym for "make check". Corrected some comments in
5400 the Makefile.
5401
5402 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5403 Makefile.
5404
5405 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5406 command called pgrep for grepping around the active processes.
5407
5408 7. Added the beginnings of support for UTF-8 character strings.
5409
5410 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5411 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5412 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5413 out for the ar command.)
5414
5415
5416 Version 3.2 12-May-00
5417 ---------------------
5418
5419 This is purely a bug fixing release.
5420
5421 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5422 of ZA. This was just one example of several cases that could provoke this bug,
5423 which was introduced by change 9 of version 2.00. The code for breaking
5424 infinite loops after an iteration that matches an empty string was't working
5425 correctly.
5426
5427 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5428 when matched against abbab (for example). After matching an empty string, it
5429 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5430 caused it to match further down the string than it should.
5431
5432 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5433 was there because it doesn't seem to be needed, and it causes trouble on some
5434 systems, as it is not a Standard C header. It has been removed.
5435
5436 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5437 were reported on the Macintosh. The changes were from
5438
5439 while ((c = *(++ptr)) != 0 && c != '\n');
5440 to
5441 while ((c = *(++ptr)) != 0 && c != '\n') ;
5442
5443 Totally extraordinary, but if that's what it takes...
5444
5445 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5446 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5447 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5448 assumes the way PCRE uses memmove() (always moving upwards).
5449
5450 6. PCRE is being used in one environment where strchr() is not available. There
5451 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5452 faster code anyway.
5453
5454
5455 Version 3.1 09-Feb-00
5456 ---------------------
5457
5458 The only change in this release is the fixing of some bugs in Makefile.in for
5459 the "install" target:
5460
5461 (1) It was failing to install pcreposix.h.
5462
5463 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5464
5465
5466 Version 3.0 01-Feb-00
5467 ---------------------
5468
5469 1. Add support for the /+ modifier to perltest (to output $` like it does in
5470 pcretest).
5471
5472 2. Add support for the /g modifier to perltest.
5473
5474 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5475 matches null strings.
5476
5477 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5478 pattern. Perl treats empty patterns specially - it reuses the most recent
5479 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5480 effect.
5481
5482 5. The POSIX interface was broken in that it was just handing over the POSIX
5483 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5484 required a bigger vector, with some working space on the end. This means that
5485 the POSIX wrapper now has to get and free some memory, and copy the results.
5486
5487 6. Added some simple autoconf support, placing the test data and the
5488 documentation in separate directories, re-organizing some of the
5489 information files, and making it build pcre-config (a GNU standard). Also added
5490 libtool support for building PCRE as a shared library, which is now the
5491 default.
5492
5493 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5494 09 are not valid octal constants. Single digits will be used for minor values
5495 less than 10.
5496
5497 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5498 existing programs that set these in the POSIX interface can use PCRE without
5499 modification.
5500
5501 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5502 return all that pcre_info() returns, plus additional data. The pcre_info()
5503 function is retained for compatibility, but is considered to be obsolete.
5504
5505 10. Added experimental recursion feature (?R) to handle one common case that
5506 Perl 5.6 will be able to do with (?p{...}).
5507
5508 11. Added support for POSIX character classes like [:alpha:], which Perl is
5509 adopting.
5510
5511
5512 Version 2.08 31-Aug-99
5513 ----------------------
5514
5515 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5516 trying to match at the startoffset position, but instead was moving forward to
5517 the next newline as if a previous match had failed.
5518
5519 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5520 and could get into a loop if a null string was matched other than at the start
5521 of the subject.
5522
5523 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5524 be distinguished at compile time, and for completeness also added PCRE_DATE.
5525
5526 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5527 in GnuWin32 environments.
5528
5529
5530 Version 2.07 29-Jul-99
5531 ----------------------
5532
5533 1. The documentation is now supplied in plain text form and HTML as well as in
5534 the form of man page sources.
5535
5536 2. C++ compilers don't like assigning (void *) values to other pointer types.
5537 In particular this affects malloc(). Although there is no problem in Standard
5538 C, I've put in casts to keep C++ compilers happy.
5539
5540 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5541 should be (const char *).
5542
5543 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5544 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5545 However, I haven't made this a standard facility. The documentation doesn't
5546 mention it, and the Makefile doesn't support it.
5547
5548 5. The Makefile now contains an "install" target, with editable destinations at
5549 the top of the file. The pcretest program is not installed.
5550
5551 6. pgrep -V now gives the PCRE version number and date.
5552
5553 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5554 causing the entire string to be ignored, instead of just the last character.
5555
5556 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5557 non-matching string, it can take a very, very long time, even for strings of
5558 quite modest length, because of the nested recursion. PCRE now does better in
5559 some of these cases. It does this by remembering the last required literal
5560 character in the pattern, and pre-searching the subject to ensure it is present
5561 before running the real match. In other words, it applies a heuristic to detect
5562 some types of certain failure quickly, and in the above example, if presented
5563 with a string that has no trailing " it gives "no match" very quickly.
5564
5565 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5566 other alternatives are tried instead.
5567
5568
5569 Version 2.06 09-Jun-99
5570 ----------------------
5571
5572 1. Change pcretest's output for amount of store used to show just the code
5573 space, because the remainder (the data block) varies in size between 32-bit and
5574 64-bit systems.
5575
5576 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5577 start matching at. This allows lookbehinds to work when searching for multiple
5578 occurrences in a string.
5579
5580 3. Added additional options to pcretest for testing multiple occurrences:
5581
5582 /+ outputs the rest of the string that follows a match
5583 /g loops for multiple occurrences, using the new startoffset argument
5584 /G loops for multiple occurrences by passing an incremented pointer
5585
5586 4. PCRE wasn't doing the "first character" optimization for patterns starting
5587 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5588 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5589 the letter 'x'. On long subject strings, this gives a significant speed-up.
5590
5591
5592 Version 2.05 21-Apr-99
5593 ----------------------
5594
5595 1. Changed the type of magic_number from int to long int so that it works
5596 properly on 16-bit systems.
5597
5598 2. Fixed a bug which caused patterns starting with .* not to work correctly
5599 when the subject string contained newline characters. PCRE was assuming
5600 anchoring for such patterns in all cases, which is not correct because .* will
5601 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5602 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5603 must be retried after every newline in the subject.
5604
5605
5606 Version 2.04 18-Feb-99
5607 ----------------------
5608
5609 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5610 computation of the store needed to hold the pattern was incorrect (too large).
5611 If such patterns were nested a few deep, this could multiply and become a real
5612 problem.
5613
5614 2. Added /M option to pcretest to show the memory requirement of a specific
5615 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5616
5617 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5618 compiled in such a way that the backtracking after subsequent failure was
5619 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5620 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5621
5622
5623 Version 2.03 02-Feb-99
5624 ----------------------
5625
5626 1. Fixed typo and small mistake in man page.
5627
5628 2. Added 4th condition (GPL supersedes if conflict) and created separate
5629 LICENCE file containing the conditions.
5630
5631 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5632 Perl, that is the internal \ allows the delimiter to be included in the
5633 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5634 the final delimiter, add \ to the end of the pattern (to test the error).
5635
5636 4. Added the convenience functions for extracting substrings after a successful
5637 match. Updated pcretest to make it able to test these functions.
5638
5639
5640 Version 2.02 14-Jan-99
5641 ----------------------
5642
5643 1. Initialized the working variables associated with each extraction so that
5644 their saving and restoring doesn't refer to uninitialized store.
5645
5646 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5647 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5648 fix the problem.
5649
5650 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5651 calls, and wasn't printing the correct value for compiling calls. Increased the
5652 default value of LOOPREPEAT, and the number of significant figures in the
5653 times.
5654
5655 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5656
5657 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5658 a building problem on Windows NT with a FAT file system.
5659
5660
5661 Version 2.01 21-Oct-98
5662 ----------------------
5663
5664 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5665 to character tables built by pcre_maketables() in the current locale. If NULL
5666 is passed, the default tables are used.
5667
5668
5669 Version 2.00 24-Sep-98
5670 ----------------------
5671
5672 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5673 it any more.
5674
5675 2. Allow quantification of (?>) groups, and make it work correctly.
5676
5677 3. The first character computation wasn't working for (?>) groups.
5678
5679 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5680 end of the subject) and add 5.005's \z, which really does match only at the
5681 very end of the subject.
5682
5683 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5684
5685 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5686 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5687 localized options. All options to pcre_study() were also removed.
5688
5689 7. Add other new features from 5.005:
5690
5691 $(?<= positive lookbehind
5692 $(?<! negative lookbehind
5693 (?imsx-imsx) added the unsetting capability
5694 such a setting is global if at outer level; local otherwise
5695 (?imsx-imsx:) non-capturing groups with option setting
5696 (?(cond)re|re) conditional pattern matching
5697
5698 A backreference to itself in a repeated group matches the previous
5699 captured string.
5700
5701 8. General tidying up of studying (both automatic and via "study")
5702 consequential on the addition of new assertions.
5703
5704 9. As in 5.005, unlimited repeated groups that could match an empty substring
5705 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5706 runtime if any iteration does actually match an empty substring.
5707
5708 10. Include the RunTest script in the distribution.
5709
5710 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5711 discrepancies, some of which were old and were also with respect to 5.004. They
5712 have now been fixed.
5713
5714
5715 Version 1.09 28-Apr-98
5716 ----------------------
5717
5718 1. A negated single character class followed by a quantifier with a minimum
5719 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5720 program crashes, or just wrong answers. This did not apply to negated classes
5721 containing more than one character, or to minima other than one.
5722
5723
5724 Version 1.08 27-Mar-98
5725 ----------------------
5726
5727 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5728
5729 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5730 latter must appear before anything that relies on it in the pattern.
5731
5732
5733 Version 1.07 16-Feb-98
5734 ----------------------
5735
5736 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5737 repeat of a potentially empty string).
5738
5739
5740 Version 1.06 23-Jan-98
5741 ----------------------
5742
5743 1. Added Markus Oberhumer's little patches for C++.
5744
5745 2. Literal strings longer than 255 characters were broken.
5746
5747
5748 Version 1.05 23-Dec-97
5749 ----------------------
5750
5751 1. Negated character classes containing more than one character were failing if
5752 PCRE_CASELESS was set at run time.
5753
5754
5755 Version 1.04 19-Dec-97
5756 ----------------------
5757
5758 1. Corrected the man page, where some "const" qualifiers had been omitted.
5759
5760 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5761 input syntax.
5762
5763 3. Fixed memory leak which occurred when a regex with back references was
5764 matched with an offsets vector that wasn't big enough. The temporary memory
5765 that is used in this case wasn't being freed if the match failed.
5766
5767 4. Tidied pcretest to ensure it frees memory that it gets.
5768
5769 5. Temporary memory was being obtained in the case where the passed offsets
5770 vector was exactly big enough.
5771
5772 6. Corrected definition of offsetof() from change 5 below.
5773
5774 7. I had screwed up change 6 below and broken the rules for the use of
5775 setjmp(). Now fixed.
5776
5777
5778 Version 1.03 18-Dec-97
5779 ----------------------
5780
5781 1. A erroneous regex with a missing opening parenthesis was correctly
5782 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5783 on some systems.
5784
5785 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5786 it was reported that one broken compiler failed on the former because "code" is
5787 also an independent variable.
5788
5789 3. The erroneous regex a[]b caused an array overrun reference.
5790
5791 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5792 fail on data ending with that character. (It was going on too far, and checking
5793 the next character, typically a binary zero.) This was specific to the
5794 optimized code for single-character negative classes.
5795
5796 5. Added a contributed patch from the TIN world which does the following:
5797
5798 + Add an undef for memmove, in case the the system defines a macro for it.
5799
5800 + Add a definition of offsetof(), in case there isn't one. (I don't know
5801 the reason behind this - offsetof() is part of the ANSI standard - but
5802 it does no harm).
5803
5804 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5805 most of the places where whitespace preceded '#'. I have given up and
5806 allowed the remaining 2 cases to be at the margin.
5807
5808 + Rename some variables in pcre to eliminate shadowing. This seems very
5809 pedantic, but does no harm, of course.
5810
5811 6. Moved the call to setjmp() into its own function, to get rid of warnings
5812 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5813
5814 7. Constructs such as \d{8,} were compiling into the equivalent of
5815 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5816 outcome, but in this particular case used more store than had been allocated,
5817 which caused the bug to be discovered because it threw up an internal error.
5818
5819 8. The debugging code in both pcre and pcretest for outputting the compiled
5820 form of a regex was going wrong in the case of back references followed by
5821 curly-bracketed repeats.
5822
5823
5824 Version 1.02 12-Dec-97
5825 ----------------------
5826
5827 1. Typos in pcre.3 and comments in the source fixed.
5828
5829 2. Applied a contributed patch to get rid of places where it used to remove
5830 'const' from variables, and fixed some signed/unsigned and uninitialized
5831 variable warnings.
5832
5833 3. Added the "runtest" target to Makefile.
5834
5835 4. Set default compiler flag to -O2 rather than just -O.
5836
5837
5838 Version 1.01 19-Nov-97
5839 ----------------------
5840
5841 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5842 like /([ab]*)*/, that is, for classes with more than one character in them.
5843
5844 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5845 as /((?>a*))*/ (a PCRE_EXTRA facility).
5846
5847
5848 Version 1.00 18-Nov-97
5849 ----------------------
5850
5851 1. Added compile-time macros to support systems such as SunOS4 which don't have
5852 memmove() or strerror() but have other things that can be used instead.
5853
5854 2. Arranged that "make clean" removes the executables.
5855
5856
5857 Version 0.99 27-Oct-97
5858 ----------------------
5859
5860 1. Fixed bug in code for optimizing classes with only one character. It was
5861 initializing a 32-byte map regardless, which could cause it to run off the end
5862 of the memory it had got.
5863
5864 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5865
5866
5867 Version 0.98 22-Oct-97
5868 ----------------------
5869
5870 1. Fixed bug in code for handling temporary memory usage when there are more
5871 back references than supplied space in the ovector. This could cause segfaults.
5872
5873
5874 Version 0.97 21-Oct-97
5875 ----------------------
5876
5877 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5878
5879 2. Optimized negated single characters not to use a bit map.
5880
5881 3. Brought error texts together as macro definitions; clarified some of them;
5882 fixed one that was wrong - it said "range out of order" when it meant "invalid
5883 escape sequence".
5884
5885 4. Changed some char * arguments to const char *.
5886
5887 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5888
5889 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5890 pcretest.
5891
5892
5893 Version 0.96 16-Oct-97
5894 ----------------------
5895
5896 1. Added a simple "pgrep" utility to the distribution.
5897
5898 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5899 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5900 where "ddd" means "one or more decimal digits".
5901
5902 3. Fixed serious bug. If a pattern had a back reference, but the call to
5903 pcre_exec() didn't supply a large enough ovector to record the related
5904 identifying subpattern, the match always failed. PCRE now remembers the number
5905 of the largest back reference, and gets some temporary memory in which to save
5906 the offsets during matching if necessary, in order to ensure that
5907 backreferences always work.
5908
5909 4. Increased the compatibility with Perl in a number of ways:
5910
5911 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5912 to request this handling. The option can be set at compile or exec time.
5913
5914 (b) $ matches before a terminating newline by default; an option
5915 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5916 mode). The option can be set at compile or exec time.
5917
5918 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5919 the same as Perl's. If the decimal number it represents is less than 10
5920 or there aren't that many previous left capturing parentheses, an octal
5921 escape is read. Inside a character class, it's always an octal escape,
5922 even if it is a single digit.
5923
5924 (d) An escaped but undefined alphabetic character is taken as a literal,
5925 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5926 escapes.
5927
5928 (e) {0} is now permitted. (The previous item is removed from the compiled
5929 pattern).
5930
5931 5. Changed all the names of code files so that the basic parts are no longer
5932 than 10 characters, and abolished the teeny "globals.c" file.
5933
5934 6. Changed the handling of character classes; they are now done with a 32-byte
5935 bit map always.
5936
5937 7. Added the -d and /D options to pcretest to make it possible to look at the
5938 internals of compilation without having to recompile pcre.
5939
5940
5941 Version 0.95 23-Sep-97
5942 ----------------------
5943
5944 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5945 \x20 at the start of a run of normal characters. These were being treated as
5946 real characters, instead of the source characters being re-checked.
5947
5948
5949 Version 0.94 18-Sep-97
5950 ----------------------
5951
5952 1. The functions are now thread-safe, with the caveat that the global variables
5953 containing pointers to malloc() and free() or alternative functions are the
5954 same for all threads.
5955
5956 2. Get pcre_study() to generate a bitmap of initial characters for non-
5957 anchored patterns when this is possible, and use it if passed to pcre_exec().
5958
5959
5960 Version 0.93 15-Sep-97
5961 ----------------------
5962
5963 1. /(b)|(:+)/ was computing an incorrect first character.
5964
5965 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5966 but not actually doing anything yet.
5967
5968 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5969 as Perl does (e.g. [-az] or [az-]).
5970
5971 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5972 all possible positions.
5973
5974 5. Split up into different modules to avoid including unneeded functions in a
5975 compiled binary. However, compile and exec are still in one module. The "study"
5976 function is split off.
5977
5978 6. The character tables are now in a separate module whose source is generated
5979 by an auxiliary program - but can then be edited by hand if required. There are
5980 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5981 toupper() in the code.
5982
5983 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5984 make them global. Abolish the function for setting them, as the caller can now
5985 set them directly.
5986
5987
5988 Version 0.92 11-Sep-97
5989 ----------------------
5990
5991 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5992 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5993
5994 2. Caseless matching was not working in character classes if the characters in
5995 the pattern were in upper case.
5996
5997 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5998
5999 4. Make PCRE_ANCHORED public and accept as a compile option.
6000
6001 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
6002 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
6003 pass them.
6004
6005 6. Give an error if bad option bits passed at compile or run time.
6006
6007 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
6008 pcretest to cause it to pass that flag.
6009
6010 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
6011 options, and the first character, if set.
6012
6013 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
6014
6015
6016 Version 0.91 10-Sep-97
6017 ----------------------
6018
6019 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
6020 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
6021
6022 2. PCRE was looping on encountering an indefinitely repeated back reference to
6023 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
6024 Perl does - treats the match as successful.
6025
6026 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5