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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1660 - (show annotations)
Fri Jul 1 09:06:38 2016 UTC (18 hours, 44 minutes ago) by zherczeg
File size: 289894 byte(s)
Fix register overwite in JIT when SSE2 acceleration is enabled.
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
19 Version 8.39 14-June-2016
20 -------------------------
21
22 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
23 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
24 misbehaved. This bug was found by the LLVM fuzzer.
25
26 2. Similar to the above, if an isolated \E was present between an item and its
27 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
28 bug was found by the LLVM fuzzer.
29
30 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
31 working correctly in UCP mode.
32
33 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
34 was set when the pmatch argument was NULL. It now returns REG_INVARG.
35
36 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
37
38 6. An empty \Q\E sequence between an item and its qualifier caused
39 pcre_compile() to misbehave when auto callouts were enabled. This bug was
40 found by the LLVM fuzzer.
41
42 7. If a pattern that was compiled with PCRE_EXTENDED started with white
43 space or a #-type comment that was followed by (?-x), which turns off
44 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
45 pcre_compile() assumed that (?-x) applied to the whole pattern and
46 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
47
48 8. A call of pcre_copy_named_substring() for a named substring whose number
49 was greater than the space in the ovector could cause a crash.
50
51 9. Yet another buffer overflow bug involved duplicate named groups with a
52 group that reset capture numbers (compare 8.38/7 below). Once again, I have
53 just allowed for more memory, even if not needed. (A proper fix is
54 implemented in PCRE2, but it involves a lot of refactoring.)
55
56 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
57 start of the match to be earlier than the end.
58
59 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
60
61 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
62 assertion, caused pcretest to generate incorrect output, and also to read
63 uninitialized memory (detected by ASAN or valgrind).
64
65 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
66 nested set of parentheses of sufficient size caused an overflow of the
67 compiling workspace (which was diagnosed, but of course is not desirable).
68
69 14. And yet another buffer overflow bug involving duplicate named groups, this
70 time nested, with a nested back reference. Yet again, I have just allowed
71 for more memory, because anything more needs all the refactoring that has
72 been done for PCRE2. An example pattern that provoked this bug is:
73 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
74 registered as CVE-2016-1283.
75
76 15. pcretest went into a loop if global matching was requested with an ovector
77 size less than 2. It now gives an error message. This bug was found by
78 afl-fuzz.
79
80 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
81 ("assertion expected") when (?(?C) was not followed by an opening
82 parenthesis.
83
84 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
85 actually affect anything, by sheer luck.
86
87 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
88 static compilation.
89
90 19. Modified the RunTest script to incorporate a valgrind suppressions file so
91 that certain errors, provoked by the SSE2 instruction set when JIT is used,
92 are ignored.
93
94 20. A racing condition is fixed in JIT reported by Mozilla.
95
96 21. Minor code refactor to avoid "array subscript is below array bounds"
97 compiler warning.
98
99 22. Minor code refactor to avoid "left shift of negative number" warning.
100
101 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
102 UCP support.
103
104 24. Refactor to avoid compiler warnings in pcrecpp.cc.
105
106 25. Refactor to fix a typo in pcre_jit_test.c
107
108 26. Patch to support compiling pcrecpp.cc with Intel compiler.
109
110
111 Version 8.38 23-November-2015
112 -----------------------------
113
114 1. If a group that contained a recursive back reference also contained a
115 forward reference subroutine call followed by a non-forward-reference
116 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
117 compile correct code, leading to undefined behaviour or an internally
118 detected error. This bug was discovered by the LLVM fuzzer.
119
120 2. Quantification of certain items (e.g. atomic back references) could cause
121 incorrect code to be compiled when recursive forward references were
122 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
123 This bug was discovered by the LLVM fuzzer.
124
125 3. A repeated conditional group whose condition was a reference by name caused
126 a buffer overflow if there was more than one group with the given name.
127 This bug was discovered by the LLVM fuzzer.
128
129 4. A recursive back reference by name within a group that had the same name as
130 another group caused a buffer overflow. For example:
131 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
132
133 5. A forward reference by name to a group whose number is the same as the
134 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
135 a buffer overflow at compile time. This bug was discovered by the LLVM
136 fuzzer.
137
138 6. A lookbehind assertion within a set of mutually recursive subpatterns could
139 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
140
141 7. Another buffer overflow bug involved duplicate named groups with a
142 reference between their definition, with a group that reset capture
143 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
144 fixed by always allowing for more memory, even if not needed. (A proper fix
145 is implemented in PCRE2, but it involves more refactoring.)
146
147 8. There was no check for integer overflow in subroutine calls such as (?123).
148
149 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
150 being treated as a literal 'l' instead of causing an error.
151
152 10. There was a buffer overflow if pcre_exec() was called with an ovector of
153 size 1. This bug was found by american fuzzy lop.
154
155 11. If a non-capturing group containing a conditional group that could match
156 an empty string was repeated, it was not identified as matching an empty
157 string itself. For example: /^(?:(?(1)x|)+)+$()/.
158
159 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
160 \a and \e in test subject lines.
161
162 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
163 instead of the EBCDIC value.
164
165 14. The handling of \c in an EBCDIC environment has been revised so that it is
166 now compatible with the specification in Perl's perlebcdic page.
167
168 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
169 ASCII/Unicode. This has now been added to the list of characters that are
170 recognized as white space in EBCDIC.
171
172 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
173 error (correctly) when used outside a class, but did not give an error
174 within a class.
175
176 17. \h within a class was incorrectly compiled in EBCDIC environments.
177
178 18. A pattern with an unmatched closing parenthesis that contained a backward
179 assertion which itself contained a forward reference caused buffer
180 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
181
182 19. JIT should return with error when the compiled pattern requires more stack
183 space than the maximum.
184
185 20. A possessively repeated conditional group that could match an empty string,
186 for example, /(?(R))*+/, was incorrectly compiled.
187
188 21. Fix infinite recursion in the JIT compiler when certain patterns such as
189 /(?:|a|){100}x/ are analysed.
190
191 22. Some patterns with character classes involving [: and \\ were incorrectly
192 compiled and could cause reading from uninitialized memory or an incorrect
193 error diagnosis.
194
195 23. Pathological patterns containing many nested occurrences of [: caused
196 pcre_compile() to run for a very long time.
197
198 24. A conditional group with only one branch has an implicit empty alternative
199 branch and must therefore be treated as potentially matching an empty
200 string.
201
202 25. If (?R was followed by - or + incorrect behaviour happened instead of a
203 diagnostic.
204
205 26. Arrange to give up on finding the minimum matching length for overly
206 complex patterns.
207
208 27. Similar to (4) above: in a pattern with duplicated named groups and an
209 occurrence of (?| it is possible for an apparently non-recursive back
210 reference to become recursive if a later named group with the relevant
211 number is encountered. This could lead to a buffer overflow. Wen Guanxing
212 from Venustech ADLAB discovered this bug.
213
214 28. If pcregrep was given the -q option with -c or -l, or when handling a
215 binary file, it incorrectly wrote output to stdout.
216
217 29. The JIT compiler did not restore the control verb head in case of *THEN
218 control verbs. This issue was found by Karl Skomski with a custom LLVM
219 fuzzer.
220
221 30. Error messages for syntax errors following \g and \k were giving inaccurate
222 offsets in the pattern.
223
224 31. Added a check for integer overflow in conditions (?(<digits>) and
225 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
226 fuzzer.
227
228 32. Handling recursive references such as (?2) when the reference is to a group
229 later in the pattern uses code that is very hacked about and error-prone.
230 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
231 give an internal error if it is obvious that compiling has gone wrong.
232
233 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
234 This issue was found by Karl Skomski with a custom LLVM fuzzer.
235
236 34. The JIT compiler should restore the control chain for empty possessive
237 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
238
239 35. Match limit check added to JIT recursion. This issue was found by Karl
240 Skomski with a custom LLVM fuzzer.
241
242 36. Yet another case similar to 27 above has been circumvented by an
243 unconditional allocation of extra memory. This issue is fixed "properly" in
244 PCRE2 by refactoring the way references are handled. Wen Guanxing
245 from Venustech ADLAB discovered this bug.
246
247 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
248 with a custom LLVM fuzzer.
249
250 38. Fixed a corner case of range optimization in JIT.
251
252 39. An incorrect error "overran compiling workspace" was given if there were
253 exactly enough group forward references such that the last one extended
254 into the workspace safety margin. The next one would have expanded the
255 workspace. The test for overflow was not including the safety margin.
256
257 40. A match limit issue is fixed in JIT which was found by Karl Skomski
258 with a custom LLVM fuzzer.
259
260 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
261 work under Windows. (Why has it taken so long for anyone to notice?)
262
263 42. In a character class such as [\W\p{Any}] where both a negative-type escape
264 ("not a word character") and a property escape were present, the property
265 escape was being ignored.
266
267 43. Fix crash caused by very long (*MARK) or (*THEN) names.
268
269 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
270 by a single ASCII character in a class item, was incorrectly compiled in
271 UCP mode. The POSIX class got lost, but only if the single character
272 followed it.
273
274 45. [:punct:] in UCP mode was matching some characters in the range 128-255
275 that should not have been matched.
276
277 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
278 class, all characters with code points greater than 255 are in the class.
279 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
280 such as \w are turned into Unicode properties), wide characters were not
281 correctly handled, and could fail to match.
282
283
284 Version 8.37 28-April-2015
285 --------------------------
286
287 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
288 for those parentheses to be closed with whatever has been captured so far.
289 However, it was failing to mark any other groups between the hightest
290 capture so far and the currrent group as "unset". Thus, the ovector for
291 those groups contained whatever was previously there. An example is the
292 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
293
294 2. If an assertion condition was quantified with a minimum of zero (an odd
295 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
296
297 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
298 unrecognized modifier, a crash could occur.
299
300 4. An attempt to do global matching in pcretest with a zero-length ovector
301 caused a crash.
302
303 5. Fixed a memory leak during matching that could occur for a subpattern
304 subroutine call (recursive or otherwise) if the number of captured groups
305 that had to be saved was greater than ten.
306
307 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
308 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
309 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
310
311 7. A UTF pattern containing a "not" match of a non-ASCII character and a
312 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
313
314 8. When a pattern is compiled, it remembers the highest back reference so that
315 when matching, if the ovector is too small, extra memory can be obtained to
316 use instead. A conditional subpattern whose condition is a check on a
317 capture having happened, such as, for example in the pattern
318 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
319 setting the highest backreference number. This mattered only if pcre_exec()
320 was called with an ovector that was too small to hold the capture, and there
321 was no other kind of back reference (a situation which is probably quite
322 rare). The effect of the bug was that the condition was always treated as
323 FALSE when the capture could not be consulted, leading to a incorrect
324 behaviour by pcre_exec(). This bug has been fixed.
325
326 9. A reference to a duplicated named group (either a back reference or a test
327 for being set in a conditional) that occurred in a part of the pattern where
328 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
329 to be incorrectly calculated, leading to overwriting.
330
331 10. A mutually recursive set of back references such as (\2)(\1) caused a
332 segfault at study time (while trying to find the minimum matching length).
333 The infinite loop is now broken (with the minimum length unset, that is,
334 zero).
335
336 11. If an assertion that was used as a condition was quantified with a minimum
337 of zero, matching went wrong. In particular, if the whole group had
338 unlimited repetition and could match an empty string, a segfault was
339 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
340 assertions to be quantified, but not if they are being used as conditions,
341 so the above pattern is faulted by Perl. PCRE has now been changed so that
342 it also rejects such patterns.
343
344 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
345 failed to allow the zero-repeat case if pcre2_exec() was called with an
346 ovector too small to capture the group.
347
348 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
349 Red Hat Product Security:
350
351 (a) A crash if /K and /F were both set with the option to save the compiled
352 pattern.
353
354 (b) Another crash if the option to print captured substrings in a callout
355 was combined with setting a null ovector, for example \O\C+ as a subject
356 string.
357
358 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
359 forward reference repeated a large (but limited) number of times within a
360 repeated outer group that has a zero minimum quantifier, caused incorrect
361 code to be compiled, leading to the error "internal error:
362 previously-checked referenced subpattern not found" when an incorrect
363 memory address was read. This bug was reported as "heap overflow",
364 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
365 CVE-2015-2325.
366
367 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
368 call within a group that also contained a recursive back reference caused
369 incorrect code to be compiled. This bug was reported as "heap overflow",
370 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
371 number CVE-2015-2326.
372
373 24. Computing the size of the JIT read-only data in advance has been a source
374 of various issues, and new ones are still appear unfortunately. To fix
375 existing and future issues, size computation is eliminated from the code,
376 and replaced by on-demand memory allocation.
377
378 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
379 adjacent to the end of the range, and the range contained characters with
380 more than one other case, caused incorrect behaviour when compiled in UTF
381 mode. In that example, the range a-j was left out of the class.
382
383 26. Fix JIT compilation of conditional blocks, which assertion
384 is converted to (*FAIL). E.g: /(?(?!))/.
385
386 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
387 discovered by the LLVM fuzzer.
388
389 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
390 when this assertion was used as a condition, for example (?(?!)a|b). In
391 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
392 error about an unsupported item.
393
394 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
395 possessification code could take exponential time to complete. A recursion
396 depth limit of 1000 has been imposed to limit the resources used by this
397 optimization.
398
399 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
400 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
401 because \S ensures they are all in the class. The code for doing this was
402 interacting badly with the code for computing the amount of space needed to
403 compile the pattern, leading to a buffer overflow. This bug was discovered
404 by the LLVM fuzzer.
405
406 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
407 other kinds of group caused stack overflow at compile time. This bug was
408 discovered by the LLVM fuzzer.
409
410 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
411 between a subroutine call and its quantifier was incorrectly compiled,
412 leading to buffer overflow or other errors. This bug was discovered by the
413 LLVM fuzzer.
414
415 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
416 assertion after (?(. The code was failing to check the character after
417 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
418 was discovered by the LLVM fuzzer.
419
420 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
421 a fixed maximum following a group that contains a subroutine reference was
422 incorrectly compiled and could trigger buffer overflow. This bug was
423 discovered by the LLVM fuzzer.
424
425 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
426 caused a stack overflow instead of the diagnosis of a non-fixed length
427 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
428
429 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
430 (e.g. /(?<=\Ka)/) could make pcregrep loop.
431
432 37. There was a similar problem to 36 in pcretest for global matches.
433
434 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
435 and a subsequent item in the pattern caused a non-match, backtracking over
436 the repeated \X did not stop, but carried on past the start of the subject,
437 causing reference to random memory and/or a segfault. There were also some
438 other cases where backtracking after \C could crash. This set of bugs was
439 discovered by the LLVM fuzzer.
440
441 39. The function for finding the minimum length of a matching string could take
442 a very long time if mutual recursion was present many times in a pattern,
443 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
444 method has been implemented. This infelicity was discovered by the LLVM
445 fuzzer.
446
447 40. Static linking against the PCRE library using the pkg-config module was
448 failing on missing pthread symbols.
449
450
451 Version 8.36 26-September-2014
452 ------------------------------
453
454 1. Got rid of some compiler warnings in the C++ modules that were shown up by
455 -Wmissing-field-initializers and -Wunused-parameter.
456
457 2. The tests for quantifiers being too big (greater than 65535) were being
458 applied after reading the number, and stupidly assuming that integer
459 overflow would give a negative number. The tests are now applied as the
460 numbers are read.
461
462 3. Tidy code in pcre_exec.c where two branches that used to be different are
463 now the same.
464
465 4. The JIT compiler did not generate match limit checks for certain
466 bracketed expressions with quantifiers. This may lead to exponential
467 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
468 issue should be resolved now.
469
470 5. Fixed an issue, which occures when nested alternatives are optimized
471 with table jumps.
472
473 6. Inserted two casts and changed some ints to size_t in the light of some
474 reported 64-bit compiler warnings (Bugzilla 1477).
475
476 7. Fixed a bug concerned with zero-minimum possessive groups that could match
477 an empty string, which sometimes were behaving incorrectly in the
478 interpreter (though correctly in the JIT matcher). This pcretest input is
479 an example:
480
481 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
482 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
483
484 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
485 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
486 for an empty string was breaking the inner loop and carrying on at a lower
487 level, when possessive repeated groups should always return to a higher
488 level as they have no backtrack points in them. The empty string test now
489 occurs at the outer level.
490
491 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
492 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
493
494 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
495 doesn't).
496
497 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
498 to hold a pattern to be incorrectly computed (too small) when there were
499 named back references to duplicated names. This could cause "internal
500 error: code overflow" or "double free or corruption" or other memory
501 handling errors.
502
503 11. When named subpatterns had the same prefixes, back references could be
504 confused. For example, in this pattern:
505
506 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
507
508 the reference to 'Name' was incorrectly treated as a reference to a
509 duplicate name.
510
511 12. A pattern such as /^s?c/mi8 where the optional character has more than
512 one "other case" was incorrectly compiled such that it would only try to
513 match starting at "c".
514
515 13. When a pattern starting with \s was studied, VT was not included in the
516 list of possible starting characters; this should have been part of the
517 8.34/18 patch.
518
519 14. If a character class started [\Qx]... where x is any character, the class
520 was incorrectly terminated at the ].
521
522 15. If a pattern that started with a caseless match for a character with more
523 than one "other case" was studied, PCRE did not set up the starting code
524 unit bit map for the list of possible characters. Now it does. This is an
525 optimization improvement, not a bug fix.
526
527 16. The Unicode data tables have been updated to Unicode 7.0.0.
528
529 17. Fixed a number of memory leaks in pcregrep.
530
531 18. Avoid a compiler warning (from some compilers) for a function call with
532 a cast that removes "const" from an lvalue by using an intermediate
533 variable (to which the compiler does not object).
534
535 19. Incorrect code was compiled if a group that contained an internal recursive
536 back reference was optional (had quantifier with a minimum of zero). This
537 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
538 caused segmentation faults because of stack overflows at compile time.
539
540 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
541 group that is quantified with an indefinite repeat, caused a compile-time
542 loop which used up all the system stack and provoked a segmentation fault.
543 This was not the same bug as 19 above.
544
545 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
546 Patch by Mike Frysinger.
547
548
549 Version 8.35 04-April-2014
550 --------------------------
551
552 1. A new flag is set, when property checks are present in an XCLASS.
553 When this flag is not set, PCRE can perform certain optimizations
554 such as studying these XCLASS-es.
555
556 2. The auto-possessification of character sets were improved: a normal
557 and an extended character set can be compared now. Furthermore
558 the JIT compiler optimizes more character set checks.
559
560 3. Got rid of some compiler warnings for potentially uninitialized variables
561 that show up only when compiled with -O2.
562
563 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
564 of a match later then the end of the match. The pcretest program was not
565 handling the case sensibly - it was outputting from the start to the next
566 binary zero. It now reports this situation in a message, and outputs the
567 text from the end to the start.
568
569 5. Fast forward search is improved in JIT. Instead of the first three
570 characters, any three characters with fixed position can be searched.
571 Search order: first, last, middle.
572
573 6. Improve character range checks in JIT. Characters are read by an inprecise
574 function now, which returns with an unknown value if the character code is
575 above a certain threshold (e.g: 256). The only limitation is that the value
576 must be bigger than the threshold as well. This function is useful when
577 the characters above the threshold are handled in the same way.
578
579 7. The macros whose names start with RAWUCHAR are placeholders for a future
580 mode in which only the bottom 21 bits of 32-bit data items are used. To
581 make this more memorable for those maintaining the code, the names have
582 been changed to start with UCHAR21, and an extensive comment has been added
583 to their definition.
584
585 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
586 to the export list in Makefile.am (they were accidentally omitted from the
587 8.34 tarball).
588
589 9. The informational output from pcretest used the phrase "starting byte set"
590 which is inappropriate for the 16-bit and 32-bit libraries. As the output
591 for "first char" and "need char" really means "non-UTF-char", I've changed
592 "byte" to "char", and slightly reworded the output. The documentation about
593 these values has also been (I hope) clarified.
594
595 10. Another JIT related optimization: use table jumps for selecting the correct
596 backtracking path, when more than four alternatives are present inside a
597 bracket.
598
599 11. Empty match is not possible, when the minimum length is greater than zero,
600 and there is no \K in the pattern. JIT should avoid empty match checks in
601 such cases.
602
603 12. In a caseless character class with UCP support, when a character with more
604 than one alternative case was not the first character of a range, not all
605 the alternative cases were added to the class. For example, s and \x{17f}
606 are both alternative cases for S: the class [RST] was handled correctly,
607 but [R-T] was not.
608
609 13. The configure.ac file always checked for pthread support when JIT was
610 enabled. This is not used in Windows, so I have put this test inside a
611 check for the presence of windows.h (which was already tested for).
612
613 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
614 The algorithm provides a way to skip certain starting offsets, and usually
615 faster than linear prefix searches.
616
617 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
618 as for 'fr_FR' and 'french'. For some reason, however, it then used the
619 Windows-specific input and output files, which have 'french' screwed in.
620 So this could never have worked. One of the problems with locales is that
621 they aren't always the same. I have now updated RunTest so that it checks
622 the output of the locale test (test 3) against three different output
623 files, and it allows the test to pass if any one of them matches. With luck
624 this should make the test pass on some versions of Solaris where it was
625 failing. Because of the uncertainty, the script did not used to stop if
626 test 3 failed; it now does. If further versions of a French locale ever
627 come to light, they can now easily be added.
628
629 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
630 there was a message during ./configure, but it did not stop. This now
631 provokes an error. The invalid example in README has been corrected.
632 If a value less than the minimum is given, the minimum value has always
633 been used, but now a warning is given.
634
635 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
636 was a bug in the test system, which is now fixed. Also, the list of various
637 configurations that are tested for each release did not have one with both
638 16/32 bits and --enable-bar-anycrlf. It now does.
639
640 18. pcretest was missing "-C bsr" for displaying the \R default setting.
641
642 19. Little endian PowerPC systems are supported now by the JIT compiler.
643
644 20. The fast forward newline mechanism could enter to an infinite loop on
645 certain invalid UTF-8 input. Although we don't support these cases
646 this issue can be fixed by a performance optimization.
647
648 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
649 not take account if existing stack usage. There is now a new global
650 variable called pcre_stack_guard that can be set to point to an external
651 function to check stack availability. It is called at the start of
652 processing every parenthesized group.
653
654 22. A typo in the code meant that in ungreedy mode the max/min qualifier
655 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
656 not match "ab".
657
658 23. When UTF was disabled, the JIT program reported some incorrect compile
659 errors. These messages are silenced now.
660
661 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
662 compiler.
663
664 25. Change all the temporary files used in RunGrepTest to be different to those
665 used by RunTest so that the tests can be run simultaneously, for example by
666 "make -j check".
667
668
669 Version 8.34 15-December-2013
670 -----------------------------
671
672 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
673 executable memory. Patch inspired by Carsten Klein.
674
675 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
676 this macro is never tested and has no effect, because the work to support
677 coverage involves only compiling and linking options and special targets in
678 the Makefile. The comment in config.h implied that defining the macro would
679 enable coverage support, which is totally false. There was also support for
680 setting this macro in the CMake files (my fault, I just copied it from
681 configure). SUPPORT_GCOV has now been removed.
682
683 3. Make a small performance improvement in strlen16() and strlen32() in
684 pcretest.
685
686 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
687 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
688 these cases). There was also one in pcretest.c.
689
690 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
691
692 6. In UTF mode, the code for checking whether a group could match an empty
693 string (which is used for indefinitely repeated groups to allow for
694 breaking an infinite loop) was broken when the group contained a repeated
695 negated single-character class with a character that occupied more than one
696 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
697 UTF-8 mode). The effect was undefined: the group might or might not be
698 deemed as matching an empty string, or the program might have crashed.
699
700 7. The code for checking whether a group could match an empty string was not
701 recognizing that \h, \H, \v, \V, and \R must match a character.
702
703 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
704 an empty string. If it can, pcretest shows this in its information output.
705
706 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
707 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
708 matched by pcre_exec() without using JIT:
709
710 (a) If the rest of the pattern did not match after a maximal run of
711 grapheme clusters, the code for backing up to try with fewer of them
712 did not always back up over a full grapheme when characters that do not
713 have the modifier quality were involved, e.g. Hangul syllables.
714
715 (b) If the match point in a subject started with modifier character, and
716 there was no match, the code could incorrectly back up beyond the match
717 point, and potentially beyond the first character in the subject,
718 leading to a segfault or an incorrect match result.
719
720 10. A conditional group with an assertion condition could lead to PCRE
721 recording an incorrect first data item for a match if no other first data
722 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
723 first data item, and therefore matched "ca" after "c" instead of at the
724 start.
725
726 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
727 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
728
729 12. The source of pcregrep now includes z/OS-specific code so that it can be
730 compiled for z/OS as part of the special z/OS distribution.
731
732 13. Added the -T and -TM options to pcretest.
733
734 14. The code in pcre_compile.c for creating the table of named capturing groups
735 has been refactored. Instead of creating the table dynamically during the
736 actual compiling pass, the information is remembered during the pre-compile
737 pass (on the stack unless there are more than 20 named groups, in which
738 case malloc() is used) and the whole table is created before the actual
739 compile happens. This has simplified the code (it is now nearly 150 lines
740 shorter) and prepared the way for better handling of references to groups
741 with duplicate names.
742
743 15. A back reference to a named subpattern when there is more than one of the
744 same name now checks them in the order in which they appear in the pattern.
745 The first one that is set is used for the reference. Previously only the
746 first one was inspected. This change makes PCRE more compatible with Perl.
747
748 16. Unicode character properties were updated from Unicode 6.3.0.
749
750 17. The compile-time code for auto-possessification has been refactored, based
751 on a patch by Zoltan Herczeg. It now happens after instead of during
752 compilation. The code is cleaner, and more cases are handled. The option
753 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
754 options in pcretest are provided to set it. It can also be set by
755 (*NO_AUTO_POSSESS) at the start of a pattern.
756
757 18. The character VT has been added to the default ("C" locale) set of
758 characters that match \s and are generally treated as white space,
759 following this same change in Perl 5.18. There is now no difference between
760 "Perl space" and "POSIX space". Whether VT is treated as white space in
761 other locales depends on the locale.
762
763 19. The code for checking named groups as conditions, either for being set or
764 for being recursed, has been refactored (this is related to 14 and 15
765 above). Processing unduplicated named groups should now be as fast at
766 numerical groups, and processing duplicated groups should be faster than
767 before.
768
769 20. Two patches to the CMake build system, by Alexander Barkov:
770
771 (1) Replace the "source" command by "." in CMakeLists.txt because
772 "source" is a bash-ism.
773
774 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
775 without these the CMake build does not work on Solaris.
776
777 21. Perl has changed its handling of \8 and \9. If there is no previously
778 encountered capturing group of those numbers, they are treated as the
779 literal characters 8 and 9 instead of a binary zero followed by the
780 literals. PCRE now does the same.
781
782 22. Following Perl, added \o{} to specify codepoints in octal, making it
783 possible to specify values greater than 0777 and also making them
784 unambiguous.
785
786 23. Perl now gives an error for missing closing braces after \x{... instead of
787 treating the string as literal. PCRE now does the same.
788
789 24. RunTest used to grumble if an inappropriate test was selected explicitly,
790 but just skip it when running all tests. This make it awkward to run ranges
791 of tests when one of them was inappropriate. Now it just skips any
792 inappropriate tests, as it always did when running all tests.
793
794 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
795 character types such as \d or \w, too many callouts were inserted, and the
796 data that they returned was rubbish.
797
798 26. In UCP mode, \s was not matching two of the characters that Perl matches,
799 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
800 were matched by \h. The code has now been refactored so that the lists of
801 the horizontal and vertical whitespace characters used for \h and \v (which
802 are defined only in one place) are now also used for \s.
803
804 27. Add JIT support for the 64 bit TileGX architecture.
805 Patch by Jiong Wang (Tilera Corporation).
806
807 28. Possessive quantifiers for classes (both explicit and automatically
808 generated) now use special opcodes instead of wrapping in ONCE brackets.
809
810 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
811 (because it's meaningless), this was not happening when PCRE_CASELESS was
812 set. Not wrong, but inefficient.
813
814 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
815 properties for \w, \d, etc) is present in a test regex. Otherwise if the
816 test contains no characters greater than 255, Perl doesn't realise it
817 should be using Unicode semantics.
818
819 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
820 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
821 does in Unicode mode.
822
823 32. Added the "forbid" facility to pcretest so that putting tests into the
824 wrong test files can sometimes be quickly detected.
825
826 33. There is now a limit (default 250) on the depth of nesting of parentheses.
827 This limit is imposed to control the amount of system stack used at compile
828 time. It can be changed at build time by --with-parens-nest-limit=xxx or
829 the equivalent in CMake.
830
831 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
832 errors. Perl warns for these when in warning mode, but PCRE has no facility
833 for giving warnings.
834
835 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
836 However, this was not working for (?!) because it is optimized to (*FAIL),
837 for which PCRE does not allow quantifiers. The optimization is now disabled
838 when a quantifier follows (?!). I can't see any use for this, but it makes
839 things uniform.
840
841 36. Perl no longer allows group names to start with digits, so I have made this
842 change also in PCRE. It simplifies the code a bit.
843
844 37. In extended mode, Perl ignores spaces before a + that indicates a
845 possessive quantifier. PCRE allowed a space before the quantifier, but not
846 before the possessive +. It now does.
847
848 38. The use of \K (reset reported match start) within a repeated possessive
849 group such as (a\Kb)*+ was not working.
850
851 40. Document that the same character tables must be used at compile time and
852 run time, and that the facility to pass tables to pcre_exec() and
853 pcre_dfa_exec() is for use only with saved/restored patterns.
854
855 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
856 features for Builds with MSVC:
857
858 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
859 builds). The .rc files can be used to set FileDescription and many other
860 attributes.
861
862 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
863 This allows higher-level build scripts which want .pdb files to avoid
864 hard-coding the exact files needed."
865
866 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
867 mean "start of word" and "end of word", respectively, as a transition aid.
868
869 43. A minimizing repeat of a class containing codepoints greater than 255 in
870 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
871 compiled to use the heap for recursion.
872
873 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
874 is configured.
875
876
877 Version 8.33 28-May-2013
878 ------------------------
879
880 1. Added 'U' to some constants that are compared to unsigned integers, to
881 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
882 variables that are added to signed variables, to ensure the result is
883 signed and can be negated.
884
885 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
886 CMake config files.
887
888 3. Revise the creation of config.h.generic so that all boolean macros are
889 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
890 overriding via -D on the command line possible.
891
892 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
893 to unsigned int is reported to make a quite noticeable speed difference in
894 a specific Windows environment. Testing on Linux did also appear to show
895 some benefit (and it is clearly not harmful). Also fixed the definition of
896 Xop which should be unsigned.
897
898 5. Related to (4), changing the definition of the intermediate variable cc
899 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
900 improvements.
901
902 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
903 unnecessary spaces.
904
905 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
906 and later.
907
908 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
909
910 9. Optimizing fast_forward_start_bits in JIT.
911
912 10. Adding support for callouts in JIT, and fixing some issues revealed
913 during this work. Namely:
914
915 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
916
917 (b) Minimum length was not checked before the matching is started.
918
919 11. The value of capture_last that is passed to callouts was incorrect in some
920 cases when there was a capture on one path that was subsequently abandoned
921 after a backtrack. Also, the capture_last value is now reset after a
922 recursion, since all captures are also reset in this case.
923
924 12. The interpreter no longer returns the "too many substrings" error in the
925 case when an overflowing capture is in a branch that is subsequently
926 abandoned after a backtrack.
927
928 13. In the pathological case when an offset vector of size 2 is used, pcretest
929 now prints out the matched string after a yield of 0 or 1.
930
931 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
932 Only supported by the JIT compiler at the moment.
933
934 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
935
936 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
937 offset of the starting point of the matching process, provided the offsets
938 vector is large enough.
939
940 17. The \A escape now records a lookbehind value of 1, though its execution
941 does not actually inspect the previous character. This is to ensure that,
942 in partial multi-segment matching, at least one character from the old
943 segment is retained when a new segment is processed. Otherwise, if there
944 are no lookbehinds in the pattern, \A might match incorrectly at the start
945 of a new segment.
946
947 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
948
949 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
950 gives some modest performance improvement in 8-bit mode.
951
952 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
953 be expressed in certain programming languages using Universal Character
954 Names.
955
956 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
957 which points out that "non characters" are not "characters that may not
958 appear in Unicode strings" but rather "characters that are reserved for
959 internal use and have only local meaning".
960
961 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
962 there was a conditional group that depended on an assertion, if the
963 assertion was false, the callout that immediately followed the alternation
964 in the condition was skipped when pcre_exec() was used for matching.
965
966 23. Allow an explicit callout to be inserted before an assertion that is the
967 condition for a conditional group, for compatibility with automatic
968 callouts, which always insert a callout at this point.
969
970 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
971 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
972
973 25. (*PRUNE) is now supported by the JIT compiler.
974
975 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
976
977 27. Fix the case where there are two or more SKIPs with arguments that may be
978 ignored.
979
980 28. (*SKIP) is now supported by the JIT compiler.
981
982 29. (*THEN) is now supported by the JIT compiler.
983
984 30. Update RunTest with additional test selector options.
985
986 31. The way PCRE handles backtracking verbs has been changed in two ways.
987
988 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
989 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
990 backtracking. In some cases this makes it more Perl-compatible, but Perl's
991 rather obscure rules do not always do the same thing.
992
993 (2) Previously, backtracking verbs were confined within assertions. This is
994 no longer the case for positive assertions, except for (*ACCEPT). Again,
995 this sometimes improves Perl compatibility, and sometimes does not.
996
997 32. A number of tests that were in test 2 because Perl did things differently
998 have been moved to test 1, because either Perl or PCRE has changed, and
999 these tests are now compatible.
1000
1001 32. Backtracking control verbs are now handled in the same way in JIT and
1002 interpreter.
1003
1004 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1005 contained a forward subroutine reference caused a compile error.
1006
1007 34. Auto-detect and optimize limited repetitions in JIT.
1008
1009 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1010 blocking (*UTF) etc.
1011
1012 36. In the interpreter, maximizing pattern repetitions for characters and
1013 character types now use tail recursion, which reduces stack usage.
1014
1015 37. The value of the max lookbehind was not correctly preserved if a compiled
1016 and saved regex was reloaded on a host of different endianness.
1017
1018 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1019 of the compiled pattern block, expand the flags field from 16 to 32 bits
1020 because it was almost full.
1021
1022 39. Try madvise first before posix_madvise.
1023
1024 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1025 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1026 original change locked out all matches of empty strings. This has been
1027 changed so that one match of an empty string per line is recognized.
1028 Subsequent searches on the same line (for colouring or for --only-matching,
1029 for example) do not recognize empty strings.
1030
1031 41. Applied a user patch to fix a number of spelling mistakes in comments.
1032
1033 42. Data lines longer than 65536 caused pcretest to crash.
1034
1035 43. Clarified the data type for length and startoffset arguments for pcre_exec
1036 and pcre_dfa_exec in the function-specific man pages, where they were
1037 explicitly stated to be in bytes, never having been updated. I also added
1038 some clarification to the pcreapi man page.
1039
1040 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1041 a segmentation fault.
1042
1043
1044 Version 8.32 30-November-2012
1045 -----------------------------
1046
1047 1. Improved JIT compiler optimizations for first character search and single
1048 character iterators.
1049
1050 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1051 Patch by Daniel Richard G.
1052
1053 3. Single character iterator optimizations in the JIT compiler.
1054
1055 4. Improved JIT compiler optimizations for character ranges.
1056
1057 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1058 Reported by Giuseppe D'Angelo.
1059
1060 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1061 of a line, was being set incorrectly in cases where .* appeared inside
1062 atomic brackets at the start of a pattern, or where there was a subsequent
1063 *PRUNE or *SKIP.
1064
1065 7. Improved instruction cache flush for POWER/PowerPC.
1066 Patch by Daniel Richard G.
1067
1068 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1069 grep:
1070
1071 (a) There is now no limit to the number of patterns to be matched.
1072
1073 (b) An error is given if a pattern is too long.
1074
1075 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1076 are now supported.
1077
1078 (d) --exclude-from and --include-from (multiple use) have been added.
1079
1080 (e) Exclusions and inclusions now apply to all files and directories, not
1081 just to those obtained from scanning a directory recursively.
1082
1083 (f) Multiple uses of -f and --file-list are now supported.
1084
1085 (g) In a Windows environment, the default for -d has been changed from
1086 "read" (the GNU grep default) to "skip", because otherwise the presence
1087 of a directory in the file list provokes an error.
1088
1089 (h) The documentation has been revised and clarified in places.
1090
1091 9. Improve the matching speed of capturing brackets.
1092
1093 10. Changed the meaning of \X so that it now matches a Unicode extended
1094 grapheme cluster.
1095
1096 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1097 out POSIX threads when JIT support is configured.
1098
1099 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1100
1101 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1102 structure could go wrong in environments where size_t is not the same size
1103 as int.
1104
1105 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1106 set.
1107
1108 15. The EBCDIC support had decayed; later updates to the code had included
1109 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1110 general tidy up of EBCDIC-related issues, and the documentation was also
1111 not quite right. There is now a test that can be run on ASCII systems to
1112 check some of the EBCDIC-related things (but is it not a full test).
1113
1114 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1115 in a small tidy to the code.
1116
1117 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1118
1119 18. If the --only-matching (-o) option in pcregrep is specified multiple
1120 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1121 substrings matched by the 1st and 2nd capturing parentheses. A separating
1122 string can be specified by --om-separator (default empty).
1123
1124 19. Improving the first n character searches.
1125
1126 20. Turn case lists for horizontal and vertical white space into macros so that
1127 they are defined only once.
1128
1129 21. This set of changes together give more compatible Unicode case-folding
1130 behaviour for characters that have more than one other case when UCP
1131 support is available.
1132
1133 (a) The Unicode property table now has offsets into a new table of sets of
1134 three or more characters that are case-equivalent. The MultiStage2.py
1135 script that generates these tables (the pcre_ucd.c file) now scans
1136 CaseFolding.txt instead of UnicodeData.txt for character case
1137 information.
1138
1139 (b) The code for adding characters or ranges of characters to a character
1140 class has been abstracted into a generalized function that also handles
1141 case-independence. In UTF-mode with UCP support, this uses the new data
1142 to handle characters with more than one other case.
1143
1144 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1145 whose other case is greater than 256 are now correctly matched
1146 caselessly. Previously, the high codepoint matched the low one, but not
1147 vice versa.
1148
1149 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1150 of the new class addition function, using character lists defined as
1151 macros alongside the case definitions of 20 above.
1152
1153 (e) Caseless back references now work with characters that have more than
1154 one other case.
1155
1156 (f) General caseless matching of characters with more than one other case
1157 is supported.
1158
1159 22. Unicode character properties were updated from Unicode 6.2.0
1160
1161 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1162
1163 24. Add support for 32-bit character strings, and UTF-32
1164
1165 25. Major JIT compiler update (code refactoring and bugfixing).
1166 Experimental Sparc 32 support is added.
1167
1168 26. Applied a modified version of Daniel Richard G's patch to create
1169 pcre.h.generic and config.h.generic by "make" instead of in the
1170 PrepareRelease script.
1171
1172 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1173 pcre_compile.c when checking for a zero character.
1174
1175 28. Introducing a native interface for JIT. Through this interface, the compiled
1176 machine code can be directly executed. The purpose of this interface is to
1177 provide fast pattern matching, so several sanity checks are not performed.
1178 However, feature tests are still performed. The new interface provides
1179 1.4x speedup compared to the old one.
1180
1181 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1182 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1183 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1184
1185 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1186 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1187 the "old" RFC 2279). Instead, it ended up passing a negative length to
1188 pcre_exec().
1189
1190 31. Add support for GCC's visibility feature to hide internal functions.
1191
1192 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1193 "unknown -C option" after outputting 0 or 1.
1194
1195 33. There is now support for generating a code coverage report for the test
1196 suite in environments where gcc is the compiler and lcov is installed. This
1197 is mainly for the benefit of the developers.
1198
1199 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1200 unaddressable using valgrind annotations, allowing valgrind to detect
1201 invalid memory accesses. This is mainly for the benefit of the developers.
1202
1203 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1204
1205 26. Give configure error if --enable-cpp but no C++ compiler found.
1206
1207
1208 Version 8.31 06-July-2012
1209 -------------------------
1210
1211 1. Fixing a wrong JIT test case and some compiler warnings.
1212
1213 2. Removed a bashism from the RunTest script.
1214
1215 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1216 to unsigned type, result still unsigned" that was given by an MS compiler
1217 on encountering the code "-sizeof(xxx)".
1218
1219 4. Partial matching support is added to the JIT compiler.
1220
1221 5. Fixed several bugs concerned with partial matching of items that consist
1222 of more than one character:
1223
1224 (a) /^(..)\1/ did not partially match "aba" because checking references was
1225 done on an "all or nothing" basis. This also applied to repeated
1226 references.
1227
1228 (b) \R did not give a hard partial match if \r was found at the end of the
1229 subject.
1230
1231 (c) \X did not give a hard partial match after matching one or more
1232 characters at the end of the subject.
1233
1234 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1235 a partial match for the string "\r".
1236
1237 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1238 a partial match for a CR character at the end of the subject string.
1239
1240 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1241 running pcretest, the text "(JIT)" added to the output whenever JIT is
1242 actually used to run the match.
1243
1244 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1245 or /S+[+] with a digit between 1 and 7.
1246
1247 8. OP_NOT now supports any UTF character not just single-byte ones.
1248
1249 9. (*MARK) control verb is now supported by the JIT compiler.
1250
1251 10. The command "./RunTest list" lists the available tests without actually
1252 running any of them. (Because I keep forgetting what they all are.)
1253
1254 11. Add PCRE_INFO_MAXLOOKBEHIND.
1255
1256 12. Applied a (slightly modified) user-supplied patch that improves performance
1257 when the heap is used for recursion (compiled with --disable-stack-for-
1258 recursion). Instead of malloc and free for each heap frame each time a
1259 logical recursion happens, frames are retained on a chain and re-used where
1260 possible. This sometimes gives as much as 30% improvement.
1261
1262 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1263 call.
1264
1265 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1266
1267 15. It is now possible to link pcretest with libedit as an alternative to
1268 libreadline.
1269
1270 16. (*COMMIT) control verb is now supported by the JIT compiler.
1271
1272 17. The Unicode data tables have been updated to Unicode 6.1.0.
1273
1274 18. Added --file-list option to pcregrep.
1275
1276 19. Added binary file support to pcregrep, including the -a, --binary-files,
1277 -I, and --text options.
1278
1279 20. The madvise function is renamed for posix_madvise for QNX compatibility
1280 reasons. Fixed by Giuseppe D'Angelo.
1281
1282 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1283
1284 22. Changed the option for creating symbolic links for 16-bit man pages from
1285 -s to -sf so that re-installing does not cause issues.
1286
1287 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1288
1289 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1290 matches in certain environments (the workspace was not being correctly
1291 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1292 some of the workspace data at the beginning of a restart.
1293
1294 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1295 was not doing so when it should - probably a typo introduced by SVN 528
1296 (change 8.10/14).
1297
1298 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1299 \w+ when the character tables indicated that \x{c4} was a word character.
1300 There were several related cases, all because the tests for doing a table
1301 lookup were testing for characters less than 127 instead of 255.
1302
1303 27. If a pattern contains capturing parentheses that are not used in a match,
1304 their slots in the ovector are set to -1. For those that are higher than
1305 any matched groups, this happens at the end of processing. In the case when
1306 there were back references that the ovector was too small to contain
1307 (causing temporary malloc'd memory to be used during matching), and the
1308 highest capturing number was not used, memory off the end of the ovector
1309 was incorrectly being set to -1. (It was using the size of the temporary
1310 memory instead of the true size.)
1311
1312 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1313 ovector size, it uses memory at the end of the block that it has got.
1314
1315 29. Check for an overlong MARK name and give an error at compile time. The
1316 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1317
1318 30. JIT compiler update.
1319
1320 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1321 Rill for the patch.
1322
1323 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1324
1325 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1326
1327 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1328 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1329 was enabled.
1330
1331 35. Improve JIT code generation for greedy plus quantifier.
1332
1333 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1334 1 to "aa" instead of to an empty string. The bug affected repeated groups
1335 that could potentially match an empty string.
1336
1337 37. Optimizing single character iterators in JIT.
1338
1339 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1340 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1341 codepoints that are too big for the mode are faulted, and in a UTF mode,
1342 disallowed codepoints are also faulted.
1343
1344 39. If PCRE was compiled with UTF support, in three places in the DFA
1345 matcher there was code that should only have been obeyed in UTF mode, but
1346 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1347 processing when bytes with values greater than 127 were present. In 16-bit
1348 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1349 both cases the values are those that cannot be the first data item in a UTF
1350 character. The three items that might have provoked this were recursions,
1351 possessively repeated groups, and atomic groups.
1352
1353 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1354 and pcregrep, because some OS require shared objects to be explicitly
1355 passed to ld, causing the link step to fail if they are not.
1356
1357 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1358 mode, patterns that started with \h* or \R* might be incorrectly matched.
1359
1360
1361 Version 8.30 04-February-2012
1362 -----------------------------
1363
1364 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1365 name is defined in ctype.h.
1366
1367 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1368 only in quite long subpatterns.
1369
1370 3. Removed the function pcre_info(), which has been obsolete and deprecated
1371 since it was replaced by pcre_fullinfo() in February 2000.
1372
1373 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1374 match a (*MARK), and the match failed at the start of the subject, a
1375 reference to memory before the start of the subject could occur. This bug
1376 was introduced by fix 17 of release 8.21.
1377
1378 5. A reference to an unset group with zero minimum repetition was giving
1379 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1380 /(another)?(\1?)test/ matched against "hello world test". This bug was
1381 introduced in release 8.13.
1382
1383 6. Add support for 16-bit character strings (a large amount of work involving
1384 many changes and refactorings).
1385
1386 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1387 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1388 from a file.
1389
1390 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1391 rounding is not applied in this particular case).
1392
1393 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1394 if they appear, or are escaped, in patterns.
1395
1396 10. Get rid of a number of -Wunused-but-set-variable warnings.
1397
1398 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1399 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1400 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1401 also returns the mark "x". This bug applied to capturing parentheses,
1402 non-capturing parentheses, and atomic parentheses. It also applied to some
1403 assertions.
1404
1405 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1406 information out of configure.ac instead of relying on pcre.h.generic, which
1407 is not stored in the repository.
1408
1409 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1410 -lreadline.
1411
1412 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1413
1414 15. Applied Graycode's patch to put the top-level frame on the stack rather
1415 than the heap when not using the stack for recursion. This gives a
1416 performance improvement in many cases when recursion is not deep.
1417
1418 16. Experimental code added to "pcretest -C" to output the stack frame size.
1419
1420
1421 Version 8.21 12-Dec-2011
1422 ------------------------
1423
1424 1. Updating the JIT compiler.
1425
1426 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1427 are added as well.
1428
1429 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1430 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1431 calling _pcre_jit_exec. Some extra comments are added.
1432
1433 4. (*MARK) settings inside atomic groups that do not contain any capturing
1434 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1435 was introduced by change 18 for 8.20.
1436
1437 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1438 ECMA-262 standard.
1439
1440 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1441 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1442 This bug was probably introduced by change 9 of 8.13.
1443
1444 7. While fixing 6 above, I noticed that a number of other items were being
1445 incorrectly rejected as "not fixed length". This arose partly because newer
1446 opcodes had not been added to the fixed-length checking code. I have (a)
1447 corrected the bug and added tests for these items, and (b) arranged for an
1448 error to occur if an unknown opcode is encountered while checking for fixed
1449 length instead of just assuming "not fixed length". The items that were
1450 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1451 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1452 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1453
1454 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1455 being incorrectly compiled and would have given unpredicatble results.
1456
1457 9. A possessively repeated subpattern with minimum repeat count greater than
1458 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1459 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1460 the first (A) could occur when it should not.
1461
1462 10. Add a cast and remove a redundant test from the code.
1463
1464 11. JIT should use pcre_malloc/pcre_free for allocation.
1465
1466 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1467 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1468 is anything other than /usr, -L is still shown).
1469
1470 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1471
1472 14. Perl does not support \N without a following name in a [] class; PCRE now
1473 also gives an error.
1474
1475 15. If a forward reference was repeated with an upper limit of around 2000,
1476 it caused the error "internal error: overran compiling workspace". The
1477 maximum number of forward references (including repeats) was limited by the
1478 internal workspace, and dependent on the LINK_SIZE. The code has been
1479 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1480 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1481 of around 200,000 forward references. While doing this, I also speeded up
1482 the filling in of repeated forward references.
1483
1484 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1485 incorrectly expecting the subject to contain another "a" after the start.
1486
1487 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1488 in the match, the SKIP should be ignored. This was not happening; instead
1489 the SKIP was being treated as NOMATCH. For patterns such as
1490 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1491 tested.
1492
1493 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1494 now much more compatible with Perl, in particular in cases where the result
1495 is a non-match for a non-anchored pattern. For example, if
1496 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1497 "m", where previously it did not return a name. A side effect of this
1498 change is that for partial matches, the last encountered mark name is
1499 returned, as for non matches. A number of tests that were previously not
1500 Perl-compatible have been moved into the Perl-compatible test files. The
1501 refactoring has had the pleasing side effect of removing one argument from
1502 the match() function, thus reducing its stack requirements.
1503
1504 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1505 subsequent uses of /S (without +) incorrectly behaved like /S+.
1506
1507 21. Retrieve executable code size support for the JIT compiler and fixing
1508 some warnings.
1509
1510 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1511 not work when the shorter character appeared right at the end of the
1512 subject string.
1513
1514 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1515 systems.
1516
1517 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1518 output it when the /M option is used in pcretest.
1519
1520 25. The CheckMan script was not being included in the distribution. Also, added
1521 an explicit "perl" to run Perl scripts from the PrepareRelease script
1522 because this is reportedly needed in Windows.
1523
1524 26. If study data was being save in a file and studying had not found a set of
1525 "starts with" bytes for the pattern, the data written to the file (though
1526 never used) was taken from uninitialized memory and so caused valgrind to
1527 complain.
1528
1529 27. Updated RunTest.bat as provided by Sheri Pierce.
1530
1531 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1532
1533 29. Computation of memory usage for the table of capturing group names was
1534 giving an unnecessarily large value.
1535
1536
1537 Version 8.20 21-Oct-2011
1538 ------------------------
1539
1540 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1541 a POSIX class. After further experiments with Perl, which convinced me that
1542 Perl has bugs and confusions, a closing square bracket is no longer allowed
1543 in a POSIX name. This bug also affected patterns with classes that started
1544 with full stops.
1545
1546 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1547 captured substring, but while checking the failing first alternative,
1548 substring 1 is temporarily captured. If the output vector supplied to
1549 pcre_exec() was not big enough for this capture, the yield of the function
1550 was still zero ("insufficient space for captured substrings"). This cannot
1551 be totally fixed without adding another stack variable, which seems a lot
1552 of expense for a edge case. However, I have improved the situation in cases
1553 such as /(a)(b)x|abc/ matched against "abc", where the return code
1554 indicates that fewer than the maximum number of slots in the ovector have
1555 been set.
1556
1557 3. Related to (2) above: when there are more back references in a pattern than
1558 slots in the output vector, pcre_exec() uses temporary memory during
1559 matching, and copies in the captures as far as possible afterwards. It was
1560 using the entire output vector, but this conflicts with the specification
1561 that only 2/3 is used for passing back captured substrings. Now it uses
1562 only the first 2/3, for compatibility. This is, of course, another edge
1563 case.
1564
1565 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1566 main code base, and can be used by building with --enable-jit. When this is
1567 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1568 runtime --no-jit option is given.
1569
1570 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1571 ovector, the return from the function was zero, implying that there were
1572 other matches that did not fit. The correct "exactly full" value is now
1573 returned.
1574
1575 6. If a subpattern that was called recursively or as a subroutine contained
1576 (*PRUNE) or any other control that caused it to give a non-standard return,
1577 invalid errors such as "Error -26 (nested recursion at the same subject
1578 position)" or even infinite loops could occur.
1579
1580 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1581 computing the minimum length on reaching *ACCEPT, and so ended up with the
1582 wrong value of 1 rather than 0. Further investigation indicates that
1583 computing a minimum subject length in the presence of *ACCEPT is difficult
1584 (think back references, subroutine calls), and so I have changed the code
1585 so that no minimum is registered for a pattern that contains *ACCEPT.
1586
1587 8. If (*THEN) was present in the first (true) branch of a conditional group,
1588 it was not handled as intended. [But see 16 below.]
1589
1590 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1591 Sheri Pierce.
1592
1593 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1594 the first byte in a match must be "a".
1595
1596 11. Change 17 for 8.13 increased the recursion depth for patterns like
1597 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1598 pattern contains any instances of (*THEN). If it does not, the old
1599 optimizations are restored. It would be nice to do this on a per-group
1600 basis, but at the moment that is not feasible.
1601
1602 12. In some environments, the output of pcretest -C is CRLF terminated. This
1603 broke RunTest's code that checks for the link size. A single white space
1604 character after the value is now allowed for.
1605
1606 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1607 For "fr", it uses the Windows-specific input and output files.
1608
1609 14. If (*THEN) appeared in a group that was called recursively or as a
1610 subroutine, it did not work as intended. [But see next item.]
1611
1612 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1613 pattern fragments (but not containing any | characters). If A and B are
1614 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1615 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1616 D. In other words, Perl considers parentheses that do not contain any |
1617 characters to be part of a surrounding alternative, whereas PCRE was
1618 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1619 differently. PCRE now behaves in the same way as Perl, except in the case
1620 of subroutine/recursion calls such as (?1) which have in any case always
1621 been different (but PCRE had them first :-).
1622
1623 16. Related to 15 above: Perl does not treat the | in a conditional group as
1624 creating alternatives. Such a group is treated in the same way as an
1625 ordinary group without any | characters when processing (*THEN). PCRE has
1626 been changed to match Perl's behaviour.
1627
1628 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1629 RunGrepTest script failed.
1630
1631 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1632 inevitable for groups that contain captures, but it can lead to a lot of
1633 stack use in large patterns. The old behaviour has been restored for atomic
1634 groups that do not contain any capturing parentheses.
1635
1636 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1637 suppress the check for a minimum subject length at run time. (If it was
1638 given to pcre_exec() or pcre_dfa_exec() it did work.)
1639
1640 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1641 fail to work when decoding hex characters in data strings in EBCDIC
1642 environments.
1643
1644 21. It appears that in at least one Mac OS environment, the isxdigit() function
1645 is implemented as a macro that evaluates to its argument more than once,
1646 contravening the C 90 Standard (I haven't checked a later standard). There
1647 was an instance in pcretest which caused it to go wrong when processing
1648 \x{...} escapes in subject strings. The has been rewritten to avoid using
1649 things like p++ in the argument of isxdigit().
1650
1651
1652 Version 8.13 16-Aug-2011
1653 ------------------------
1654
1655 1. The Unicode data tables have been updated to Unicode 6.0.0.
1656
1657 2. Two minor typos in pcre_internal.h have been fixed.
1658
1659 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1660 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1661 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1662
1663 4. There were a number of related bugs in the code for matching backrefences
1664 caselessly in UTF-8 mode when codes for the characters concerned were
1665 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1666 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1667 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1668 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1669 2-byte code at the end of the subject (it thought there wasn't enough data
1670 left).
1671
1672 5. Comprehensive information about what went wrong is now returned by
1673 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1674 as the output vector has at least 2 elements. The offset of the start of
1675 the failing character and a reason code are placed in the vector.
1676
1677 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1678 now returned is for the first byte of the failing character, instead of the
1679 last byte inspected. This is an incompatible change, but I hope it is small
1680 enough not to be a problem. It makes the returned offset consistent with
1681 pcre_exec() and pcre_dfa_exec().
1682
1683 7. pcretest now gives a text phrase as well as the error number when
1684 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1685 failure, the offset and reason code are output.
1686
1687 8. When \R was used with a maximizing quantifier it failed to skip backwards
1688 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1689 back over a single character (\n). This seems wrong (because it treated the
1690 two characters as a single entity when going forwards), conflicts with the
1691 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1692 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1693 has been changed.
1694
1695 9. Some internal refactoring has changed the processing so that the handling
1696 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1697 time (the PCRE_DOTALL option was changed this way some time ago: version
1698 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1699 which was always a bit of a fudge. It also means that there is one less
1700 argument for the match() function, which reduces its stack requirements
1701 slightly. This change also fixes an incompatibility with Perl: the pattern
1702 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1703
1704 10. More internal refactoring has drastically reduced the number of recursive
1705 calls to match() for possessively repeated groups such as (abc)++ when
1706 using pcre_exec().
1707
1708 11. While implementing 10, a number of bugs in the handling of groups were
1709 discovered and fixed:
1710
1711 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1712 (a|)*(?1) gave a compile-time internal error.
1713 ((a|)+)+ did not notice that the outer group could match an empty string.
1714 (^a|^)+ was not marked as anchored.
1715 (.*a|.*)+ was not marked as matching at start or after a newline.
1716
1717 12. Yet more internal refactoring has removed another argument from the match()
1718 function. Special calls to this function are now indicated by setting a
1719 value in a variable in the "match data" data block.
1720
1721 13. Be more explicit in pcre_study() instead of relying on "default" for
1722 opcodes that mean there is no starting character; this means that when new
1723 ones are added and accidentally left out of pcre_study(), testing should
1724 pick them up.
1725
1726 14. The -s option of pcretest has been documented for ages as being an old
1727 synonym of -m (show memory usage). I have changed it to mean "force study
1728 for every regex", that is, assume /S for every regex. This is similar to -i
1729 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1730 using it. It makes it easier to run collections of tests with and without
1731 study enabled, and thereby test pcre_study() more easily. All the standard
1732 tests are now run with and without -s (but some patterns can be marked as
1733 "never study" - see 20 below).
1734
1735 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1736 restoration of the capturing data to the outer values was not happening
1737 correctly.
1738
1739 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1740 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1741 pattern had matched an empty string, and so incorrectly returned a no
1742 match.
1743
1744 17. There was optimizing code for the last branch of non-capturing parentheses,
1745 and also for the obeyed branch of a conditional subexpression, which used
1746 tail recursion to cut down on stack usage. Unfortunately, now that there is
1747 the possibility of (*THEN) occurring in these branches, tail recursion is
1748 no longer possible because the return has to be checked for (*THEN). These
1749 two optimizations have therefore been removed. [But see 8.20/11 above.]
1750
1751 18. If a pattern containing \R was studied, it was assumed that \R always
1752 matched two bytes, thus causing the minimum subject length to be
1753 incorrectly computed because \R can also match just one byte.
1754
1755 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1756 was incorrectly computed.
1757
1758 20. If /S is present twice on a test pattern in pcretest input, it now
1759 *disables* studying, thereby overriding the use of -s on the command line
1760 (see 14 above). This is necessary for one or two tests to keep the output
1761 identical in both cases.
1762
1763 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1764 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1765
1766 22. When an atomic group that contained a capturing parenthesis was
1767 successfully matched, but the branch in which it appeared failed, the
1768 capturing was not being forgotten if a higher numbered group was later
1769 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1770 group 1 to "a", when in fact it should be unset. This applied to multi-
1771 branched capturing and non-capturing groups, repeated or not, and also to
1772 positive assertions (capturing in negative assertions does not happen
1773 in PCRE) and also to nested atomic groups.
1774
1775 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1776 subject after a captured substring, to make it easier to tell which of a
1777 number of identical substrings has been captured.
1778
1779 24. The way atomic groups are processed by pcre_exec() has been changed so that
1780 if they are repeated, backtracking one repetition now resets captured
1781 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1782 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1783 "aaa". Previously, it would have been "a". As part of this code
1784 refactoring, the way recursive calls are handled has also been changed.
1785
1786 25. If an assertion condition captured any substrings, they were not passed
1787 back unless some other capturing happened later. For example, if
1788 (?(?=(a))a) was matched against "a", no capturing was returned.
1789
1790 26. When studying a pattern that contained subroutine calls or assertions,
1791 the code for finding the minimum length of a possible match was handling
1792 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1793 group 1 called group 2 while simultaneously a separate group 2 called group
1794 1). A stack overflow occurred in this case. I have fixed this by limiting
1795 the recursion depth to 10.
1796
1797 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1798 Fortmann. This supports explicit test numbers on the command line, and has
1799 argument validation and error reporting.
1800
1801 28. An instance of \X with an unlimited repeat could fail if at any point the
1802 first character it looked at was a mark character.
1803
1804 29. Some minor code refactoring concerning Unicode properties and scripts
1805 should reduce the stack requirement of match() slightly.
1806
1807 30. Added the '=' option to pcretest to check the setting of unused capturing
1808 slots at the end of the pattern, which are documented as being -1, but are
1809 not included in the return count.
1810
1811 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1812 compiled something random. Now it gives a compile-time error (as does
1813 Perl).
1814
1815 32. A *MARK encountered during the processing of a positive assertion is now
1816 recorded and passed back (compatible with Perl).
1817
1818 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1819 had alternative anchored branches, the search for a second match in a line
1820 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1821 matched the line "0102" twice. The same bug affected patterns that started
1822 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1823 twice.
1824
1825 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1826 does, and because of capturing effects, quantifying parenthesized
1827 assertions may at times be useful. Quantifiers are now allowed for
1828 parenthesized assertions.
1829
1830 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1831
1832 36. \g was being checked for fancy things in a character class, when it should
1833 just be a literal "g".
1834
1835 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1836 appearance of a nested POSIX class supersedes an apparent external class.
1837 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1838 unescaped square brackets may also appear as part of class names. For
1839 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1840 more like Perl. (But see 8.20/1 above.)
1841
1842 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1843 was because it thought it was \N{name}, which is not supported).
1844
1845 39. Add minix to OS list not supporting the -S option in pcretest.
1846
1847 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1848 cannot analyze patterns in sufficient detail to catch mutual recursions
1849 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1850 subgroup is called recursively as a subpattern for a second time at the
1851 same position in the subject string. In previous releases this might have
1852 been caught by the recursion limit, or it might have run out of stack.
1853
1854 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1855 happen only once. PCRE was, however incorrectly giving a compile time error
1856 "recursive call could loop indefinitely" because it cannot analyze the
1857 pattern in sufficient detail. The compile time test no longer happens when
1858 PCRE is compiling a conditional subpattern, but actual runaway loops are
1859 now caught at runtime (see 40 above).
1860
1861 42. It seems that Perl allows any characters other than a closing parenthesis
1862 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1863 has been changed to be the same.
1864
1865 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1866 as not to get warnings when autogen.sh is called. Also changed
1867 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1868
1869 44. To help people who use pcregrep to scan files containing exceedingly long
1870 lines, the following changes have been made:
1871
1872 (a) The default value of the buffer size parameter has been increased from
1873 8K to 20K. (The actual buffer used is three times this size.)
1874
1875 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1876 PCRE is built.
1877
1878 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1879 to be set at run time.
1880
1881 (d) Numerical values in pcregrep options can be followed by K or M, for
1882 example --buffer-size=50K.
1883
1884 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1885 given and the return code is set to 2.
1886
1887 45. Add a pointer to the latest mark to the callout data block.
1888
1889 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1890 partial match of an empty string instead of no match. This was specific to
1891 the use of ".".
1892
1893 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1894 complete match instead of a partial match. This bug was dependent on both
1895 the PCRE_UTF8 and PCRE_DOTALL options being set.
1896
1897 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1898 starting byte set, because \b was not being ignored.
1899
1900
1901 Version 8.12 15-Jan-2011
1902 ------------------------
1903
1904 1. Fixed some typos in the markup of the man pages, and wrote a script that
1905 checks for such things as part of the documentation building process.
1906
1907 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1908 --match-limit and --recursion-limit options (added for 8.11). In
1909 particular, this made one of the standard tests fail. (The integer value
1910 went into the wrong half of a long int.)
1911
1912 3. If the --colour option was given to pcregrep with -v (invert match), it
1913 did strange things, either producing crazy output, or crashing. It should,
1914 of course, ignore a request for colour when reporting lines that do not
1915 match.
1916
1917 4. Another pcregrep bug caused similar problems if --colour was specified with
1918 -M (multiline) and the pattern match finished with a line ending.
1919
1920 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1921 matched in multiline mode, the following line was shown as part of the
1922 match. This seems wrong, so I have changed it.
1923
1924 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1925 the check for further matches in the same line (so they could be coloured)
1926 to overrun the end of the current line. If another match was found, it was
1927 incorrectly shown (and then shown again when found in the next line).
1928
1929 7. If pcregrep was compiled under Windows, there was a reference to the
1930 function pcregrep_exit() before it was defined. I am assuming this was
1931 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1932 reported by a user. I've moved the definition above the reference.
1933
1934
1935 Version 8.11 10-Dec-2010
1936 ------------------------
1937
1938 1. (*THEN) was not working properly if there were untried alternatives prior
1939 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1940 backtracked to try for "b" instead of moving to the next alternative branch
1941 at the same level (in this case, to look for "c"). The Perl documentation
1942 is clear that when (*THEN) is backtracked onto, it goes to the "next
1943 alternative in the innermost enclosing group".
1944
1945 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1946 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1947 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1948 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1949 (*THEN).
1950
1951 3. If \s appeared in a character class, it removed the VT character from
1952 the class, even if it had been included by some previous item, for example
1953 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1954 of \s, but is part of the POSIX "space" class.)
1955
1956 4. A partial match never returns an empty string (because you can always
1957 match an empty string at the end of the subject); however the checking for
1958 an empty string was starting at the "start of match" point. This has been
1959 changed to the "earliest inspected character" point, because the returned
1960 data for a partial match starts at this character. This means that, for
1961 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1962 (previously it gave "no match").
1963
1964 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1965 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1966 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1967 has an implication that the given string is incomplete (because a partial
1968 match is preferred over a full match). For this reason, these items now
1969 give a partial match in this situation. [Aside: previously, the one case
1970 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1971 match rather than a full match, which was wrong by the old rules, but is
1972 now correct.]
1973
1974 6. There was a bug in the handling of #-introduced comments, recognized when
1975 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1976 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1977 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1978 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1979 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1980 places in pcre_compile().
1981
1982 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1983 comments when looking ahead for named forward references to subpatterns,
1984 the only newline sequence it recognized was NL. It now handles newlines
1985 according to the set newline convention.
1986
1987 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1988 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1989 cater for a lack of strerror(). These oversights have been fixed.
1990
1991 9. Added --match-limit and --recursion-limit to pcregrep.
1992
1993 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1994
1995 11. When the -o option was used, pcregrep was setting a return code of 1, even
1996 when matches were found, and --line-buffered was not being honoured.
1997
1998 12. Added an optional parentheses number to the -o and --only-matching options
1999 of pcregrep.
2000
2001 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2002 can match an empty string. The code to do it in pcretest and pcredemo
2003 needed fixing:
2004
2005 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2006 only one byte after an empty string match just before CRLF (this case
2007 just got forgotten; "any" and "anycrlf" were OK).
2008
2009 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2010 mode when an empty string match preceded an ASCII character followed by
2011 a non-ASCII character. (The code for advancing by one character rather
2012 than one byte was nonsense.)
2013
2014 (c) The pcredemo.c sample program did not have any code at all to handle
2015 the cases when CRLF is a valid newline sequence.
2016
2017 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2018 as a starting offset was within the subject string. There is now a new
2019 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2020 negative or greater than the length of the string. In order to test this,
2021 pcretest is extended to allow the setting of negative starting offsets.
2022
2023 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2024 starting offset points to the beginning of a UTF-8 character was
2025 unnecessarily clumsy. I tidied it up.
2026
2027 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2028 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2029
2030 17. Nobody had reported that the --include_dir option, which was added in
2031 release 7.7 should have been called --include-dir (hyphen, not underscore)
2032 for compatibility with GNU grep. I have changed it to --include-dir, but
2033 left --include_dir as an undocumented synonym, and the same for
2034 --exclude-dir, though that is not available in GNU grep, at least as of
2035 release 2.5.4.
2036
2037 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2038 characters from a string of bytes) have been redefined so as not to use
2039 loops, in order to improve performance in some environments. At the same
2040 time, I abstracted some of the common code into auxiliary macros to save
2041 repetition (this should not affect the compiled code).
2042
2043 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2044 compile-time error is now given if \c is not followed by an ASCII
2045 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2046 different, and any byte value is allowed.)
2047
2048 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2049 START_OPTIMIZE option, which is now allowed at compile time - but just
2050 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2051 to pcregrep and other applications that have no direct access to PCRE
2052 options. The new /Y option in pcretest sets this option when calling
2053 pcre_compile().
2054
2055 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2056 back references. Groups containing recursive back references were forced to
2057 be atomic by that change, but in the case of named groups, the amount of
2058 memory required was incorrectly computed, leading to "Failed: internal
2059 error: code overflow". This has been fixed.
2060
2061 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2062 pcretest.c, to avoid build problems in some Borland environments.
2063
2064
2065 Version 8.10 25-Jun-2010
2066 ------------------------
2067
2068 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2069 THEN.
2070
2071 2. (*ACCEPT) was not working when inside an atomic group.
2072
2073 3. Inside a character class, \B is treated as a literal by default, but
2074 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2075 causes the error). The code is unchanged, but I tidied the documentation.
2076
2077 4. Inside a character class, PCRE always treated \R and \X as literals,
2078 whereas Perl faults them if its -w option is set. I have changed PCRE so
2079 that it faults them when PCRE_EXTRA is set.
2080
2081 5. Added support for \N, which always matches any character other than
2082 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2083
2084 6. When compiling pcregrep with newer versions of gcc which may have
2085 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2086 declared with attribute warn_unused_result" were given. Just casting the
2087 result to (void) does not stop the warnings; a more elaborate fudge is
2088 needed. I've used a macro to implement this.
2089
2090 7. Minor change to pcretest.c to avoid a compiler warning.
2091
2092 8. Added four artifical Unicode properties to help with an option to make
2093 \s etc use properties (see next item). The new properties are: Xan
2094 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2095
2096 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2097 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2098 this option. Modified pcretest to add /W to test this facility. Added
2099 REG_UCP to make it available via the POSIX interface.
2100
2101 10. Added --line-buffered to pcregrep.
2102
2103 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2104 studied, and the match started with a letter with a code point greater than
2105 127 whose first byte was different to the first byte of the other case of
2106 the letter, the other case of this starting letter was not recognized
2107 (#976).
2108
2109 12. If a pattern that was studied started with a repeated Unicode property
2110 test, for example, \p{Nd}+, there was the theoretical possibility of
2111 setting up an incorrect bitmap of starting bytes, but fortunately it could
2112 not have actually happened in practice until change 8 above was made (it
2113 added property types that matched character-matching opcodes).
2114
2115 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2116 possible starting bytes for non-anchored patterns.
2117
2118 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2119 \R, and also a number of cases that involve Unicode properties, both
2120 explicit and implicit when PCRE_UCP is set.
2121
2122 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2123 input, it could crash or give wrong results if characters with values
2124 greater than 0xc0 were present in the subject string. (Detail: it assumed
2125 UTF-8 input when processing these items.)
2126
2127 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2128 size_t is 64-bit (#991).
2129
2130 17. Added a check for running out of memory when PCRE is compiled with
2131 --disable-stack-for-recursion (#990).
2132
2133 18. If the last data line in a file for pcretest does not have a newline on
2134 the end, a newline was missing in the output.
2135
2136 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2137 less than 128) in its various bitmaps. However, there is a facility for
2138 generating tables according to the current locale when PCRE is compiled. It
2139 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2140 characters, are recognized by isspace() and therefore were getting set in
2141 these tables, and indeed these tables seem to approximate to ISO 8859. This
2142 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2143 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2144 which of course cannot start UTF-8 characters. I have changed the code so
2145 that only real ASCII characters (less than 128) and the correct starting
2146 bytes for UTF-8 encodings are set for characters greater than 127 when in
2147 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2148 altogether.)
2149
2150 20. Added the /T option to pcretest so as to be able to run tests with non-
2151 standard character tables, thus making it possible to include the tests
2152 used for 19 above in the standard set of tests.
2153
2154 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2155 reference to a subpattern the other side of a comment that contains an
2156 opening parenthesis caused either an internal compiling error, or a
2157 reference to the wrong subpattern.
2158
2159
2160 Version 8.02 19-Mar-2010
2161 ------------------------
2162
2163 1. The Unicode data tables have been updated to Unicode 5.2.0.
2164
2165 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2166 configured.
2167
2168 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2169 original author of that file, following a query about its status.
2170
2171 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2172 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2173
2174 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2175 quantifier applied to a forward-referencing subroutine call, could compile
2176 incorrect code or give the error "internal error: previously-checked
2177 referenced subpattern not found".
2178
2179 6. Both MS Visual Studio and Symbian OS have problems with initializing
2180 variables to point to external functions. For these systems, therefore,
2181 pcre_malloc etc. are now initialized to local functions that call the
2182 relevant global functions.
2183
2184 7. There were two entries missing in the vectors called coptable and poptable
2185 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2186 I've fixed the data, and added a kludgy way of testing at compile time that
2187 the lengths are correct (equal to the number of opcodes).
2188
2189 8. Following on from 7, I added a similar kludge to check the length of the
2190 eint vector in pcreposix.c.
2191
2192 9. Error texts for pcre_compile() are held as one long string to avoid too
2193 much relocation at load time. To find a text, the string is searched,
2194 counting zeros. There was no check for running off the end of the string,
2195 which could happen if a new error number was added without updating the
2196 string.
2197
2198 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2199
2200 11. \K was not working if it appeared in an atomic group or in a group that
2201 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2202 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2203 the assertion is positive, but not if it is negative.
2204
2205 12. Change 11 fortuitously reduced the size of the stack frame used in the
2206 "match()" function of pcre_exec.c by one pointer. Forthcoming
2207 implementation of support for (*MARK) will need an extra pointer on the
2208 stack; I have reserved it now, so that the stack frame size does not
2209 decrease.
2210
2211 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2212 item in branch that calls a recursion is a subroutine call - as in the
2213 second branch in the above example - was incorrectly given the compile-
2214 time error "recursive call could loop indefinitely" because pcre_compile()
2215 was not correctly checking the subroutine for matching a non-empty string.
2216
2217 14. The checks for overrunning compiling workspace could trigger after an
2218 overrun had occurred. This is a "should never occur" error, but it can be
2219 triggered by pathological patterns such as hundreds of nested parentheses.
2220 The checks now trigger 100 bytes before the end of the workspace.
2221
2222 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2223
2224
2225 Version 8.01 19-Jan-2010
2226 ------------------------
2227
2228 1. If a pattern contained a conditional subpattern with only one branch (in
2229 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2230 computed the wrong minimum data length (which is of course zero for such
2231 subpatterns). This could cause incorrect "no match" results.
2232
2233 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2234 the pattern is reset in the first branch, pcre_compile() failed with
2235 "internal error: code overflow at offset...". This happened only when
2236 the reset was to the original external option setting. (An optimization
2237 abstracts leading options settings into an external setting, which was the
2238 cause of this.)
2239
2240 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2241 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2242 assertion pattern did not match (meaning that the assertion was true), it
2243 was incorrectly treated as false if the SKIP had been reached during the
2244 matching. This also applied to assertions used as conditions.
2245
2246 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2247 assertion subpattern, including such a pattern used as a condition,
2248 unpredictable results occurred, instead of the error return
2249 PCRE_ERROR_DFA_UITEM.
2250
2251 5. The C++ GlobalReplace function was not working like Perl for the special
2252 situation when an empty string is matched. It now does the fancy magic
2253 stuff that is necessary.
2254
2255 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2256 removed. (These were left over from very, very early versions of PCRE.)
2257
2258 7. Some cosmetic changes to the code to make life easier when compiling it
2259 as part of something else:
2260
2261 (a) Change DEBUG to PCRE_DEBUG.
2262
2263 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2264 called "current" as "current_branch", to prevent a collision with the
2265 Linux macro when compiled as a kernel module.
2266
2267 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2268 prevent a collision with the Linux macro when compiled as a kernel
2269 module.
2270
2271 8. In pcre_compile() there are some checks for integer overflows that used to
2272 cast potentially large values to (double). This has been changed to that
2273 when building, a check for int64_t is made, and if it is found, it is used
2274 instead, thus avoiding the use of floating point arithmetic. (There is no
2275 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2276 double.
2277
2278 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2279 2005 (difference between two addresses compared to an unsigned value).
2280
2281 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2282 custom one, because of the following reported problem in Windows:
2283
2284 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2285 under Win32.
2286 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2287 therefore missing the function definition.
2288 - The compiler thus generates a "C" signature for the test function.
2289 - The linker fails to find the "C" function.
2290 - PCRE fails to configure if asked to do so against libbz2.
2291
2292 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2293 messages were output:
2294
2295 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2296 rerunning libtoolize, to keep the correct libtool macros in-tree.
2297 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2298
2299 I have done both of these things.
2300
2301 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2302 most of the time, it *can* run out if it is given a pattern that contains a
2303 runaway infinite recursion. I updated the discussion in the pcrestack man
2304 page.
2305
2306 13. Now that we have gone to the x.xx style of version numbers, the minor
2307 version may start with zero. Using 08 or 09 is a bad idea because users
2308 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2309 interpreted as invalid octal numbers. I've updated the previous comment in
2310 configure.ac, and also added a check that gives an error if 08 or 09 are
2311 used.
2312
2313 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2314 causing partial matching to fail when the end of the subject matched \W
2315 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2316
2317 15. There were some discrepancies between the declarations in pcre_internal.h
2318 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2319 their definitions. The declarations used "const uschar *" and the
2320 definitions used USPTR. Even though USPTR is normally defined as "const
2321 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2322 reported that: "This difference in casting confuses some C++ compilers, for
2323 example, SunCC recognizes above declarations as different functions and
2324 generates broken code for hbpcre." I have changed the declarations to use
2325 USPTR.
2326
2327 16. GNU libtool is named differently on some systems. The autogen.sh script now
2328 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2329 (FreeBSD).
2330
2331 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2332 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2333 comment: "Figure out how to create a longlong from a string: strtoll and
2334 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2335 instance, but it only takes 2 args instead of 3!"
2336
2337 18. A subtle bug concerned with back references has been fixed by a change of
2338 specification, with a corresponding code fix. A pattern such as
2339 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2340 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2341 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2342 same bug. Such groups have to be quantified to be useful, or contained
2343 inside another quantified group. (If there's no repetition, the reference
2344 can never match.) The problem arises because, having left the group and
2345 moved on to the rest of the pattern, a later failure that backtracks into
2346 the group uses the captured value from the final iteration of the group
2347 rather than the correct earlier one. I have fixed this in PCRE by forcing
2348 any group that contains a reference to itself to be an atomic group; that
2349 is, there cannot be any backtracking into it once it has completed. This is
2350 similar to recursive and subroutine calls.
2351
2352
2353 Version 8.00 19-Oct-09
2354 ----------------------
2355
2356 1. The table for translating pcre_compile() error codes into POSIX error codes
2357 was out-of-date, and there was no check on the pcre_compile() error code
2358 being within the table. This could lead to an OK return being given in
2359 error.
2360
2361 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2362 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2363 in a Windows environment.
2364
2365 3. The pcregrep --count option prints the count for each file even when it is
2366 zero, as does GNU grep. However, pcregrep was also printing all files when
2367 --files-with-matches was added. Now, when both options are given, it prints
2368 counts only for those files that have at least one match. (GNU grep just
2369 prints the file name in this circumstance, but including the count seems
2370 more useful - otherwise, why use --count?) Also ensured that the
2371 combination -clh just lists non-zero counts, with no names.
2372
2373 4. The long form of the pcregrep -F option was incorrectly implemented as
2374 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2375 but it seems right to fix it, and I didn't think it was worth preserving
2376 the old behaviour.
2377
2378 5. The command line items --regex=pattern and --regexp=pattern were not
2379 recognized by pcregrep, which required --regex pattern or --regexp pattern
2380 (with a space rather than an '='). The man page documented the '=' forms,
2381 which are compatible with GNU grep; these now work.
2382
2383 6. No libpcreposix.pc file was created for pkg-config; there was just
2384 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2385
2386 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2387 when UCP support is not needed, by modifying the Python script that
2388 generates it from Unicode data files. This should not matter if the module
2389 is correctly used as a library, but I received one complaint about 50K of
2390 unwanted data. My guess is that the person linked everything into his
2391 program rather than using a library. Anyway, it does no harm.
2392
2393 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2394 was a minimum greater than 1 for a wide character in a possessive
2395 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2396 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2397 character. Chaos in the form of incorrect output or a compiling loop could
2398 result.
2399
2400 9. The restrictions on what a pattern can contain when partial matching is
2401 requested for pcre_exec() have been removed. All patterns can now be
2402 partially matched by this function. In addition, if there are at least two
2403 slots in the offset vector, the offset of the earliest inspected character
2404 for the match and the offset of the end of the subject are set in them when
2405 PCRE_ERROR_PARTIAL is returned.
2406
2407 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2408 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2409 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2410 and may be more useful for multi-segment matching.
2411
2412 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2413 used to be given if ever the end of the subject was reached; now it is
2414 given only if matching could not proceed because another character was
2415 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2416 string "Z", which now yields "no match" instead of "partial match". In the
2417 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2418 final character ended with (*FAIL).
2419
2420 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2421 if the pattern had a "must contain" character that was already found in the
2422 earlier partial match, unless partial matching was again requested. For
2423 example, with the pattern /dog.(body)?/, the "must contain" character is
2424 "g". If the first part-match was for the string "dog", restarting with
2425 "sbody" failed. This bug has been fixed.
2426
2427 13. The string returned by pcre_dfa_exec() after a partial match has been
2428 changed so that it starts at the first inspected character rather than the
2429 first character of the match. This makes a difference only if the pattern
2430 starts with a lookbehind assertion or \b or \B (\K is not supported by
2431 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2432 matching functions compatible, and I think it's the right thing to do.
2433
2434 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2435 so that the demonstration program is easily available in environments where
2436 PCRE has not been installed from source.
2437
2438 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2439 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2440 library.
2441
2442 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2443 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2444 is not the first non-POSIX option to be added. Clearly some people find
2445 these options useful.
2446
2447 17. If a caller to the POSIX matching function regexec() passes a non-zero
2448 value for nmatch with a NULL value for pmatch, the value of
2449 nmatch is forced to zero.
2450
2451 18. RunGrepTest did not have a test for the availability of the -u option of
2452 the diff command, as RunTest does. It now checks in the same way as
2453 RunTest, and also checks for the -b option.
2454
2455 19. If an odd number of negated classes containing just a single character
2456 interposed, within parentheses, between a forward reference to a named
2457 subpattern and the definition of the subpattern, compilation crashed with
2458 an internal error, complaining that it could not find the referenced
2459 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2460 [The bug was that it was starting one character too far in when skipping
2461 over the character class, thus treating the ] as data rather than
2462 terminating the class. This meant it could skip too much.]
2463
2464 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2465 /g option in pcretest when the pattern contains \K, which makes it possible
2466 to have an empty string match not at the start, even when the pattern is
2467 anchored. Updated pcretest and pcredemo to use this option.
2468
2469 21. If the maximum number of capturing subpatterns in a recursion was greater
2470 than the maximum at the outer level, the higher number was returned, but
2471 with unset values at the outer level. The correct (outer level) value is
2472 now given.
2473
2474 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2475 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2476 make it do so. The string so far is captured, making this feature
2477 compatible with Perl.
2478
2479 23. The tests have been re-organized, adding tests 11 and 12, to make it
2480 possible to check the Perl 5.10 features against Perl 5.10.
2481
2482 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2483 pattern matches a fixed length string. PCRE did not allow this; now it
2484 does. Neither allows recursion.
2485
2486 25. I finally figured out how to implement a request to provide the minimum
2487 length of subject string that was needed in order to match a given pattern.
2488 (It was back references and recursion that I had previously got hung up
2489 on.) This code has now been added to pcre_study(); it finds a lower bound
2490 to the length of subject needed. It is not necessarily the greatest lower
2491 bound, but using it to avoid searching strings that are too short does give
2492 some useful speed-ups. The value is available to calling programs via
2493 pcre_fullinfo().
2494
2495 26. While implementing 25, I discovered to my embarrassment that pcretest had
2496 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2497 study optimizations had never been tested with that matching function.
2498 Oops. What is worse, even when it was passed study data, there was a bug in
2499 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2500 were also very few tests of studied patterns with pcre_dfa_exec().
2501
2502 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2503 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2504 on the other side of the coin, they are no longer allowed to have different
2505 names, because these cannot be distinguished in PCRE, and this has caused
2506 confusion. (This is a difference from Perl.)
2507
2508 28. When duplicate subpattern names are present (necessarily with different
2509 numbers, as required by 27 above), and a test is made by name in a
2510 conditional pattern, either for a subpattern having been matched, or for
2511 recursion in such a pattern, all the associated numbered subpatterns are
2512 tested, and the overall condition is true if the condition is true for any
2513 one of them. This is the way Perl works, and is also more like the way
2514 testing by number works.
2515
2516
2517 Version 7.9 11-Apr-09
2518 ---------------------
2519
2520 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2521 (pcretest), all targets were linked against these libraries. This included
2522 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2523 libraries. This caused unwanted dependencies to be created. This problem
2524 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2525 pcretest is linked with readline.
2526
2527 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2528 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2529 moved outside it again, because FALSE and TRUE are already defined in AIX,
2530 but BOOL is not.
2531
2532 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2533 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2534
2535 4. The pcregrep documentation said spaces were inserted as well as colons (or
2536 hyphens) following file names and line numbers when outputting matching
2537 lines. This is not true; no spaces are inserted. I have also clarified the
2538 wording for the --colour (or --color) option.
2539
2540 5. In pcregrep, when --colour was used with -o, the list of matching strings
2541 was not coloured; this is different to GNU grep, so I have changed it to be
2542 the same.
2543
2544 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2545 each matching line was coloured. Now it goes on to look for further matches
2546 of any of the test patterns, which is the same behaviour as GNU grep.
2547
2548 7. A pattern that could match an empty string could cause pcregrep to loop; it
2549 doesn't make sense to accept an empty string match in pcregrep, so I have
2550 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2551 seems to be how GNU grep behaves. [But see later change 40 for release
2552 8.33.]
2553
2554 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2555 start or after a newline", because the conditional assertion was not being
2556 correctly handled. The rule now is that both the assertion and what follows
2557 in the first alternative must satisfy the test.
2558
2559 9. If auto-callout was enabled in a pattern with a conditional group whose
2560 condition was an assertion, PCRE could crash during matching, both with
2561 pcre_exec() and pcre_dfa_exec().
2562
2563 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2564 used for matching.
2565
2566 11. Unicode property support in character classes was not working for
2567 characters (bytes) greater than 127 when not in UTF-8 mode.
2568
2569 12. Added the -M command line option to pcretest.
2570
2571 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2572
2573 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2574
2575 16. Added comments and documentation about mis-use of no_arg in the C++
2576 wrapper.
2577
2578 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2579 from Martin Jerabek that uses macro names for all relevant character and
2580 string constants.
2581
2582 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2583 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2584 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2585 these, but not everybody uses configure.
2586
2587 19. A conditional group that had only one branch was not being correctly
2588 recognized as an item that could match an empty string. This meant that an
2589 enclosing group might also not be so recognized, causing infinite looping
2590 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2591 with the subject "ab", where knowledge that the repeated group can match
2592 nothing is needed in order to break the loop.
2593
2594 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2595 exec(), but without supplying a callout function, matching went wrong.
2596
2597 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2598 leak if the size of the offset vector was greater than 30. When the vector
2599 is smaller, the saved offsets during recursion go onto a local stack
2600 vector, but for larger vectors malloc() is used. It was failing to free
2601 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2602 error, in fact).
2603
2604 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2605 heapframe that is used only when UTF-8 support is enabled. This caused no
2606 problem, but was untidy.
2607
2608 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2609 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2610 included within another project.
2611
2612 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2613 slightly modified by me:
2614
2615 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2616 not building pcregrep.
2617
2618 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2619 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2620
2621 25. Forward references, both numeric and by name, in patterns that made use of
2622 duplicate group numbers, could behave incorrectly or give incorrect errors,
2623 because when scanning forward to find the reference group, PCRE was not
2624 taking into account the duplicate group numbers. A pattern such as
2625 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2626
2627 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2628 the feature of a custom pointer more persuasive (as requested by a user).
2629
2630 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2631 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2632 pre-defined.
2633
2634 28. Added support for (*UTF8) at the start of a pattern.
2635
2636 29. Arrange for flags added by the "release type" setting in CMake to be shown
2637 in the configuration summary.
2638
2639
2640 Version 7.8 05-Sep-08
2641 ---------------------
2642
2643 1. Replaced UCP searching code with optimized version as implemented for Ad
2644 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2645 stage table and inline lookup instead of a function, giving speed ups of 2
2646 to 5 times on some simple patterns that I tested. Permission was given to
2647 distribute the MultiStage2.py script that generates the tables (it's not in
2648 the tarball, but is in the Subversion repository).
2649
2650 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2651 scripts.
2652
2653 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2654 a group with a zero qualifier. The result of the study could be incorrect,
2655 or the function might crash, depending on the pattern.
2656
2657 4. Caseless matching was not working for non-ASCII characters in back
2658 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2659 It now works when Unicode Property Support is available.
2660
2661 5. In pcretest, an escape such as \x{de} in the data was always generating
2662 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2663 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2664 truncation.
2665
2666 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2667
2668 7. Added two (int) casts to pcregrep when printing the difference of two
2669 pointers, in case they are 64-bit values.
2670
2671 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2672 test 2 if it fails.
2673
2674 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2675 and a #define of that name to empty if it is not externally set. This is to
2676 allow users of MSVC to set it if necessary.
2677
2678 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2679 the convenience functions in the pcre_get.c source file.
2680
2681 11. An option change at the start of a pattern that had top-level alternatives
2682 could cause overwriting and/or a crash. This command provoked a crash in
2683 some environments:
2684
2685 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2686
2687 This potential security problem was recorded as CVE-2008-2371.
2688
2689 12. For a pattern where the match had to start at the beginning or immediately
2690 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2691 pcre_dfa_exec() could read past the end of the passed subject if there was
2692 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2693 pcretest so that it places the subject at the end of its malloc-ed buffer.
2694
2695 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2696 exec() might read past the end of the data buffer in UTF-8 mode.
2697
2698 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2699 the data contained the byte 0x85 as part of a UTF-8 character within its
2700 first line. This applied both to normal and DFA matching.
2701
2702 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2703 /^[^d]*?$/8 failed to match "abc".
2704
2705 16. Added a missing copyright notice to pcrecpp_internal.h.
2706
2707 17. Make it more clear in the documentation that values returned from
2708 pcre_exec() in ovector are byte offsets, not character counts.
2709
2710 18. Tidied a few places to stop certain compilers from issuing warnings.
2711
2712 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2713 supplied by Stefan Weber. I made a further small update for 7.8 because
2714 there is a change of source arrangements: the pcre_searchfuncs.c module is
2715 replaced by pcre_ucd.c.
2716
2717
2718 Version 7.7 07-May-08
2719 ---------------------
2720
2721 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2722 a string to a long long, pretend we don't even have a long long." This is
2723 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2724
2725 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2726 pre-7.6 versions, which defined a global no_arg variable instead of putting
2727 it in the RE class. (See also #8 below.)
2728
2729 3. Remove a line of dead code, identified by coverity and reported by Nuno
2730 Lopes.
2731
2732 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2733
2734 (1) The include/exclude patterns were being applied to the whole pathnames
2735 of files, instead of just to the final components.
2736
2737 (2) If there was more than one level of directory, the subdirectories were
2738 skipped unless they satisfied the include/exclude conditions. This is
2739 inconsistent with GNU grep (and could even be seen as contrary to the
2740 pcregrep specification - which I improved to make it absolutely clear).
2741 The action now is always to scan all levels of directory, and just
2742 apply the include/exclude patterns to regular files.
2743
2744 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2745 --exclude_dir in the tests to avoid scanning .svn directories.
2746
2747 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2748 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2749 doesn't support NULs in patterns.
2750
2751 7. Added some missing "const"s to declarations of static tables in
2752 pcre_compile.c and pcre_dfa_exec.c.
2753
2754 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2755 caused by fix #2 above. (Subsequently also a second patch to fix the
2756 first patch. And a third patch - this was a messy problem.)
2757
2758 9. Applied Craig's patch to remove the use of push_back().
2759
2760 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2761 matching function regexec().
2762
2763 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2764 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2765 references. PCRE supports relative numbers with this syntax (I don't think
2766 Oniguruma does).
2767
2768 12. Previously, a group with a zero repeat such as (...){0} was completely
2769 omitted from the compiled regex. However, this means that if the group
2770 was called as a subroutine from elsewhere in the pattern, things went wrong
2771 (an internal error was given). Such groups are now left in the compiled
2772 pattern, with a new opcode that causes them to be skipped at execution
2773 time.
2774
2775 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2776 to the way PCRE behaves:
2777
2778 (a) A lone ] character is dis-allowed (Perl treats it as data).
2779
2780 (b) A back reference to an unmatched subpattern matches an empty string
2781 (Perl fails the current match path).
2782
2783 (c) A data ] in a character class must be notated as \] because if the
2784 first data character in a class is ], it defines an empty class. (In
2785 Perl it is not possible to have an empty class.) The empty class []
2786 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2787 The negative empty class [^] matches any one character, independently
2788 of the DOTALL setting.
2789
2790 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2791 non-existent subpattern following a character class starting with ']' and
2792 containing () gave an internal compiling error instead of "reference to
2793 non-existent subpattern". Fortunately, when the pattern did exist, the
2794 compiled code was correct. (When scanning forwards to check for the
2795 existence of the subpattern, it was treating the data ']' as terminating
2796 the class, so got the count wrong. When actually compiling, the reference
2797 was subsequently set up correctly.)
2798
2799 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2800 it was being rejected as not supported by pcre_dfa_exec(), even though
2801 other assertions are supported. I have made pcre_dfa_exec() support
2802 (*FAIL).
2803
2804 16. The implementation of 13c above involved the invention of a new opcode,
2805 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2806 cannot be changed at match time, I realized I could make a small
2807 improvement to matching performance by compiling OP_ALLANY instead of
2808 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2809 on the OP_ANY path.
2810
2811 17. Compiling pcretest on Windows with readline support failed without the
2812 following two fixes: (1) Make the unistd.h include conditional on
2813 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2814
2815 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2816 ncurses library to be included for pcretest when ReadLine support is
2817 requested, but also to allow for it to be overridden. This patch came from
2818 Daniel Bergström.
2819
2820 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2821 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2822 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2823 spotting this.
2824
2825
2826 Version 7.6 28-Jan-08
2827 ---------------------
2828
2829 1. A character class containing a very large number of characters with
2830 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2831 overflow.
2832
2833 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2834 HAVE_LONG_LONG is not defined.
2835
2836 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2837 bring them up to date and include new features. This patch includes:
2838
2839 - Fixed PH's badly added libz and libbz2 support.
2840 - Fixed a problem with static linking.
2841 - Added pcredemo. [But later removed - see 7 below.]
2842 - Fixed dftables problem and added an option.
2843 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2844 HAVE_LONG_LONG.
2845 - Added readline support for pcretest.
2846 - Added an listing of the option settings after cmake has run.
2847
2848 4. A user submitted a patch to Makefile that makes it easy to create
2849 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2850 Makefile.am that cause it to include this special target, without
2851 affecting anything else. Note that the same mingw target plus all
2852 the other distribution libraries and programs are now supported
2853 when configuring with CMake (see 6 below) instead of with
2854 Configure/Make.
2855
2856 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2857 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2858 exported in the Windows port". It has not yet been confirmed that the patch
2859 solves the problem, but it does no harm.
2860
2861 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2862 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2863 with CMake, and also correct the comment about stack recursion.
2864
2865 7. Remove the automatic building of pcredemo from the ./configure system and
2866 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2867 of a program that users should build themselves after PCRE is installed, so
2868 building it automatically is not really right. What is more, it gave
2869 trouble in some build environments.
2870
2871 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2872
2873
2874 Version 7.5 10-Jan-08
2875 ---------------------
2876
2877 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2878 values in parens when parsing an RE using the C++ wrapper."
2879
2880 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2881 Characters greater than 255 were excluded from the class instead of being
2882 included.
2883
2884 3. The same bug as (2) above applied to negated POSIX classes such as
2885 [:^space:].
2886
2887 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2888 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2889 I have changed it.
2890
2891 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2892 first named subpattern) and a construct such as (?&a) would reference the
2893 first named subpattern whose name started with "a" (in other words, the
2894 length check was missing). Both these problems are fixed. "Subpattern name
2895 expected" is now given for (?&) (a zero-length name), and this patch also
2896 makes it give the same error for \k'' (previously it complained that that
2897 was a reference to a non-existent subpattern).
2898
2899 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2900 this is right because (?- can be followed by option settings as well as by
2901 digits. I have, however, made the messages clearer.
2902
2903 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2904 than the number used in the conditional) now cause a compile-time error.
2905 This is actually not compatible with Perl, which accepts such patterns, but
2906 treats the conditional as always being FALSE (as PCRE used to), but it
2907 seems to me that giving a diagnostic is better.
2908
2909 8. Change "alphameric" to the more common word "alphanumeric" in comments
2910 and messages.
2911
2912 9. Fix two occurrences of "backslash" in comments that should have been
2913 "backspace".
2914
2915 10. Remove two redundant lines of code that can never be obeyed (their function
2916 was moved elsewhere).
2917
2918 11. The program that makes PCRE's Unicode character property table had a bug
2919 which caused it to generate incorrect table entries for sequences of
2920 characters that have the same character type, but are in different scripts.
2921 It amalgamated them into a single range, with the script of the first of
2922 them. In other words, some characters were in the wrong script. There were
2923 thirteen such cases, affecting characters in the following ranges:
2924
2925 U+002b0 - U+002c1
2926 U+0060c - U+0060d
2927 U+0061e - U+00612
2928 U+0064b - U+0065e
2929 U+0074d - U+0076d
2930 U+01800 - U+01805
2931 U+01d00 - U+01d77
2932 U+01d9b - U+01dbf
2933 U+0200b - U+0200f
2934 U+030fc - U+030fe
2935 U+03260 - U+0327f
2936 U+0fb46 - U+0fbb1
2937 U+10450 - U+1049d
2938
2939 12. The -o option (show only the matching part of a line) for pcregrep was not
2940 compatible with GNU grep in that, if there was more than one match in a
2941 line, it showed only the first of them. It now behaves in the same way as
2942 GNU grep.
2943
2944 13. If the -o and -v options were combined for pcregrep, it printed a blank
2945 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2946 does the same. The return code can be used to tell if there were any
2947 non-matching lines.
2948
2949 14. Added --file-offsets and --line-offsets to pcregrep.
2950
2951 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2952 infinitely looping recursion. The bug was that positive lookaheads were not
2953 being skipped when checking for a possible empty match (negative lookaheads
2954 and both kinds of lookbehind were skipped).
2955
2956 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2957 inclusion of <windows.h> to before rather than after the definition of
2958 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2959
2960 17. Specifying a possessive quantifier with a specific limit for a Unicode
2961 character property caused pcre_compile() to compile bad code, which led at
2962 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2963 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2964 caused the error; without that there was no problem.
2965
2966 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2967
2968 19. Added --enable-pcretest-libreadline.
2969
2970 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2971 RE::GlobalReplace(). As a result, the number of replacements returned was
2972 double what it should be. I removed one of the increments, but Craig sent a
2973 later patch that removed the other one (the right fix) and added unit tests
2974 that check the return values (which was not done before).
2975
2976 21. Several CMake things:
2977
2978 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2979 the names libpcre and libpcreposix, not just pcre and pcreposix.
2980
2981 (2) The above change means that pcretest and pcregrep are now correctly
2982 linked with the newly-built libraries, not previously installed ones.
2983
2984 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2985
2986 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2987 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2988 UTF-8 newline character). The key issue is that the pattern starts .*;
2989 this means that the match must be either at the beginning, or after a
2990 newline. The bug was in the code for advancing after a failed match and
2991 checking that the new position followed a newline. It was not taking
2992 account of UTF-8 characters correctly.
2993
2994 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2995 character classes. PCRE was not treating the sequence [:...:] as a
2996 character class unless the ... were all letters. Perl, however, seems to
2997 allow any characters between [: and :], though of course it rejects as
2998 unknown any "names" that contain non-letters, because all the known class
2999 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3000 for example, whereas PCRE did not - it did not recognize a POSIX character
3001 class. This seemed a bit dangerous, so the code has been changed to be
3002 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3003 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3004 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3005 Perl does, and where it didn't before.
3006
3007 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3008 Windows environments %n is disabled by default.
3009
3010
3011 Version 7.4 21-Sep-07
3012 ---------------------
3013
3014 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3015 means that a class such as [\s] counted as "explicit reference to CR or
3016 LF". That isn't really right - the whole point of the change was to try to
3017 help when there was an actual mention of one of the two characters. So now
3018 the change happens only if \r or \n (or a literal CR or LF) character is
3019 encountered.
3020
3021 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3022 of both had grown to the point where there were only 3 bits left.
3023 Fortunately, there was spare space in the data structure, and so I have
3024 moved the internal flags into a new 16-bit field to free up more option
3025 bits.
3026
3027 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3028 but did not set the internal JCHANGED flag - either of these is enough to
3029 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3030 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3031 start sets both bits.
3032
3033 4. Added options (at build time, compile time, exec time) to change \R from
3034 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3035
3036 5. doc/pcresyntax.html was missing from the distribution.
3037
3038 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3039 compatibility, even though it is no longer used.
3040
3041 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3042 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3043 windows.h file is present (where different names are used). [This was
3044 reversed later after testing - see 16 below.]
3045
3046 8. Changed all #include <config.h> to #include "config.h". There were also
3047 some further <pcre.h> cases that I changed to "pcre.h".
3048
3049 9. When pcregrep was used with the --colour option, it missed the line ending
3050 sequence off the lines that it output.
3051
3052 10. It was pointed out to me that arrays of string pointers cause lots of
3053 relocations when a shared library is dynamically loaded. A technique of
3054 using a single long string with a table of offsets can drastically reduce
3055 these. I have refactored PCRE in four places to do this. The result is
3056 dramatic:
3057
3058 Originally: 290
3059 After changing UCP table: 187
3060 After changing error message table: 43
3061 After changing table of "verbs" 36
3062 After changing table of Posix names 22
3063
3064 Thanks to the folks working on Gregex for glib for this insight.
3065
3066 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3067 unicode-properties was also set.
3068
3069 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3070
3071 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3072 checked only for CRLF.
3073
3074 14. Added casts to pcretest.c to avoid compiler warnings.
3075
3076 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3077
3078 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3079 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3080 entirely. This removes changes made in 7 above.
3081
3082 17. The CMake files have been updated, and there is now more information about
3083 building with CMake in the NON-UNIX-USE document.
3084
3085
3086 Version 7.3 28-Aug-07
3087 ---------------------
3088
3089 1. In the rejigging of the build system that eventually resulted in 7.1, the
3090 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3091 brackets there is not right, since it causes compilers to look for an
3092 installed pcre.h, not the version that is in the source that is being
3093 compiled (which of course may be different). I have changed it back to:
3094
3095 #include "pcre.h"
3096
3097 I have a vague recollection that the change was concerned with compiling in
3098 different directories, but in the new build system, that is taken care of
3099 by the VPATH setting the Makefile.
3100
3101 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3102 when the subject happened to end in the byte 0x85 (e.g. if the last
3103 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3104 characters but of course it shouldn't be taken as a newline when it is part
3105 of another character. The bug was that, for an unlimited repeat of . in
3106 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3107 characters when looking for a newline.
3108
3109 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3110
3111 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3112 in debug output.
3113
3114 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3115 long printing in the pcrecpp unittest when running under MinGW.
3116
3117 6. ESC_K was left out of the EBCDIC table.
3118
3119 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3120 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3121 limit also applies to "virtual nesting" when a pattern is recursive, and in
3122 this case 1000 isn't so big. I have been able to remove this limit at the
3123 expense of backing off one optimization in certain circumstances. Normally,
3124 when pcre_exec() would call its internal match() function recursively and
3125 immediately return the result unconditionally, it uses a "tail recursion"
3126 feature to save stack. However, when a subpattern that can match an empty
3127 string has an unlimited repetition quantifier, it no longer makes this
3128 optimization. That gives it a stack frame in which to save the data for
3129 checking that an empty string has been matched. Previously this was taken
3130 from the 1000-entry workspace that had been reserved. So now there is no
3131 explicit limit, but more stack is used.
3132
3133 8. Applied Daniel's patches to solve problems with the import/export magic
3134 syntax that is required for Windows, and which was going wrong for the
3135 pcreposix and pcrecpp parts of the library. These were overlooked when this
3136 problem was solved for the main library.
3137
3138 9. There were some crude static tests to avoid integer overflow when computing
3139 the size of patterns that contain repeated groups with explicit upper
3140 limits. As the maximum quantifier is 65535, the maximum group length was
3141 set at 30,000 so that the product of these two numbers did not overflow a
3142 32-bit integer. However, it turns out that people want to use groups that
3143 are longer than 30,000 bytes (though not repeat them that many times).
3144 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3145 made it possible to implement the integer overflow checks in a much more
3146 dynamic way, which I have now done. The artificial limitation on group
3147 length has been removed - we now have only the limit on the total length of
3148 the compiled pattern, which depends on the LINK_SIZE setting.
3149
3150 10. Fixed a bug in the documentation for get/copy named substring when
3151 duplicate names are permitted. If none of the named substrings are set, the
3152 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3153 empty string.
3154
3155 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3156 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3157 because the ] is interpreted as the first data character and the
3158 terminating ] is not found. PCRE has been made compatible with Perl in this
3159 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3160 cause memory overwriting.
3161
3162 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3163 string has been matched (to stop an infinite loop). It was not recognizing
3164 a conditional subpattern that could match an empty string if that
3165 subpattern was within another subpattern. For example, it looped when
3166 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3167 condition was not nested. This bug has been fixed.
3168
3169 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3170 past the start of the subject in the presence of bytes with the top bit
3171 set, for example "\x8aBCD".
3172
3173 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3174 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3175
3176 14. Optimized (?!) to (*FAIL).
3177
3178 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3179 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3180 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3181 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3182 does: it's just the validity check that is more restrictive.
3183
3184 16. Inserted checks for integer overflows during escape sequence (backslash)
3185 processing, and also fixed erroneous offset values for syntax errors during
3186 backslash processing.
3187
3188 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3189 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3190
3191 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3192 caused an overrun.
3193
3194 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3195 something other than just ASCII characters) inside a group that had an
3196 unlimited repeat caused a loop at compile time (while checking to see
3197 whether the group could match an empty string).
3198
3199 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3200 [\P{Any}] did so. (Error in the code for printing property names.)
3201
3202 21. An orphan \E inside a character class could cause a crash.
3203
3204 22. A repeated capturing bracket such as (A)? could cause a wild memory
3205 reference during compilation.
3206
3207 23. There are several functions in pcre_compile() that scan along a compiled
3208 expression for various reasons (e.g. to see if it's fixed length for look
3209 behind). There were bugs in these functions when a repeated \p or \P was
3210 present in the pattern. These operators have additional parameters compared
3211 with \d, etc, and these were not being taken into account when moving along
3212 the compiled data. Specifically:
3213
3214 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3215 length.
3216
3217 (b) An item such as \pL+ within a repeated group could cause crashes or
3218 loops.
3219
3220 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3221 "reference to non-existent subpattern" error.
3222
3223 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3224
3225 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3226 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3227
3228 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3229
3230 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3231 character were causing crashes (broken optimization).
3232
3233 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3234 \p or \P) caused a compile-time loop.
3235
3236 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3237 break. For example, the unstudied pattern [\r\n]A does not match the string
3238 "\r\nA" because change 7.0/46 below moves the current point on by two
3239 characters after failing to match at the start. However, the pattern \nA
3240 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3241 the same is true. There doesn't seem any very clean way out of this, but
3242 what I have chosen to do makes the common cases work: PCRE now takes note
3243 of whether there can be an explicit match for \r or \n anywhere in the
3244 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3245 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3246 pattern has explicit CR or LF references.
3247
3248 29. Added (*CR) etc for changing newline setting at start of pattern.
3249
3250
3251 Version 7.2 19-Jun-07
3252 ---------------------
3253
3254 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3255 which is apparently normally available under Windows.
3256
3257 2. Re-jig the pcregrep tests with different newline settings in an attempt
3258 to make them independent of the local environment's newline setting.
3259
3260 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3261
3262 4. Some of the "internals" tests were previously cut out when the link size
3263 was not 2, because the output contained actual offsets. The recent new
3264 "Z" feature of pcretest means that these can be cut out, making the tests
3265 usable with all link sizes.
3266
3267 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3268 stack recursion. This gives a massive performance boost under BSD, but just
3269 a small improvement under Linux. However, it saves one field in the frame
3270 in all cases.
3271
3272 6. Added more features from the forthcoming Perl 5.10:
3273
3274 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3275 recursion call. It refers to the nth most recently opened parentheses.
3276
3277 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3278 to be opened parentheses.
3279
3280 (c) Conditions that refer to capturing parentheses can be specified
3281 relatively, for example, (?(-2)... or (?(+3)...
3282
3283 (d) \K resets the start of the current match so that everything before
3284 is not part of it.
3285
3286 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3287
3288 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3289 reference syntax.
3290
3291 (g) (?| introduces a group in which the numbering of parentheses in each
3292 alternative starts with the same number.
3293
3294 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3295
3296 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3297 PCRE_INFO_JCHANGED.
3298
3299 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3300 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3301 for detecting groups that can match an empty string.
3302
3303 9. A pattern with a very large number of alternatives (more than several
3304 hundred) was running out of internal workspace during the pre-compile
3305 phase, where pcre_compile() figures out how much memory will be needed. A
3306 bit of new cunning has reduced the workspace needed for groups with
3307 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3308 workspace instead of running out of the 4096 that are available.
3309
3310 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3311
3312 11. Applied patch from Google to remove an optimization that didn't quite work.
3313 The report of the bug said:
3314
3315 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3316 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3317 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3318
3319 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3320 it matched the wrong number of bytes.
3321
3322
3323 Version 7.1 24-Apr-07
3324 ---------------------
3325
3326 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3327 that is more "standard", making use of automake and other Autotools. There
3328 is some re-arrangement of the files and adjustment of comments consequent
3329 on this.
3330
3331 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3332 for recursive directory scanning broke on some systems because the files
3333 are not scanned in any specific order and on different systems the order
3334 was different. A call to "sort" has been inserted into RunGrepTest for the
3335 approprate test as a short-term fix. In the longer term there may be an
3336 alternative.
3337
3338 3. I had an email from Eric Raymond about problems translating some of PCRE's
3339 man pages to HTML (despite the fact that I distribute HTML pages, some
3340 people do their own conversions for various reasons). The problems
3341 concerned the use of low-level troff macros .br and .in. I have therefore
3342 removed all such uses from the man pages (some were redundant, some could
3343 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3344 HTML has been updated to handle .nf/.fi and to complain if it encounters
3345 .br or .in.
3346
3347 4. Updated comments in configure.ac that get placed in config.h.in and also
3348 arranged for config.h to be included in the distribution, with the name
3349 config.h.generic, for the benefit of those who have to compile without
3350 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3351
3352 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3353 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3354 makevp.bat for the current PCRE, using the additional files
3355 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3356
3357 6. A Windows user reported a minor discrepancy with test 2, which turned out
3358 to be caused by a trailing space on an input line that had got lost in his
3359 copy. The trailing space was an accident, so I've just removed it.
3360
3361 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3362 that is needed.
3363
3364 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3365 as "const" (a) because they are and (b) because it helps the PHP
3366 maintainers who have recently made a script to detect big data structures
3367 in the php code that should be moved to the .rodata section. I remembered
3368 to update Builducptable as well, so it won't revert if ucptable.h is ever
3369 re-created.
3370
3371 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3372 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3373 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3374 support is not required. This saves 1.5-2K of code, which is important in
3375 some applications.
3376
3377 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3378 so as not to refer to the tables, even though these functions will never be
3379 called when UTF-8 support is disabled. Otherwise there are problems with a
3380 shared library.
3381
3382 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3383
3384 (a) It was defining its arguments as char * instead of void *.
3385
3386 (b) It was assuming that all moves were upwards in memory; this was true
3387 a long time ago when I wrote it, but is no longer the case.
3388
3389 The emulated memove() is provided for those environments that have neither
3390 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3391 is clearly not the case, as these two bugs were recently reported.
3392
3393 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3394 and Detrail to create the HTML documentation, the .txt form of the man
3395 pages, and it removes trailing spaces from listed files. It also creates
3396 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3397 case, it wraps all the #defines with #ifndefs. This script should be run
3398 before "make dist".
3399
3400 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3401 with Unicode property support.
3402
3403 (a) For a maximizing quantifier, if the two different cases of the
3404 character were of different lengths in their UTF-8 codings (there are
3405 some cases like this - I found 11), and the matching function had to
3406 back up over a mixture of the two cases, it incorrectly assumed they
3407 were both the same length.
3408
3409 (b) When PCRE was configured to use the heap rather than the stack for
3410 recursion during matching, it was not correctly preserving the data for
3411 the other case of a UTF-8 character when checking ahead for a match
3412 while processing a minimizing repeat. If the check also involved
3413 matching a wide character, but failed, corruption could cause an
3414 erroneous result when trying to check for a repeat of the original
3415 character.
3416
3417 13. Some tidying changes to the testing mechanism:
3418
3419 (a) The RunTest script now detects the internal link size and whether there
3420 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3421 values substituted by "configure". (The RunGrepTest script already did
3422 this for UTF-8.) The configure.ac script no longer substitutes the
3423 relevant variables.
3424
3425 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3426 with length and offset values. This means that the output is different
3427 for different internal link sizes. Test 2 is skipped for link sizes
3428 other than 2 because of this, bypassing the problem. Unfortunately,
3429 there was also a test in test 3 (the locale tests) that used /B and
3430 failed for link sizes other than 2. Rather than cut the whole test out,
3431 I have added a new /Z option to pcretest that replaces the length and
3432 offset values with spaces. This is now used to make test 3 independent
3433 of link size. (Test 2 will be tidied up later.)
3434
3435 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3436 segmentation fault instead of returning the appropriate error message.
3437
3438 15. In multiline mode when the newline sequence was set to "any", the pattern
3439 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3440 This doesn't seem right; it now treats the CRLF combination as the line
3441 ending, and so does not match in that case. It's only a pattern such as ^$
3442 that would hit this one: something like ^ABC$ would have failed after \r
3443 and then tried again after \r\n.
3444
3445 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3446 in an attempt to make files that differ only in their line terminators
3447 compare equal. This works on Linux.
3448
3449 17. Under certain error circumstances pcregrep might try to free random memory
3450 as it exited. This is now fixed, thanks to valgrind.
3451
3452 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3453 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3454 was because its rules for how to advance for /g after matching an empty
3455 string at the end of a line did not allow for this case. They now check for
3456 it specially.
3457
3458 20. pcretest is supposed to handle patterns and data of any length, by
3459 extending its buffers when necessary. It was getting this wrong when the
3460 buffer for a data line had to be extended.
3461
3462 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3463 CRLF as a newline sequence.
3464
3465 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3466 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3467 I have nevertheless tidied it up.
3468
3469 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3470
3471 24. Added a man page for pcre-config.
3472
3473
3474 Version 7.0 19-Dec-06
3475 ---------------------
3476
3477 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3478 moving to gcc 4.1.1.
3479
3480 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3481 sys/time.h, which is documented as needed for this function. It doesn't
3482 seem to matter on Linux, but it showed up on some releases of OS X.
3483
3484 3. It seems that there are systems where bytes whose values are greater than
3485 127 match isprint() in the "C" locale. The "C" locale should be the
3486 default when a C program starts up. In most systems, only ASCII printing
3487 characters match isprint(). This difference caused the output from pcretest
3488 to vary, making some of the tests fail. I have changed pcretest so that:
3489
3490 (a) When it is outputting text in the compiled version of a pattern, bytes
3491 other than 32-126 are always shown as hex escapes.
3492
3493 (b) When it is outputting text that is a matched part of a subject string,
3494 it does the same, unless a different locale has been set for the match
3495 (using the /L modifier). In this case, it uses isprint() to decide.
3496
3497 4. Fixed a major bug that caused incorrect computation of the amount of memory
3498 required for a compiled pattern when options that changed within the
3499 pattern affected the logic of the preliminary scan that determines the
3500 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3501 that the computed length was too small. The symptoms of this bug were
3502 either the PCRE error "internal error: code overflow" from pcre_compile(),
3503 or a glibc crash with a message such as "pcretest: free(): invalid next
3504 size (fast)". Examples of patterns that provoked this bug (shown in
3505 pcretest format) are:
3506
3507 /(?-x: )/x
3508 /(?x)(?-x: \s*#\s*)/
3509 /((?i)[\x{c0}])/8
3510 /(?i:[\x{c0}])/8
3511
3512 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3513 is now done differently.
3514
3515 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3516 wrapper classes; (b) implement a new function in the C++ scanner that is
3517 more efficient than the old way of doing things because it avoids levels of
3518 recursion in the regex matching; (c) add a paragraph to the documentation
3519 for the FullMatch() function.
3520
3521 6. The escape sequence \n was being treated as whatever was defined as
3522 "newline". Not only was this contrary to the documentation, which states
3523 that \n is character 10 (hex 0A), but it also went horribly wrong when
3524 "newline" was defined as CRLF. This has been fixed.
3525
3526 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3527 was being set to -1 for the "end of line" case (supposedly a value that no
3528 character can have). Though this value is never used (the check for end of
3529 line is "zero bytes in current character"), it caused compiler complaints.
3530 I've changed it to 0xffffffff.
3531
3532 8. In pcre_version.c, the version string was being built by a sequence of
3533 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3534 string (as it is for production releases) called a macro with an empty
3535 argument. The C standard says the result of this is undefined. The gcc
3536 compiler treats it as an empty string (which was what was wanted) but it is
3537 reported that Visual C gives an error. The source has been hacked around to
3538 avoid this problem.
3539
3540 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3541 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3542 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3543 of them did).
3544
3545 10. Originally, pcretest opened its input and output without "b"; then I was
3546 told that "b" was needed in some environments, so it was added for release
3547 5.0 to both the input and output. (It makes no difference on Unix-like
3548 systems.) Later I was told that it is wrong for the input on Windows. I've
3549 now abstracted the modes into two macros, to make it easier to fiddle with
3550 them, and removed "b" from the input mode under Windows.
3551
3552 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3553
3554 12. Added -help and --help to pcretest as an official way of being reminded
3555 of the options.
3556
3557 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3558 and pcrecpp.cc because they annoy compilers at high warning levels.
3559
3560 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3561
3562 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3563 scripts are not C programs :-) and which was not noticed because it works
3564 on Linux.
3565
3566 16. pcretest is supposed to handle any length of pattern and data line (as one
3567 line or as a continued sequence of lines) by extending its input buffer if
3568 necessary. This feature was broken for very long pattern lines, leading to
3569 a string of junk being passed to pcre_compile() if the pattern was longer
3570 than about 50K.
3571
3572 17. I have done a major re-factoring of the way pcre_compile() computes the
3573 amount of memory needed for a compiled pattern. Previously, there was code
3574 that made a preliminary scan of the pattern in order to do this. That was
3575 OK when PCRE was new, but as the facilities have expanded, it has become
3576 harder and harder to keep it in step with the real compile phase, and there
3577 have been a number of bugs (see for example, 4 above). I have now found a
3578 cunning way of running the real compile function in a "fake" mode that
3579 enables it to compute how much memory it would need, while actually only
3580 ever using a few hundred bytes of working memory and without too many
3581 tests of the mode. This should make future maintenance and development
3582 easier. A side effect of this work is that the limit of 200 on the nesting
3583 depth of parentheses has been removed (though this was never a serious
3584 limitation, I suspect). However, there is a downside: pcre_compile() now
3585 runs more slowly than before (30% or more, depending on the pattern). I
3586 hope this isn't a big issue. There is no effect on runtime performance.
3587
3588 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3589 newline (only possible for the last line of a file) and it was a
3590 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3591
3592 19. Added additional timing features to pcretest. (1) The -tm option now times
3593 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3594 separate command line item, by a number that specifies the number of
3595 repeats to use when timing. The default is 50000; this gives better
3596 precision, but takes uncomfortably long for very large patterns.
3597
3598 20. Extended pcre_study() to be more clever in cases where a branch of a
3599 subpattern has no definite first character. For example, (a*|b*)[cd] would
3600 previously give no result from pcre_study(). Now it recognizes that the
3601 first character must be a, b, c, or d.
3602
3603 21. There was an incorrect error "recursive call could loop indefinitely" if
3604 a subpattern (or the entire pattern) that was being tested for matching an
3605 empty string contained only one non-empty item after a nested subpattern.
3606 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3607 incorrectly, because the \d was being skipped in the check.
3608
3609 22. The pcretest program now has a new pattern option /B and a command line
3610 option -b, which is equivalent to adding /B to every pattern. This causes
3611 it to show the compiled bytecode, without the additional information that
3612 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3613 is the same as /B/I).
3614
3615 23. A new optimization is now able automatically to treat some sequences such
3616 as a*b as a*+b. More specifically, if something simple (such as a character
3617 or a simple class like \d) has an unlimited quantifier, and is followed by
3618 something that cannot possibly match the quantified thing, the quantifier
3619 is automatically "possessified".
3620
3621 24. A recursive reference to a subpattern whose number was greater than 39
3622 went wrong under certain circumstances in UTF-8 mode. This bug could also
3623 have affected the operation of pcre_study().
3624
3625 25. Realized that a little bit of performance could be had by replacing
3626 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3627
3628 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3629
3630 27. Possessive quantifiers such as a++ were previously implemented by turning
3631 them into atomic groups such as ($>a+). Now they have their own opcodes,
3632 which improves performance. This includes the automatically created ones
3633 from 23 above.
3634
3635 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3636 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3637 the first matched character to be a colon. This applied both to named and
3638 numbered groups.
3639
3640 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3641
3642 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3643 building PCRE on Windows easier, so I have included it in the distribution.
3644
3645 31. There is now a check in pcretest against a ridiculously large number being
3646 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3647 loop, the loop is abandoned.
3648
3649 32. Forward references to subpatterns in conditions such as (?(2)...) where
3650 subpattern 2 is defined later cause pcre_compile() to search forwards in
3651 the pattern for the relevant set of parentheses. This search went wrong
3652 when there were unescaped parentheses in a character class, parentheses
3653 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3654
3655 33. "Subroutine" calls and backreferences were previously restricted to
3656 referencing subpatterns earlier in the regex. This restriction has now
3657 been removed.
3658
3659 34. Added a number of extra features that are going to be in Perl 5.10. On the
3660 whole, these are just syntactic alternatives for features that PCRE had
3661 previously implemented using the Python syntax or my own invention. The
3662 other formats are all retained for compatibility.
3663
3664 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3665 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3666 also .NET compatible.
3667
3668 (b) A recursion or subroutine call to a named group can now be defined as
3669 (?&name) as well as (?P>name).
3670
3671 (c) A backreference to a named group can now be defined as \k<name> or
3672 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3673 5.10, are also .NET compatible.
3674
3675 (d) A conditional reference to a named group can now use the syntax
3676 (?(<name>) or (?('name') as well as (?(name).
3677
3678 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3679 groups (named and numbered) that are never evaluated inline, but can be
3680 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3681 is always false. There may be only one alternative in such a group.
3682
3683 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3684 as the simple (?(R). The condition is true only if the most recent
3685 recursion is that of the given number or name. It does not search out
3686 through the entire recursion stack.
3687
3688 (g) The escape \gN or \g{N} has been added, where N is a positive or
3689 negative number, specifying an absolute or relative reference.
3690
3691 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3692 some "unreachable code" warnings.
3693
3694 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3695 things, this adds five new scripts.
3696
3697 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3698 There were also incompatibilities regarding the handling of \Q..\E inside
3699 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3700 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3701
3702 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3703 matches an empty string, and forcibly breaks the loop. There were bugs in
3704 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3705 against aaaa the result was just "a" rather than "aaaa", for example. Two
3706 separate and independent bugs (that affected different cases) have been
3707 fixed.
3708
3709 39. Refactored the code to abolish the use of different opcodes for small
3710 capturing bracket numbers. This is a tidy that I avoided doing when I
3711 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3712 The new approach is not only tidier, it makes it possible to reduce the
3713 memory needed to fix the previous bug (38).
3714
3715 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3716 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3717 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3718 mode.
3719
3720 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3721 report.
3722
3723 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3724 copy construction and assignment in the C++ wrapper.
3725
3726 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3727 couple of bugs that could have given wrong results in the "--newline=crlf"
3728 case.
3729
3730 44. Added a number of casts and did some reorganization of signed/unsigned int
3731 variables following suggestions from Dair Grant. Also renamed the variable
3732 "this" as "item" because it is a C++ keyword.
3733
3734 45. Arranged for dftables to add
3735
3736 #include "pcre_internal.h"
3737
3738 to pcre_chartables.c because without it, gcc 4.x may remove the array
3739 definition from the final binary if PCRE is built into a static library and
3740 dead code stripping is activated.
3741
3742 46. For an unanchored pattern, if a match attempt fails at the start of a
3743 newline sequence, and the newline setting is CRLF or ANY, and the next two
3744 characters are CRLF, advance by two characters instead of one.
3745
3746
3747 Version 6.7 04-Jul-06
3748 ---------------------
3749
3750 1. In order to handle tests when input lines are enormously long, pcretest has
3751 been re-factored so that it automatically extends its buffers when
3752 necessary. The code is crude, but this _is_ just a test program. The
3753 default size has been increased from 32K to 50K.
3754
3755 2. The code in pcre_study() was using the value of the re argument before
3756 testing it for NULL. (Of course, in any sensible call of the function, it
3757 won't be NULL.)
3758
3759 3. The memmove() emulation function in pcre_internal.h, which is used on
3760 systems that lack both memmove() and bcopy() - that is, hardly ever -
3761 was missing a "static" storage class specifier.
3762
3763 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3764 containing an extended class (one that cannot be represented by a bitmap
3765 because it contains high-valued characters or Unicode property items, e.g.
3766 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3767 pattern, but PCRE should not loop if you do not (it no longer does).
3768 [Detail: two cases were found: (a) a repeated subpattern containing an
3769 extended class; (b) a recursive reference to a subpattern that followed a
3770 previous extended class. It wasn't skipping over the extended class
3771 correctly when UTF-8 mode was not set.]
3772
3773 5. A negated single-character class was not being recognized as fixed-length
3774 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3775 compile error "lookbehind assertion is not fixed length".
3776
3777 6. The RunPerlTest auxiliary script was showing an unexpected difference
3778 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3779 write a Perl script that can interpret lines of an input file either as
3780 byte characters or as UTF-8, which is what "perltest" was being required to
3781 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3782 can't do is switch easily at run time between having the "use utf8;" pragma
3783 or not. In the end, I fudged it by using the RunPerlTest script to insert
3784 "use utf8;" explicitly for the UTF-8 tests.
3785
3786 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3787 the end of the subject string, contrary to the documentation and to what
3788 Perl does. This was true of both matching functions. Now it matches only at
3789 the start of the subject and immediately after *internal* newlines.
3790
3791 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3792 a pointer to an int instead of a pointer to an unsigned long int. This
3793 caused problems on 64-bit systems.
3794
3795 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3796 instance of the 'standard' template library not being so standard".
3797
3798 10. There was no check on the number of named subpatterns nor the maximum
3799 length of a subpattern name. The product of these values is used to compute
3800 the size of the memory block for a compiled pattern. By supplying a very
3801 long subpattern name and a large number of named subpatterns, the size
3802 computation could be caused to overflow. This is now prevented by limiting
3803 the length of names to 32 characters, and the number of named subpatterns
3804 to 10,000.
3805
3806 11. Subpatterns that are repeated with specific counts have to be replicated in
3807 the compiled pattern. The size of memory for this was computed from the
3808 length of the subpattern and the repeat count. The latter is limited to
3809 65535, but there was no limit on the former, meaning that integer overflow
3810 could in principle occur. The compiled length of a repeated subpattern is
3811 now limited to 30,000 bytes in order to prevent this.
3812
3813 12. Added the optional facility to have named substrings with the same name.
3814
3815 13. Added the ability to use a named substring as a condition, using the
3816 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3817 are numbers (not recommended). Forward references are permitted.
3818
3819 14. Added forward references in named backreferences (if you see what I mean).
3820
3821 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3822 pattern could run off the end of the subject. For example, the pattern
3823 "(?s)(.{1,5})"8 did this with the subject "ab".
3824
3825 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3826 PCRE_CASELESS was set when matching characters that were quantified with ?
3827 or *.
3828
3829 17. A character class other than a single negated character that had a minimum
3830 but no maximum quantifier - for example [ab]{6,} - was not handled
3831 correctly by pce_dfa_exec(). It would match only one character.
3832
3833 18. A valid (though odd) pattern that looked like a POSIX character
3834 class but used an invalid character after [ (for example [[,abc,]]) caused
3835 pcre_compile() to give the error "Failed: internal error: code overflow" or
3836 in some cases to crash with a glibc free() error. This could even happen if
3837 the pattern terminated after [[ but there just happened to be a sequence of
3838 letters, a binary zero, and a closing ] in the memory that followed.
3839
3840 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3841 over the years. Originally (before any Unicode support), just the bottom 8
3842 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3843 output from "man perlunicode" includes this:
3844
3845 The regular expression compiler produces polymorphic opcodes. That
3846 is, the pattern adapts to the data and automatically switches to
3847 the Unicode character scheme when presented with Unicode data--or
3848 instead uses a traditional byte scheme when presented with byte
3849 data.
3850
3851 Sadly, a wide octal escape does not cause a switch, and in a string with
3852 no other multibyte characters, these octal escapes are treated as before.
3853 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3854 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3855 Unicode string.
3856
3857 I have not perpetrated such confusion in PCRE. Up till now, it took just
3858 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3859 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3860 translate to the appropriate multibyte character.
3861
3862 29. Applied some refactoring to reduce the number of warnings from Microsoft
3863 and Borland compilers. This has included removing the fudge introduced
3864 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3865 a warning about an unused variable.
3866
3867 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3868 characters since release 4.0, because Perl (from release 5.004) does not.
3869 [Or at least, is documented not to: some releases seem to be in conflict
3870 with the documentation.] However, when a pattern was studied with
3871 pcre_study() and all its branches started with \s, PCRE still included VT
3872 as a possible starting character. Of course, this did no harm; it just
3873 caused an unnecessary match attempt.
3874
3875 22. Removed a now-redundant internal flag bit that recorded the fact that case
3876 dependency changed within the pattern. This was once needed for "required
3877 byte" processing, but is no longer used. This recovers a now-scarce options
3878 bit. Also moved the least significant internal flag bit to the most-
3879 significant bit of the word, which was not previously used (hangover from
3880 the days when it was an int rather than a uint) to free up another bit for
3881 the future.
3882
3883 23. Added support for CRLF line endings as well as CR and LF. As well as the
3884 default being selectable at build time, it can now be changed at runtime
3885 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3886 specify that it is scanning data with non-default line endings.
3887
3888 24. Changed the definition of CXXLINK to make it agree with the definition of
3889 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3890
3891 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3892 recursions. This makes a big different to stack usage for some patterns.
3893
3894 26. If a subpattern containing a named recursion or subroutine reference such
3895 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3896 the space required for the compiled pattern went wrong and gave too small a
3897 value. Depending on the environment, this could lead to "Failed: internal
3898 error: code overflow at offset 49" or "glibc detected double free or
3899 corruption" errors.
3900
3901 27. Applied patches from Google (a) to support the new newline modes and (b) to
3902 advance over multibyte UTF-8 characters in GlobalReplace.
3903
3904 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3905 difference for some implementation of PCRE in some Windows version.
3906
3907 29. Added some extra testing facilities to pcretest:
3908
3909 \q<number> in a data line sets the "match limit" value
3910 \Q<number> in a data line sets the "match recursion limt" value
3911 -S <number> sets the stack size, where <number> is in megabytes
3912
3913 The -S option isn't available for Windows.
3914
3915
3916 Version 6.6 06-Feb-06
3917 ---------------------
3918
3919 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3920 in pcreposix.h. I have copied the definition from pcre.h.
3921
3922 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3923 because pcre.h is no longer a built file.
3924
3925 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3926 not normally included in the compiled code.
3927
3928
3929 Version 6.5 01-Feb-06
3930 ---------------------
3931
3932 1. When using the partial match feature with pcre_dfa_exec(), it was not
3933 anchoring the second and subsequent partial matches at the new starting
3934 point. This could lead to incorrect results. For example, with the pattern
3935 /1234/, partially matching against "123" and then "a4" gave a match.
3936
3937 2. Changes to pcregrep:
3938
3939 (a) All non-match returns from pcre_exec() were being treated as failures
3940 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3941 error message is output. Some extra information is given for the
3942 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3943 probably the only errors that are likely to be caused by users (by
3944 specifying a regex that has nested indefinite repeats, for instance).
3945 If there are more than 20 of these errors, pcregrep is abandoned.
3946
3947 (b) A binary zero was treated as data while matching, but terminated the
3948 output line if it was written out. This has been fixed: binary zeroes
3949 are now no different to any other data bytes.
3950
3951 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3952 used to set a locale for matching. The --locale=xxxx long option has
3953 been added (no short equivalent) to specify a locale explicitly on the
3954 pcregrep command, overriding the environment variables.
3955
3956 (d) When -B was used with -n, some line numbers in the output were one less
3957 than they should have been.
3958
3959 (e) Added the -o (--only-matching) option.
3960
3961 (f) If -A or -C was used with -c (count only), some lines of context were
3962 accidentally printed for the final match.
3963
3964 (g) Added the -H (--with-filename) option.
3965
3966 (h) The combination of options -rh failed to suppress file names for files
3967 that were found from directory arguments.
3968
3969 (i) Added the -D (--devices) and -d (--directories) options.
3970
3971 (j) Added the -F (--fixed-strings) option.
3972
3973 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3974
3975 (l) Added the --colo(u)r option.
3976
3977 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3978 is not present by default.
3979
3980 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3981 items such as (?R) or (?1), when the recursion could match a number of
3982 alternatives. If it matched one of the alternatives, but subsequently,
3983 outside the recursion, there was a failure, the code tried to back up into
3984 the recursion. However, because of the way PCRE is implemented, this is not
3985 possible, and the result was an incorrect result from the match.
3986
3987 In order to prevent this happening, the specification of recursion has
3988 been changed so that all such subpatterns are automatically treated as
3989 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3990
3991 4. I had overlooked the fact that, in some locales, there are characters for
3992 which isalpha() is true but neither isupper() nor islower() are true. In
3993 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3994 and ordfeminine) are like this. This affected the treatment of \w and \W
3995 when they appeared in character classes, but not when they appeared outside
3996 a character class. The bit map for "word" characters is now created
3997 separately from the results of isalnum() instead of just taking it from the
3998 upper, lower, and digit maps. (Plus the underscore character, of course.)
3999
4000 5. The above bug also affected the handling of POSIX character classes such as
4001 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4002 permanent tables. Instead, the bit maps for such a class were previously
4003 created as the appropriate unions of the upper, lower, and digit bitmaps.
4004 Now they are created by subtraction from the [[:word:]] class, which has
4005 its own bitmap.
4006
4007 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4008 It is created by subtracting the vertical space characters (\x09, \x0a,
4009 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4010 subtraction was done in the overall bitmap for a character class, meaning
4011 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4012 be recognized. This bug has been fixed.
4013
4014 7. Patches from the folks at Google:
4015
4016 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4017 real life, but is still worth protecting against".
4018
4019 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4020 regular expressions".
4021
4022 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4023 have it.
4024
4025 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4026 "configure" and the latter not, in order to fix a problem somebody had
4027 with compiling the Arg class on HP-UX.
4028
4029 (e) Improve the error-handling of the C++ wrapper a little bit.
4030
4031 (f) New tests for checking recursion limiting.
4032
4033 8. The pcre_memmove() function, which is used only if the environment does not
4034 have a standard memmove() function (and is therefore rarely compiled),
4035 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4036 returning a result (though PCRE never actually uses the result).
4037
4038 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4039 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4040 returned instead of calling malloc() with an overflowing number that would
4041 most likely cause subsequent chaos.
4042
4043 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4044
4045 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4046 with this option is matched, the nmatch and pmatch options of regexec() are
4047 ignored.
4048
4049 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4050 provided in case anyone wants to the the POSIX interface with UTF-8
4051 strings.
4052
4053 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4054 C++ linking (needed for some HP-UX environments).
4055
4056 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4057 (unused parameter) and in the pcre_printint() function (omitted "default"
4058 switch label when the default is to do nothing).
4059
4060 15. Added some code to make it possible, when PCRE is compiled as a C++
4061 library, to replace subject pointers for pcre_exec() with a smart pointer
4062 class, thus making it possible to process discontinuous strings.
4063
4064 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4065 much the same function. They were added by different people who were trying
4066 to make PCRE easy to compile on non-Unix systems. It has been suggested
4067 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4068 for compiling on Windows systems. I have therefore replaced it with
4069 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4070 defaults to "extern" for C or "extern C" for C++, which works fine on
4071 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4072 SCOPE with something explicit in config.h. In addition:
4073
4074 (a) pcreposix.h still had just "extern" instead of either of these macros;
4075 I have replaced it with PCRE_DATA_SCOPE.
4076
4077 (b) Functions such as _pcre_xclass(), which are internal to the library,
4078 but external in the C sense, all had PCRE_EXPORT in their definitions.
4079 This is apparently wrong for the Windows case, so I have removed it.
4080 (It makes no difference on Unix-like systems.)
4081
4082 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4083 of recursive calls to match(). This is different to MATCH_LIMIT because
4084 that limits the total number of calls to match(), not all of which increase
4085 the depth of recursion. Limiting the recursion depth limits the amount of
4086 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4087 when PCRE is compiled, and changed at run time. A patch from Google adds
4088 this functionality to the C++ interface.
4089
4090 18. Changes to the handling of Unicode character properties:
4091
4092 (a) Updated the table to Unicode 4.1.0.
4093
4094 (b) Recognize characters that are not in the table as "Cn" (undefined).
4095
4096 (c) I revised the way the table is implemented to a much improved format
4097 which includes recognition of ranges. It now supports the ranges that
4098 are defined in UnicodeData.txt, and it also amalgamates other
4099 characters into ranges. This has reduced the number of entries in the
4100 table from around 16,000 to around 3,000, thus reducing its size
4101 considerably. I realized I did not need to use a tree structure after
4102 all - a binary chop search is just as efficient. Having reduced the
4103 number of entries, I extended their size from 6 bytes to 8 bytes to
4104 allow for more data.
4105
4106 (d) Added support for Unicode script names via properties such as \p{Han}.
4107
4108 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4109 matching that character.
4110
4111 20. When matching a repeated Unicode property with a minimum greater than zero,
4112 (for example \pL{2,}), PCRE could look past the end of the subject if it
4113 reached it while seeking the minimum number of characters. This could
4114 happen only if some of the characters were more than one byte long, because
4115 there is a check for at least the minimum number of bytes.
4116
4117 21. Refactored the implementation of \p and \P so as to be more general, to
4118 allow for more different types of property in future. This has changed the
4119 compiled form incompatibly. Anybody with saved compiled patterns that use
4120 \p or \P will have to recompile them.
4121
4122 22. Added "Any" and "L&" to the supported property types.
4123
4124 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4125 but give a compile time error if the value is greater than 0xff.
4126
4127 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4128 accidentally not being installed or uninstalled.
4129
4130 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4131 made were to insert the current release number. This seemed silly, because
4132 it made things harder for people building PCRE on systems that don't run
4133 "configure". I have turned pcre.h into a distributed file, no longer built
4134 by "configure", with the version identification directly included. There is
4135 no longer a pcre.h.in file.
4136
4137 However, this change necessitated a change to the pcre-config script as
4138 well. It is built from pcre-config.in, and one of the substitutions was the
4139 release number. I have updated configure.ac so that ./configure now finds
4140 the release number by grepping pcre.h.
4141
4142 26. Added the ability to run the tests under valgrind.
4143
4144
4145 Version 6.4 05-Sep-05
4146 ---------------------
4147
4148 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4149 "--" to be printed when multiple files were scanned, even when none of the
4150 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4151 consider it to be a bug, and have restored the previous behaviour.
4152
4153 2. A couple of code tidies to get rid of compiler warnings.
4154
4155 3. The pcretest program used to cheat by referring to symbols in the library
4156 whose names begin with _pcre_. These are internal symbols that are not
4157 really supposed to be visible externally, and in some environments it is
4158 possible to suppress them. The cheating is now confined to including
4159 certain files from the library's source, which is a bit cleaner.
4160
4161 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4162 file's purpose clearer.
4163
4164 5. Reorganized pcre_ucp_findchar().
4165
4166
4167 Version 6.3 15-Aug-05
4168 ---------------------
4169
4170 1. The file libpcre.pc.in did not have general read permission in the tarball.
4171
4172 2. There were some problems when building without C++ support:
4173
4174 (a) If C++ support was not built, "make install" and "make test" still
4175 tried to test it.
4176
4177 (b) There were problems when the value of CXX was explicitly set. Some
4178 changes have been made to try to fix these, and ...
4179
4180 (c) --disable-cpp can now be used to explicitly disable C++ support.
4181
4182 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4183 backslash in a target when C++ was disabled. This confuses some
4184 versions of "make", apparently. Using an intermediate variable solves
4185 this. (Same for CPP_LOBJ.)
4186
4187 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4188 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4189 necessary on certain architectures.
4190
4191 4. Added a setting of -export-symbols-regex to the link command to remove
4192 those symbols that are exported in the C sense, but actually are local
4193 within the library, and not documented. Their names all begin with
4194 "_pcre_". This is not a perfect job, because (a) we have to except some
4195 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4196 available (and never for static libraries). I have made a note to try to
4197 find a way round (a) in the future.
4198
4199
4200 Version 6.2 01-Aug-05
4201 ---------------------
4202
4203 1. There was no test for integer overflow of quantifier values. A construction
4204 such as {1111111111111111} would give undefined results. What is worse, if
4205 a minimum quantifier for a parenthesized subpattern overflowed and became
4206 negative, the calculation of the memory size went wrong. This could have
4207 led to memory overwriting.
4208
4209 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4210
4211 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4212 operating environments where this matters.
4213
4214 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4215 PCRE options from within the C++ wrapper.
4216
4217 5. Named capturing subpatterns were not being correctly counted when a pattern
4218 was compiled. This caused two problems: (a) If there were more than 100
4219 such subpatterns, the calculation of the memory needed for the whole
4220 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4221 back references of the form \12, where the number was greater than 9, were
4222 not recognized as back references, even though there were sufficient
4223 previous subpatterns.
4224
4225 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4226 versions of gcc, e.g. 2.95.4.
4227
4228
4229 Version 6.1 21-Jun-05
4230 ---------------------
4231
4232 1. There was one reference to the variable "posix" in pcretest.c that was not
4233 surrounded by "#if !defined NOPOSIX".
4234
4235 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4236 the cross-check on the old pcre_info() function, for the benefit of the
4237 cut-down version of PCRE that is currently imported into Exim.
4238
4239 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4240 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4241 patterns that start (?i) but I don't think that matters. The use of (?i) is
4242 just an example; this all applies to the other options as well.
4243
4244 4. Since libtool seems to echo the compile commands it is issuing, the output
4245 from "make" can be reduced a bit by putting "@" in front of each libtool
4246 compile command.
4247
4248 5. Patch from the folks at Google for configure.in to be a bit more thorough
4249 in checking for a suitable C++ installation before trying to compile the
4250 C++ stuff. This should fix a reported problem when a compiler was present,
4251 but no suitable headers.
4252
4253 6. The man pages all had just "PCRE" as their title. I have changed them to
4254 be the relevant file name. I have also arranged that these names are
4255 retained in the file doc/pcre.txt, which is a concatenation in text format
4256 of all the man pages except the little individual ones for each function.
4257
4258 7. The NON-UNIX-USE file had not been updated for the different set of source
4259 files that come with release 6. I also added a few comments about the C++
4260 wrapper.
4261
4262
4263 Version 6.0 07-Jun-05
4264 ---------------------
4265
4266 1. Some minor internal re-organization to help with my DFA experiments.
4267
4268 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4269 didn't matter for the library itself when fully configured, but did matter
4270 when compiling without UCP support, or within Exim, where the ucp files are
4271 not imported.
4272
4273 3. Refactoring of the library code to split up the various functions into
4274 different source modules. The addition of the new DFA matching code (see
4275 below) to a single monolithic source would have made it really too
4276 unwieldy, quite apart from causing all the code to be include in a
4277 statically linked application, when only some functions are used. This is
4278 relevant even without the DFA addition now that patterns can be compiled in
4279 one application and matched in another.
4280
4281 The downside of splitting up is that there have to be some external
4282 functions and data tables that are used internally in different modules of
4283 the library but which are not part of the API. These have all had their
4284 names changed to start with "_pcre_" so that they are unlikely to clash
4285 with other external names.
4286
4287 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4288 a different (DFA) algorithm. Although it is slower than the original
4289 function, it does have some advantages for certain types of matching
4290 problem.
4291
4292 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4293 including restarting after a partial match.
4294
4295 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4296 defined when compiling for Windows was sent to me. I have put it into the
4297 code, though I have no means of testing or verifying it.
4298
4299 7. Added the pcre_refcount() auxiliary function.
4300
4301 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4302 match before or at the first newline in the subject string. In pcretest,
4303 the /f option on a pattern can be used to set this.
4304
4305 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4306 would behave wrongly. This has been present in PCRE since release 4.0.
4307
4308 10. A number of changes to the pcregrep command:
4309
4310 (a) Refactored how -x works; insert ^(...)$ instead of setting
4311 PCRE_ANCHORED and checking the length, in preparation for adding
4312 something similar for -w.
4313
4314 (b) Added the -w (match as a word) option.
4315
4316 (c) Refactored the way lines are read and buffered so as to have more
4317 than one at a time available.
4318
4319 (d) Implemented a pcregrep test script.
4320
4321 (e) Added the -M (multiline match) option. This allows patterns to match
4322 over several lines of the subject. The buffering ensures that at least
4323 8K, or the rest of the document (whichever is the shorter) is available
4324 for matching (and similarly the previous 8K for lookbehind assertions).
4325
4326 (f) Changed the --help output so that it now says
4327
4328 -w, --word-regex(p)
4329
4330 instead of two lines, one with "regex" and the other with "regexp"
4331 because that confused at least one person since the short forms are the
4332 same. (This required a bit of code, as the output is generated
4333 automatically from a table. It wasn't just a text change.)
4334
4335 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4336 option but starts with a hyphen. Could be a pattern or a path name
4337 starting with a hyphen, for instance.
4338
4339 (h) "-" can be given as a file name to represent stdin.
4340
4341 (i) When file names are being printed, "(standard input)" is used for
4342 the standard input, for compatibility with GNU grep. Previously
4343 "<stdin>" was used.
4344
4345 (j) The option --label=xxx can be used to supply a name to be used for
4346 stdin when file names are being printed. There is no short form.
4347
4348 (k) Re-factored the options decoding logic because we are going to add
4349 two more options that take data. Such options can now be given in four
4350 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4351
4352 (l) Added the -A, -B, and -C options for requesting that lines of context
4353 around matches be printed.
4354
4355 (m) Added the -L option to print the names of files that do not contain
4356 any matching lines, that is, the complement of -l.
4357
4358 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4359 continue to scan other files.
4360
4361 (o) The -s option was incorrectly implemented. For compatibility with other
4362 greps, it now suppresses the error message for a non-existent or non-
4363 accessible file (but not the return code). There is a new option called
4364 -q that suppresses the output of matching lines, which was what -s was
4365 previously doing.
4366
4367 (p) Added --include and --exclude options to specify files for inclusion
4368 and exclusion when recursing.
4369
4370 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4371 Hopefully, it now does.
4372
4373 12. Missing cast in pcre_study().
4374
4375 13. Added an "uninstall" target to the makefile.
4376
4377 14. Replaced "extern" in the function prototypes in Makefile.in with
4378 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4379 world, but is set differently for Windows.
4380
4381 15. Added a second compiling function called pcre_compile2(). The only
4382 difference is that it has an extra argument, which is a pointer to an
4383 integer error code. When there is a compile-time failure, this is set
4384 non-zero, in addition to the error test pointer being set to point to an
4385 error message. The new argument may be NULL if no error number is required
4386 (but then you may as well call pcre_compile(), which is now just a
4387 wrapper). This facility is provided because some applications need a
4388 numeric error indication, but it has also enabled me to tidy up the way
4389 compile-time errors are handled in the POSIX wrapper.
4390
4391 16. Added VPATH=.libs to the makefile; this should help when building with one
4392 prefix path and installing with another. (Or so I'm told by someone who
4393 knows more about this stuff than I do.)
4394
4395 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4396 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4397 match everything, including newlines. This is not POSIX-compatible, but
4398 somebody wanted the feature. From pcretest it can be activated by using
4399 both the P and the s flags.
4400
4401 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4402
4403 19. libpcre.pc was being incorrectly installed as executable.
4404
4405 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4406 it now also looks for '\r' so that it will work unmodified on Windows.
4407
4408 21. Added Google's contributed C++ wrapper to the distribution.
4409
4410 22. Added some untidy missing memory free() calls in pcretest, to keep
4411 Electric Fence happy when testing.
4412
4413
4414
4415 Version 5.0 13-Sep-04
4416 ---------------------
4417
4418 1. Internal change: literal characters are no longer packed up into items
4419 containing multiple characters in a single byte-string. Each character
4420 is now matched using a separate opcode. However, there may be more than one
4421 byte in the character in UTF-8 mode.
4422
4423 2. The pcre_callout_block structure has two new fields: pattern_position and
4424 next_item_length. These contain the offset in the pattern to the next match
4425 item, and its length, respectively.
4426
4427 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4428 insertion of callouts before each pattern item. Added the /C option to
4429 pcretest to make use of this.
4430
4431 4. On the advice of a Windows user, the lines
4432
4433 #if defined(_WIN32) || defined(WIN32)
4434 _setmode( _fileno( stdout ), 0x8000 );
4435 #endif /* defined(_WIN32) || defined(WIN32) */
4436
4437 have been added to the source of pcretest. This apparently does useful
4438 magic in relation to line terminators.
4439
4440 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4441 for the benefit of those environments where the "b" makes a difference.
4442
4443 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4444 to know about it. I have put a hack into configure.in that adds in code
4445 to set GCC=yes if CC=icc. This seems to end up at a point in the
4446 generated configure script that is early enough to affect the setting of
4447 compiler options, which is what is needed, but I have no means of testing
4448 whether it really works. (The user who reported this had patched the
4449 generated configure script, which of course I cannot do.)
4450
4451 LATER: After change 22 below (new libtool files), the configure script
4452 seems to know about icc (and also ecc). Therefore, I have commented out
4453 this hack in configure.in.
4454
4455 7. Added support for pkg-config (2 patches were sent in).
4456
4457 8. Negated POSIX character classes that used a combination of internal tables
4458 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4459 [[:^ascii]]. Typically, they would match almost any characters. The other
4460 POSIX classes were not broken in this way.
4461
4462 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4463 to find the match, as PCRE was deluded into thinking that the match had to
4464 start at the start point or following a newline. The same bug applied to
4465 patterns with negative forward assertions or any backward assertions
4466 preceding ".*" at the start, unless the pattern required a fixed first
4467 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4468
4469 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4470 starting at the last subject character, bytes beyond the end of the subject
4471 string were read.
4472
4473 11. Renamed the variable "class" as "classbits" to make life easier for C++
4474 users. (Previously there was a macro definition, but it apparently wasn't
4475 enough.)
4476
4477 12. Added the new field "tables" to the extra data so that tables can be passed
4478 in at exec time, or the internal tables can be re-selected. This allows
4479 a compiled regex to be saved and re-used at a later time by a different
4480 program that might have everything at different addresses.
4481
4482 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4483 -R library as well as a -L library.
4484
4485 14. The debugging options of pcretest (-d on the command line or D on a
4486 pattern) showed incorrect output for anything following an extended class
4487 that contained multibyte characters and which was followed by a quantifier.
4488
4489 15. Added optional support for general category Unicode character properties
4490 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4491 support. It adds about 90K to the size of the library. The meanings of the
4492 inbuilt class escapes such as \d and \s have NOT been changed.
4493
4494 16. Updated pcredemo.c to include calls to free() to release the memory for the
4495 compiled pattern.
4496
4497 17. The generated file chartables.c was being created in the source directory
4498 instead of in the building directory. This caused the build to fail if the
4499 source directory was different from the building directory, and was
4500 read-only.
4501
4502 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4503 file. No doubt somebody will tell me if they don't make sense... Also added
4504 Dan Mooney's comments about building on OpenVMS.
4505
4506 19. Added support for partial matching via the PCRE_PARTIAL option for
4507 pcre_exec() and the \P data escape in pcretest.
4508
4509 20. Extended pcretest with 3 new pattern features:
4510
4511 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4512 write the compiled pattern to the file whose name is "rest-of-line".
4513 This is a straight binary dump of the data, with the saved pointer to
4514 the character tables forced to be NULL. The study data, if any, is
4515 written too. After writing, pcretest reads a new pattern.
4516
4517 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4518 compiled pattern from the given file. There must not be any
4519 occurrences of "<" in the file name (pretty unlikely); if there are,
4520 pcretest will instead treat the initial "<" as a pattern delimiter.
4521 After reading in the pattern, pcretest goes on to read data lines as
4522 usual.
4523
4524 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4525 and 16-bit fields in a compiled pattern, to simulate a pattern that
4526 was compiled on a host of opposite endianness.
4527
4528 21. The pcre-exec() function can now cope with patterns that were compiled on
4529 hosts of opposite endianness, with this restriction:
4530
4531 As for any compiled expression that is saved and used later, the tables
4532 pointer field cannot be preserved; the extra_data field in the arguments
4533 to pcre_exec() should be used to pass in a tables address if a value
4534 other than the default internal tables were used at compile time.
4535
4536 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4537 now diagnosed as an error. Previously, most of the time, a negative number
4538 would have been treated as zero, but if in addition "ovector" was passed as
4539 NULL, a crash could occur.
4540
4541 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4542 new versions from the libtool 1.5 distribution (the last one is a copy of
4543 a file called libtool.m4). This seems to have fixed the need to patch
4544 "configure" to support Darwin 1.3 (which I used to do). However, I still
4545 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4546 workstation).
4547
4548 24. Changed the PCRE licence to be the more standard "BSD" licence.
4549
4550
4551 Version 4.5 01-Dec-03
4552 ---------------------
4553
4554 1. There has been some re-arrangement of the code for the match() function so
4555 that it can be compiled in a version that does not call itself recursively.
4556 Instead, it keeps those local variables that need separate instances for
4557 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4558 needs to "recurse". Keeping track of where control must go is done by means
4559 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4560 hide most of the details from the main code, and operates only if
4561 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4562 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4563 operating.
4564
4565 To make it easier for callers to provide specially tailored get/free
4566 functions for this usage, two new functions, pcre_stack_malloc, and
4567 pcre_stack_free, are used. They are always called in strict stacking order,
4568 and the size of block requested is always the same.
4569
4570 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4571 PCRE has been compiled to use the stack or the heap for recursion. The
4572 -C option of pcretest uses this to show which version is compiled.
4573
4574 A new data escape \S, is added to pcretest; it causes the amounts of store
4575 obtained and freed by both kinds of malloc/free at match time to be added
4576 to the output.
4577
4578 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4579 what's available on my current Linux desktop machine.
4580
4581 3. When matching a UTF-8 string, the test for a valid string at the start has
4582 been extended. If start_offset is not zero, PCRE now checks that it points
4583 to a byte that is the start of a UTF-8 character. If not, it returns
4584 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4585 this is necessary because there may be backward assertions in the pattern.
4586 When matching the same subject several times, it may save resources to use
4587 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4588
4589 4. The code for checking the validity of UTF-8 strings has been tightened so
4590 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4591 containing "overlong sequences".
4592
4593 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4594 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4595 should have been "&", but it just so happened that all the cases this let
4596 through by mistake were picked up later in the function.
4597
4598 6. I had used a variable called "isblank" - this is a C99 function, causing
4599 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4600
4601 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4602 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4603 so the version is shown; (c) stop "make test" echoing "./RunTest".
4604
4605 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4606
4607 9. The prototype for memmove() for systems that don't have it was using
4608 size_t, but the inclusion of the header that defines size_t was later. I've
4609 moved the #includes for the C headers earlier to avoid this.
4610
4611 10. Added some adjustments to the code to make it easier to compiler on certain
4612 special systems:
4613
4614 (a) Some "const" qualifiers were missing.
4615 (b) Added the macro EXPORT before all exported functions; by default this
4616 is defined to be empty.
4617 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4618 that it reads its output file name as an argument instead of writing
4619 to the standard output and assuming this can be redirected.
4620
4621 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4622 class containing characters with values greater than 255, PCRE compilation
4623 went into a loop.
4624
4625 12. A recursive reference to a subpattern that was within another subpattern
4626 that had a minimum quantifier of zero caused PCRE to crash. For example,
4627 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4628 recursion. If the recursively-called subpattern itself had a zero repeat,
4629 that was OK.
4630
4631 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4632 buffer into which it was copied (for escape processing) was still set at
4633 1024, so long lines caused crashes.
4634
4635 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4636 "internal error: code overflow...". This applied to any character class
4637 that was followed by a possessive quantifier.
4638
4639 15. Modified the Makefile to add libpcre.la as a prerequisite for
4640 libpcreposix.la because I was told this is needed for a parallel build to
4641 work.
4642
4643 16. If a pattern that contained .* following optional items at the start was
4644 studied, the wrong optimizing data was generated, leading to matching
4645 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4646 matching string must start with a or b or c. The correct conclusion for
4647 this pattern is that a match can start with any character.
4648
4649
4650 Version 4.4 13-Aug-03
4651 ---------------------
4652
4653 1. In UTF-8 mode, a character class containing characters with values between
4654 127 and 255 was not handled correctly if the compiled pattern was studied.
4655 In fixing this, I have also improved the studying algorithm for such
4656 classes (slightly).
4657
4658 2. Three internal functions had redundant arguments passed to them. Removal
4659 might give a very teeny performance improvement.
4660
4661 3. Documentation bug: the value of the capture_top field in a callout is *one
4662 more than* the number of the hightest numbered captured substring.
4663
4664 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4665 in incorrectly linking with a previously installed version. They now link
4666 explicitly with libpcre.la.
4667
4668 5. configure.in no longer needs to recognize Cygwin specially.
4669
4670 6. A problem in pcre.in for Windows platforms is fixed.
4671
4672 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4673 pcretest, it used to include the size of the study block as part of its
4674 output. Unfortunately, the structure contains a field that has a different
4675 size on different hardware architectures. This meant that the tests that
4676 showed this size failed. As the block is currently always of a fixed size,
4677 this information isn't actually particularly useful in pcretest output, so
4678 I have just removed it.
4679
4680 8. Three pre-processor statements accidentally did not start in column 1.
4681 Sadly, there are *still* compilers around that complain, even though
4682 standard C has not required this for well over a decade. Sigh.
4683
4684 9. In pcretest, the code for checking callouts passed small integers in the
4685 callout_data field, which is a void * field. However, some picky compilers
4686 complained about the casts involved for this on 64-bit systems. Now
4687 pcretest passes the address of the small integer instead, which should get
4688 rid of the warnings.
4689
4690 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4691 both compile and run time, and gives an error if an invalid UTF-8 sequence
4692 is found. There is a option for disabling this check in cases where the
4693 string is known to be correct and/or the maximum performance is wanted.
4694
4695 11. In response to a bug report, I changed one line in Makefile.in from
4696
4697 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4698 to
4699 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4700
4701 to look similar to other lines, but I have no way of telling whether this
4702 is the right thing to do, as I do not use Windows. No doubt I'll get told
4703 if it's wrong...
4704
4705
4706 Version 4.3 21-May-03
4707 ---------------------
4708
4709 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4710 Makefile.
4711
4712 2. Some refactoring to improve the quality of the code:
4713
4714 (i) The utf8_table... variables are now declared "const".
4715
4716 (ii) The code for \cx, which used the "case flipping" table to upper case
4717 lower case letters, now just substracts 32. This is ASCII-specific,
4718 but the whole concept of \cx is ASCII-specific, so it seems
4719 reasonable.
4720
4721 (iii) PCRE was using its character types table to recognize decimal and
4722 hexadecimal digits in the pattern. This is silly, because it handles
4723 only 0-9, a-f, and A-F, but the character types table is locale-
4724 specific, which means strange things might happen. A private
4725 table is now used for this - though it costs 256 bytes, a table is
4726 much faster than multiple explicit tests. Of course, the standard
4727 character types table is still used for matching digits in subject
4728 strings against \d.
4729
4730 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4731 ending in _t are). So I've renamed it as ESC_tee.
4732
4733 3. The first argument for regexec() in the POSIX wrapper should have been
4734 defined as "const".
4735
4736 4. Changed pcretest to use malloc() for its buffers so that they can be
4737 Electric Fenced for debugging.
4738
4739 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4740 to read one or more bytes before the start of the subject string. Often this
4741 had no effect on PCRE's behaviour, but in some circumstances it could
4742 provoke a segmentation fault.
4743
4744 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4745 to try to read one or more bytes before the start of the subject string.
4746
4747 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4748 UTF-8 support could misbehave in various ways if the subject string
4749 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4750 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4751 back over UTF-8 characters.)
4752
4753
4754 Version 4.2 14-Apr-03
4755 ---------------------
4756
4757 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4758
4759 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4760 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4761 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4762 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4763 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4764 and BUILD_EXEEXT
4765 Note: automatic setting of the BUILD variables is not yet working
4766 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4767 compile-time but not at link-time
4768 [LINK]: use for linking executables only
4769 make different versions for Windows and non-Windows
4770 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4771 libraries
4772 [LINK_FOR_BUILD]: new variable
4773 [OBJEXT]: use throughout
4774 [EXEEXT]: use throughout
4775 <winshared>: new target
4776 <wininstall>: new target
4777 <dftables.o>: use native compiler
4778 <dftables>: use native linker
4779 <install>: handle Windows platform correctly
4780 <clean>: ditto
4781 <check>: ditto
4782 copy DLL to top builddir before testing
4783
4784 As part of these changes, -no-undefined was removed again. This was reported
4785 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4786 in any case.
4787
4788 3. Some tidies to get rid of compiler warnings:
4789
4790 . In the match_data structure, match_limit was an unsigned long int, whereas
4791 match_call_count was an int. I've made them both unsigned long ints.
4792
4793 . In pcretest the fact that a const uschar * doesn't automatically cast to
4794 a void * provoked a warning.
4795
4796 . Turning on some more compiler warnings threw up some "shadow" variables
4797 and a few more missing casts.
4798
4799 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4800 option, a class that contained a single character with a value between 128
4801 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4802
4803 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4804 option, a class that contained several characters, but with at least one
4805 whose value was between 128 and 255 caused PCRE to crash.
4806
4807
4808 Version 4.1 12-Mar-03
4809 ---------------------
4810
4811 1. Compiling with gcc -pedantic found a couple of places where casts were
4812 needed, and a string in dftables.c that was longer than standard compilers are
4813 required to support.
4814
4815 2. Compiling with Sun's compiler found a few more places where the code could
4816 be tidied up in order to avoid warnings.
4817
4818 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4819 first of these names is deprecated in the latest Autoconf in favour of the name
4820 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4821 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4822 analogy I have changed it to CFLAGS_FOR_BUILD.
4823
4824 4. Added -no-undefined to the linking command in the Makefile, because this is
4825 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4826 linking step for the pcreposix library.
4827
4828 5. PCRE was failing to diagnose the case of two named groups with the same
4829 name.
4830
4831 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4832 literal character that is needed in the subject for a match, and scans along to
4833 ensure that it is present before embarking on the full matching process. This
4834 saves time in cases of nested unlimited repeats that are never going to match.
4835 Problem: the scan can take a lot of time if the subject is very long (e.g.
4836 megabytes), thus penalizing straightforward matches. It is now done only if the
4837 amount of subject to be scanned is less than 1000 bytes.
4838
4839 7. A lesser problem with the same optimization is that it was recording the
4840 first character of an anchored pattern as "needed", thus provoking a search
4841 right along the subject, even when the first match of the pattern was going to
4842 fail. The "needed" character is now not set for anchored patterns, unless it
4843 follows something in the pattern that is of non-fixed length. Thus, it still
4844 fulfils its original purpose of finding quick non-matches in cases of nested
4845 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4846
4847
4848 Version 4.0 17-Feb-03
4849 ---------------------
4850
4851 1. If a comment in an extended regex that started immediately after a meta-item
4852 extended to the end of string, PCRE compiled incorrect data. This could lead to
4853 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4854
4855 2. Moved to autoconf 2.53 and libtool 1.4.2.
4856
4857 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4858 the special perltest8 script is no longer needed - all the tests can be run
4859 from a single perltest script.
4860
4861 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4862 by \s. It has now been removed in PCRE. This means it isn't recognized as
4863 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4864 class [:space:] *does* include VT, thereby creating a mess.
4865
4866 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4867 space and tab.
4868
4869 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4870 its new features into the main test script, reducing the number of scripts.
4871
4872 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4873 were backward compatible, and made the (?i) apply to the whole pattern, as if
4874 /i were given. Now it behaves more logically, and applies the option setting
4875 only to what follows. PCRE has been changed to follow suit. However, if it
4876 finds options settings right at the start of the pattern, it extracts them into
4877 the global options, as before. Thus, they show up in the info data.
4878
4879 8. Added support for the \Q...\E escape sequence. Characters in between are
4880 treated as literals. This is slightly different from Perl in that $ and @ are
4881 also handled as literals inside the quotes. In Perl, they will cause variable
4882 interpolation. Note the following examples:
4883
4884 Pattern PCRE matches Perl matches
4885
4886 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4887 \Qabc\$xyz\E abc\$xyz abc\$xyz
4888 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4889
4890 For compatibility with Perl, \Q...\E sequences are recognized inside character
4891 classes as well as outside them.
4892
4893 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4894 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4895 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4896 signed/unsigned warnings.
4897
4898 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4899 option for pcretest, so I've replaced it by a simple function that does just
4900 that job.
4901
4902 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4903 "pcregrep -".
4904
4905 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4906 Java package. This provides some syntactic sugar for simple cases of what my
4907 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4908 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4909 item, you can use this simplified notation. Note that only makes sense with
4910 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4911 greediness, whatever the setting of the PCRE_UNGREEDY option.
4912
4913 13. A change of greediness default within a pattern was not taking effect at
4914 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4915 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4916 was abstracted outside.
4917
4918 14. PCRE now supports the \G assertion. It is true when the current matching
4919 position is at the start point of the match. This differs from \A when the
4920 starting offset is non-zero. Used with the /g option of pcretest (or similar
4921 code), it works in the same way as it does for Perl's /g option. If all
4922 alternatives of a regex begin with \G, the expression is anchored to the start
4923 match position, and the "anchored" flag is set in the compiled expression.
4924
4925 15. Some bugs concerning the handling of certain option changes within patterns
4926 have been fixed. These applied to options other than (?ims). For example,
4927 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4928 been the other way round. Some of this was related to change 7 above.
4929
4930 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4931 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4932 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4933 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4934
4935 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4936 mode. Unlike ".", it always matches newline, whatever the setting of
4937 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4938 assertions. Perl allows it, but it doesn't (in general) work because it can't
4939 calculate the length of the lookbehind. At least, that's the case for Perl
4940 5.8.0 - I've been told they are going to document that it doesn't work in
4941 future.
4942
4943 18. Added an error diagnosis for escapes that PCRE does not support: these are
4944 \L, \l, \N, \P, \p, \U, \u, and \X.
4945
4946 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4947 reading past the end of the pattern in cases such as /[abcd/.
4948
4949 20. PCRE was getting more memory than necessary for patterns with classes that
4950 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4951
4952 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4953 compiling PCRE for use with Virtual Pascal.
4954
4955 22. Small fix to the Makefile to make it work properly if the build is done
4956 outside the source tree.
4957
4958 23. Added a new extension: a condition to go with recursion. If a conditional
4959 subpattern starts with (?(R) the "true" branch is used if recursion has
4960 happened, whereas the "false" branch is used only at the top level.
4961
4962 24. When there was a very long string of literal characters (over 255 bytes
4963 without UTF support, over 250 bytes with UTF support), the computation of how
4964 much memory was required could be incorrect, leading to segfaults or other
4965 strange effects.
4966
4967 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4968 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4969 there was a subsequent back reference to those brackets. This meant that, for
4970 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4971 possible to check for precisely this case. All we can do is abandon the
4972 optimization if .* occurs inside capturing brackets when there are any back
4973 references whatsoever. (See below for a better fix that came later.)
4974
4975 26. The handling of the optimization for finding the first character of a
4976 non-anchored pattern, and for finding a character that is required later in the
4977 match were failing in some cases. This didn't break the matching; it just
4978 failed to optimize when it could. The way this is done has been re-implemented.
4979
4980 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4981
4982 28. Added a new feature that provides some of the functionality that Perl
4983 provides with (?{...}). The facility is termed a "callout". The way it is done
4984 in PCRE is for the caller to provide an optional function, by setting
4985 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4986 global variable. By default it is unset, which disables all calling out. To get
4987 the function called, the regex must include (?C) at appropriate points. This
4988 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4989 This provides a means of identifying different callout points. When PCRE
4990 reaches such a point in the regex, if pcre_callout has been set, the external
4991 function is called. It is provided with data in a structure called
4992 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4993 matching continues; if it returns a non-zero value, the match at the current
4994 point fails. However, backtracking will occur if possible. [This was changed
4995 later and other features added - see item 49 below.]
4996
4997 29. pcretest is upgraded to test the callout functionality. It provides a
4998 callout function that displays information. By default, it shows the start of
4999 the match and the current position in the text. There are some new data escapes
5000 to vary what happens:
5001
5002 \C+ in addition, show current contents of captured substrings
5003 \C- do not supply a callout function
5004 \C!n return 1 when callout number n is reached
5005 \C!n!m return 1 when callout number n is reached for the mth time
5006
5007 30. If pcregrep was called with the -l option and just a single file name, it
5008 output "<stdin>" if a match was found, instead of the file name.
5009
5010 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5011 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5012 pcre_exec(). This saves a malloc/free per call. The default value of
5013 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5014 when configuring.
5015
5016 32. The default maximum size of a compiled pattern is 64K. There have been a
5017 few cases of people hitting this limit. The code now uses macros to handle the
5018 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5019 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5020 configuring. Tests 2 and 5 work only with 2-byte links because they output
5021 debugging information about compiled patterns.
5022
5023 33. Internal code re-arrangements:
5024
5025 (a) Moved the debugging function for printing out a compiled regex into
5026 its own source file (printint.c) and used #include to pull it into
5027 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5028 separate copies.
5029
5030 (b) Defined the list of op-code names for debugging as a macro in
5031 internal.h so that it is next to the definition of the opcodes.
5032
5033 (c) Defined a table of op-code lengths for simpler skipping along compiled
5034 code. This is again a macro in internal.h so that it is next to the
5035 definition of the opcodes.
5036
5037 34. Added support for recursive calls to individual subpatterns, along the
5038 lines of Robin Houston's patch (but implemented somewhat differently).
5039
5040 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5041 allow it to read and process whole directories in Win32. This code was
5042 contributed by Lionel Fourquaux; it has not been tested by me.
5043
5044 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5045 used to name a group. Names consist of alphanumerics and underscores, and must
5046 be unique. Back references use the syntax (?P=name) and recursive calls use
5047 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5048 numbers. The function pcre_fullinfo() can be used after compilation to extract
5049 a name/number map. There are three relevant calls:
5050
5051 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5052 PCRE_INFO_NAMECOUNT yields the number of entries
5053 PCRE_INFO_NAMETABLE yields a pointer to the map.
5054
5055 The map is a vector of fixed-size entries. The size of each entry depends on
5056 the length of the longest name used. The first two bytes of each entry are the
5057 group number, most significant byte first. There follows the corresponding
5058 name, zero terminated. The names are in alphabetical order.
5059
5060 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5061 case instead of 255. Making it the same both with and without UTF-8 support
5062 means that the same test output works with both.
5063
5064 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5065 calling malloc() with a zero argument.
5066
5067 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5068 optimization. I've improved things by keeping a bitmap of backreferences with
5069 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5070 fact referenced, the optimization can be applied. It is unlikely that a
5071 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5072 the match to follow \n) will appear inside brackets with a number greater than
5073 31, but if it does, any back reference > 31 suppresses the optimization.
5074
5075 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5076 of disabling numbered capturing parentheses. Any opening parenthesis that is
5077 not followed by ? behaves as if it were followed by ?: but named parentheses
5078 can still be used for capturing (and they will acquire numbers in the usual
5079 way).
5080
5081 41. Redesigned the return codes from the match() function into yes/no/error so
5082 that errors can be passed back from deep inside the nested calls. A malloc
5083 failure while inside a recursive subpattern call now causes the
5084 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5085
5086 42. It is now possible to set a limit on the number of times the match()
5087 function is called in a call to pcre_exec(). This facility makes it possible to
5088 limit the amount of recursion and backtracking, though not in a directly
5089 obvious way, because the match() function is used in a number of different
5090 circumstances. The count starts from zero for each position in the subject
5091 string (for non-anchored patterns). The default limit is, for compatibility, a
5092 large number, namely 10 000 000. You can change this in two ways:
5093
5094 (a) When configuring PCRE before making, you can use --with-match-limit=n
5095 to set a default value for the compiled library.
5096
5097 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5098 a different value is set. See 45 below.
5099
5100 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5101
5102 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5103 of things that can be changed at compile time. The first argument specifies
5104 what is wanted and the second points to where the information is to be placed.
5105 The current list of available information is:
5106
5107 PCRE_CONFIG_UTF8
5108
5109 The output is an integer that is set to one if UTF-8 support is available;
5110 otherwise it is set to zero.
5111
5112 PCRE_CONFIG_NEWLINE
5113
5114 The output is an integer that it set to the value of the code that is used for
5115 newline. It is either LF (10) or CR (13).
5116
5117 PCRE_CONFIG_LINK_SIZE
5118
5119 The output is an integer that contains the number of bytes used for internal
5120 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5121
5122 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5123
5124 The output is an integer that contains the threshold above which the POSIX
5125 interface uses malloc() for output vectors. See item 31 above.
5126
5127 PCRE_CONFIG_MATCH_LIMIT
5128
5129 The output is an unsigned integer that contains the default limit of the number
5130 of match() calls in a pcre_exec() execution. See 42 above.
5131
5132 44. pcretest has been upgraded by the addition of the -C option. This causes it
5133 to extract all the available output from the new pcre_config() function, and to
5134 output it. The program then exits immediately.
5135
5136 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5137 order to support additional features. One way would have been to define
5138 pcre_exec2() (for example) with extra arguments, but this would not have been
5139 extensible, and would also have required all calls to the original function to
5140 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5141 is used for passing in "extra" data from pcre_study().
5142
5143 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5144 contains the following fields:
5145
5146 flags a bitmap indicating which of the following fields are set
5147 study_data opaque data from pcre_study()
5148 match_limit a way of specifying a limit on match() calls for a specific
5149 call to pcre_exec()
5150 callout_data data for callouts (see 49 below)
5151
5152 The flag bits are also defined in pcre.h, and are
5153
5154 PCRE_EXTRA_STUDY_DATA
5155 PCRE_EXTRA_MATCH_LIMIT
5156 PCRE_EXTRA_CALLOUT_DATA
5157
5158 The pcre_study() function now returns one of these new pcre_extra blocks, with
5159 the actual study data pointed to by the study_data field, and the
5160 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5161 before. That is, this change is entirely upwards-compatible and requires no
5162 change to existing code.
5163
5164 If you want to pass in additional data to pcre_exec(), you can either place it
5165 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5166 block.
5167
5168 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5169 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5170 times with different match limits, until it finds the minimum value needed for
5171 pcre_exec() to complete. The value is then output. This can be instructive; for
5172 most simple matches the number is quite small, but for pathological cases it
5173 gets very large very quickly.
5174
5175 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5176 returns the size of the data block pointed to by the study_data field in a
5177 pcre_extra block, that is, the value that was passed as the argument to
5178 pcre_malloc() when PCRE was getting memory in which to place the information
5179 created by pcre_study(). The fourth argument should point to a size_t variable.
5180 pcretest has been extended so that this information is shown after a successful
5181 pcre_study() call when information about the compiled regex is being displayed.
5182
5183 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5184 because what follows is always an absolute path. (Later: it turns out that this
5185 is more than cosmetic for MinGW, because it doesn't like empty path
5186 components.)
5187
5188 49. Some changes have been made to the callout feature (see 28 above):
5189
5190 (i) A callout function now has three choices for what it returns:
5191
5192 0 => success, carry on matching
5193 > 0 => failure at this point, but backtrack if possible
5194 < 0 => serious error, return this value from pcre_exec()
5195
5196 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5197 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5198 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5199 use by callout functions. It will never be used by PCRE itself.
5200
5201 (ii) The pcre_extra structure (see 45 above) has a void * field called
5202 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5203 pcre_callout_block structure has a field of the same name. The contents of
5204 the field passed in the pcre_extra structure are passed to the callout
5205 function in the corresponding field in the callout block. This makes it
5206 easier to use the same callout-containing regex from multiple threads. For
5207 testing, the pcretest program has a new data escape
5208
5209 \C*n pass the number n (may be negative) as callout_data
5210
5211 If the callout function in pcretest receives a non-zero value as
5212 callout_data, it returns that value.
5213
5214 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5215 there were some redundant $(CFLAGS) in commands that are now specified as
5216 $(LINK), which already includes $(CFLAGS).
5217
5218 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5219 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5220 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5221 one-byte characters throughout. Note that case-insensitive matching applies
5222 only to characters whose values are less than 256. PCRE doesn't support the
5223 notion of cases for higher-valued characters.
5224
5225 (i) A character class whose characters are all within 0-255 is handled as
5226 a bit map, and the map is inverted for negative classes. Previously, a
5227 character > 255 always failed to match such a class; however it should
5228 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5229
5230 (ii) A negated character class with a single character < 255 is coded as
5231 "not this character" (OP_NOT). This wasn't working properly when the test
5232 character was multibyte, either singly or repeated.
5233
5234 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5235 mode, for example: \x{100}{2,3}.
5236
5237 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5238 singly or repeated) now correctly test multibyte characters. However,
5239 PCRE doesn't recognize any characters with values greater than 255 as
5240 digits, spaces, or word characters. Such characters always match \D, \S,
5241 and \W, and never match \d, \s, or \w.
5242
5243 (v) Classes may now contain characters and character ranges with values
5244 greater than 255. For example: [ab\x{100}-\x{400}].
5245
5246 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5247 PCRE in UTF-8 mode.
5248
5249 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5250 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5251 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5252 value.)
5253
5254 53. The single man page has become too large. I have therefore split it up into
5255 a number of separate man pages. These also give rise to individual HTML pages;
5256 these are now put in a separate directory, and there is an index.html page that
5257 lists them all. Some hyperlinking between the pages has been installed.
5258
5259 54. Added convenience functions for handling named capturing parentheses.
5260
5261 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5262 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5263 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5264 are faulted.
5265
5266 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5267 calling configure. These values are used when compiling the dftables.c program
5268 which is run to generate the source of the default character tables. They
5269 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5270 you will need to set these values.
5271
5272 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5273
5274
5275 Version 3.9 02-Jan-02
5276 ---------------------
5277
5278 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5279
5280 2. If --disable-static was given, the building process failed when trying to
5281 build pcretest and pcregrep. (For some reason it was using libtool to compile
5282 them, which is not right, as they aren't part of the library.)
5283
5284
5285 Version 3.8 18-Dec-01
5286 ---------------------
5287
5288 1. The experimental UTF-8 code was completely screwed up. It was packing the
5289 bytes in the wrong order. How dumb can you get?
5290
5291
5292 Version 3.7 29-Oct-01
5293 ---------------------
5294
5295 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5296 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5297 this didn't happen under Solaris 8, where I normally test things.
5298
5299 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5300 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5301 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5302 this fix an hour or so after the initial 3.7 release.)
5303
5304
5305 Version 3.6 23-Oct-01
5306 ---------------------
5307
5308 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5309 offsets passed as NULL with zero offset count.
5310
5311 2. The config.guess and config.sub files had not been updated when I moved to
5312 the latest autoconf.
5313
5314
5315 Version 3.5 15-Aug-01
5316 ---------------------
5317
5318 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5319 had been forgotten.
5320
5321 2. By using declared but undefined structures, we can avoid using "void"
5322 definitions in pcre.h while keeping the internal definitions of the structures
5323 private.
5324
5325 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5326 user point of view, this means that both static and shared libraries are built
5327 by default, but this can be individually controlled. More of the work of
5328 handling this static/shared cases is now inside libtool instead of PCRE's make
5329 file.
5330
5331 4. The pcretest utility is now installed along with pcregrep because it is
5332 useful for users (to test regexs) and by doing this, it automatically gets
5333 relinked by libtool. The documentation has been turned into a man page, so
5334 there are now .1, .txt, and .html versions in /doc.
5335
5336 5. Upgrades to pcregrep:
5337 (i) Added long-form option names like gnu grep.
5338 (ii) Added --help to list all options with an explanatory phrase.
5339 (iii) Added -r, --recursive to recurse into sub-directories.
5340 (iv) Added -f, --file to read patterns from a file.
5341
5342 6. pcre_exec() was referring to its "code" argument before testing that
5343 argument for NULL (and giving an error if it was NULL).
5344
5345 7. Upgraded Makefile.in to allow for compiling in a different directory from
5346 the source directory.
5347
5348 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5349 options bits, the pointer it was passed was to an int instead of to an unsigned
5350 long int. This mattered only on 64-bit systems.
5351
5352 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5353 generated) instead of pcre.in, which it its source. Also made the same change
5354 in several of the .c files.
5355
5356 10. A new release of gcc defines printf() as a macro, which broke pcretest
5357 because it had an ifdef in the middle of a string argument for printf(). Fixed
5358 by using separate calls to printf().
5359
5360 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5361 script, to force use of CR or LF instead of \n in the source. On non-Unix
5362 systems, the value can be set in config.h.
5363
5364 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5365 absolute limit. Changed the text of the error message to make this clear, and
5366 likewise updated the man page.
5367
5368 13. The limit of 99 on the number of capturing subpatterns has been removed.
5369 The new limit is 65535, which I hope will not be a "real" limit.
5370
5371
5372 Version 3.4 22-Aug-00
5373 ---------------------
5374
5375 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5376
5377 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5378
5379
5380 Version 3.3 01-Aug-00
5381 ---------------------
5382
5383 1. If an octal character was given, but the value was greater than \377, it
5384 was not getting masked to the least significant bits, as documented. This could
5385 lead to crashes in some systems.
5386
5387 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5388 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5389
5390 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5391 These just pass their arguments on to (pcre_free)(), but they are provided
5392 because some uses of PCRE bind it to non-C systems that can call its functions,
5393 but cannot call free() or pcre_free() directly.
5394
5395 4. Add "make test" as a synonym for "make check". Corrected some comments in
5396 the Makefile.
5397
5398 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5399 Makefile.
5400
5401 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5402 command called pgrep for grepping around the active processes.
5403
5404 7. Added the beginnings of support for UTF-8 character strings.
5405
5406 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5407 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5408 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5409 out for the ar command.)
5410
5411
5412 Version 3.2 12-May-00
5413 ---------------------
5414
5415 This is purely a bug fixing release.
5416
5417 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5418 of ZA. This was just one example of several cases that could provoke this bug,
5419 which was introduced by change 9 of version 2.00. The code for breaking
5420 infinite loops after an iteration that matches an empty string was't working
5421 correctly.
5422
5423 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5424 when matched against abbab (for example). After matching an empty string, it
5425 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5426 caused it to match further down the string than it should.
5427
5428 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5429 was there because it doesn't seem to be needed, and it causes trouble on some
5430 systems, as it is not a Standard C header. It has been removed.
5431
5432 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5433 were reported on the Macintosh. The changes were from
5434
5435 while ((c = *(++ptr)) != 0 && c != '\n');
5436 to
5437 while ((c = *(++ptr)) != 0 && c != '\n') ;
5438
5439 Totally extraordinary, but if that's what it takes...
5440
5441 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5442 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5443 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5444 assumes the way PCRE uses memmove() (always moving upwards).
5445
5446 6. PCRE is being used in one environment where strchr() is not available. There
5447 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5448 faster code anyway.
5449
5450
5451 Version 3.1 09-Feb-00
5452 ---------------------
5453
5454 The only change in this release is the fixing of some bugs in Makefile.in for
5455 the "install" target:
5456
5457 (1) It was failing to install pcreposix.h.
5458
5459 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5460
5461
5462 Version 3.0 01-Feb-00
5463 ---------------------
5464
5465 1. Add support for the /+ modifier to perltest (to output $` like it does in
5466 pcretest).
5467
5468 2. Add support for the /g modifier to perltest.
5469
5470 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5471 matches null strings.
5472
5473 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5474 pattern. Perl treats empty patterns specially - it reuses the most recent
5475 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5476 effect.
5477
5478 5. The POSIX interface was broken in that it was just handing over the POSIX
5479 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5480 required a bigger vector, with some working space on the end. This means that
5481 the POSIX wrapper now has to get and free some memory, and copy the results.
5482
5483 6. Added some simple autoconf support, placing the test data and the
5484 documentation in separate directories, re-organizing some of the
5485 information files, and making it build pcre-config (a GNU standard). Also added
5486 libtool support for building PCRE as a shared library, which is now the
5487 default.
5488
5489 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5490 09 are not valid octal constants. Single digits will be used for minor values
5491 less than 10.
5492
5493 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5494 existing programs that set these in the POSIX interface can use PCRE without
5495 modification.
5496
5497 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5498 return all that pcre_info() returns, plus additional data. The pcre_info()
5499 function is retained for compatibility, but is considered to be obsolete.
5500
5501 10. Added experimental recursion feature (?R) to handle one common case that
5502 Perl 5.6 will be able to do with (?p{...}).
5503
5504 11. Added support for POSIX character classes like [:alpha:], which Perl is
5505 adopting.
5506
5507
5508 Version 2.08 31-Aug-99
5509 ----------------------
5510
5511 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5512 trying to match at the startoffset position, but instead was moving forward to
5513 the next newline as if a previous match had failed.
5514
5515 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5516 and could get into a loop if a null string was matched other than at the start
5517 of the subject.
5518
5519 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5520 be distinguished at compile time, and for completeness also added PCRE_DATE.
5521
5522 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5523 in GnuWin32 environments.
5524
5525
5526 Version 2.07 29-Jul-99
5527 ----------------------
5528
5529 1. The documentation is now supplied in plain text form and HTML as well as in
5530 the form of man page sources.
5531
5532 2. C++ compilers don't like assigning (void *) values to other pointer types.
5533 In particular this affects malloc(). Although there is no problem in Standard
5534 C, I've put in casts to keep C++ compilers happy.
5535
5536 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5537 should be (const char *).
5538
5539 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5540 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5541 However, I haven't made this a standard facility. The documentation doesn't
5542 mention it, and the Makefile doesn't support it.
5543
5544 5. The Makefile now contains an "install" target, with editable destinations at
5545 the top of the file. The pcretest program is not installed.
5546
5547 6. pgrep -V now gives the PCRE version number and date.
5548
5549 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5550 causing the entire string to be ignored, instead of just the last character.
5551
5552 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5553 non-matching string, it can take a very, very long time, even for strings of
5554 quite modest length, because of the nested recursion. PCRE now does better in
5555 some of these cases. It does this by remembering the last required literal
5556 character in the pattern, and pre-searching the subject to ensure it is present
5557 before running the real match. In other words, it applies a heuristic to detect
5558 some types of certain failure quickly, and in the above example, if presented
5559 with a string that has no trailing " it gives "no match" very quickly.
5560
5561 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5562 other alternatives are tried instead.
5563
5564
5565 Version 2.06 09-Jun-99
5566 ----------------------
5567
5568 1. Change pcretest's output for amount of store used to show just the code
5569 space, because the remainder (the data block) varies in size between 32-bit and
5570 64-bit systems.
5571
5572 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5573 start matching at. This allows lookbehinds to work when searching for multiple
5574 occurrences in a string.
5575
5576 3. Added additional options to pcretest for testing multiple occurrences:
5577
5578 /+ outputs the rest of the string that follows a match
5579 /g loops for multiple occurrences, using the new startoffset argument
5580 /G loops for multiple occurrences by passing an incremented pointer
5581
5582 4. PCRE wasn't doing the "first character" optimization for patterns starting
5583 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5584 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5585 the letter 'x'. On long subject strings, this gives a significant speed-up.
5586
5587
5588 Version 2.05 21-Apr-99
5589 ----------------------
5590
5591 1. Changed the type of magic_number from int to long int so that it works
5592 properly on 16-bit systems.
5593
5594 2. Fixed a bug which caused patterns starting with .* not to work correctly
5595 when the subject string contained newline characters. PCRE was assuming
5596 anchoring for such patterns in all cases, which is not correct because .* will
5597 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5598 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5599 must be retried after every newline in the subject.
5600
5601
5602 Version 2.04 18-Feb-99
5603 ----------------------
5604
5605 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5606 computation of the store needed to hold the pattern was incorrect (too large).
5607 If such patterns were nested a few deep, this could multiply and become a real
5608 problem.
5609
5610 2. Added /M option to pcretest to show the memory requirement of a specific
5611 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5612
5613 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5614 compiled in such a way that the backtracking after subsequent failure was
5615 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5616 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5617
5618
5619 Version 2.03 02-Feb-99
5620 ----------------------
5621
5622 1. Fixed typo and small mistake in man page.
5623
5624 2. Added 4th condition (GPL supersedes if conflict) and created separate
5625 LICENCE file containing the conditions.
5626
5627 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5628 Perl, that is the internal \ allows the delimiter to be included in the
5629 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5630 the final delimiter, add \ to the end of the pattern (to test the error).
5631
5632 4. Added the convenience functions for extracting substrings after a successful
5633 match. Updated pcretest to make it able to test these functions.
5634
5635
5636 Version 2.02 14-Jan-99
5637 ----------------------
5638
5639 1. Initialized the working variables associated with each extraction so that
5640 their saving and restoring doesn't refer to uninitialized store.
5641
5642 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5643 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5644 fix the problem.
5645
5646 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5647 calls, and wasn't printing the correct value for compiling calls. Increased the
5648 default value of LOOPREPEAT, and the number of significant figures in the
5649 times.
5650
5651 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5652
5653 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5654 a building problem on Windows NT with a FAT file system.
5655
5656
5657 Version 2.01 21-Oct-98
5658 ----------------------
5659
5660 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5661 to character tables built by pcre_maketables() in the current locale. If NULL
5662 is passed, the default tables are used.
5663
5664
5665 Version 2.00 24-Sep-98
5666 ----------------------
5667
5668 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5669 it any more.
5670
5671 2. Allow quantification of (?>) groups, and make it work correctly.
5672
5673 3. The first character computation wasn't working for (?>) groups.
5674
5675 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5676 end of the subject) and add 5.005's \z, which really does match only at the
5677 very end of the subject.
5678
5679 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5680
5681 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5682 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5683 localized options. All options to pcre_study() were also removed.
5684
5685 7. Add other new features from 5.005:
5686
5687 $(?<= positive lookbehind
5688 $(?<! negative lookbehind
5689 (?imsx-imsx) added the unsetting capability
5690 such a setting is global if at outer level; local otherwise
5691 (?imsx-imsx:) non-capturing groups with option setting
5692 (?(cond)re|re) conditional pattern matching
5693
5694 A backreference to itself in a repeated group matches the previous
5695 captured string.
5696
5697 8. General tidying up of studying (both automatic and via "study")
5698 consequential on the addition of new assertions.
5699
5700 9. As in 5.005, unlimited repeated groups that could match an empty substring
5701 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5702 runtime if any iteration does actually match an empty substring.
5703
5704 10. Include the RunTest script in the distribution.
5705
5706 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5707 discrepancies, some of which were old and were also with respect to 5.004. They
5708 have now been fixed.
5709
5710
5711 Version 1.09 28-Apr-98
5712 ----------------------
5713
5714 1. A negated single character class followed by a quantifier with a minimum
5715 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5716 program crashes, or just wrong answers. This did not apply to negated classes
5717 containing more than one character, or to minima other than one.
5718
5719
5720 Version 1.08 27-Mar-98
5721 ----------------------
5722
5723 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5724
5725 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5726 latter must appear before anything that relies on it in the pattern.
5727
5728
5729 Version 1.07 16-Feb-98
5730 ----------------------
5731
5732 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5733 repeat of a potentially empty string).
5734
5735
5736 Version 1.06 23-Jan-98
5737 ----------------------
5738
5739 1. Added Markus Oberhumer's little patches for C++.
5740
5741 2. Literal strings longer than 255 characters were broken.
5742
5743
5744 Version 1.05 23-Dec-97
5745 ----------------------
5746
5747 1. Negated character classes containing more than one character were failing if
5748 PCRE_CASELESS was set at run time.
5749
5750
5751 Version 1.04 19-Dec-97
5752 ----------------------
5753
5754 1. Corrected the man page, where some "const" qualifiers had been omitted.
5755
5756 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5757 input syntax.
5758
5759 3. Fixed memory leak which occurred when a regex with back references was
5760 matched with an offsets vector that wasn't big enough. The temporary memory
5761 that is used in this case wasn't being freed if the match failed.
5762
5763 4. Tidied pcretest to ensure it frees memory that it gets.
5764
5765 5. Temporary memory was being obtained in the case where the passed offsets
5766 vector was exactly big enough.
5767
5768 6. Corrected definition of offsetof() from change 5 below.
5769
5770 7. I had screwed up change 6 below and broken the rules for the use of
5771 setjmp(). Now fixed.
5772
5773
5774 Version 1.03 18-Dec-97
5775 ----------------------
5776
5777 1. A erroneous regex with a missing opening parenthesis was correctly
5778 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5779 on some systems.
5780
5781 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5782 it was reported that one broken compiler failed on the former because "code" is
5783 also an independent variable.
5784
5785 3. The erroneous regex a[]b caused an array overrun reference.
5786
5787 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5788 fail on data ending with that character. (It was going on too far, and checking
5789 the next character, typically a binary zero.) This was specific to the
5790 optimized code for single-character negative classes.
5791
5792 5. Added a contributed patch from the TIN world which does the following:
5793
5794 + Add an undef for memmove, in case the the system defines a macro for it.
5795
5796 + Add a definition of offsetof(), in case there isn't one. (I don't know
5797 the reason behind this - offsetof() is part of the ANSI standard - but
5798 it does no harm).
5799
5800 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5801 most of the places where whitespace preceded '#'. I have given up and
5802 allowed the remaining 2 cases to be at the margin.
5803
5804 + Rename some variables in pcre to eliminate shadowing. This seems very
5805 pedantic, but does no harm, of course.
5806
5807 6. Moved the call to setjmp() into its own function, to get rid of warnings
5808 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5809
5810 7. Constructs such as \d{8,} were compiling into the equivalent of
5811 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5812 outcome, but in this particular case used more store than had been allocated,
5813 which caused the bug to be discovered because it threw up an internal error.
5814
5815 8. The debugging code in both pcre and pcretest for outputting the compiled
5816 form of a regex was going wrong in the case of back references followed by
5817 curly-bracketed repeats.
5818
5819
5820 Version 1.02 12-Dec-97
5821 ----------------------
5822
5823 1. Typos in pcre.3 and comments in the source fixed.
5824
5825 2. Applied a contributed patch to get rid of places where it used to remove
5826 'const' from variables, and fixed some signed/unsigned and uninitialized
5827 variable warnings.
5828
5829 3. Added the "runtest" target to Makefile.
5830
5831 4. Set default compiler flag to -O2 rather than just -O.
5832
5833
5834 Version 1.01 19-Nov-97
5835 ----------------------
5836
5837 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5838 like /([ab]*)*/, that is, for classes with more than one character in them.
5839
5840 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5841 as /((?>a*))*/ (a PCRE_EXTRA facility).
5842
5843
5844 Version 1.00 18-Nov-97
5845 ----------------------
5846
5847 1. Added compile-time macros to support systems such as SunOS4 which don't have
5848 memmove() or strerror() but have other things that can be used instead.
5849
5850 2. Arranged that "make clean" removes the executables.
5851
5852
5853 Version 0.99 27-Oct-97
5854 ----------------------
5855
5856 1. Fixed bug in code for optimizing classes with only one character. It was
5857 initializing a 32-byte map regardless, which could cause it to run off the end
5858 of the memory it had got.
5859
5860 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5861
5862
5863 Version 0.98 22-Oct-97
5864 ----------------------
5865
5866 1. Fixed bug in code for handling temporary memory usage when there are more
5867 back references than supplied space in the ovector. This could cause segfaults.
5868
5869
5870 Version 0.97 21-Oct-97
5871 ----------------------
5872
5873 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5874
5875 2. Optimized negated single characters not to use a bit map.
5876
5877 3. Brought error texts together as macro definitions; clarified some of them;
5878 fixed one that was wrong - it said "range out of order" when it meant "invalid
5879 escape sequence".
5880
5881 4. Changed some char * arguments to const char *.
5882
5883 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5884
5885 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5886 pcretest.
5887
5888
5889 Version 0.96 16-Oct-97
5890 ----------------------
5891
5892 1. Added a simple "pgrep" utility to the distribution.
5893
5894 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5895 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5896 where "ddd" means "one or more decimal digits".
5897
5898 3. Fixed serious bug. If a pattern had a back reference, but the call to
5899 pcre_exec() didn't supply a large enough ovector to record the related
5900 identifying subpattern, the match always failed. PCRE now remembers the number
5901 of the largest back reference, and gets some temporary memory in which to save
5902 the offsets during matching if necessary, in order to ensure that
5903 backreferences always work.
5904
5905 4. Increased the compatibility with Perl in a number of ways:
5906
5907 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5908 to request this handling. The option can be set at compile or exec time.
5909
5910 (b) $ matches before a terminating newline by default; an option
5911 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5912 mode). The option can be set at compile or exec time.
5913
5914 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5915 the same as Perl's. If the decimal number it represents is less than 10
5916 or there aren't that many previous left capturing parentheses, an octal
5917 escape is read. Inside a character class, it's always an octal escape,
5918 even if it is a single digit.
5919
5920 (d) An escaped but undefined alphabetic character is taken as a literal,
5921 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5922 escapes.
5923
5924 (e) {0} is now permitted. (The previous item is removed from the compiled
5925 pattern).
5926
5927 5. Changed all the names of code files so that the basic parts are no longer
5928 than 10 characters, and abolished the teeny "globals.c" file.
5929
5930 6. Changed the handling of character classes; they are now done with a 32-byte
5931 bit map always.
5932
5933 7. Added the -d and /D options to pcretest to make it possible to look at the
5934 internals of compilation without having to recompile pcre.
5935
5936
5937 Version 0.95 23-Sep-97
5938 ----------------------
5939
5940 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5941 \x20 at the start of a run of normal characters. These were being treated as
5942 real characters, instead of the source characters being re-checked.
5943
5944
5945 Version 0.94 18-Sep-97
5946 ----------------------
5947
5948 1. The functions are now thread-safe, with the caveat that the global variables
5949 containing pointers to malloc() and free() or alternative functions are the
5950 same for all threads.
5951
5952 2. Get pcre_study() to generate a bitmap of initial characters for non-
5953 anchored patterns when this is possible, and use it if passed to pcre_exec().
5954
5955
5956 Version 0.93 15-Sep-97
5957 ----------------------
5958
5959 1. /(b)|(:+)/ was computing an incorrect first character.
5960
5961 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5962 but not actually doing anything yet.
5963
5964 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5965 as Perl does (e.g. [-az] or [az-]).
5966
5967 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5968 all possible positions.
5969
5970 5. Split up into different modules to avoid including unneeded functions in a
5971 compiled binary. However, compile and exec are still in one module. The "study"
5972 function is split off.
5973
5974 6. The character tables are now in a separate module whose source is generated
5975 by an auxiliary program - but can then be edited by hand if required. There are
5976 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5977 toupper() in the code.
5978
5979 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5980 make them global. Abolish the function for setting them, as the caller can now
5981 set them directly.
5982
5983
5984 Version 0.92 11-Sep-97
5985 ----------------------
5986
5987 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5988 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5989
5990 2. Caseless matching was not working in character classes if the characters in
5991 the pattern were in upper case.
5992
5993 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5994
5995 4. Make PCRE_ANCHORED public and accept as a compile option.
5996
5997 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5998 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5999 pass them.
6000
6001 6. Give an error if bad option bits passed at compile or run time.
6002
6003 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
6004 pcretest to cause it to pass that flag.
6005
6006 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
6007 options, and the first character, if set.
6008
6009 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
6010
6011
6012 Version 0.91 10-Sep-97
6013 ----------------------
6014
6015 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
6016 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
6017
6018 2. PCRE was looping on encountering an indefinitely repeated back reference to
6019 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
6020 Perl does - treats the match as successful.
6021
6022 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5