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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1625 - (show annotations)
Sat Feb 6 16:54:14 2016 UTC (11 hours, 6 minutes ago) by ph10
File size: 287749 byte(s)
Error occurred while calculating annotation data.
Fix pcretest bad behaviour for callout in lookbehind.
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 8.39 xx-xxxxxx-201x
8 ---------------------------
9
10 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
11 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
12 misbehaved. This bug was found by the LLVM fuzzer.
13
14 2. Similar to the above, if an isolated \E was present between an item and its
15 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
16 bug was found by the LLVM fuzzer.
17
18 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
19 working correctly in UCP mode.
20
21 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
22 was set when the pmatch argument was NULL. It now returns REG_INVARG.
23
24 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
25
26 6. An empty \Q\E sequence between an item and its qualifier caused
27 pcre_compile() to misbehave when auto callouts were enabled. This bug was
28 found by the LLVM fuzzer.
29
30 7. If a pattern that was compiled with PCRE_EXTENDED started with white
31 space or a #-type comment that was followed by (?-x), which turns off
32 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
33 pcre_compile() assumed that (?-x) applied to the whole pattern and
34 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
35
36 8. An call of pcre_copy_named_substring() for a named substring whose number
37 was greater than the space in the ovector could cause a crash.
38
39 9. Yet another buffer overflow bug involved duplicate named groups with a
40 group that reset capture numbers (compare 8.38/7 below). Once again, I have
41 just allowed for more memory, even if not needed. (A proper fix is
42 implemented in PCRE2, but it involves a lot of refactoring.)
43
44 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
45 start of the match to be earlier than the end.
46
47 11. JIT code generator for assertion matching is refactored to a separate
48 function.
49
50 12. Assertion code generator in JIT has been optimized.
51
52 13. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
53 assertion, caused pcre2test to generate incorrect output, and also to read
54 uninitialized memory (detected by ASAN or valgrind).
55
56
57 Version 8.38 23-November-2015
58 -----------------------------
59
60 1. If a group that contained a recursive back reference also contained a
61 forward reference subroutine call followed by a non-forward-reference
62 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
63 compile correct code, leading to undefined behaviour or an internally
64 detected error. This bug was discovered by the LLVM fuzzer.
65
66 2. Quantification of certain items (e.g. atomic back references) could cause
67 incorrect code to be compiled when recursive forward references were
68 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
69 This bug was discovered by the LLVM fuzzer.
70
71 3. A repeated conditional group whose condition was a reference by name caused
72 a buffer overflow if there was more than one group with the given name.
73 This bug was discovered by the LLVM fuzzer.
74
75 4. A recursive back reference by name within a group that had the same name as
76 another group caused a buffer overflow. For example:
77 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
78
79 5. A forward reference by name to a group whose number is the same as the
80 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
81 a buffer overflow at compile time. This bug was discovered by the LLVM
82 fuzzer.
83
84 6. A lookbehind assertion within a set of mutually recursive subpatterns could
85 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
86
87 7. Another buffer overflow bug involved duplicate named groups with a
88 reference between their definition, with a group that reset capture
89 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
90 fixed by always allowing for more memory, even if not needed. (A proper fix
91 is implemented in PCRE2, but it involves more refactoring.)
92
93 8. There was no check for integer overflow in subroutine calls such as (?123).
94
95 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
96 being treated as a literal 'l' instead of causing an error.
97
98 10. There was a buffer overflow if pcre_exec() was called with an ovector of
99 size 1. This bug was found by american fuzzy lop.
100
101 11. If a non-capturing group containing a conditional group that could match
102 an empty string was repeated, it was not identified as matching an empty
103 string itself. For example: /^(?:(?(1)x|)+)+$()/.
104
105 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
106 \a and \e in test subject lines.
107
108 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
109 instead of the EBCDIC value.
110
111 14. The handling of \c in an EBCDIC environment has been revised so that it is
112 now compatible with the specification in Perl's perlebcdic page.
113
114 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
115 ASCII/Unicode. This has now been added to the list of characters that are
116 recognized as white space in EBCDIC.
117
118 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
119 error (correctly) when used outside a class, but did not give an error
120 within a class.
121
122 17. \h within a class was incorrectly compiled in EBCDIC environments.
123
124 18. A pattern with an unmatched closing parenthesis that contained a backward
125 assertion which itself contained a forward reference caused buffer
126 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
127
128 19. JIT should return with error when the compiled pattern requires more stack
129 space than the maximum.
130
131 20. A possessively repeated conditional group that could match an empty string,
132 for example, /(?(R))*+/, was incorrectly compiled.
133
134 21. Fix infinite recursion in the JIT compiler when certain patterns such as
135 /(?:|a|){100}x/ are analysed.
136
137 22. Some patterns with character classes involving [: and \\ were incorrectly
138 compiled and could cause reading from uninitialized memory or an incorrect
139 error diagnosis.
140
141 23. Pathological patterns containing many nested occurrences of [: caused
142 pcre_compile() to run for a very long time.
143
144 24. A conditional group with only one branch has an implicit empty alternative
145 branch and must therefore be treated as potentially matching an empty
146 string.
147
148 25. If (?R was followed by - or + incorrect behaviour happened instead of a
149 diagnostic.
150
151 26. Arrange to give up on finding the minimum matching length for overly
152 complex patterns.
153
154 27. Similar to (4) above: in a pattern with duplicated named groups and an
155 occurrence of (?| it is possible for an apparently non-recursive back
156 reference to become recursive if a later named group with the relevant
157 number is encountered. This could lead to a buffer overflow. Wen Guanxing
158 from Venustech ADLAB discovered this bug.
159
160 28. If pcregrep was given the -q option with -c or -l, or when handling a
161 binary file, it incorrectly wrote output to stdout.
162
163 29. The JIT compiler did not restore the control verb head in case of *THEN
164 control verbs. This issue was found by Karl Skomski with a custom LLVM
165 fuzzer.
166
167 30. Error messages for syntax errors following \g and \k were giving inaccurate
168 offsets in the pattern.
169
170 31. Added a check for integer overflow in conditions (?(<digits>) and
171 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
172 fuzzer.
173
174 32. Handling recursive references such as (?2) when the reference is to a group
175 later in the pattern uses code that is very hacked about and error-prone.
176 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
177 give an internal error if it is obvious that compiling has gone wrong.
178
179 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
180 This issue was found by Karl Skomski with a custom LLVM fuzzer.
181
182 34. The JIT compiler should restore the control chain for empty possessive
183 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
184
185 35. Match limit check added to JIT recursion. This issue was found by Karl
186 Skomski with a custom LLVM fuzzer.
187
188 36. Yet another case similar to 27 above has been circumvented by an
189 unconditional allocation of extra memory. This issue is fixed "properly" in
190 PCRE2 by refactoring the way references are handled. Wen Guanxing
191 from Venustech ADLAB discovered this bug.
192
193 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
194 with a custom LLVM fuzzer.
195
196 38. Fixed a corner case of range optimization in JIT.
197
198 39. An incorrect error "overran compiling workspace" was given if there were
199 exactly enough group forward references such that the last one extended
200 into the workspace safety margin. The next one would have expanded the
201 workspace. The test for overflow was not including the safety margin.
202
203 40. A match limit issue is fixed in JIT which was found by Karl Skomski
204 with a custom LLVM fuzzer.
205
206 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
207 work under Windows. (Why has it taken so long for anyone to notice?)
208
209 42. In a character class such as [\W\p{Any}] where both a negative-type escape
210 ("not a word character") and a property escape were present, the property
211 escape was being ignored.
212
213 43. Fix crash caused by very long (*MARK) or (*THEN) names.
214
215 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
216 by a single ASCII character in a class item, was incorrectly compiled in
217 UCP mode. The POSIX class got lost, but only if the single character
218 followed it.
219
220 45. [:punct:] in UCP mode was matching some characters in the range 128-255
221 that should not have been matched.
222
223 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
224 class, all characters with code points greater than 255 are in the class.
225 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
226 such as \w are turned into Unicode properties), wide characters were not
227 correctly handled, and could fail to match.
228
229
230 Version 8.37 28-April-2015
231 --------------------------
232
233 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
234 for those parentheses to be closed with whatever has been captured so far.
235 However, it was failing to mark any other groups between the hightest
236 capture so far and the currrent group as "unset". Thus, the ovector for
237 those groups contained whatever was previously there. An example is the
238 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
239
240 2. If an assertion condition was quantified with a minimum of zero (an odd
241 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
242
243 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
244 unrecognized modifier, a crash could occur.
245
246 4. An attempt to do global matching in pcretest with a zero-length ovector
247 caused a crash.
248
249 5. Fixed a memory leak during matching that could occur for a subpattern
250 subroutine call (recursive or otherwise) if the number of captured groups
251 that had to be saved was greater than ten.
252
253 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
254 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
255 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
256
257 7. A UTF pattern containing a "not" match of a non-ASCII character and a
258 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
259
260 8. When a pattern is compiled, it remembers the highest back reference so that
261 when matching, if the ovector is too small, extra memory can be obtained to
262 use instead. A conditional subpattern whose condition is a check on a
263 capture having happened, such as, for example in the pattern
264 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
265 setting the highest backreference number. This mattered only if pcre_exec()
266 was called with an ovector that was too small to hold the capture, and there
267 was no other kind of back reference (a situation which is probably quite
268 rare). The effect of the bug was that the condition was always treated as
269 FALSE when the capture could not be consulted, leading to a incorrect
270 behaviour by pcre_exec(). This bug has been fixed.
271
272 9. A reference to a duplicated named group (either a back reference or a test
273 for being set in a conditional) that occurred in a part of the pattern where
274 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
275 to be incorrectly calculated, leading to overwriting.
276
277 10. A mutually recursive set of back references such as (\2)(\1) caused a
278 segfault at study time (while trying to find the minimum matching length).
279 The infinite loop is now broken (with the minimum length unset, that is,
280 zero).
281
282 11. If an assertion that was used as a condition was quantified with a minimum
283 of zero, matching went wrong. In particular, if the whole group had
284 unlimited repetition and could match an empty string, a segfault was
285 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
286 assertions to be quantified, but not if they are being used as conditions,
287 so the above pattern is faulted by Perl. PCRE has now been changed so that
288 it also rejects such patterns.
289
290 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
291 failed to allow the zero-repeat case if pcre2_exec() was called with an
292 ovector too small to capture the group.
293
294 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
295 Red Hat Product Security:
296
297 (a) A crash if /K and /F were both set with the option to save the compiled
298 pattern.
299
300 (b) Another crash if the option to print captured substrings in a callout
301 was combined with setting a null ovector, for example \O\C+ as a subject
302 string.
303
304 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
305 forward reference repeated a large (but limited) number of times within a
306 repeated outer group that has a zero minimum quantifier, caused incorrect
307 code to be compiled, leading to the error "internal error:
308 previously-checked referenced subpattern not found" when an incorrect
309 memory address was read. This bug was reported as "heap overflow",
310 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
311 CVE-2015-2325.
312
313 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
314 call within a group that also contained a recursive back reference caused
315 incorrect code to be compiled. This bug was reported as "heap overflow",
316 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
317 number CVE-2015-2326.
318
319 24. Computing the size of the JIT read-only data in advance has been a source
320 of various issues, and new ones are still appear unfortunately. To fix
321 existing and future issues, size computation is eliminated from the code,
322 and replaced by on-demand memory allocation.
323
324 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
325 adjacent to the end of the range, and the range contained characters with
326 more than one other case, caused incorrect behaviour when compiled in UTF
327 mode. In that example, the range a-j was left out of the class.
328
329 26. Fix JIT compilation of conditional blocks, which assertion
330 is converted to (*FAIL). E.g: /(?(?!))/.
331
332 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
333 discovered by the LLVM fuzzer.
334
335 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
336 when this assertion was used as a condition, for example (?(?!)a|b). In
337 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
338 error about an unsupported item.
339
340 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
341 possessification code could take exponential time to complete. A recursion
342 depth limit of 1000 has been imposed to limit the resources used by this
343 optimization.
344
345 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
346 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
347 because \S ensures they are all in the class. The code for doing this was
348 interacting badly with the code for computing the amount of space needed to
349 compile the pattern, leading to a buffer overflow. This bug was discovered
350 by the LLVM fuzzer.
351
352 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
353 other kinds of group caused stack overflow at compile time. This bug was
354 discovered by the LLVM fuzzer.
355
356 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
357 between a subroutine call and its quantifier was incorrectly compiled,
358 leading to buffer overflow or other errors. This bug was discovered by the
359 LLVM fuzzer.
360
361 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
362 assertion after (?(. The code was failing to check the character after
363 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
364 was discovered by the LLVM fuzzer.
365
366 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
367 a fixed maximum following a group that contains a subroutine reference was
368 incorrectly compiled and could trigger buffer overflow. This bug was
369 discovered by the LLVM fuzzer.
370
371 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
372 caused a stack overflow instead of the diagnosis of a non-fixed length
373 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
374
375 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
376 (e.g. /(?<=\Ka)/) could make pcregrep loop.
377
378 37. There was a similar problem to 36 in pcretest for global matches.
379
380 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
381 and a subsequent item in the pattern caused a non-match, backtracking over
382 the repeated \X did not stop, but carried on past the start of the subject,
383 causing reference to random memory and/or a segfault. There were also some
384 other cases where backtracking after \C could crash. This set of bugs was
385 discovered by the LLVM fuzzer.
386
387 39. The function for finding the minimum length of a matching string could take
388 a very long time if mutual recursion was present many times in a pattern,
389 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
390 method has been implemented. This infelicity was discovered by the LLVM
391 fuzzer.
392
393 40. Static linking against the PCRE library using the pkg-config module was
394 failing on missing pthread symbols.
395
396
397 Version 8.36 26-September-2014
398 ------------------------------
399
400 1. Got rid of some compiler warnings in the C++ modules that were shown up by
401 -Wmissing-field-initializers and -Wunused-parameter.
402
403 2. The tests for quantifiers being too big (greater than 65535) were being
404 applied after reading the number, and stupidly assuming that integer
405 overflow would give a negative number. The tests are now applied as the
406 numbers are read.
407
408 3. Tidy code in pcre_exec.c where two branches that used to be different are
409 now the same.
410
411 4. The JIT compiler did not generate match limit checks for certain
412 bracketed expressions with quantifiers. This may lead to exponential
413 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
414 issue should be resolved now.
415
416 5. Fixed an issue, which occures when nested alternatives are optimized
417 with table jumps.
418
419 6. Inserted two casts and changed some ints to size_t in the light of some
420 reported 64-bit compiler warnings (Bugzilla 1477).
421
422 7. Fixed a bug concerned with zero-minimum possessive groups that could match
423 an empty string, which sometimes were behaving incorrectly in the
424 interpreter (though correctly in the JIT matcher). This pcretest input is
425 an example:
426
427 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
428 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
429
430 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
431 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
432 for an empty string was breaking the inner loop and carrying on at a lower
433 level, when possessive repeated groups should always return to a higher
434 level as they have no backtrack points in them. The empty string test now
435 occurs at the outer level.
436
437 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
438 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
439
440 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
441 doesn't).
442
443 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
444 to hold a pattern to be incorrectly computed (too small) when there were
445 named back references to duplicated names. This could cause "internal
446 error: code overflow" or "double free or corruption" or other memory
447 handling errors.
448
449 11. When named subpatterns had the same prefixes, back references could be
450 confused. For example, in this pattern:
451
452 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
453
454 the reference to 'Name' was incorrectly treated as a reference to a
455 duplicate name.
456
457 12. A pattern such as /^s?c/mi8 where the optional character has more than
458 one "other case" was incorrectly compiled such that it would only try to
459 match starting at "c".
460
461 13. When a pattern starting with \s was studied, VT was not included in the
462 list of possible starting characters; this should have been part of the
463 8.34/18 patch.
464
465 14. If a character class started [\Qx]... where x is any character, the class
466 was incorrectly terminated at the ].
467
468 15. If a pattern that started with a caseless match for a character with more
469 than one "other case" was studied, PCRE did not set up the starting code
470 unit bit map for the list of possible characters. Now it does. This is an
471 optimization improvement, not a bug fix.
472
473 16. The Unicode data tables have been updated to Unicode 7.0.0.
474
475 17. Fixed a number of memory leaks in pcregrep.
476
477 18. Avoid a compiler warning (from some compilers) for a function call with
478 a cast that removes "const" from an lvalue by using an intermediate
479 variable (to which the compiler does not object).
480
481 19. Incorrect code was compiled if a group that contained an internal recursive
482 back reference was optional (had quantifier with a minimum of zero). This
483 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
484 caused segmentation faults because of stack overflows at compile time.
485
486 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
487 group that is quantified with an indefinite repeat, caused a compile-time
488 loop which used up all the system stack and provoked a segmentation fault.
489 This was not the same bug as 19 above.
490
491 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
492 Patch by Mike Frysinger.
493
494
495 Version 8.35 04-April-2014
496 --------------------------
497
498 1. A new flag is set, when property checks are present in an XCLASS.
499 When this flag is not set, PCRE can perform certain optimizations
500 such as studying these XCLASS-es.
501
502 2. The auto-possessification of character sets were improved: a normal
503 and an extended character set can be compared now. Furthermore
504 the JIT compiler optimizes more character set checks.
505
506 3. Got rid of some compiler warnings for potentially uninitialized variables
507 that show up only when compiled with -O2.
508
509 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
510 of a match later then the end of the match. The pcretest program was not
511 handling the case sensibly - it was outputting from the start to the next
512 binary zero. It now reports this situation in a message, and outputs the
513 text from the end to the start.
514
515 5. Fast forward search is improved in JIT. Instead of the first three
516 characters, any three characters with fixed position can be searched.
517 Search order: first, last, middle.
518
519 6. Improve character range checks in JIT. Characters are read by an inprecise
520 function now, which returns with an unknown value if the character code is
521 above a certain threshold (e.g: 256). The only limitation is that the value
522 must be bigger than the threshold as well. This function is useful when
523 the characters above the threshold are handled in the same way.
524
525 7. The macros whose names start with RAWUCHAR are placeholders for a future
526 mode in which only the bottom 21 bits of 32-bit data items are used. To
527 make this more memorable for those maintaining the code, the names have
528 been changed to start with UCHAR21, and an extensive comment has been added
529 to their definition.
530
531 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
532 to the export list in Makefile.am (they were accidentally omitted from the
533 8.34 tarball).
534
535 9. The informational output from pcretest used the phrase "starting byte set"
536 which is inappropriate for the 16-bit and 32-bit libraries. As the output
537 for "first char" and "need char" really means "non-UTF-char", I've changed
538 "byte" to "char", and slightly reworded the output. The documentation about
539 these values has also been (I hope) clarified.
540
541 10. Another JIT related optimization: use table jumps for selecting the correct
542 backtracking path, when more than four alternatives are present inside a
543 bracket.
544
545 11. Empty match is not possible, when the minimum length is greater than zero,
546 and there is no \K in the pattern. JIT should avoid empty match checks in
547 such cases.
548
549 12. In a caseless character class with UCP support, when a character with more
550 than one alternative case was not the first character of a range, not all
551 the alternative cases were added to the class. For example, s and \x{17f}
552 are both alternative cases for S: the class [RST] was handled correctly,
553 but [R-T] was not.
554
555 13. The configure.ac file always checked for pthread support when JIT was
556 enabled. This is not used in Windows, so I have put this test inside a
557 check for the presence of windows.h (which was already tested for).
558
559 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
560 The algorithm provides a way to skip certain starting offsets, and usually
561 faster than linear prefix searches.
562
563 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
564 as for 'fr_FR' and 'french'. For some reason, however, it then used the
565 Windows-specific input and output files, which have 'french' screwed in.
566 So this could never have worked. One of the problems with locales is that
567 they aren't always the same. I have now updated RunTest so that it checks
568 the output of the locale test (test 3) against three different output
569 files, and it allows the test to pass if any one of them matches. With luck
570 this should make the test pass on some versions of Solaris where it was
571 failing. Because of the uncertainty, the script did not used to stop if
572 test 3 failed; it now does. If further versions of a French locale ever
573 come to light, they can now easily be added.
574
575 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
576 there was a message during ./configure, but it did not stop. This now
577 provokes an error. The invalid example in README has been corrected.
578 If a value less than the minimum is given, the minimum value has always
579 been used, but now a warning is given.
580
581 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
582 was a bug in the test system, which is now fixed. Also, the list of various
583 configurations that are tested for each release did not have one with both
584 16/32 bits and --enable-bar-anycrlf. It now does.
585
586 18. pcretest was missing "-C bsr" for displaying the \R default setting.
587
588 19. Little endian PowerPC systems are supported now by the JIT compiler.
589
590 20. The fast forward newline mechanism could enter to an infinite loop on
591 certain invalid UTF-8 input. Although we don't support these cases
592 this issue can be fixed by a performance optimization.
593
594 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
595 not take account if existing stack usage. There is now a new global
596 variable called pcre_stack_guard that can be set to point to an external
597 function to check stack availability. It is called at the start of
598 processing every parenthesized group.
599
600 22. A typo in the code meant that in ungreedy mode the max/min qualifier
601 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
602 not match "ab".
603
604 23. When UTF was disabled, the JIT program reported some incorrect compile
605 errors. These messages are silenced now.
606
607 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
608 compiler.
609
610 25. Change all the temporary files used in RunGrepTest to be different to those
611 used by RunTest so that the tests can be run simultaneously, for example by
612 "make -j check".
613
614
615 Version 8.34 15-December-2013
616 -----------------------------
617
618 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
619 executable memory. Patch inspired by Carsten Klein.
620
621 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
622 this macro is never tested and has no effect, because the work to support
623 coverage involves only compiling and linking options and special targets in
624 the Makefile. The comment in config.h implied that defining the macro would
625 enable coverage support, which is totally false. There was also support for
626 setting this macro in the CMake files (my fault, I just copied it from
627 configure). SUPPORT_GCOV has now been removed.
628
629 3. Make a small performance improvement in strlen16() and strlen32() in
630 pcretest.
631
632 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
633 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
634 these cases). There was also one in pcretest.c.
635
636 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
637
638 6. In UTF mode, the code for checking whether a group could match an empty
639 string (which is used for indefinitely repeated groups to allow for
640 breaking an infinite loop) was broken when the group contained a repeated
641 negated single-character class with a character that occupied more than one
642 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
643 UTF-8 mode). The effect was undefined: the group might or might not be
644 deemed as matching an empty string, or the program might have crashed.
645
646 7. The code for checking whether a group could match an empty string was not
647 recognizing that \h, \H, \v, \V, and \R must match a character.
648
649 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
650 an empty string. If it can, pcretest shows this in its information output.
651
652 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
653 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
654 matched by pcre_exec() without using JIT:
655
656 (a) If the rest of the pattern did not match after a maximal run of
657 grapheme clusters, the code for backing up to try with fewer of them
658 did not always back up over a full grapheme when characters that do not
659 have the modifier quality were involved, e.g. Hangul syllables.
660
661 (b) If the match point in a subject started with modifier character, and
662 there was no match, the code could incorrectly back up beyond the match
663 point, and potentially beyond the first character in the subject,
664 leading to a segfault or an incorrect match result.
665
666 10. A conditional group with an assertion condition could lead to PCRE
667 recording an incorrect first data item for a match if no other first data
668 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
669 first data item, and therefore matched "ca" after "c" instead of at the
670 start.
671
672 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
673 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
674
675 12. The source of pcregrep now includes z/OS-specific code so that it can be
676 compiled for z/OS as part of the special z/OS distribution.
677
678 13. Added the -T and -TM options to pcretest.
679
680 14. The code in pcre_compile.c for creating the table of named capturing groups
681 has been refactored. Instead of creating the table dynamically during the
682 actual compiling pass, the information is remembered during the pre-compile
683 pass (on the stack unless there are more than 20 named groups, in which
684 case malloc() is used) and the whole table is created before the actual
685 compile happens. This has simplified the code (it is now nearly 150 lines
686 shorter) and prepared the way for better handling of references to groups
687 with duplicate names.
688
689 15. A back reference to a named subpattern when there is more than one of the
690 same name now checks them in the order in which they appear in the pattern.
691 The first one that is set is used for the reference. Previously only the
692 first one was inspected. This change makes PCRE more compatible with Perl.
693
694 16. Unicode character properties were updated from Unicode 6.3.0.
695
696 17. The compile-time code for auto-possessification has been refactored, based
697 on a patch by Zoltan Herczeg. It now happens after instead of during
698 compilation. The code is cleaner, and more cases are handled. The option
699 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
700 options in pcretest are provided to set it. It can also be set by
701 (*NO_AUTO_POSSESS) at the start of a pattern.
702
703 18. The character VT has been added to the default ("C" locale) set of
704 characters that match \s and are generally treated as white space,
705 following this same change in Perl 5.18. There is now no difference between
706 "Perl space" and "POSIX space". Whether VT is treated as white space in
707 other locales depends on the locale.
708
709 19. The code for checking named groups as conditions, either for being set or
710 for being recursed, has been refactored (this is related to 14 and 15
711 above). Processing unduplicated named groups should now be as fast at
712 numerical groups, and processing duplicated groups should be faster than
713 before.
714
715 20. Two patches to the CMake build system, by Alexander Barkov:
716
717 (1) Replace the "source" command by "." in CMakeLists.txt because
718 "source" is a bash-ism.
719
720 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
721 without these the CMake build does not work on Solaris.
722
723 21. Perl has changed its handling of \8 and \9. If there is no previously
724 encountered capturing group of those numbers, they are treated as the
725 literal characters 8 and 9 instead of a binary zero followed by the
726 literals. PCRE now does the same.
727
728 22. Following Perl, added \o{} to specify codepoints in octal, making it
729 possible to specify values greater than 0777 and also making them
730 unambiguous.
731
732 23. Perl now gives an error for missing closing braces after \x{... instead of
733 treating the string as literal. PCRE now does the same.
734
735 24. RunTest used to grumble if an inappropriate test was selected explicitly,
736 but just skip it when running all tests. This make it awkward to run ranges
737 of tests when one of them was inappropriate. Now it just skips any
738 inappropriate tests, as it always did when running all tests.
739
740 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
741 character types such as \d or \w, too many callouts were inserted, and the
742 data that they returned was rubbish.
743
744 26. In UCP mode, \s was not matching two of the characters that Perl matches,
745 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
746 were matched by \h. The code has now been refactored so that the lists of
747 the horizontal and vertical whitespace characters used for \h and \v (which
748 are defined only in one place) are now also used for \s.
749
750 27. Add JIT support for the 64 bit TileGX architecture.
751 Patch by Jiong Wang (Tilera Corporation).
752
753 28. Possessive quantifiers for classes (both explicit and automatically
754 generated) now use special opcodes instead of wrapping in ONCE brackets.
755
756 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
757 (because it's meaningless), this was not happening when PCRE_CASELESS was
758 set. Not wrong, but inefficient.
759
760 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
761 properties for \w, \d, etc) is present in a test regex. Otherwise if the
762 test contains no characters greater than 255, Perl doesn't realise it
763 should be using Unicode semantics.
764
765 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
766 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
767 does in Unicode mode.
768
769 32. Added the "forbid" facility to pcretest so that putting tests into the
770 wrong test files can sometimes be quickly detected.
771
772 33. There is now a limit (default 250) on the depth of nesting of parentheses.
773 This limit is imposed to control the amount of system stack used at compile
774 time. It can be changed at build time by --with-parens-nest-limit=xxx or
775 the equivalent in CMake.
776
777 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
778 errors. Perl warns for these when in warning mode, but PCRE has no facility
779 for giving warnings.
780
781 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
782 However, this was not working for (?!) because it is optimized to (*FAIL),
783 for which PCRE does not allow quantifiers. The optimization is now disabled
784 when a quantifier follows (?!). I can't see any use for this, but it makes
785 things uniform.
786
787 36. Perl no longer allows group names to start with digits, so I have made this
788 change also in PCRE. It simplifies the code a bit.
789
790 37. In extended mode, Perl ignores spaces before a + that indicates a
791 possessive quantifier. PCRE allowed a space before the quantifier, but not
792 before the possessive +. It now does.
793
794 38. The use of \K (reset reported match start) within a repeated possessive
795 group such as (a\Kb)*+ was not working.
796
797 40. Document that the same character tables must be used at compile time and
798 run time, and that the facility to pass tables to pcre_exec() and
799 pcre_dfa_exec() is for use only with saved/restored patterns.
800
801 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
802 features for Builds with MSVC:
803
804 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
805 builds). The .rc files can be used to set FileDescription and many other
806 attributes.
807
808 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
809 This allows higher-level build scripts which want .pdb files to avoid
810 hard-coding the exact files needed."
811
812 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
813 mean "start of word" and "end of word", respectively, as a transition aid.
814
815 43. A minimizing repeat of a class containing codepoints greater than 255 in
816 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
817 compiled to use the heap for recursion.
818
819 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
820 is configured.
821
822
823 Version 8.33 28-May-2013
824 ------------------------
825
826 1. Added 'U' to some constants that are compared to unsigned integers, to
827 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
828 variables that are added to signed variables, to ensure the result is
829 signed and can be negated.
830
831 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
832 CMake config files.
833
834 3. Revise the creation of config.h.generic so that all boolean macros are
835 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
836 overriding via -D on the command line possible.
837
838 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
839 to unsigned int is reported to make a quite noticeable speed difference in
840 a specific Windows environment. Testing on Linux did also appear to show
841 some benefit (and it is clearly not harmful). Also fixed the definition of
842 Xop which should be unsigned.
843
844 5. Related to (4), changing the definition of the intermediate variable cc
845 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
846 improvements.
847
848 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
849 unnecessary spaces.
850
851 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
852 and later.
853
854 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
855
856 9. Optimizing fast_forward_start_bits in JIT.
857
858 10. Adding support for callouts in JIT, and fixing some issues revealed
859 during this work. Namely:
860
861 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
862
863 (b) Minimum length was not checked before the matching is started.
864
865 11. The value of capture_last that is passed to callouts was incorrect in some
866 cases when there was a capture on one path that was subsequently abandoned
867 after a backtrack. Also, the capture_last value is now reset after a
868 recursion, since all captures are also reset in this case.
869
870 12. The interpreter no longer returns the "too many substrings" error in the
871 case when an overflowing capture is in a branch that is subsequently
872 abandoned after a backtrack.
873
874 13. In the pathological case when an offset vector of size 2 is used, pcretest
875 now prints out the matched string after a yield of 0 or 1.
876
877 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
878 Only supported by the JIT compiler at the moment.
879
880 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
881
882 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
883 offset of the starting point of the matching process, provided the offsets
884 vector is large enough.
885
886 17. The \A escape now records a lookbehind value of 1, though its execution
887 does not actually inspect the previous character. This is to ensure that,
888 in partial multi-segment matching, at least one character from the old
889 segment is retained when a new segment is processed. Otherwise, if there
890 are no lookbehinds in the pattern, \A might match incorrectly at the start
891 of a new segment.
892
893 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
894
895 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
896 gives some modest performance improvement in 8-bit mode.
897
898 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
899 be expressed in certain programming languages using Universal Character
900 Names.
901
902 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
903 which points out that "non characters" are not "characters that may not
904 appear in Unicode strings" but rather "characters that are reserved for
905 internal use and have only local meaning".
906
907 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
908 there was a conditional group that depended on an assertion, if the
909 assertion was false, the callout that immediately followed the alternation
910 in the condition was skipped when pcre_exec() was used for matching.
911
912 23. Allow an explicit callout to be inserted before an assertion that is the
913 condition for a conditional group, for compatibility with automatic
914 callouts, which always insert a callout at this point.
915
916 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
917 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
918
919 25. (*PRUNE) is now supported by the JIT compiler.
920
921 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
922
923 27. Fix the case where there are two or more SKIPs with arguments that may be
924 ignored.
925
926 28. (*SKIP) is now supported by the JIT compiler.
927
928 29. (*THEN) is now supported by the JIT compiler.
929
930 30. Update RunTest with additional test selector options.
931
932 31. The way PCRE handles backtracking verbs has been changed in two ways.
933
934 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
935 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
936 backtracking. In some cases this makes it more Perl-compatible, but Perl's
937 rather obscure rules do not always do the same thing.
938
939 (2) Previously, backtracking verbs were confined within assertions. This is
940 no longer the case for positive assertions, except for (*ACCEPT). Again,
941 this sometimes improves Perl compatibility, and sometimes does not.
942
943 32. A number of tests that were in test 2 because Perl did things differently
944 have been moved to test 1, because either Perl or PCRE has changed, and
945 these tests are now compatible.
946
947 32. Backtracking control verbs are now handled in the same way in JIT and
948 interpreter.
949
950 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
951 contained a forward subroutine reference caused a compile error.
952
953 34. Auto-detect and optimize limited repetitions in JIT.
954
955 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
956 blocking (*UTF) etc.
957
958 36. In the interpreter, maximizing pattern repetitions for characters and
959 character types now use tail recursion, which reduces stack usage.
960
961 37. The value of the max lookbehind was not correctly preserved if a compiled
962 and saved regex was reloaded on a host of different endianness.
963
964 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
965 of the compiled pattern block, expand the flags field from 16 to 32 bits
966 because it was almost full.
967
968 39. Try madvise first before posix_madvise.
969
970 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
971 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
972 original change locked out all matches of empty strings. This has been
973 changed so that one match of an empty string per line is recognized.
974 Subsequent searches on the same line (for colouring or for --only-matching,
975 for example) do not recognize empty strings.
976
977 41. Applied a user patch to fix a number of spelling mistakes in comments.
978
979 42. Data lines longer than 65536 caused pcretest to crash.
980
981 43. Clarified the data type for length and startoffset arguments for pcre_exec
982 and pcre_dfa_exec in the function-specific man pages, where they were
983 explicitly stated to be in bytes, never having been updated. I also added
984 some clarification to the pcreapi man page.
985
986 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
987 a segmentation fault.
988
989
990 Version 8.32 30-November-2012
991 -----------------------------
992
993 1. Improved JIT compiler optimizations for first character search and single
994 character iterators.
995
996 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
997 Patch by Daniel Richard G.
998
999 3. Single character iterator optimizations in the JIT compiler.
1000
1001 4. Improved JIT compiler optimizations for character ranges.
1002
1003 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1004 Reported by Giuseppe D'Angelo.
1005
1006 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1007 of a line, was being set incorrectly in cases where .* appeared inside
1008 atomic brackets at the start of a pattern, or where there was a subsequent
1009 *PRUNE or *SKIP.
1010
1011 7. Improved instruction cache flush for POWER/PowerPC.
1012 Patch by Daniel Richard G.
1013
1014 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1015 grep:
1016
1017 (a) There is now no limit to the number of patterns to be matched.
1018
1019 (b) An error is given if a pattern is too long.
1020
1021 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1022 are now supported.
1023
1024 (d) --exclude-from and --include-from (multiple use) have been added.
1025
1026 (e) Exclusions and inclusions now apply to all files and directories, not
1027 just to those obtained from scanning a directory recursively.
1028
1029 (f) Multiple uses of -f and --file-list are now supported.
1030
1031 (g) In a Windows environment, the default for -d has been changed from
1032 "read" (the GNU grep default) to "skip", because otherwise the presence
1033 of a directory in the file list provokes an error.
1034
1035 (h) The documentation has been revised and clarified in places.
1036
1037 9. Improve the matching speed of capturing brackets.
1038
1039 10. Changed the meaning of \X so that it now matches a Unicode extended
1040 grapheme cluster.
1041
1042 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1043 out POSIX threads when JIT support is configured.
1044
1045 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1046
1047 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1048 structure could go wrong in environments where size_t is not the same size
1049 as int.
1050
1051 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1052 set.
1053
1054 15. The EBCDIC support had decayed; later updates to the code had included
1055 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1056 general tidy up of EBCDIC-related issues, and the documentation was also
1057 not quite right. There is now a test that can be run on ASCII systems to
1058 check some of the EBCDIC-related things (but is it not a full test).
1059
1060 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1061 in a small tidy to the code.
1062
1063 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1064
1065 18. If the --only-matching (-o) option in pcregrep is specified multiple
1066 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1067 substrings matched by the 1st and 2nd capturing parentheses. A separating
1068 string can be specified by --om-separator (default empty).
1069
1070 19. Improving the first n character searches.
1071
1072 20. Turn case lists for horizontal and vertical white space into macros so that
1073 they are defined only once.
1074
1075 21. This set of changes together give more compatible Unicode case-folding
1076 behaviour for characters that have more than one other case when UCP
1077 support is available.
1078
1079 (a) The Unicode property table now has offsets into a new table of sets of
1080 three or more characters that are case-equivalent. The MultiStage2.py
1081 script that generates these tables (the pcre_ucd.c file) now scans
1082 CaseFolding.txt instead of UnicodeData.txt for character case
1083 information.
1084
1085 (b) The code for adding characters or ranges of characters to a character
1086 class has been abstracted into a generalized function that also handles
1087 case-independence. In UTF-mode with UCP support, this uses the new data
1088 to handle characters with more than one other case.
1089
1090 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1091 whose other case is greater than 256 are now correctly matched
1092 caselessly. Previously, the high codepoint matched the low one, but not
1093 vice versa.
1094
1095 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1096 of the new class addition function, using character lists defined as
1097 macros alongside the case definitions of 20 above.
1098
1099 (e) Caseless back references now work with characters that have more than
1100 one other case.
1101
1102 (f) General caseless matching of characters with more than one other case
1103 is supported.
1104
1105 22. Unicode character properties were updated from Unicode 6.2.0
1106
1107 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1108
1109 24. Add support for 32-bit character strings, and UTF-32
1110
1111 25. Major JIT compiler update (code refactoring and bugfixing).
1112 Experimental Sparc 32 support is added.
1113
1114 26. Applied a modified version of Daniel Richard G's patch to create
1115 pcre.h.generic and config.h.generic by "make" instead of in the
1116 PrepareRelease script.
1117
1118 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1119 pcre_compile.c when checking for a zero character.
1120
1121 28. Introducing a native interface for JIT. Through this interface, the compiled
1122 machine code can be directly executed. The purpose of this interface is to
1123 provide fast pattern matching, so several sanity checks are not performed.
1124 However, feature tests are still performed. The new interface provides
1125 1.4x speedup compared to the old one.
1126
1127 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1128 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1129 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1130
1131 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1132 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1133 the "old" RFC 2279). Instead, it ended up passing a negative length to
1134 pcre_exec().
1135
1136 31. Add support for GCC's visibility feature to hide internal functions.
1137
1138 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1139 "unknown -C option" after outputting 0 or 1.
1140
1141 33. There is now support for generating a code coverage report for the test
1142 suite in environments where gcc is the compiler and lcov is installed. This
1143 is mainly for the benefit of the developers.
1144
1145 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1146 unaddressable using valgrind annotations, allowing valgrind to detect
1147 invalid memory accesses. This is mainly for the benefit of the developers.
1148
1149 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1150
1151 26. Give configure error if --enable-cpp but no C++ compiler found.
1152
1153
1154 Version 8.31 06-July-2012
1155 -------------------------
1156
1157 1. Fixing a wrong JIT test case and some compiler warnings.
1158
1159 2. Removed a bashism from the RunTest script.
1160
1161 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1162 to unsigned type, result still unsigned" that was given by an MS compiler
1163 on encountering the code "-sizeof(xxx)".
1164
1165 4. Partial matching support is added to the JIT compiler.
1166
1167 5. Fixed several bugs concerned with partial matching of items that consist
1168 of more than one character:
1169
1170 (a) /^(..)\1/ did not partially match "aba" because checking references was
1171 done on an "all or nothing" basis. This also applied to repeated
1172 references.
1173
1174 (b) \R did not give a hard partial match if \r was found at the end of the
1175 subject.
1176
1177 (c) \X did not give a hard partial match after matching one or more
1178 characters at the end of the subject.
1179
1180 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1181 a partial match for the string "\r".
1182
1183 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1184 a partial match for a CR character at the end of the subject string.
1185
1186 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1187 running pcretest, the text "(JIT)" added to the output whenever JIT is
1188 actually used to run the match.
1189
1190 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1191 or /S+[+] with a digit between 1 and 7.
1192
1193 8. OP_NOT now supports any UTF character not just single-byte ones.
1194
1195 9. (*MARK) control verb is now supported by the JIT compiler.
1196
1197 10. The command "./RunTest list" lists the available tests without actually
1198 running any of them. (Because I keep forgetting what they all are.)
1199
1200 11. Add PCRE_INFO_MAXLOOKBEHIND.
1201
1202 12. Applied a (slightly modified) user-supplied patch that improves performance
1203 when the heap is used for recursion (compiled with --disable-stack-for-
1204 recursion). Instead of malloc and free for each heap frame each time a
1205 logical recursion happens, frames are retained on a chain and re-used where
1206 possible. This sometimes gives as much as 30% improvement.
1207
1208 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1209 call.
1210
1211 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1212
1213 15. It is now possible to link pcretest with libedit as an alternative to
1214 libreadline.
1215
1216 16. (*COMMIT) control verb is now supported by the JIT compiler.
1217
1218 17. The Unicode data tables have been updated to Unicode 6.1.0.
1219
1220 18. Added --file-list option to pcregrep.
1221
1222 19. Added binary file support to pcregrep, including the -a, --binary-files,
1223 -I, and --text options.
1224
1225 20. The madvise function is renamed for posix_madvise for QNX compatibility
1226 reasons. Fixed by Giuseppe D'Angelo.
1227
1228 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1229
1230 22. Changed the option for creating symbolic links for 16-bit man pages from
1231 -s to -sf so that re-installing does not cause issues.
1232
1233 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1234
1235 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1236 matches in certain environments (the workspace was not being correctly
1237 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1238 some of the workspace data at the beginning of a restart.
1239
1240 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1241 was not doing so when it should - probably a typo introduced by SVN 528
1242 (change 8.10/14).
1243
1244 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1245 \w+ when the character tables indicated that \x{c4} was a word character.
1246 There were several related cases, all because the tests for doing a table
1247 lookup were testing for characters less than 127 instead of 255.
1248
1249 27. If a pattern contains capturing parentheses that are not used in a match,
1250 their slots in the ovector are set to -1. For those that are higher than
1251 any matched groups, this happens at the end of processing. In the case when
1252 there were back references that the ovector was too small to contain
1253 (causing temporary malloc'd memory to be used during matching), and the
1254 highest capturing number was not used, memory off the end of the ovector
1255 was incorrectly being set to -1. (It was using the size of the temporary
1256 memory instead of the true size.)
1257
1258 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1259 ovector size, it uses memory at the end of the block that it has got.
1260
1261 29. Check for an overlong MARK name and give an error at compile time. The
1262 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1263
1264 30. JIT compiler update.
1265
1266 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1267 Rill for the patch.
1268
1269 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1270
1271 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1272
1273 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1274 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1275 was enabled.
1276
1277 35. Improve JIT code generation for greedy plus quantifier.
1278
1279 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1280 1 to "aa" instead of to an empty string. The bug affected repeated groups
1281 that could potentially match an empty string.
1282
1283 37. Optimizing single character iterators in JIT.
1284
1285 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1286 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1287 codepoints that are too big for the mode are faulted, and in a UTF mode,
1288 disallowed codepoints are also faulted.
1289
1290 39. If PCRE was compiled with UTF support, in three places in the DFA
1291 matcher there was code that should only have been obeyed in UTF mode, but
1292 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1293 processing when bytes with values greater than 127 were present. In 16-bit
1294 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1295 both cases the values are those that cannot be the first data item in a UTF
1296 character. The three items that might have provoked this were recursions,
1297 possessively repeated groups, and atomic groups.
1298
1299 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1300 and pcregrep, because some OS require shared objects to be explicitly
1301 passed to ld, causing the link step to fail if they are not.
1302
1303 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1304 mode, patterns that started with \h* or \R* might be incorrectly matched.
1305
1306
1307 Version 8.30 04-February-2012
1308 -----------------------------
1309
1310 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1311 name is defined in ctype.h.
1312
1313 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1314 only in quite long subpatterns.
1315
1316 3. Removed the function pcre_info(), which has been obsolete and deprecated
1317 since it was replaced by pcre_fullinfo() in February 2000.
1318
1319 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1320 match a (*MARK), and the match failed at the start of the subject, a
1321 reference to memory before the start of the subject could occur. This bug
1322 was introduced by fix 17 of release 8.21.
1323
1324 5. A reference to an unset group with zero minimum repetition was giving
1325 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1326 /(another)?(\1?)test/ matched against "hello world test". This bug was
1327 introduced in release 8.13.
1328
1329 6. Add support for 16-bit character strings (a large amount of work involving
1330 many changes and refactorings).
1331
1332 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1333 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1334 from a file.
1335
1336 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1337 rounding is not applied in this particular case).
1338
1339 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1340 if they appear, or are escaped, in patterns.
1341
1342 10. Get rid of a number of -Wunused-but-set-variable warnings.
1343
1344 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1345 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1346 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1347 also returns the mark "x". This bug applied to capturing parentheses,
1348 non-capturing parentheses, and atomic parentheses. It also applied to some
1349 assertions.
1350
1351 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1352 information out of configure.ac instead of relying on pcre.h.generic, which
1353 is not stored in the repository.
1354
1355 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1356 -lreadline.
1357
1358 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1359
1360 15. Applied Graycode's patch to put the top-level frame on the stack rather
1361 than the heap when not using the stack for recursion. This gives a
1362 performance improvement in many cases when recursion is not deep.
1363
1364 16. Experimental code added to "pcretest -C" to output the stack frame size.
1365
1366
1367 Version 8.21 12-Dec-2011
1368 ------------------------
1369
1370 1. Updating the JIT compiler.
1371
1372 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1373 are added as well.
1374
1375 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1376 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1377 calling _pcre_jit_exec. Some extra comments are added.
1378
1379 4. (*MARK) settings inside atomic groups that do not contain any capturing
1380 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1381 was introduced by change 18 for 8.20.
1382
1383 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1384 ECMA-262 standard.
1385
1386 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1387 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1388 This bug was probably introduced by change 9 of 8.13.
1389
1390 7. While fixing 6 above, I noticed that a number of other items were being
1391 incorrectly rejected as "not fixed length". This arose partly because newer
1392 opcodes had not been added to the fixed-length checking code. I have (a)
1393 corrected the bug and added tests for these items, and (b) arranged for an
1394 error to occur if an unknown opcode is encountered while checking for fixed
1395 length instead of just assuming "not fixed length". The items that were
1396 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1397 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1398 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1399
1400 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1401 being incorrectly compiled and would have given unpredicatble results.
1402
1403 9. A possessively repeated subpattern with minimum repeat count greater than
1404 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1405 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1406 the first (A) could occur when it should not.
1407
1408 10. Add a cast and remove a redundant test from the code.
1409
1410 11. JIT should use pcre_malloc/pcre_free for allocation.
1411
1412 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1413 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1414 is anything other than /usr, -L is still shown).
1415
1416 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1417
1418 14. Perl does not support \N without a following name in a [] class; PCRE now
1419 also gives an error.
1420
1421 15. If a forward reference was repeated with an upper limit of around 2000,
1422 it caused the error "internal error: overran compiling workspace". The
1423 maximum number of forward references (including repeats) was limited by the
1424 internal workspace, and dependent on the LINK_SIZE. The code has been
1425 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1426 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1427 of around 200,000 forward references. While doing this, I also speeded up
1428 the filling in of repeated forward references.
1429
1430 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1431 incorrectly expecting the subject to contain another "a" after the start.
1432
1433 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1434 in the match, the SKIP should be ignored. This was not happening; instead
1435 the SKIP was being treated as NOMATCH. For patterns such as
1436 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1437 tested.
1438
1439 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1440 now much more compatible with Perl, in particular in cases where the result
1441 is a non-match for a non-anchored pattern. For example, if
1442 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1443 "m", where previously it did not return a name. A side effect of this
1444 change is that for partial matches, the last encountered mark name is
1445 returned, as for non matches. A number of tests that were previously not
1446 Perl-compatible have been moved into the Perl-compatible test files. The
1447 refactoring has had the pleasing side effect of removing one argument from
1448 the match() function, thus reducing its stack requirements.
1449
1450 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1451 subsequent uses of /S (without +) incorrectly behaved like /S+.
1452
1453 21. Retrieve executable code size support for the JIT compiler and fixing
1454 some warnings.
1455
1456 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1457 not work when the shorter character appeared right at the end of the
1458 subject string.
1459
1460 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1461 systems.
1462
1463 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1464 output it when the /M option is used in pcretest.
1465
1466 25. The CheckMan script was not being included in the distribution. Also, added
1467 an explicit "perl" to run Perl scripts from the PrepareRelease script
1468 because this is reportedly needed in Windows.
1469
1470 26. If study data was being save in a file and studying had not found a set of
1471 "starts with" bytes for the pattern, the data written to the file (though
1472 never used) was taken from uninitialized memory and so caused valgrind to
1473 complain.
1474
1475 27. Updated RunTest.bat as provided by Sheri Pierce.
1476
1477 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1478
1479 29. Computation of memory usage for the table of capturing group names was
1480 giving an unnecessarily large value.
1481
1482
1483 Version 8.20 21-Oct-2011
1484 ------------------------
1485
1486 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1487 a POSIX class. After further experiments with Perl, which convinced me that
1488 Perl has bugs and confusions, a closing square bracket is no longer allowed
1489 in a POSIX name. This bug also affected patterns with classes that started
1490 with full stops.
1491
1492 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1493 captured substring, but while checking the failing first alternative,
1494 substring 1 is temporarily captured. If the output vector supplied to
1495 pcre_exec() was not big enough for this capture, the yield of the function
1496 was still zero ("insufficient space for captured substrings"). This cannot
1497 be totally fixed without adding another stack variable, which seems a lot
1498 of expense for a edge case. However, I have improved the situation in cases
1499 such as /(a)(b)x|abc/ matched against "abc", where the return code
1500 indicates that fewer than the maximum number of slots in the ovector have
1501 been set.
1502
1503 3. Related to (2) above: when there are more back references in a pattern than
1504 slots in the output vector, pcre_exec() uses temporary memory during
1505 matching, and copies in the captures as far as possible afterwards. It was
1506 using the entire output vector, but this conflicts with the specification
1507 that only 2/3 is used for passing back captured substrings. Now it uses
1508 only the first 2/3, for compatibility. This is, of course, another edge
1509 case.
1510
1511 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1512 main code base, and can be used by building with --enable-jit. When this is
1513 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1514 runtime --no-jit option is given.
1515
1516 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1517 ovector, the return from the function was zero, implying that there were
1518 other matches that did not fit. The correct "exactly full" value is now
1519 returned.
1520
1521 6. If a subpattern that was called recursively or as a subroutine contained
1522 (*PRUNE) or any other control that caused it to give a non-standard return,
1523 invalid errors such as "Error -26 (nested recursion at the same subject
1524 position)" or even infinite loops could occur.
1525
1526 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1527 computing the minimum length on reaching *ACCEPT, and so ended up with the
1528 wrong value of 1 rather than 0. Further investigation indicates that
1529 computing a minimum subject length in the presence of *ACCEPT is difficult
1530 (think back references, subroutine calls), and so I have changed the code
1531 so that no minimum is registered for a pattern that contains *ACCEPT.
1532
1533 8. If (*THEN) was present in the first (true) branch of a conditional group,
1534 it was not handled as intended. [But see 16 below.]
1535
1536 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1537 Sheri Pierce.
1538
1539 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1540 the first byte in a match must be "a".
1541
1542 11. Change 17 for 8.13 increased the recursion depth for patterns like
1543 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1544 pattern contains any instances of (*THEN). If it does not, the old
1545 optimizations are restored. It would be nice to do this on a per-group
1546 basis, but at the moment that is not feasible.
1547
1548 12. In some environments, the output of pcretest -C is CRLF terminated. This
1549 broke RunTest's code that checks for the link size. A single white space
1550 character after the value is now allowed for.
1551
1552 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1553 For "fr", it uses the Windows-specific input and output files.
1554
1555 14. If (*THEN) appeared in a group that was called recursively or as a
1556 subroutine, it did not work as intended. [But see next item.]
1557
1558 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1559 pattern fragments (but not containing any | characters). If A and B are
1560 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1561 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1562 D. In other words, Perl considers parentheses that do not contain any |
1563 characters to be part of a surrounding alternative, whereas PCRE was
1564 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1565 differently. PCRE now behaves in the same way as Perl, except in the case
1566 of subroutine/recursion calls such as (?1) which have in any case always
1567 been different (but PCRE had them first :-).
1568
1569 16. Related to 15 above: Perl does not treat the | in a conditional group as
1570 creating alternatives. Such a group is treated in the same way as an
1571 ordinary group without any | characters when processing (*THEN). PCRE has
1572 been changed to match Perl's behaviour.
1573
1574 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1575 RunGrepTest script failed.
1576
1577 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1578 inevitable for groups that contain captures, but it can lead to a lot of
1579 stack use in large patterns. The old behaviour has been restored for atomic
1580 groups that do not contain any capturing parentheses.
1581
1582 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1583 suppress the check for a minimum subject length at run time. (If it was
1584 given to pcre_exec() or pcre_dfa_exec() it did work.)
1585
1586 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1587 fail to work when decoding hex characters in data strings in EBCDIC
1588 environments.
1589
1590 21. It appears that in at least one Mac OS environment, the isxdigit() function
1591 is implemented as a macro that evaluates to its argument more than once,
1592 contravening the C 90 Standard (I haven't checked a later standard). There
1593 was an instance in pcretest which caused it to go wrong when processing
1594 \x{...} escapes in subject strings. The has been rewritten to avoid using
1595 things like p++ in the argument of isxdigit().
1596
1597
1598 Version 8.13 16-Aug-2011
1599 ------------------------
1600
1601 1. The Unicode data tables have been updated to Unicode 6.0.0.
1602
1603 2. Two minor typos in pcre_internal.h have been fixed.
1604
1605 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1606 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1607 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1608
1609 4. There were a number of related bugs in the code for matching backrefences
1610 caselessly in UTF-8 mode when codes for the characters concerned were
1611 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1612 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1613 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1614 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1615 2-byte code at the end of the subject (it thought there wasn't enough data
1616 left).
1617
1618 5. Comprehensive information about what went wrong is now returned by
1619 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1620 as the output vector has at least 2 elements. The offset of the start of
1621 the failing character and a reason code are placed in the vector.
1622
1623 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1624 now returned is for the first byte of the failing character, instead of the
1625 last byte inspected. This is an incompatible change, but I hope it is small
1626 enough not to be a problem. It makes the returned offset consistent with
1627 pcre_exec() and pcre_dfa_exec().
1628
1629 7. pcretest now gives a text phrase as well as the error number when
1630 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1631 failure, the offset and reason code are output.
1632
1633 8. When \R was used with a maximizing quantifier it failed to skip backwards
1634 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1635 back over a single character (\n). This seems wrong (because it treated the
1636 two characters as a single entity when going forwards), conflicts with the
1637 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1638 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1639 has been changed.
1640
1641 9. Some internal refactoring has changed the processing so that the handling
1642 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1643 time (the PCRE_DOTALL option was changed this way some time ago: version
1644 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1645 which was always a bit of a fudge. It also means that there is one less
1646 argument for the match() function, which reduces its stack requirements
1647 slightly. This change also fixes an incompatibility with Perl: the pattern
1648 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1649
1650 10. More internal refactoring has drastically reduced the number of recursive
1651 calls to match() for possessively repeated groups such as (abc)++ when
1652 using pcre_exec().
1653
1654 11. While implementing 10, a number of bugs in the handling of groups were
1655 discovered and fixed:
1656
1657 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1658 (a|)*(?1) gave a compile-time internal error.
1659 ((a|)+)+ did not notice that the outer group could match an empty string.
1660 (^a|^)+ was not marked as anchored.
1661 (.*a|.*)+ was not marked as matching at start or after a newline.
1662
1663 12. Yet more internal refactoring has removed another argument from the match()
1664 function. Special calls to this function are now indicated by setting a
1665 value in a variable in the "match data" data block.
1666
1667 13. Be more explicit in pcre_study() instead of relying on "default" for
1668 opcodes that mean there is no starting character; this means that when new
1669 ones are added and accidentally left out of pcre_study(), testing should
1670 pick them up.
1671
1672 14. The -s option of pcretest has been documented for ages as being an old
1673 synonym of -m (show memory usage). I have changed it to mean "force study
1674 for every regex", that is, assume /S for every regex. This is similar to -i
1675 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1676 using it. It makes it easier to run collections of tests with and without
1677 study enabled, and thereby test pcre_study() more easily. All the standard
1678 tests are now run with and without -s (but some patterns can be marked as
1679 "never study" - see 20 below).
1680
1681 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1682 restoration of the capturing data to the outer values was not happening
1683 correctly.
1684
1685 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1686 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1687 pattern had matched an empty string, and so incorrectly returned a no
1688 match.
1689
1690 17. There was optimizing code for the last branch of non-capturing parentheses,
1691 and also for the obeyed branch of a conditional subexpression, which used
1692 tail recursion to cut down on stack usage. Unfortunately, now that there is
1693 the possibility of (*THEN) occurring in these branches, tail recursion is
1694 no longer possible because the return has to be checked for (*THEN). These
1695 two optimizations have therefore been removed. [But see 8.20/11 above.]
1696
1697 18. If a pattern containing \R was studied, it was assumed that \R always
1698 matched two bytes, thus causing the minimum subject length to be
1699 incorrectly computed because \R can also match just one byte.
1700
1701 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1702 was incorrectly computed.
1703
1704 20. If /S is present twice on a test pattern in pcretest input, it now
1705 *disables* studying, thereby overriding the use of -s on the command line
1706 (see 14 above). This is necessary for one or two tests to keep the output
1707 identical in both cases.
1708
1709 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1710 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1711
1712 22. When an atomic group that contained a capturing parenthesis was
1713 successfully matched, but the branch in which it appeared failed, the
1714 capturing was not being forgotten if a higher numbered group was later
1715 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1716 group 1 to "a", when in fact it should be unset. This applied to multi-
1717 branched capturing and non-capturing groups, repeated or not, and also to
1718 positive assertions (capturing in negative assertions does not happen
1719 in PCRE) and also to nested atomic groups.
1720
1721 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1722 subject after a captured substring, to make it easier to tell which of a
1723 number of identical substrings has been captured.
1724
1725 24. The way atomic groups are processed by pcre_exec() has been changed so that
1726 if they are repeated, backtracking one repetition now resets captured
1727 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1728 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1729 "aaa". Previously, it would have been "a". As part of this code
1730 refactoring, the way recursive calls are handled has also been changed.
1731
1732 25. If an assertion condition captured any substrings, they were not passed
1733 back unless some other capturing happened later. For example, if
1734 (?(?=(a))a) was matched against "a", no capturing was returned.
1735
1736 26. When studying a pattern that contained subroutine calls or assertions,
1737 the code for finding the minimum length of a possible match was handling
1738 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1739 group 1 called group 2 while simultaneously a separate group 2 called group
1740 1). A stack overflow occurred in this case. I have fixed this by limiting
1741 the recursion depth to 10.
1742
1743 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1744 Fortmann. This supports explicit test numbers on the command line, and has
1745 argument validation and error reporting.
1746
1747 28. An instance of \X with an unlimited repeat could fail if at any point the
1748 first character it looked at was a mark character.
1749
1750 29. Some minor code refactoring concerning Unicode properties and scripts
1751 should reduce the stack requirement of match() slightly.
1752
1753 30. Added the '=' option to pcretest to check the setting of unused capturing
1754 slots at the end of the pattern, which are documented as being -1, but are
1755 not included in the return count.
1756
1757 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1758 compiled something random. Now it gives a compile-time error (as does
1759 Perl).
1760
1761 32. A *MARK encountered during the processing of a positive assertion is now
1762 recorded and passed back (compatible with Perl).
1763
1764 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1765 had alternative anchored branches, the search for a second match in a line
1766 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1767 matched the line "0102" twice. The same bug affected patterns that started
1768 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1769 twice.
1770
1771 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1772 does, and because of capturing effects, quantifying parenthesized
1773 assertions may at times be useful. Quantifiers are now allowed for
1774 parenthesized assertions.
1775
1776 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1777
1778 36. \g was being checked for fancy things in a character class, when it should
1779 just be a literal "g".
1780
1781 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1782 appearance of a nested POSIX class supersedes an apparent external class.
1783 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1784 unescaped square brackets may also appear as part of class names. For
1785 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1786 more like Perl. (But see 8.20/1 above.)
1787
1788 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1789 was because it thought it was \N{name}, which is not supported).
1790
1791 39. Add minix to OS list not supporting the -S option in pcretest.
1792
1793 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1794 cannot analyze patterns in sufficient detail to catch mutual recursions
1795 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1796 subgroup is called recursively as a subpattern for a second time at the
1797 same position in the subject string. In previous releases this might have
1798 been caught by the recursion limit, or it might have run out of stack.
1799
1800 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1801 happen only once. PCRE was, however incorrectly giving a compile time error
1802 "recursive call could loop indefinitely" because it cannot analyze the
1803 pattern in sufficient detail. The compile time test no longer happens when
1804 PCRE is compiling a conditional subpattern, but actual runaway loops are
1805 now caught at runtime (see 40 above).
1806
1807 42. It seems that Perl allows any characters other than a closing parenthesis
1808 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1809 has been changed to be the same.
1810
1811 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1812 as not to get warnings when autogen.sh is called. Also changed
1813 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1814
1815 44. To help people who use pcregrep to scan files containing exceedingly long
1816 lines, the following changes have been made:
1817
1818 (a) The default value of the buffer size parameter has been increased from
1819 8K to 20K. (The actual buffer used is three times this size.)
1820
1821 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1822 PCRE is built.
1823
1824 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1825 to be set at run time.
1826
1827 (d) Numerical values in pcregrep options can be followed by K or M, for
1828 example --buffer-size=50K.
1829
1830 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1831 given and the return code is set to 2.
1832
1833 45. Add a pointer to the latest mark to the callout data block.
1834
1835 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1836 partial match of an empty string instead of no match. This was specific to
1837 the use of ".".
1838
1839 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1840 complete match instead of a partial match. This bug was dependent on both
1841 the PCRE_UTF8 and PCRE_DOTALL options being set.
1842
1843 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1844 starting byte set, because \b was not being ignored.
1845
1846
1847 Version 8.12 15-Jan-2011
1848 ------------------------
1849
1850 1. Fixed some typos in the markup of the man pages, and wrote a script that
1851 checks for such things as part of the documentation building process.
1852
1853 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1854 --match-limit and --recursion-limit options (added for 8.11). In
1855 particular, this made one of the standard tests fail. (The integer value
1856 went into the wrong half of a long int.)
1857
1858 3. If the --colour option was given to pcregrep with -v (invert match), it
1859 did strange things, either producing crazy output, or crashing. It should,
1860 of course, ignore a request for colour when reporting lines that do not
1861 match.
1862
1863 4. Another pcregrep bug caused similar problems if --colour was specified with
1864 -M (multiline) and the pattern match finished with a line ending.
1865
1866 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1867 matched in multiline mode, the following line was shown as part of the
1868 match. This seems wrong, so I have changed it.
1869
1870 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1871 the check for further matches in the same line (so they could be coloured)
1872 to overrun the end of the current line. If another match was found, it was
1873 incorrectly shown (and then shown again when found in the next line).
1874
1875 7. If pcregrep was compiled under Windows, there was a reference to the
1876 function pcregrep_exit() before it was defined. I am assuming this was
1877 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1878 reported by a user. I've moved the definition above the reference.
1879
1880
1881 Version 8.11 10-Dec-2010
1882 ------------------------
1883
1884 1. (*THEN) was not working properly if there were untried alternatives prior
1885 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1886 backtracked to try for "b" instead of moving to the next alternative branch
1887 at the same level (in this case, to look for "c"). The Perl documentation
1888 is clear that when (*THEN) is backtracked onto, it goes to the "next
1889 alternative in the innermost enclosing group".
1890
1891 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1892 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1893 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1894 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1895 (*THEN).
1896
1897 3. If \s appeared in a character class, it removed the VT character from
1898 the class, even if it had been included by some previous item, for example
1899 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1900 of \s, but is part of the POSIX "space" class.)
1901
1902 4. A partial match never returns an empty string (because you can always
1903 match an empty string at the end of the subject); however the checking for
1904 an empty string was starting at the "start of match" point. This has been
1905 changed to the "earliest inspected character" point, because the returned
1906 data for a partial match starts at this character. This means that, for
1907 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1908 (previously it gave "no match").
1909
1910 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1911 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1912 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1913 has an implication that the given string is incomplete (because a partial
1914 match is preferred over a full match). For this reason, these items now
1915 give a partial match in this situation. [Aside: previously, the one case
1916 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1917 match rather than a full match, which was wrong by the old rules, but is
1918 now correct.]
1919
1920 6. There was a bug in the handling of #-introduced comments, recognized when
1921 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1922 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1923 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1924 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1925 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1926 places in pcre_compile().
1927
1928 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1929 comments when looking ahead for named forward references to subpatterns,
1930 the only newline sequence it recognized was NL. It now handles newlines
1931 according to the set newline convention.
1932
1933 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1934 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1935 cater for a lack of strerror(). These oversights have been fixed.
1936
1937 9. Added --match-limit and --recursion-limit to pcregrep.
1938
1939 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1940
1941 11. When the -o option was used, pcregrep was setting a return code of 1, even
1942 when matches were found, and --line-buffered was not being honoured.
1943
1944 12. Added an optional parentheses number to the -o and --only-matching options
1945 of pcregrep.
1946
1947 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1948 can match an empty string. The code to do it in pcretest and pcredemo
1949 needed fixing:
1950
1951 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1952 only one byte after an empty string match just before CRLF (this case
1953 just got forgotten; "any" and "anycrlf" were OK).
1954
1955 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1956 mode when an empty string match preceded an ASCII character followed by
1957 a non-ASCII character. (The code for advancing by one character rather
1958 than one byte was nonsense.)
1959
1960 (c) The pcredemo.c sample program did not have any code at all to handle
1961 the cases when CRLF is a valid newline sequence.
1962
1963 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1964 as a starting offset was within the subject string. There is now a new
1965 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1966 negative or greater than the length of the string. In order to test this,
1967 pcretest is extended to allow the setting of negative starting offsets.
1968
1969 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1970 starting offset points to the beginning of a UTF-8 character was
1971 unnecessarily clumsy. I tidied it up.
1972
1973 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1974 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1975
1976 17. Nobody had reported that the --include_dir option, which was added in
1977 release 7.7 should have been called --include-dir (hyphen, not underscore)
1978 for compatibility with GNU grep. I have changed it to --include-dir, but
1979 left --include_dir as an undocumented synonym, and the same for
1980 --exclude-dir, though that is not available in GNU grep, at least as of
1981 release 2.5.4.
1982
1983 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1984 characters from a string of bytes) have been redefined so as not to use
1985 loops, in order to improve performance in some environments. At the same
1986 time, I abstracted some of the common code into auxiliary macros to save
1987 repetition (this should not affect the compiled code).
1988
1989 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1990 compile-time error is now given if \c is not followed by an ASCII
1991 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1992 different, and any byte value is allowed.)
1993
1994 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1995 START_OPTIMIZE option, which is now allowed at compile time - but just
1996 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1997 to pcregrep and other applications that have no direct access to PCRE
1998 options. The new /Y option in pcretest sets this option when calling
1999 pcre_compile().
2000
2001 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2002 back references. Groups containing recursive back references were forced to
2003 be atomic by that change, but in the case of named groups, the amount of
2004 memory required was incorrectly computed, leading to "Failed: internal
2005 error: code overflow". This has been fixed.
2006
2007 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2008 pcretest.c, to avoid build problems in some Borland environments.
2009
2010
2011 Version 8.10 25-Jun-2010
2012 ------------------------
2013
2014 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2015 THEN.
2016
2017 2. (*ACCEPT) was not working when inside an atomic group.
2018
2019 3. Inside a character class, \B is treated as a literal by default, but
2020 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2021 causes the error). The code is unchanged, but I tidied the documentation.
2022
2023 4. Inside a character class, PCRE always treated \R and \X as literals,
2024 whereas Perl faults them if its -w option is set. I have changed PCRE so
2025 that it faults them when PCRE_EXTRA is set.
2026
2027 5. Added support for \N, which always matches any character other than
2028 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2029
2030 6. When compiling pcregrep with newer versions of gcc which may have
2031 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2032 declared with attribute warn_unused_result" were given. Just casting the
2033 result to (void) does not stop the warnings; a more elaborate fudge is
2034 needed. I've used a macro to implement this.
2035
2036 7. Minor change to pcretest.c to avoid a compiler warning.
2037
2038 8. Added four artifical Unicode properties to help with an option to make
2039 \s etc use properties (see next item). The new properties are: Xan
2040 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2041
2042 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2043 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2044 this option. Modified pcretest to add /W to test this facility. Added
2045 REG_UCP to make it available via the POSIX interface.
2046
2047 10. Added --line-buffered to pcregrep.
2048
2049 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2050 studied, and the match started with a letter with a code point greater than
2051 127 whose first byte was different to the first byte of the other case of
2052 the letter, the other case of this starting letter was not recognized
2053 (#976).
2054
2055 12. If a pattern that was studied started with a repeated Unicode property
2056 test, for example, \p{Nd}+, there was the theoretical possibility of
2057 setting up an incorrect bitmap of starting bytes, but fortunately it could
2058 not have actually happened in practice until change 8 above was made (it
2059 added property types that matched character-matching opcodes).
2060
2061 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2062 possible starting bytes for non-anchored patterns.
2063
2064 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2065 \R, and also a number of cases that involve Unicode properties, both
2066 explicit and implicit when PCRE_UCP is set.
2067
2068 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2069 input, it could crash or give wrong results if characters with values
2070 greater than 0xc0 were present in the subject string. (Detail: it assumed
2071 UTF-8 input when processing these items.)
2072
2073 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2074 size_t is 64-bit (#991).
2075
2076 17. Added a check for running out of memory when PCRE is compiled with
2077 --disable-stack-for-recursion (#990).
2078
2079 18. If the last data line in a file for pcretest does not have a newline on
2080 the end, a newline was missing in the output.
2081
2082 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2083 less than 128) in its various bitmaps. However, there is a facility for
2084 generating tables according to the current locale when PCRE is compiled. It
2085 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2086 characters, are recognized by isspace() and therefore were getting set in
2087 these tables, and indeed these tables seem to approximate to ISO 8859. This
2088 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2089 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2090 which of course cannot start UTF-8 characters. I have changed the code so
2091 that only real ASCII characters (less than 128) and the correct starting
2092 bytes for UTF-8 encodings are set for characters greater than 127 when in
2093 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2094 altogether.)
2095
2096 20. Added the /T option to pcretest so as to be able to run tests with non-
2097 standard character tables, thus making it possible to include the tests
2098 used for 19 above in the standard set of tests.
2099
2100 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2101 reference to a subpattern the other side of a comment that contains an
2102 opening parenthesis caused either an internal compiling error, or a
2103 reference to the wrong subpattern.
2104
2105
2106 Version 8.02 19-Mar-2010
2107 ------------------------
2108
2109 1. The Unicode data tables have been updated to Unicode 5.2.0.
2110
2111 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2112 configured.
2113
2114 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2115 original author of that file, following a query about its status.
2116
2117 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2118 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2119
2120 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2121 quantifier applied to a forward-referencing subroutine call, could compile
2122 incorrect code or give the error "internal error: previously-checked
2123 referenced subpattern not found".
2124
2125 6. Both MS Visual Studio and Symbian OS have problems with initializing
2126 variables to point to external functions. For these systems, therefore,
2127 pcre_malloc etc. are now initialized to local functions that call the
2128 relevant global functions.
2129
2130 7. There were two entries missing in the vectors called coptable and poptable
2131 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2132 I've fixed the data, and added a kludgy way of testing at compile time that
2133 the lengths are correct (equal to the number of opcodes).
2134
2135 8. Following on from 7, I added a similar kludge to check the length of the
2136 eint vector in pcreposix.c.
2137
2138 9. Error texts for pcre_compile() are held as one long string to avoid too
2139 much relocation at load time. To find a text, the string is searched,
2140 counting zeros. There was no check for running off the end of the string,
2141 which could happen if a new error number was added without updating the
2142 string.
2143
2144 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2145
2146 11. \K was not working if it appeared in an atomic group or in a group that
2147 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2148 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2149 the assertion is positive, but not if it is negative.
2150
2151 12. Change 11 fortuitously reduced the size of the stack frame used in the
2152 "match()" function of pcre_exec.c by one pointer. Forthcoming
2153 implementation of support for (*MARK) will need an extra pointer on the
2154 stack; I have reserved it now, so that the stack frame size does not
2155 decrease.
2156
2157 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2158 item in branch that calls a recursion is a subroutine call - as in the
2159 second branch in the above example - was incorrectly given the compile-
2160 time error "recursive call could loop indefinitely" because pcre_compile()
2161 was not correctly checking the subroutine for matching a non-empty string.
2162
2163 14. The checks for overrunning compiling workspace could trigger after an
2164 overrun had occurred. This is a "should never occur" error, but it can be
2165 triggered by pathological patterns such as hundreds of nested parentheses.
2166 The checks now trigger 100 bytes before the end of the workspace.
2167
2168 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2169
2170
2171 Version 8.01 19-Jan-2010
2172 ------------------------
2173
2174 1. If a pattern contained a conditional subpattern with only one branch (in
2175 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2176 computed the wrong minimum data length (which is of course zero for such
2177 subpatterns). This could cause incorrect "no match" results.
2178
2179 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2180 the pattern is reset in the first branch, pcre_compile() failed with
2181 "internal error: code overflow at offset...". This happened only when
2182 the reset was to the original external option setting. (An optimization
2183 abstracts leading options settings into an external setting, which was the
2184 cause of this.)
2185
2186 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2187 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2188 assertion pattern did not match (meaning that the assertion was true), it
2189 was incorrectly treated as false if the SKIP had been reached during the
2190 matching. This also applied to assertions used as conditions.
2191
2192 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2193 assertion subpattern, including such a pattern used as a condition,
2194 unpredictable results occurred, instead of the error return
2195 PCRE_ERROR_DFA_UITEM.
2196
2197 5. The C++ GlobalReplace function was not working like Perl for the special
2198 situation when an empty string is matched. It now does the fancy magic
2199 stuff that is necessary.
2200
2201 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2202 removed. (These were left over from very, very early versions of PCRE.)
2203
2204 7. Some cosmetic changes to the code to make life easier when compiling it
2205 as part of something else:
2206
2207 (a) Change DEBUG to PCRE_DEBUG.
2208
2209 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2210 called "current" as "current_branch", to prevent a collision with the
2211 Linux macro when compiled as a kernel module.
2212
2213 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2214 prevent a collision with the Linux macro when compiled as a kernel
2215 module.
2216
2217 8. In pcre_compile() there are some checks for integer overflows that used to
2218 cast potentially large values to (double). This has been changed to that
2219 when building, a check for int64_t is made, and if it is found, it is used
2220 instead, thus avoiding the use of floating point arithmetic. (There is no
2221 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2222 double.
2223
2224 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2225 2005 (difference between two addresses compared to an unsigned value).
2226
2227 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2228 custom one, because of the following reported problem in Windows:
2229
2230 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2231 under Win32.
2232 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2233 therefore missing the function definition.
2234 - The compiler thus generates a "C" signature for the test function.
2235 - The linker fails to find the "C" function.
2236 - PCRE fails to configure if asked to do so against libbz2.
2237
2238 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2239 messages were output:
2240
2241 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2242 rerunning libtoolize, to keep the correct libtool macros in-tree.
2243 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2244
2245 I have done both of these things.
2246
2247 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2248 most of the time, it *can* run out if it is given a pattern that contains a
2249 runaway infinite recursion. I updated the discussion in the pcrestack man
2250 page.
2251
2252 13. Now that we have gone to the x.xx style of version numbers, the minor
2253 version may start with zero. Using 08 or 09 is a bad idea because users
2254 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2255 interpreted as invalid octal numbers. I've updated the previous comment in
2256 configure.ac, and also added a check that gives an error if 08 or 09 are
2257 used.
2258
2259 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2260 causing partial matching to fail when the end of the subject matched \W
2261 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2262
2263 15. There were some discrepancies between the declarations in pcre_internal.h
2264 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2265 their definitions. The declarations used "const uschar *" and the
2266 definitions used USPTR. Even though USPTR is normally defined as "const
2267 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2268 reported that: "This difference in casting confuses some C++ compilers, for
2269 example, SunCC recognizes above declarations as different functions and
2270 generates broken code for hbpcre." I have changed the declarations to use
2271 USPTR.
2272
2273 16. GNU libtool is named differently on some systems. The autogen.sh script now
2274 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2275 (FreeBSD).
2276
2277 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2278 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2279 comment: "Figure out how to create a longlong from a string: strtoll and
2280 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2281 instance, but it only takes 2 args instead of 3!"
2282
2283 18. A subtle bug concerned with back references has been fixed by a change of
2284 specification, with a corresponding code fix. A pattern such as
2285 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2286 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2287 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2288 same bug. Such groups have to be quantified to be useful, or contained
2289 inside another quantified group. (If there's no repetition, the reference
2290 can never match.) The problem arises because, having left the group and
2291 moved on to the rest of the pattern, a later failure that backtracks into
2292 the group uses the captured value from the final iteration of the group
2293 rather than the correct earlier one. I have fixed this in PCRE by forcing
2294 any group that contains a reference to itself to be an atomic group; that
2295 is, there cannot be any backtracking into it once it has completed. This is
2296 similar to recursive and subroutine calls.
2297
2298
2299 Version 8.00 19-Oct-09
2300 ----------------------
2301
2302 1. The table for translating pcre_compile() error codes into POSIX error codes
2303 was out-of-date, and there was no check on the pcre_compile() error code
2304 being within the table. This could lead to an OK return being given in
2305 error.
2306
2307 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2308 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2309 in a Windows environment.
2310
2311 3. The pcregrep --count option prints the count for each file even when it is
2312 zero, as does GNU grep. However, pcregrep was also printing all files when
2313 --files-with-matches was added. Now, when both options are given, it prints
2314 counts only for those files that have at least one match. (GNU grep just
2315 prints the file name in this circumstance, but including the count seems
2316 more useful - otherwise, why use --count?) Also ensured that the
2317 combination -clh just lists non-zero counts, with no names.
2318
2319 4. The long form of the pcregrep -F option was incorrectly implemented as
2320 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2321 but it seems right to fix it, and I didn't think it was worth preserving
2322 the old behaviour.
2323
2324 5. The command line items --regex=pattern and --regexp=pattern were not
2325 recognized by pcregrep, which required --regex pattern or --regexp pattern
2326 (with a space rather than an '='). The man page documented the '=' forms,
2327 which are compatible with GNU grep; these now work.
2328
2329 6. No libpcreposix.pc file was created for pkg-config; there was just
2330 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2331
2332 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2333 when UCP support is not needed, by modifying the Python script that
2334 generates it from Unicode data files. This should not matter if the module
2335 is correctly used as a library, but I received one complaint about 50K of
2336 unwanted data. My guess is that the person linked everything into his
2337 program rather than using a library. Anyway, it does no harm.
2338
2339 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2340 was a minimum greater than 1 for a wide character in a possessive
2341 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2342 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2343 character. Chaos in the form of incorrect output or a compiling loop could
2344 result.
2345
2346 9. The restrictions on what a pattern can contain when partial matching is
2347 requested for pcre_exec() have been removed. All patterns can now be
2348 partially matched by this function. In addition, if there are at least two
2349 slots in the offset vector, the offset of the earliest inspected character
2350 for the match and the offset of the end of the subject are set in them when
2351 PCRE_ERROR_PARTIAL is returned.
2352
2353 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2354 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2355 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2356 and may be more useful for multi-segment matching.
2357
2358 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2359 used to be given if ever the end of the subject was reached; now it is
2360 given only if matching could not proceed because another character was
2361 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2362 string "Z", which now yields "no match" instead of "partial match". In the
2363 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2364 final character ended with (*FAIL).
2365
2366 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2367 if the pattern had a "must contain" character that was already found in the
2368 earlier partial match, unless partial matching was again requested. For
2369 example, with the pattern /dog.(body)?/, the "must contain" character is
2370 "g". If the first part-match was for the string "dog", restarting with
2371 "sbody" failed. This bug has been fixed.
2372
2373 13. The string returned by pcre_dfa_exec() after a partial match has been
2374 changed so that it starts at the first inspected character rather than the
2375 first character of the match. This makes a difference only if the pattern
2376 starts with a lookbehind assertion or \b or \B (\K is not supported by
2377 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2378 matching functions compatible, and I think it's the right thing to do.
2379
2380 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2381 so that the demonstration program is easily available in environments where
2382 PCRE has not been installed from source.
2383
2384 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2385 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2386 library.
2387
2388 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2389 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2390 is not the first non-POSIX option to be added. Clearly some people find
2391 these options useful.
2392
2393 17. If a caller to the POSIX matching function regexec() passes a non-zero
2394 value for nmatch with a NULL value for pmatch, the value of
2395 nmatch is forced to zero.
2396
2397 18. RunGrepTest did not have a test for the availability of the -u option of
2398 the diff command, as RunTest does. It now checks in the same way as
2399 RunTest, and also checks for the -b option.
2400
2401 19. If an odd number of negated classes containing just a single character
2402 interposed, within parentheses, between a forward reference to a named
2403 subpattern and the definition of the subpattern, compilation crashed with
2404 an internal error, complaining that it could not find the referenced
2405 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2406 [The bug was that it was starting one character too far in when skipping
2407 over the character class, thus treating the ] as data rather than
2408 terminating the class. This meant it could skip too much.]
2409
2410 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2411 /g option in pcretest when the pattern contains \K, which makes it possible
2412 to have an empty string match not at the start, even when the pattern is
2413 anchored. Updated pcretest and pcredemo to use this option.
2414
2415 21. If the maximum number of capturing subpatterns in a recursion was greater
2416 than the maximum at the outer level, the higher number was returned, but
2417 with unset values at the outer level. The correct (outer level) value is
2418 now given.
2419
2420 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2421 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2422 make it do so. The string so far is captured, making this feature
2423 compatible with Perl.
2424
2425 23. The tests have been re-organized, adding tests 11 and 12, to make it
2426 possible to check the Perl 5.10 features against Perl 5.10.
2427
2428 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2429 pattern matches a fixed length string. PCRE did not allow this; now it
2430 does. Neither allows recursion.
2431
2432 25. I finally figured out how to implement a request to provide the minimum
2433 length of subject string that was needed in order to match a given pattern.
2434 (It was back references and recursion that I had previously got hung up
2435 on.) This code has now been added to pcre_study(); it finds a lower bound
2436 to the length of subject needed. It is not necessarily the greatest lower
2437 bound, but using it to avoid searching strings that are too short does give
2438 some useful speed-ups. The value is available to calling programs via
2439 pcre_fullinfo().
2440
2441 26. While implementing 25, I discovered to my embarrassment that pcretest had
2442 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2443 study optimizations had never been tested with that matching function.
2444 Oops. What is worse, even when it was passed study data, there was a bug in
2445 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2446 were also very few tests of studied patterns with pcre_dfa_exec().
2447
2448 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2449 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2450 on the other side of the coin, they are no longer allowed to have different
2451 names, because these cannot be distinguished in PCRE, and this has caused
2452 confusion. (This is a difference from Perl.)
2453
2454 28. When duplicate subpattern names are present (necessarily with different
2455 numbers, as required by 27 above), and a test is made by name in a
2456 conditional pattern, either for a subpattern having been matched, or for
2457 recursion in such a pattern, all the associated numbered subpatterns are
2458 tested, and the overall condition is true if the condition is true for any
2459 one of them. This is the way Perl works, and is also more like the way
2460 testing by number works.
2461
2462
2463 Version 7.9 11-Apr-09
2464 ---------------------
2465
2466 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2467 (pcretest), all targets were linked against these libraries. This included
2468 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2469 libraries. This caused unwanted dependencies to be created. This problem
2470 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2471 pcretest is linked with readline.
2472
2473 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2474 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2475 moved outside it again, because FALSE and TRUE are already defined in AIX,
2476 but BOOL is not.
2477
2478 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2479 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2480
2481 4. The pcregrep documentation said spaces were inserted as well as colons (or
2482 hyphens) following file names and line numbers when outputting matching
2483 lines. This is not true; no spaces are inserted. I have also clarified the
2484 wording for the --colour (or --color) option.
2485
2486 5. In pcregrep, when --colour was used with -o, the list of matching strings
2487 was not coloured; this is different to GNU grep, so I have changed it to be
2488 the same.
2489
2490 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2491 each matching line was coloured. Now it goes on to look for further matches
2492 of any of the test patterns, which is the same behaviour as GNU grep.
2493
2494 7. A pattern that could match an empty string could cause pcregrep to loop; it
2495 doesn't make sense to accept an empty string match in pcregrep, so I have
2496 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2497 seems to be how GNU grep behaves. [But see later change 40 for release
2498 8.33.]
2499
2500 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2501 start or after a newline", because the conditional assertion was not being
2502 correctly handled. The rule now is that both the assertion and what follows
2503 in the first alternative must satisfy the test.
2504
2505 9. If auto-callout was enabled in a pattern with a conditional group whose
2506 condition was an assertion, PCRE could crash during matching, both with
2507 pcre_exec() and pcre_dfa_exec().
2508
2509 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2510 used for matching.
2511
2512 11. Unicode property support in character classes was not working for
2513 characters (bytes) greater than 127 when not in UTF-8 mode.
2514
2515 12. Added the -M command line option to pcretest.
2516
2517 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2518
2519 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2520
2521 16. Added comments and documentation about mis-use of no_arg in the C++
2522 wrapper.
2523
2524 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2525 from Martin Jerabek that uses macro names for all relevant character and
2526 string constants.
2527
2528 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2529 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2530 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2531 these, but not everybody uses configure.
2532
2533 19. A conditional group that had only one branch was not being correctly
2534 recognized as an item that could match an empty string. This meant that an
2535 enclosing group might also not be so recognized, causing infinite looping
2536 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2537 with the subject "ab", where knowledge that the repeated group can match
2538 nothing is needed in order to break the loop.
2539
2540 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2541 exec(), but without supplying a callout function, matching went wrong.
2542
2543 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2544 leak if the size of the offset vector was greater than 30. When the vector
2545 is smaller, the saved offsets during recursion go onto a local stack
2546 vector, but for larger vectors malloc() is used. It was failing to free
2547 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2548 error, in fact).
2549
2550 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2551 heapframe that is used only when UTF-8 support is enabled. This caused no
2552 problem, but was untidy.
2553
2554 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2555 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2556 included within another project.
2557
2558 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2559 slightly modified by me:
2560
2561 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2562 not building pcregrep.
2563
2564 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2565 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2566
2567 25. Forward references, both numeric and by name, in patterns that made use of
2568 duplicate group numbers, could behave incorrectly or give incorrect errors,
2569 because when scanning forward to find the reference group, PCRE was not
2570 taking into account the duplicate group numbers. A pattern such as
2571 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2572
2573 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2574 the feature of a custom pointer more persuasive (as requested by a user).
2575
2576 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2577 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2578 pre-defined.
2579
2580 28. Added support for (*UTF8) at the start of a pattern.
2581
2582 29. Arrange for flags added by the "release type" setting in CMake to be shown
2583 in the configuration summary.
2584
2585
2586 Version 7.8 05-Sep-08
2587 ---------------------
2588
2589 1. Replaced UCP searching code with optimized version as implemented for Ad
2590 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2591 stage table and inline lookup instead of a function, giving speed ups of 2
2592 to 5 times on some simple patterns that I tested. Permission was given to
2593 distribute the MultiStage2.py script that generates the tables (it's not in
2594 the tarball, but is in the Subversion repository).
2595
2596 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2597 scripts.
2598
2599 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2600 a group with a zero qualifier. The result of the study could be incorrect,
2601 or the function might crash, depending on the pattern.
2602
2603 4. Caseless matching was not working for non-ASCII characters in back
2604 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2605 It now works when Unicode Property Support is available.
2606
2607 5. In pcretest, an escape such as \x{de} in the data was always generating
2608 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2609 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2610 truncation.
2611
2612 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2613
2614 7. Added two (int) casts to pcregrep when printing the difference of two
2615 pointers, in case they are 64-bit values.
2616
2617 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2618 test 2 if it fails.
2619
2620 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2621 and a #define of that name to empty if it is not externally set. This is to
2622 allow users of MSVC to set it if necessary.
2623
2624 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2625 the convenience functions in the pcre_get.c source file.
2626
2627 11. An option change at the start of a pattern that had top-level alternatives
2628 could cause overwriting and/or a crash. This command provoked a crash in
2629 some environments:
2630
2631 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2632
2633 This potential security problem was recorded as CVE-2008-2371.
2634
2635 12. For a pattern where the match had to start at the beginning or immediately
2636 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2637 pcre_dfa_exec() could read past the end of the passed subject if there was
2638 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2639 pcretest so that it places the subject at the end of its malloc-ed buffer.
2640
2641 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2642 exec() might read past the end of the data buffer in UTF-8 mode.
2643
2644 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2645 the data contained the byte 0x85 as part of a UTF-8 character within its
2646 first line. This applied both to normal and DFA matching.
2647
2648 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2649 /^[^d]*?$/8 failed to match "abc".
2650
2651 16. Added a missing copyright notice to pcrecpp_internal.h.
2652
2653 17. Make it more clear in the documentation that values returned from
2654 pcre_exec() in ovector are byte offsets, not character counts.
2655
2656 18. Tidied a few places to stop certain compilers from issuing warnings.
2657
2658 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2659 supplied by Stefan Weber. I made a further small update for 7.8 because
2660 there is a change of source arrangements: the pcre_searchfuncs.c module is
2661 replaced by pcre_ucd.c.
2662
2663
2664 Version 7.7 07-May-08
2665 ---------------------
2666
2667 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2668 a string to a long long, pretend we don't even have a long long." This is
2669 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2670
2671 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2672 pre-7.6 versions, which defined a global no_arg variable instead of putting
2673 it in the RE class. (See also #8 below.)
2674
2675 3. Remove a line of dead code, identified by coverity and reported by Nuno
2676 Lopes.
2677
2678 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2679
2680 (1) The include/exclude patterns were being applied to the whole pathnames
2681 of files, instead of just to the final components.
2682
2683 (2) If there was more than one level of directory, the subdirectories were
2684 skipped unless they satisfied the include/exclude conditions. This is
2685 inconsistent with GNU grep (and could even be seen as contrary to the
2686 pcregrep specification - which I improved to make it absolutely clear).
2687 The action now is always to scan all levels of directory, and just
2688 apply the include/exclude patterns to regular files.
2689
2690 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2691 --exclude_dir in the tests to avoid scanning .svn directories.
2692
2693 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2694 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2695 doesn't support NULs in patterns.
2696
2697 7. Added some missing "const"s to declarations of static tables in
2698 pcre_compile.c and pcre_dfa_exec.c.
2699
2700 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2701 caused by fix #2 above. (Subsequently also a second patch to fix the
2702 first patch. And a third patch - this was a messy problem.)
2703
2704 9. Applied Craig's patch to remove the use of push_back().
2705
2706 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2707 matching function regexec().
2708
2709 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2710 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2711 references. PCRE supports relative numbers with this syntax (I don't think
2712 Oniguruma does).
2713
2714 12. Previously, a group with a zero repeat such as (...){0} was completely
2715 omitted from the compiled regex. However, this means that if the group
2716 was called as a subroutine from elsewhere in the pattern, things went wrong
2717 (an internal error was given). Such groups are now left in the compiled
2718 pattern, with a new opcode that causes them to be skipped at execution
2719 time.
2720
2721 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2722 to the way PCRE behaves:
2723
2724 (a) A lone ] character is dis-allowed (Perl treats it as data).
2725
2726 (b) A back reference to an unmatched subpattern matches an empty string
2727 (Perl fails the current match path).
2728
2729 (c) A data ] in a character class must be notated as \] because if the
2730 first data character in a class is ], it defines an empty class. (In
2731 Perl it is not possible to have an empty class.) The empty class []
2732 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2733 The negative empty class [^] matches any one character, independently
2734 of the DOTALL setting.
2735
2736 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2737 non-existent subpattern following a character class starting with ']' and
2738 containing () gave an internal compiling error instead of "reference to
2739 non-existent subpattern". Fortunately, when the pattern did exist, the
2740 compiled code was correct. (When scanning forwards to check for the
2741 existence of the subpattern, it was treating the data ']' as terminating
2742 the class, so got the count wrong. When actually compiling, the reference
2743 was subsequently set up correctly.)
2744
2745 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2746 it was being rejected as not supported by pcre_dfa_exec(), even though
2747 other assertions are supported. I have made pcre_dfa_exec() support
2748 (*FAIL).
2749
2750 16. The implementation of 13c above involved the invention of a new opcode,
2751 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2752 cannot be changed at match time, I realized I could make a small
2753 improvement to matching performance by compiling OP_ALLANY instead of
2754 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2755 on the OP_ANY path.
2756
2757 17. Compiling pcretest on Windows with readline support failed without the
2758 following two fixes: (1) Make the unistd.h include conditional on
2759 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2760
2761 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2762 ncurses library to be included for pcretest when ReadLine support is
2763 requested, but also to allow for it to be overridden. This patch came from
2764 Daniel Bergström.
2765
2766 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2767 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2768 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2769 spotting this.
2770
2771
2772 Version 7.6 28-Jan-08
2773 ---------------------
2774
2775 1. A character class containing a very large number of characters with
2776 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2777 overflow.
2778
2779 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2780 HAVE_LONG_LONG is not defined.
2781
2782 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2783 bring them up to date and include new features. This patch includes:
2784
2785 - Fixed PH's badly added libz and libbz2 support.
2786 - Fixed a problem with static linking.
2787 - Added pcredemo. [But later removed - see 7 below.]
2788 - Fixed dftables problem and added an option.
2789 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2790 HAVE_LONG_LONG.
2791 - Added readline support for pcretest.
2792 - Added an listing of the option settings after cmake has run.
2793
2794 4. A user submitted a patch to Makefile that makes it easy to create
2795 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2796 Makefile.am that cause it to include this special target, without
2797 affecting anything else. Note that the same mingw target plus all
2798 the other distribution libraries and programs are now supported
2799 when configuring with CMake (see 6 below) instead of with
2800 Configure/Make.
2801
2802 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2803 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2804 exported in the Windows port". It has not yet been confirmed that the patch
2805 solves the problem, but it does no harm.
2806
2807 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2808 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2809 with CMake, and also correct the comment about stack recursion.
2810
2811 7. Remove the automatic building of pcredemo from the ./configure system and
2812 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2813 of a program that users should build themselves after PCRE is installed, so
2814 building it automatically is not really right. What is more, it gave
2815 trouble in some build environments.
2816
2817 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2818
2819
2820 Version 7.5 10-Jan-08
2821 ---------------------
2822
2823 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2824 values in parens when parsing an RE using the C++ wrapper."
2825
2826 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2827 Characters greater than 255 were excluded from the class instead of being
2828 included.
2829
2830 3. The same bug as (2) above applied to negated POSIX classes such as
2831 [:^space:].
2832
2833 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2834 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2835 I have changed it.
2836
2837 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2838 first named subpattern) and a construct such as (?&a) would reference the
2839 first named subpattern whose name started with "a" (in other words, the
2840 length check was missing). Both these problems are fixed. "Subpattern name
2841 expected" is now given for (?&) (a zero-length name), and this patch also
2842 makes it give the same error for \k'' (previously it complained that that
2843 was a reference to a non-existent subpattern).
2844
2845 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2846 this is right because (?- can be followed by option settings as well as by
2847 digits. I have, however, made the messages clearer.
2848
2849 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2850 than the number used in the conditional) now cause a compile-time error.
2851 This is actually not compatible with Perl, which accepts such patterns, but
2852 treats the conditional as always being FALSE (as PCRE used to), but it
2853 seems to me that giving a diagnostic is better.
2854
2855 8. Change "alphameric" to the more common word "alphanumeric" in comments
2856 and messages.
2857
2858 9. Fix two occurrences of "backslash" in comments that should have been
2859 "backspace".
2860
2861 10. Remove two redundant lines of code that can never be obeyed (their function
2862 was moved elsewhere).
2863
2864 11. The program that makes PCRE's Unicode character property table had a bug
2865 which caused it to generate incorrect table entries for sequences of
2866 characters that have the same character type, but are in different scripts.
2867 It amalgamated them into a single range, with the script of the first of
2868 them. In other words, some characters were in the wrong script. There were
2869 thirteen such cases, affecting characters in the following ranges:
2870
2871 U+002b0 - U+002c1
2872 U+0060c - U+0060d
2873 U+0061e - U+00612
2874 U+0064b - U+0065e
2875 U+0074d - U+0076d
2876 U+01800 - U+01805
2877 U+01d00 - U+01d77
2878 U+01d9b - U+01dbf
2879 U+0200b - U+0200f
2880 U+030fc - U+030fe
2881 U+03260 - U+0327f
2882 U+0fb46 - U+0fbb1
2883 U+10450 - U+1049d
2884
2885 12. The -o option (show only the matching part of a line) for pcregrep was not
2886 compatible with GNU grep in that, if there was more than one match in a
2887 line, it showed only the first of them. It now behaves in the same way as
2888 GNU grep.
2889
2890 13. If the -o and -v options were combined for pcregrep, it printed a blank
2891 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2892 does the same. The return code can be used to tell if there were any
2893 non-matching lines.
2894
2895 14. Added --file-offsets and --line-offsets to pcregrep.
2896
2897 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2898 infinitely looping recursion. The bug was that positive lookaheads were not
2899 being skipped when checking for a possible empty match (negative lookaheads
2900 and both kinds of lookbehind were skipped).
2901
2902 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2903 inclusion of <windows.h> to before rather than after the definition of
2904 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2905
2906 17. Specifying a possessive quantifier with a specific limit for a Unicode
2907 character property caused pcre_compile() to compile bad code, which led at
2908 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2909 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2910 caused the error; without that there was no problem.
2911
2912 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2913
2914 19. Added --enable-pcretest-libreadline.
2915
2916 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2917 RE::GlobalReplace(). As a result, the number of replacements returned was
2918 double what it should be. I removed one of the increments, but Craig sent a
2919 later patch that removed the other one (the right fix) and added unit tests
2920 that check the return values (which was not done before).
2921
2922 21. Several CMake things:
2923
2924 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2925 the names libpcre and libpcreposix, not just pcre and pcreposix.
2926
2927 (2) The above change means that pcretest and pcregrep are now correctly
2928 linked with the newly-built libraries, not previously installed ones.
2929
2930 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2931
2932 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2933 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2934 UTF-8 newline character). The key issue is that the pattern starts .*;
2935 this means that the match must be either at the beginning, or after a
2936 newline. The bug was in the code for advancing after a failed match and
2937 checking that the new position followed a newline. It was not taking
2938 account of UTF-8 characters correctly.
2939
2940 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2941 character classes. PCRE was not treating the sequence [:...:] as a
2942 character class unless the ... were all letters. Perl, however, seems to
2943 allow any characters between [: and :], though of course it rejects as
2944 unknown any "names" that contain non-letters, because all the known class
2945 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2946 for example, whereas PCRE did not - it did not recognize a POSIX character
2947 class. This seemed a bit dangerous, so the code has been changed to be
2948 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2949 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2950 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2951 Perl does, and where it didn't before.
2952
2953 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2954 Windows environments %n is disabled by default.
2955
2956
2957 Version 7.4 21-Sep-07
2958 ---------------------
2959
2960 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2961 means that a class such as [\s] counted as "explicit reference to CR or
2962 LF". That isn't really right - the whole point of the change was to try to
2963 help when there was an actual mention of one of the two characters. So now
2964 the change happens only if \r or \n (or a literal CR or LF) character is
2965 encountered.
2966
2967 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2968 of both had grown to the point where there were only 3 bits left.
2969 Fortunately, there was spare space in the data structure, and so I have
2970 moved the internal flags into a new 16-bit field to free up more option
2971 bits.
2972
2973 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2974 but did not set the internal JCHANGED flag - either of these is enough to
2975 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2976 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2977 start sets both bits.
2978
2979 4. Added options (at build time, compile time, exec time) to change \R from
2980 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2981
2982 5. doc/pcresyntax.html was missing from the distribution.
2983
2984 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2985 compatibility, even though it is no longer used.
2986
2987 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2988 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2989 windows.h file is present (where different names are used). [This was
2990 reversed later after testing - see 16 below.]
2991
2992 8. Changed all #include <config.h> to #include "config.h". There were also
2993 some further <pcre.h> cases that I changed to "pcre.h".
2994
2995 9. When pcregrep was used with the --colour option, it missed the line ending
2996 sequence off the lines that it output.
2997
2998 10. It was pointed out to me that arrays of string pointers cause lots of
2999 relocations when a shared library is dynamically loaded. A technique of
3000 using a single long string with a table of offsets can drastically reduce
3001 these. I have refactored PCRE in four places to do this. The result is
3002 dramatic:
3003
3004 Originally: 290
3005 After changing UCP table: 187
3006 After changing error message table: 43
3007 After changing table of "verbs" 36
3008 After changing table of Posix names 22
3009
3010 Thanks to the folks working on Gregex for glib for this insight.
3011
3012 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3013 unicode-properties was also set.
3014
3015 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3016
3017 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3018 checked only for CRLF.
3019
3020 14. Added casts to pcretest.c to avoid compiler warnings.
3021
3022 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3023
3024 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3025 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3026 entirely. This removes changes made in 7 above.
3027
3028 17. The CMake files have been updated, and there is now more information about
3029 building with CMake in the NON-UNIX-USE document.
3030
3031
3032 Version 7.3 28-Aug-07
3033 ---------------------
3034
3035 1. In the rejigging of the build system that eventually resulted in 7.1, the
3036 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3037 brackets there is not right, since it causes compilers to look for an
3038 installed pcre.h, not the version that is in the source that is being
3039 compiled (which of course may be different). I have changed it back to:
3040
3041 #include "pcre.h"
3042
3043 I have a vague recollection that the change was concerned with compiling in
3044 different directories, but in the new build system, that is taken care of
3045 by the VPATH setting the Makefile.
3046
3047 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3048 when the subject happened to end in the byte 0x85 (e.g. if the last
3049 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3050 characters but of course it shouldn't be taken as a newline when it is part
3051 of another character. The bug was that, for an unlimited repeat of . in
3052 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3053 characters when looking for a newline.
3054
3055 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3056
3057 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3058 in debug output.
3059
3060 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3061 long printing in the pcrecpp unittest when running under MinGW.
3062
3063 6. ESC_K was left out of the EBCDIC table.
3064
3065 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3066 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3067 limit also applies to "virtual nesting" when a pattern is recursive, and in
3068 this case 1000 isn't so big. I have been able to remove this limit at the
3069 expense of backing off one optimization in certain circumstances. Normally,
3070 when pcre_exec() would call its internal match() function recursively and
3071 immediately return the result unconditionally, it uses a "tail recursion"
3072 feature to save stack. However, when a subpattern that can match an empty
3073 string has an unlimited repetition quantifier, it no longer makes this
3074 optimization. That gives it a stack frame in which to save the data for
3075 checking that an empty string has been matched. Previously this was taken
3076 from the 1000-entry workspace that had been reserved. So now there is no
3077 explicit limit, but more stack is used.
3078
3079 8. Applied Daniel's patches to solve problems with the import/export magic
3080 syntax that is required for Windows, and which was going wrong for the
3081 pcreposix and pcrecpp parts of the library. These were overlooked when this
3082 problem was solved for the main library.
3083
3084 9. There were some crude static tests to avoid integer overflow when computing
3085 the size of patterns that contain repeated groups with explicit upper
3086 limits. As the maximum quantifier is 65535, the maximum group length was
3087 set at 30,000 so that the product of these two numbers did not overflow a
3088 32-bit integer. However, it turns out that people want to use groups that
3089 are longer than 30,000 bytes (though not repeat them that many times).
3090 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3091 made it possible to implement the integer overflow checks in a much more
3092 dynamic way, which I have now done. The artificial limitation on group
3093 length has been removed - we now have only the limit on the total length of
3094 the compiled pattern, which depends on the LINK_SIZE setting.
3095
3096 10. Fixed a bug in the documentation for get/copy named substring when
3097 duplicate names are permitted. If none of the named substrings are set, the
3098 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3099 empty string.
3100
3101 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3102 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3103 because the ] is interpreted as the first data character and the
3104 terminating ] is not found. PCRE has been made compatible with Perl in this
3105 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3106 cause memory overwriting.
3107
3108 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3109 string has been matched (to stop an infinite loop). It was not recognizing
3110 a conditional subpattern that could match an empty string if that
3111 subpattern was within another subpattern. For example, it looped when
3112 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3113 condition was not nested. This bug has been fixed.
3114
3115 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3116 past the start of the subject in the presence of bytes with the top bit
3117 set, for example "\x8aBCD".
3118
3119 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3120 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3121
3122 14. Optimized (?!) to (*FAIL).
3123
3124 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3125 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3126 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3127 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3128 does: it's just the validity check that is more restrictive.
3129
3130 16. Inserted checks for integer overflows during escape sequence (backslash)
3131 processing, and also fixed erroneous offset values for syntax errors during
3132 backslash processing.
3133
3134 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3135 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3136
3137 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3138 caused an overrun.
3139
3140 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3141 something other than just ASCII characters) inside a group that had an
3142 unlimited repeat caused a loop at compile time (while checking to see
3143 whether the group could match an empty string).
3144
3145 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3146 [\P{Any}] did so. (Error in the code for printing property names.)
3147
3148 21. An orphan \E inside a character class could cause a crash.
3149
3150 22. A repeated capturing bracket such as (A)? could cause a wild memory
3151 reference during compilation.
3152
3153 23. There are several functions in pcre_compile() that scan along a compiled
3154 expression for various reasons (e.g. to see if it's fixed length for look
3155 behind). There were bugs in these functions when a repeated \p or \P was
3156 present in the pattern. These operators have additional parameters compared
3157 with \d, etc, and these were not being taken into account when moving along
3158 the compiled data. Specifically:
3159
3160 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3161 length.
3162
3163 (b) An item such as \pL+ within a repeated group could cause crashes or
3164 loops.
3165
3166 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3167 "reference to non-existent subpattern" error.
3168
3169 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3170
3171 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3172 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3173
3174 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3175
3176 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3177 character were causing crashes (broken optimization).
3178
3179 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3180 \p or \P) caused a compile-time loop.
3181
3182 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3183 break. For example, the unstudied pattern [\r\n]A does not match the string
3184 "\r\nA" because change 7.0/46 below moves the current point on by two
3185 characters after failing to match at the start. However, the pattern \nA
3186 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3187 the same is true. There doesn't seem any very clean way out of this, but
3188 what I have chosen to do makes the common cases work: PCRE now takes note
3189 of whether there can be an explicit match for \r or \n anywhere in the
3190 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3191 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3192 pattern has explicit CR or LF references.
3193
3194 29. Added (*CR) etc for changing newline setting at start of pattern.
3195
3196
3197 Version 7.2 19-Jun-07
3198 ---------------------
3199
3200 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3201 which is apparently normally available under Windows.
3202
3203 2. Re-jig the pcregrep tests with different newline settings in an attempt
3204 to make them independent of the local environment's newline setting.
3205
3206 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3207
3208 4. Some of the "internals" tests were previously cut out when the link size
3209 was not 2, because the output contained actual offsets. The recent new
3210 "Z" feature of pcretest means that these can be cut out, making the tests
3211 usable with all link sizes.
3212
3213 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3214 stack recursion. This gives a massive performance boost under BSD, but just
3215 a small improvement under Linux. However, it saves one field in the frame
3216 in all cases.
3217
3218 6. Added more features from the forthcoming Perl 5.10:
3219
3220 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3221 recursion call. It refers to the nth most recently opened parentheses.
3222
3223 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3224 to be opened parentheses.
3225
3226 (c) Conditions that refer to capturing parentheses can be specified
3227 relatively, for example, (?(-2)... or (?(+3)...
3228
3229 (d) \K resets the start of the current match so that everything before
3230 is not part of it.
3231
3232 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3233
3234 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3235 reference syntax.
3236
3237 (g) (?| introduces a group in which the numbering of parentheses in each
3238 alternative starts with the same number.
3239
3240 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3241
3242 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3243 PCRE_INFO_JCHANGED.
3244
3245 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3246 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3247 for detecting groups that can match an empty string.
3248
3249 9. A pattern with a very large number of alternatives (more than several
3250 hundred) was running out of internal workspace during the pre-compile
3251 phase, where pcre_compile() figures out how much memory will be needed. A
3252 bit of new cunning has reduced the workspace needed for groups with
3253 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3254 workspace instead of running out of the 4096 that are available.
3255
3256 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3257
3258 11. Applied patch from Google to remove an optimization that didn't quite work.
3259 The report of the bug said:
3260
3261 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3262 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3263 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3264
3265 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3266 it matched the wrong number of bytes.
3267
3268
3269 Version 7.1 24-Apr-07
3270 ---------------------
3271
3272 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3273 that is more "standard", making use of automake and other Autotools. There
3274 is some re-arrangement of the files and adjustment of comments consequent
3275 on this.
3276
3277 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3278 for recursive directory scanning broke on some systems because the files
3279 are not scanned in any specific order and on different systems the order
3280 was different. A call to "sort" has been inserted into RunGrepTest for the
3281 approprate test as a short-term fix. In the longer term there may be an
3282 alternative.
3283
3284 3. I had an email from Eric Raymond about problems translating some of PCRE's
3285 man pages to HTML (despite the fact that I distribute HTML pages, some
3286 people do their own conversions for various reasons). The problems
3287 concerned the use of low-level troff macros .br and .in. I have therefore
3288 removed all such uses from the man pages (some were redundant, some could
3289 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3290 HTML has been updated to handle .nf/.fi and to complain if it encounters
3291 .br or .in.
3292
3293 4. Updated comments in configure.ac that get placed in config.h.in and also
3294 arranged for config.h to be included in the distribution, with the name
3295 config.h.generic, for the benefit of those who have to compile without
3296 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3297
3298 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3299 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3300 makevp.bat for the current PCRE, using the additional files
3301 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3302
3303 6. A Windows user reported a minor discrepancy with test 2, which turned out
3304 to be caused by a trailing space on an input line that had got lost in his
3305 copy. The trailing space was an accident, so I've just removed it.
3306
3307 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3308 that is needed.
3309
3310 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3311 as "const" (a) because they are and (b) because it helps the PHP
3312 maintainers who have recently made a script to detect big data structures
3313 in the php code that should be moved to the .rodata section. I remembered
3314 to update Builducptable as well, so it won't revert if ucptable.h is ever
3315 re-created.
3316
3317 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3318 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3319 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3320 support is not required. This saves 1.5-2K of code, which is important in
3321 some applications.
3322
3323 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3324 so as not to refer to the tables, even though these functions will never be
3325 called when UTF-8 support is disabled. Otherwise there are problems with a
3326 shared library.
3327
3328 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3329
3330 (a) It was defining its arguments as char * instead of void *.
3331
3332 (b) It was assuming that all moves were upwards in memory; this was true
3333 a long time ago when I wrote it, but is no longer the case.
3334
3335 The emulated memove() is provided for those environments that have neither
3336 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3337 is clearly not the case, as these two bugs were recently reported.
3338
3339 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3340 and Detrail to create the HTML documentation, the .txt form of the man
3341 pages, and it removes trailing spaces from listed files. It also creates
3342 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3343 case, it wraps all the #defines with #ifndefs. This script should be run
3344 before "make dist".
3345
3346 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3347 with Unicode property support.
3348
3349 (a) For a maximizing quantifier, if the two different cases of the
3350 character were of different lengths in their UTF-8 codings (there are
3351 some cases like this - I found 11), and the matching function had to
3352 back up over a mixture of the two cases, it incorrectly assumed they
3353 were both the same length.
3354
3355 (b) When PCRE was configured to use the heap rather than the stack for
3356 recursion during matching, it was not correctly preserving the data for
3357 the other case of a UTF-8 character when checking ahead for a match
3358 while processing a minimizing repeat. If the check also involved
3359 matching a wide character, but failed, corruption could cause an
3360 erroneous result when trying to check for a repeat of the original
3361 character.
3362
3363 13. Some tidying changes to the testing mechanism:
3364
3365 (a) The RunTest script now detects the internal link size and whether there
3366 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3367 values substituted by "configure". (The RunGrepTest script already did
3368 this for UTF-8.) The configure.ac script no longer substitutes the
3369 relevant variables.
3370
3371 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3372 with length and offset values. This means that the output is different
3373 for different internal link sizes. Test 2 is skipped for link sizes
3374 other than 2 because of this, bypassing the problem. Unfortunately,
3375 there was also a test in test 3 (the locale tests) that used /B and
3376 failed for link sizes other than 2. Rather than cut the whole test out,
3377 I have added a new /Z option to pcretest that replaces the length and
3378 offset values with spaces. This is now used to make test 3 independent
3379 of link size. (Test 2 will be tidied up later.)
3380
3381 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3382 segmentation fault instead of returning the appropriate error message.
3383
3384 15. In multiline mode when the newline sequence was set to "any", the pattern
3385 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3386 This doesn't seem right; it now treats the CRLF combination as the line
3387 ending, and so does not match in that case. It's only a pattern such as ^$
3388 that would hit this one: something like ^ABC$ would have failed after \r
3389 and then tried again after \r\n.
3390
3391 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3392 in an attempt to make files that differ only in their line terminators
3393 compare equal. This works on Linux.
3394
3395 17. Under certain error circumstances pcregrep might try to free random memory
3396 as it exited. This is now fixed, thanks to valgrind.
3397
3398 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3399 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3400 was because its rules for how to advance for /g after matching an empty
3401 string at the end of a line did not allow for this case. They now check for
3402 it specially.
3403
3404 20. pcretest is supposed to handle patterns and data of any length, by
3405 extending its buffers when necessary. It was getting this wrong when the
3406 buffer for a data line had to be extended.
3407
3408 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3409 CRLF as a newline sequence.
3410
3411 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3412 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3413 I have nevertheless tidied it up.
3414
3415 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3416
3417 24. Added a man page for pcre-config.
3418
3419
3420 Version 7.0 19-Dec-06
3421 ---------------------
3422
3423 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3424 moving to gcc 4.1.1.
3425
3426 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3427 sys/time.h, which is documented as needed for this function. It doesn't
3428 seem to matter on Linux, but it showed up on some releases of OS X.
3429
3430 3. It seems that there are systems where bytes whose values are greater than
3431 127 match isprint() in the "C" locale. The "C" locale should be the
3432 default when a C program starts up. In most systems, only ASCII printing
3433 characters match isprint(). This difference caused the output from pcretest
3434 to vary, making some of the tests fail. I have changed pcretest so that:
3435
3436 (a) When it is outputting text in the compiled version of a pattern, bytes
3437 other than 32-126 are always shown as hex escapes.
3438
3439 (b) When it is outputting text that is a matched part of a subject string,
3440 it does the same, unless a different locale has been set for the match
3441 (using the /L modifier). In this case, it uses isprint() to decide.
3442
3443 4. Fixed a major bug that caused incorrect computation of the amount of memory
3444 required for a compiled pattern when options that changed within the
3445 pattern affected the logic of the preliminary scan that determines the
3446 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3447 that the computed length was too small. The symptoms of this bug were
3448 either the PCRE error "internal error: code overflow" from pcre_compile(),
3449 or a glibc crash with a message such as "pcretest: free(): invalid next
3450 size (fast)". Examples of patterns that provoked this bug (shown in
3451 pcretest format) are:
3452
3453 /(?-x: )/x
3454 /(?x)(?-x: \s*#\s*)/
3455 /((?i)[\x{c0}])/8
3456 /(?i:[\x{c0}])/8
3457
3458 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3459 is now done differently.
3460
3461 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3462 wrapper classes; (b) implement a new function in the C++ scanner that is
3463 more efficient than the old way of doing things because it avoids levels of
3464 recursion in the regex matching; (c) add a paragraph to the documentation
3465 for the FullMatch() function.
3466
3467 6. The escape sequence \n was being treated as whatever was defined as
3468 "newline". Not only was this contrary to the documentation, which states
3469 that \n is character 10 (hex 0A), but it also went horribly wrong when
3470 "newline" was defined as CRLF. This has been fixed.
3471
3472 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3473 was being set to -1 for the "end of line" case (supposedly a value that no
3474 character can have). Though this value is never used (the check for end of
3475 line is "zero bytes in current character"), it caused compiler complaints.
3476 I've changed it to 0xffffffff.
3477
3478 8. In pcre_version.c, the version string was being built by a sequence of
3479 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3480 string (as it is for production releases) called a macro with an empty
3481 argument. The C standard says the result of this is undefined. The gcc
3482 compiler treats it as an empty string (which was what was wanted) but it is
3483 reported that Visual C gives an error. The source has been hacked around to
3484 avoid this problem.
3485
3486 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3487 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3488 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3489 of them did).
3490
3491 10. Originally, pcretest opened its input and output without "b"; then I was
3492 told that "b" was needed in some environments, so it was added for release
3493 5.0 to both the input and output. (It makes no difference on Unix-like
3494 systems.) Later I was told that it is wrong for the input on Windows. I've
3495 now abstracted the modes into two macros, to make it easier to fiddle with
3496 them, and removed "b" from the input mode under Windows.
3497
3498 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3499
3500 12. Added -help and --help to pcretest as an official way of being reminded
3501 of the options.
3502
3503 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3504 and pcrecpp.cc because they annoy compilers at high warning levels.
3505
3506 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3507
3508 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3509 scripts are not C programs :-) and which was not noticed because it works
3510 on Linux.
3511
3512 16. pcretest is supposed to handle any length of pattern and data line (as one
3513 line or as a continued sequence of lines) by extending its input buffer if
3514 necessary. This feature was broken for very long pattern lines, leading to
3515 a string of junk being passed to pcre_compile() if the pattern was longer
3516 than about 50K.
3517
3518 17. I have done a major re-factoring of the way pcre_compile() computes the
3519 amount of memory needed for a compiled pattern. Previously, there was code
3520 that made a preliminary scan of the pattern in order to do this. That was
3521 OK when PCRE was new, but as the facilities have expanded, it has become
3522 harder and harder to keep it in step with the real compile phase, and there
3523 have been a number of bugs (see for example, 4 above). I have now found a
3524 cunning way of running the real compile function in a "fake" mode that
3525 enables it to compute how much memory it would need, while actually only
3526 ever using a few hundred bytes of working memory and without too many
3527 tests of the mode. This should make future maintenance and development
3528 easier. A side effect of this work is that the limit of 200 on the nesting
3529 depth of parentheses has been removed (though this was never a serious
3530 limitation, I suspect). However, there is a downside: pcre_compile() now
3531 runs more slowly than before (30% or more, depending on the pattern). I
3532 hope this isn't a big issue. There is no effect on runtime performance.
3533
3534 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3535 newline (only possible for the last line of a file) and it was a
3536 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3537
3538 19. Added additional timing features to pcretest. (1) The -tm option now times
3539 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3540 separate command line item, by a number that specifies the number of
3541 repeats to use when timing. The default is 50000; this gives better
3542 precision, but takes uncomfortably long for very large patterns.
3543
3544 20. Extended pcre_study() to be more clever in cases where a branch of a
3545 subpattern has no definite first character. For example, (a*|b*)[cd] would
3546 previously give no result from pcre_study(). Now it recognizes that the
3547 first character must be a, b, c, or d.
3548
3549 21. There was an incorrect error "recursive call could loop indefinitely" if
3550 a subpattern (or the entire pattern) that was being tested for matching an
3551 empty string contained only one non-empty item after a nested subpattern.
3552 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3553 incorrectly, because the \d was being skipped in the check.
3554
3555 22. The pcretest program now has a new pattern option /B and a command line
3556 option -b, which is equivalent to adding /B to every pattern. This causes
3557 it to show the compiled bytecode, without the additional information that
3558 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3559 is the same as /B/I).
3560
3561 23. A new optimization is now able automatically to treat some sequences such
3562 as a*b as a*+b. More specifically, if something simple (such as a character
3563 or a simple class like \d) has an unlimited quantifier, and is followed by
3564 something that cannot possibly match the quantified thing, the quantifier
3565 is automatically "possessified".
3566
3567 24. A recursive reference to a subpattern whose number was greater than 39
3568 went wrong under certain circumstances in UTF-8 mode. This bug could also
3569 have affected the operation of pcre_study().
3570
3571 25. Realized that a little bit of performance could be had by replacing
3572 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3573
3574 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3575
3576 27. Possessive quantifiers such as a++ were previously implemented by turning
3577 them into atomic groups such as ($>a+). Now they have their own opcodes,
3578 which improves performance. This includes the automatically created ones
3579 from 23 above.
3580
3581 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3582 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3583 the first matched character to be a colon. This applied both to named and
3584 numbered groups.
3585
3586 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3587
3588 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3589 building PCRE on Windows easier, so I have included it in the distribution.
3590
3591 31. There is now a check in pcretest against a ridiculously large number being
3592 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3593 loop, the loop is abandoned.
3594
3595 32. Forward references to subpatterns in conditions such as (?(2)...) where
3596 subpattern 2 is defined later cause pcre_compile() to search forwards in
3597 the pattern for the relevant set of parentheses. This search went wrong
3598 when there were unescaped parentheses in a character class, parentheses
3599 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3600
3601 33. "Subroutine" calls and backreferences were previously restricted to
3602 referencing subpatterns earlier in the regex. This restriction has now
3603 been removed.
3604
3605 34. Added a number of extra features that are going to be in Perl 5.10. On the
3606 whole, these are just syntactic alternatives for features that PCRE had
3607 previously implemented using the Python syntax or my own invention. The
3608 other formats are all retained for compatibility.
3609
3610 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3611 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3612 also .NET compatible.
3613
3614 (b) A recursion or subroutine call to a named group can now be defined as
3615 (?&name) as well as (?P>name).
3616
3617 (c) A backreference to a named group can now be defined as \k<name> or
3618 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3619 5.10, are also .NET compatible.
3620
3621 (d) A conditional reference to a named group can now use the syntax
3622 (?(<name>) or (?('name') as well as (?(name).
3623
3624 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3625 groups (named and numbered) that are never evaluated inline, but can be
3626 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3627 is always false. There may be only one alternative in such a group.
3628
3629 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3630 as the simple (?(R). The condition is true only if the most recent
3631 recursion is that of the given number or name. It does not search out
3632 through the entire recursion stack.
3633
3634 (g) The escape \gN or \g{N} has been added, where N is a positive or
3635 negative number, specifying an absolute or relative reference.
3636
3637 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3638 some "unreachable code" warnings.
3639
3640 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3641 things, this adds five new scripts.
3642
3643 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3644 There were also incompatibilities regarding the handling of \Q..\E inside
3645 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3646 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3647
3648 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3649 matches an empty string, and forcibly breaks the loop. There were bugs in
3650 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3651 against aaaa the result was just "a" rather than "aaaa", for example. Two
3652 separate and independent bugs (that affected different cases) have been
3653 fixed.
3654
3655 39. Refactored the code to abolish the use of different opcodes for small
3656 capturing bracket numbers. This is a tidy that I avoided doing when I
3657 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3658 The new approach is not only tidier, it makes it possible to reduce the
3659 memory needed to fix the previous bug (38).
3660
3661 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3662 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3663 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3664 mode.
3665
3666 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3667 report.
3668
3669 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3670 copy construction and assignment in the C++ wrapper.
3671
3672 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3673 couple of bugs that could have given wrong results in the "--newline=crlf"
3674 case.
3675
3676 44. Added a number of casts and did some reorganization of signed/unsigned int
3677 variables following suggestions from Dair Grant. Also renamed the variable
3678 "this" as "item" because it is a C++ keyword.
3679
3680 45. Arranged for dftables to add
3681
3682 #include "pcre_internal.h"
3683
3684 to pcre_chartables.c because without it, gcc 4.x may remove the array
3685 definition from the final binary if PCRE is built into a static library and
3686 dead code stripping is activated.
3687
3688 46. For an unanchored pattern, if a match attempt fails at the start of a
3689 newline sequence, and the newline setting is CRLF or ANY, and the next two
3690 characters are CRLF, advance by two characters instead of one.
3691
3692
3693 Version 6.7 04-Jul-06
3694 ---------------------
3695
3696 1. In order to handle tests when input lines are enormously long, pcretest has
3697 been re-factored so that it automatically extends its buffers when
3698 necessary. The code is crude, but this _is_ just a test program. The
3699 default size has been increased from 32K to 50K.
3700
3701 2. The code in pcre_study() was using the value of the re argument before
3702 testing it for NULL. (Of course, in any sensible call of the function, it
3703 won't be NULL.)
3704
3705 3. The memmove() emulation function in pcre_internal.h, which is used on
3706 systems that lack both memmove() and bcopy() - that is, hardly ever -
3707 was missing a "static" storage class specifier.
3708
3709 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3710 containing an extended class (one that cannot be represented by a bitmap
3711 because it contains high-valued characters or Unicode property items, e.g.
3712 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3713 pattern, but PCRE should not loop if you do not (it no longer does).
3714 [Detail: two cases were found: (a) a repeated subpattern containing an
3715 extended class; (b) a recursive reference to a subpattern that followed a
3716 previous extended class. It wasn't skipping over the extended class
3717 correctly when UTF-8 mode was not set.]
3718
3719 5. A negated single-character class was not being recognized as fixed-length
3720 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3721 compile error "lookbehind assertion is not fixed length".
3722
3723 6. The RunPerlTest auxiliary script was showing an unexpected difference
3724 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3725 write a Perl script that can interpret lines of an input file either as
3726 byte characters or as UTF-8, which is what "perltest" was being required to
3727 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3728 can't do is switch easily at run time between having the "use utf8;" pragma
3729 or not. In the end, I fudged it by using the RunPerlTest script to insert
3730 "use utf8;" explicitly for the UTF-8 tests.
3731
3732 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3733 the end of the subject string, contrary to the documentation and to what
3734 Perl does. This was true of both matching functions. Now it matches only at
3735 the start of the subject and immediately after *internal* newlines.
3736
3737 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3738 a pointer to an int instead of a pointer to an unsigned long int. This
3739 caused problems on 64-bit systems.
3740
3741 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3742 instance of the 'standard' template library not being so standard".
3743
3744 10. There was no check on the number of named subpatterns nor the maximum
3745 length of a subpattern name. The product of these values is used to compute
3746 the size of the memory block for a compiled pattern. By supplying a very
3747 long subpattern name and a large number of named subpatterns, the size
3748 computation could be caused to overflow. This is now prevented by limiting
3749 the length of names to 32 characters, and the number of named subpatterns
3750 to 10,000.
3751
3752 11. Subpatterns that are repeated with specific counts have to be replicated in
3753 the compiled pattern. The size of memory for this was computed from the
3754 length of the subpattern and the repeat count. The latter is limited to
3755 65535, but there was no limit on the former, meaning that integer overflow
3756 could in principle occur. The compiled length of a repeated subpattern is
3757 now limited to 30,000 bytes in order to prevent this.
3758
3759 12. Added the optional facility to have named substrings with the same name.
3760
3761 13. Added the ability to use a named substring as a condition, using the
3762 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3763 are numbers (not recommended). Forward references are permitted.
3764
3765 14. Added forward references in named backreferences (if you see what I mean).
3766
3767 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3768 pattern could run off the end of the subject. For example, the pattern
3769 "(?s)(.{1,5})"8 did this with the subject "ab".
3770
3771 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3772 PCRE_CASELESS was set when matching characters that were quantified with ?
3773 or *.
3774
3775 17. A character class other than a single negated character that had a minimum
3776 but no maximum quantifier - for example [ab]{6,} - was not handled
3777 correctly by pce_dfa_exec(). It would match only one character.
3778
3779 18. A valid (though odd) pattern that looked like a POSIX character
3780 class but used an invalid character after [ (for example [[,abc,]]) caused
3781 pcre_compile() to give the error "Failed: internal error: code overflow" or
3782 in some cases to crash with a glibc free() error. This could even happen if
3783 the pattern terminated after [[ but there just happened to be a sequence of
3784 letters, a binary zero, and a closing ] in the memory that followed.
3785
3786 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3787 over the years. Originally (before any Unicode support), just the bottom 8
3788 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3789 output from "man perlunicode" includes this:
3790
3791 The regular expression compiler produces polymorphic opcodes. That
3792 is, the pattern adapts to the data and automatically switches to
3793 the Unicode character scheme when presented with Unicode data--or
3794 instead uses a traditional byte scheme when presented with byte
3795 data.
3796
3797 Sadly, a wide octal escape does not cause a switch, and in a string with
3798 no other multibyte characters, these octal escapes are treated as before.
3799 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3800 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3801 Unicode string.
3802
3803 I have not perpetrated such confusion in PCRE. Up till now, it took just
3804 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3805 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3806 translate to the appropriate multibyte character.
3807
3808 29. Applied some refactoring to reduce the number of warnings from Microsoft
3809 and Borland compilers. This has included removing the fudge introduced
3810 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3811 a warning about an unused variable.
3812
3813 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3814 characters since release 4.0, because Perl (from release 5.004) does not.
3815 [Or at least, is documented not to: some releases seem to be in conflict
3816 with the documentation.] However, when a pattern was studied with
3817 pcre_study() and all its branches started with \s, PCRE still included VT
3818 as a possible starting character. Of course, this did no harm; it just
3819 caused an unnecessary match attempt.
3820
3821 22. Removed a now-redundant internal flag bit that recorded the fact that case
3822 dependency changed within the pattern. This was once needed for "required
3823 byte" processing, but is no longer used. This recovers a now-scarce options
3824 bit. Also moved the least significant internal flag bit to the most-
3825 significant bit of the word, which was not previously used (hangover from
3826 the days when it was an int rather than a uint) to free up another bit for
3827 the future.
3828
3829 23. Added support for CRLF line endings as well as CR and LF. As well as the
3830 default being selectable at build time, it can now be changed at runtime
3831 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3832 specify that it is scanning data with non-default line endings.
3833
3834 24. Changed the definition of CXXLINK to make it agree with the definition of
3835 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3836
3837 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3838 recursions. This makes a big different to stack usage for some patterns.
3839
3840 26. If a subpattern containing a named recursion or subroutine reference such
3841 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3842 the space required for the compiled pattern went wrong and gave too small a
3843 value. Depending on the environment, this could lead to "Failed: internal
3844 error: code overflow at offset 49" or "glibc detected double free or
3845 corruption" errors.
3846
3847 27. Applied patches from Google (a) to support the new newline modes and (b) to
3848 advance over multibyte UTF-8 characters in GlobalReplace.
3849
3850 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3851 difference for some implementation of PCRE in some Windows version.
3852
3853 29. Added some extra testing facilities to pcretest:
3854
3855 \q<number> in a data line sets the "match limit" value
3856 \Q<number> in a data line sets the "match recursion limt" value
3857 -S <number> sets the stack size, where <number> is in megabytes
3858
3859 The -S option isn't available for Windows.
3860
3861
3862 Version 6.6 06-Feb-06
3863 ---------------------
3864
3865 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3866 in pcreposix.h. I have copied the definition from pcre.h.
3867
3868 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3869 because pcre.h is no longer a built file.
3870
3871 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3872 not normally included in the compiled code.
3873
3874
3875 Version 6.5 01-Feb-06
3876 ---------------------
3877
3878 1. When using the partial match feature with pcre_dfa_exec(), it was not
3879 anchoring the second and subsequent partial matches at the new starting
3880 point. This could lead to incorrect results. For example, with the pattern
3881 /1234/, partially matching against "123" and then "a4" gave a match.
3882
3883 2. Changes to pcregrep:
3884
3885 (a) All non-match returns from pcre_exec() were being treated as failures
3886 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3887 error message is output. Some extra information is given for the
3888 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3889 probably the only errors that are likely to be caused by users (by
3890 specifying a regex that has nested indefinite repeats, for instance).
3891 If there are more than 20 of these errors, pcregrep is abandoned.
3892
3893 (b) A binary zero was treated as data while matching, but terminated the
3894 output line if it was written out. This has been fixed: binary zeroes
3895 are now no different to any other data bytes.
3896
3897 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3898 used to set a locale for matching. The --locale=xxxx long option has
3899 been added (no short equivalent) to specify a locale explicitly on the
3900 pcregrep command, overriding the environment variables.
3901
3902 (d) When -B was used with -n, some line numbers in the output were one less
3903 than they should have been.
3904
3905 (e) Added the -o (--only-matching) option.
3906
3907 (f) If -A or -C was used with -c (count only), some lines of context were
3908 accidentally printed for the final match.
3909
3910 (g) Added the -H (--with-filename) option.
3911
3912 (h) The combination of options -rh failed to suppress file names for files
3913 that were found from directory arguments.
3914
3915 (i) Added the -D (--devices) and -d (--directories) options.
3916
3917 (j) Added the -F (--fixed-strings) option.
3918
3919 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3920
3921 (l) Added the --colo(u)r option.
3922
3923 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3924 is not present by default.
3925
3926 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3927 items such as (?R) or (?1), when the recursion could match a number of
3928 alternatives. If it matched one of the alternatives, but subsequently,
3929 outside the recursion, there was a failure, the code tried to back up into
3930 the recursion. However, because of the way PCRE is implemented, this is not
3931 possible, and the result was an incorrect result from the match.
3932
3933 In order to prevent this happening, the specification of recursion has
3934 been changed so that all such subpatterns are automatically treated as
3935 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3936
3937 4. I had overlooked the fact that, in some locales, there are characters for
3938 which isalpha() is true but neither isupper() nor islower() are true. In
3939 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3940 and ordfeminine) are like this. This affected the treatment of \w and \W
3941 when they appeared in character classes, but not when they appeared outside
3942 a character class. The bit map for "word" characters is now created
3943 separately from the results of isalnum() instead of just taking it from the
3944 upper, lower, and digit maps. (Plus the underscore character, of course.)
3945
3946 5. The above bug also affected the handling of POSIX character classes such as
3947 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3948 permanent tables. Instead, the bit maps for such a class were previously
3949 created as the appropriate unions of the upper, lower, and digit bitmaps.
3950 Now they are created by subtraction from the [[:word:]] class, which has
3951 its own bitmap.
3952
3953 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3954 It is created by subtracting the vertical space characters (\x09, \x0a,
3955 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3956 subtraction was done in the overall bitmap for a character class, meaning
3957 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3958 be recognized. This bug has been fixed.
3959
3960 7. Patches from the folks at Google:
3961
3962 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3963 real life, but is still worth protecting against".
3964
3965 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3966 regular expressions".
3967
3968 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3969 have it.
3970
3971 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3972 "configure" and the latter not, in order to fix a problem somebody had
3973 with compiling the Arg class on HP-UX.
3974
3975 (e) Improve the error-handling of the C++ wrapper a little bit.
3976
3977 (f) New tests for checking recursion limiting.
3978
3979 8. The pcre_memmove() function, which is used only if the environment does not
3980 have a standard memmove() function (and is therefore rarely compiled),
3981 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3982 returning a result (though PCRE never actually uses the result).
3983
3984 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3985 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3986 returned instead of calling malloc() with an overflowing number that would
3987 most likely cause subsequent chaos.
3988
3989 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3990
3991 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3992 with this option is matched, the nmatch and pmatch options of regexec() are
3993 ignored.
3994
3995 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3996 provided in case anyone wants to the the POSIX interface with UTF-8
3997 strings.
3998
3999 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4000 C++ linking (needed for some HP-UX environments).
4001
4002 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4003 (unused parameter) and in the pcre_printint() function (omitted "default"
4004 switch label when the default is to do nothing).
4005
4006 15. Added some code to make it possible, when PCRE is compiled as a C++
4007 library, to replace subject pointers for pcre_exec() with a smart pointer
4008 class, thus making it possible to process discontinuous strings.
4009
4010 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4011 much the same function. They were added by different people who were trying
4012 to make PCRE easy to compile on non-Unix systems. It has been suggested
4013 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4014 for compiling on Windows systems. I have therefore replaced it with
4015 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4016 defaults to "extern" for C or "extern C" for C++, which works fine on
4017 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4018 SCOPE with something explicit in config.h. In addition:
4019
4020 (a) pcreposix.h still had just "extern" instead of either of these macros;
4021 I have replaced it with PCRE_DATA_SCOPE.
4022
4023 (b) Functions such as _pcre_xclass(), which are internal to the library,
4024 but external in the C sense, all had PCRE_EXPORT in their definitions.
4025 This is apparently wrong for the Windows case, so I have removed it.
4026 (It makes no difference on Unix-like systems.)
4027
4028 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4029 of recursive calls to match(). This is different to MATCH_LIMIT because
4030 that limits the total number of calls to match(), not all of which increase
4031 the depth of recursion. Limiting the recursion depth limits the amount of
4032 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4033 when PCRE is compiled, and changed at run time. A patch from Google adds
4034 this functionality to the C++ interface.
4035
4036 18. Changes to the handling of Unicode character properties:
4037
4038 (a) Updated the table to Unicode 4.1.0.
4039
4040 (b) Recognize characters that are not in the table as "Cn" (undefined).
4041
4042 (c) I revised the way the table is implemented to a much improved format
4043 which includes recognition of ranges. It now supports the ranges that
4044 are defined in UnicodeData.txt, and it also amalgamates other
4045 characters into ranges. This has reduced the number of entries in the
4046 table from around 16,000 to around 3,000, thus reducing its size
4047 considerably. I realized I did not need to use a tree structure after
4048 all - a binary chop search is just as efficient. Having reduced the
4049 number of entries, I extended their size from 6 bytes to 8 bytes to
4050 allow for more data.
4051
4052 (d) Added support for Unicode script names via properties such as \p{Han}.
4053
4054 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4055 matching that character.
4056
4057 20. When matching a repeated Unicode property with a minimum greater than zero,
4058 (for example \pL{2,}), PCRE could look past the end of the subject if it
4059 reached it while seeking the minimum number of characters. This could
4060 happen only if some of the characters were more than one byte long, because
4061 there is a check for at least the minimum number of bytes.
4062
4063 21. Refactored the implementation of \p and \P so as to be more general, to
4064 allow for more different types of property in future. This has changed the
4065 compiled form incompatibly. Anybody with saved compiled patterns that use
4066 \p or \P will have to recompile them.
4067
4068 22. Added "Any" and "L&" to the supported property types.
4069
4070 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4071 but give a compile time error if the value is greater than 0xff.
4072
4073 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4074 accidentally not being installed or uninstalled.
4075
4076 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4077 made were to insert the current release number. This seemed silly, because
4078 it made things harder for people building PCRE on systems that don't run
4079 "configure". I have turned pcre.h into a distributed file, no longer built
4080 by "configure", with the version identification directly included. There is
4081 no longer a pcre.h.in file.
4082
4083 However, this change necessitated a change to the pcre-config script as
4084 well. It is built from pcre-config.in, and one of the substitutions was the
4085 release number. I have updated configure.ac so that ./configure now finds
4086 the release number by grepping pcre.h.
4087
4088 26. Added the ability to run the tests under valgrind.
4089
4090
4091 Version 6.4 05-Sep-05
4092 ---------------------
4093
4094 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4095 "--" to be printed when multiple files were scanned, even when none of the
4096 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4097 consider it to be a bug, and have restored the previous behaviour.
4098
4099 2. A couple of code tidies to get rid of compiler warnings.
4100
4101 3. The pcretest program used to cheat by referring to symbols in the library
4102 whose names begin with _pcre_. These are internal symbols that are not
4103 really supposed to be visible externally, and in some environments it is
4104 possible to suppress them. The cheating is now confined to including
4105 certain files from the library's source, which is a bit cleaner.
4106
4107 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4108 file's purpose clearer.
4109
4110 5. Reorganized pcre_ucp_findchar().
4111
4112
4113 Version 6.3 15-Aug-05
4114 ---------------------
4115
4116 1. The file libpcre.pc.in did not have general read permission in the tarball.
4117
4118 2. There were some problems when building without C++ support:
4119
4120 (a) If C++ support was not built, "make install" and "make test" still
4121 tried to test it.
4122
4123 (b) There were problems when the value of CXX was explicitly set. Some
4124 changes have been made to try to fix these, and ...
4125
4126 (c) --disable-cpp can now be used to explicitly disable C++ support.
4127
4128 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4129 backslash in a target when C++ was disabled. This confuses some
4130 versions of "make", apparently. Using an intermediate variable solves
4131 this. (Same for CPP_LOBJ.)
4132
4133 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4134 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4135 necessary on certain architectures.
4136
4137 4. Added a setting of -export-symbols-regex to the link command to remove
4138 those symbols that are exported in the C sense, but actually are local
4139 within the library, and not documented. Their names all begin with
4140 "_pcre_". This is not a perfect job, because (a) we have to except some
4141 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4142 available (and never for static libraries). I have made a note to try to
4143 find a way round (a) in the future.
4144
4145
4146 Version 6.2 01-Aug-05
4147 ---------------------
4148
4149 1. There was no test for integer overflow of quantifier values. A construction
4150 such as {1111111111111111} would give undefined results. What is worse, if
4151 a minimum quantifier for a parenthesized subpattern overflowed and became
4152 negative, the calculation of the memory size went wrong. This could have
4153 led to memory overwriting.
4154
4155 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4156
4157 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4158 operating environments where this matters.
4159
4160 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4161 PCRE options from within the C++ wrapper.
4162
4163 5. Named capturing subpatterns were not being correctly counted when a pattern
4164 was compiled. This caused two problems: (a) If there were more than 100
4165 such subpatterns, the calculation of the memory needed for the whole
4166 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4167 back references of the form \12, where the number was greater than 9, were
4168 not recognized as back references, even though there were sufficient
4169 previous subpatterns.
4170
4171 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4172 versions of gcc, e.g. 2.95.4.
4173
4174
4175 Version 6.1 21-Jun-05
4176 ---------------------
4177
4178 1. There was one reference to the variable "posix" in pcretest.c that was not
4179 surrounded by "#if !defined NOPOSIX".
4180
4181 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4182 the cross-check on the old pcre_info() function, for the benefit of the
4183 cut-down version of PCRE that is currently imported into Exim.
4184
4185 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4186 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4187 patterns that start (?i) but I don't think that matters. The use of (?i) is
4188 just an example; this all applies to the other options as well.
4189
4190 4. Since libtool seems to echo the compile commands it is issuing, the output
4191 from "make" can be reduced a bit by putting "@" in front of each libtool
4192 compile command.
4193
4194 5. Patch from the folks at Google for configure.in to be a bit more thorough
4195 in checking for a suitable C++ installation before trying to compile the
4196 C++ stuff. This should fix a reported problem when a compiler was present,
4197 but no suitable headers.
4198
4199 6. The man pages all had just "PCRE" as their title. I have changed them to
4200 be the relevant file name. I have also arranged that these names are
4201 retained in the file doc/pcre.txt, which is a concatenation in text format
4202 of all the man pages except the little individual ones for each function.
4203
4204 7. The NON-UNIX-USE file had not been updated for the different set of source
4205 files that come with release 6. I also added a few comments about the C++
4206 wrapper.
4207
4208
4209 Version 6.0 07-Jun-05
4210 ---------------------
4211
4212 1. Some minor internal re-organization to help with my DFA experiments.
4213
4214 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4215 didn't matter for the library itself when fully configured, but did matter
4216 when compiling without UCP support, or within Exim, where the ucp files are
4217 not imported.
4218
4219 3. Refactoring of the library code to split up the various functions into
4220 different source modules. The addition of the new DFA matching code (see
4221 below) to a single monolithic source would have made it really too
4222 unwieldy, quite apart from causing all the code to be include in a
4223 statically linked application, when only some functions are used. This is
4224 relevant even without the DFA addition now that patterns can be compiled in
4225 one application and matched in another.
4226
4227 The downside of splitting up is that there have to be some external
4228 functions and data tables that are used internally in different modules of
4229 the library but which are not part of the API. These have all had their
4230 names changed to start with "_pcre_" so that they are unlikely to clash
4231 with other external names.
4232
4233 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4234 a different (DFA) algorithm. Although it is slower than the original
4235 function, it does have some advantages for certain types of matching
4236 problem.
4237
4238 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4239 including restarting after a partial match.
4240
4241 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4242 defined when compiling for Windows was sent to me. I have put it into the
4243 code, though I have no means of testing or verifying it.
4244
4245 7. Added the pcre_refcount() auxiliary function.
4246
4247 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4248 match before or at the first newline in the subject string. In pcretest,
4249 the /f option on a pattern can be used to set this.
4250
4251 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4252 would behave wrongly. This has been present in PCRE since release 4.0.
4253
4254 10. A number of changes to the pcregrep command:
4255
4256 (a) Refactored how -x works; insert ^(...)$ instead of setting
4257 PCRE_ANCHORED and checking the length, in preparation for adding
4258 something similar for -w.
4259
4260 (b) Added the -w (match as a word) option.
4261
4262 (c) Refactored the way lines are read and buffered so as to have more
4263 than one at a time available.
4264
4265 (d) Implemented a pcregrep test script.
4266
4267 (e) Added the -M (multiline match) option. This allows patterns to match
4268 over several lines of the subject. The buffering ensures that at least
4269 8K, or the rest of the document (whichever is the shorter) is available
4270 for matching (and similarly the previous 8K for lookbehind assertions).
4271
4272 (f) Changed the --help output so that it now says
4273
4274 -w, --word-regex(p)
4275
4276 instead of two lines, one with "regex" and the other with "regexp"
4277 because that confused at least one person since the short forms are the
4278 same. (This required a bit of code, as the output is generated
4279 automatically from a table. It wasn't just a text change.)
4280
4281 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4282 option but starts with a hyphen. Could be a pattern or a path name
4283 starting with a hyphen, for instance.
4284
4285 (h) "-" can be given as a file name to represent stdin.
4286
4287 (i) When file names are being printed, "(standard input)" is used for
4288 the standard input, for compatibility with GNU grep. Previously
4289 "<stdin>" was used.
4290
4291 (j) The option --label=xxx can be used to supply a name to be used for
4292 stdin when file names are being printed. There is no short form.
4293
4294 (k) Re-factored the options decoding logic because we are going to add
4295 two more options that take data. Such options can now be given in four
4296 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4297
4298 (l) Added the -A, -B, and -C options for requesting that lines of context
4299 around matches be printed.
4300
4301 (m) Added the -L option to print the names of files that do not contain
4302 any matching lines, that is, the complement of -l.
4303
4304 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4305 continue to scan other files.
4306
4307 (o) The -s option was incorrectly implemented. For compatibility with other
4308 greps, it now suppresses the error message for a non-existent or non-
4309 accessible file (but not the return code). There is a new option called
4310 -q that suppresses the output of matching lines, which was what -s was
4311 previously doing.
4312
4313 (p) Added --include and --exclude options to specify files for inclusion
4314 and exclusion when recursing.
4315
4316 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4317 Hopefully, it now does.
4318
4319 12. Missing cast in pcre_study().
4320
4321 13. Added an "uninstall" target to the makefile.
4322
4323 14. Replaced "extern" in the function prototypes in Makefile.in with
4324 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4325 world, but is set differently for Windows.
4326
4327 15. Added a second compiling function called pcre_compile2(). The only
4328 difference is that it has an extra argument, which is a pointer to an
4329 integer error code. When there is a compile-time failure, this is set
4330 non-zero, in addition to the error test pointer being set to point to an
4331 error message. The new argument may be NULL if no error number is required
4332 (but then you may as well call pcre_compile(), which is now just a
4333 wrapper). This facility is provided because some applications need a
4334 numeric error indication, but it has also enabled me to tidy up the way
4335 compile-time errors are handled in the POSIX wrapper.
4336
4337 16. Added VPATH=.libs to the makefile; this should help when building with one
4338 prefix path and installing with another. (Or so I'm told by someone who
4339 knows more about this stuff than I do.)
4340
4341 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4342 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4343 match everything, including newlines. This is not POSIX-compatible, but
4344 somebody wanted the feature. From pcretest it can be activated by using
4345 both the P and the s flags.
4346
4347 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4348
4349 19. libpcre.pc was being incorrectly installed as executable.
4350
4351 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4352 it now also looks for '\r' so that it will work unmodified on Windows.
4353
4354 21. Added Google's contributed C++ wrapper to the distribution.
4355
4356 22. Added some untidy missing memory free() calls in pcretest, to keep
4357 Electric Fence happy when testing.
4358
4359
4360
4361 Version 5.0 13-Sep-04
4362 ---------------------
4363
4364 1. Internal change: literal characters are no longer packed up into items
4365 containing multiple characters in a single byte-string. Each character
4366 is now matched using a separate opcode. However, there may be more than one
4367 byte in the character in UTF-8 mode.
4368
4369 2. The pcre_callout_block structure has two new fields: pattern_position and
4370 next_item_length. These contain the offset in the pattern to the next match
4371 item, and its length, respectively.
4372
4373 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4374 insertion of callouts before each pattern item. Added the /C option to
4375 pcretest to make use of this.
4376
4377 4. On the advice of a Windows user, the lines
4378
4379 #if defined(_WIN32) || defined(WIN32)
4380 _setmode( _fileno( stdout ), 0x8000 );
4381 #endif /* defined(_WIN32) || defined(WIN32) */
4382
4383 have been added to the source of pcretest. This apparently does useful
4384 magic in relation to line terminators.
4385
4386 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4387 for the benefit of those environments where the "b" makes a difference.
4388
4389 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4390 to know about it. I have put a hack into configure.in that adds in code
4391 to set GCC=yes if CC=icc. This seems to end up at a point in the
4392 generated configure script that is early enough to affect the setting of
4393 compiler options, which is what is needed, but I have no means of testing
4394 whether it really works. (The user who reported this had patched the
4395 generated configure script, which of course I cannot do.)
4396
4397 LATER: After change 22 below (new libtool files), the configure script
4398 seems to know about icc (and also ecc). Therefore, I have commented out
4399 this hack in configure.in.
4400
4401 7. Added support for pkg-config (2 patches were sent in).
4402
4403 8. Negated POSIX character classes that used a combination of internal tables
4404 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4405 [[:^ascii]]. Typically, they would match almost any characters. The other
4406 POSIX classes were not broken in this way.
4407
4408 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4409 to find the match, as PCRE was deluded into thinking that the match had to
4410 start at the start point or following a newline. The same bug applied to
4411 patterns with negative forward assertions or any backward assertions
4412 preceding ".*" at the start, unless the pattern required a fixed first
4413 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4414
4415 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4416 starting at the last subject character, bytes beyond the end of the subject
4417 string were read.
4418
4419 11. Renamed the variable "class" as "classbits" to make life easier for C++
4420 users. (Previously there was a macro definition, but it apparently wasn't
4421 enough.)
4422
4423 12. Added the new field "tables" to the extra data so that tables can be passed
4424 in at exec time, or the internal tables can be re-selected. This allows
4425 a compiled regex to be saved and re-used at a later time by a different
4426 program that might have everything at different addresses.
4427
4428 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4429 -R library as well as a -L library.
4430
4431 14. The debugging options of pcretest (-d on the command line or D on a
4432 pattern) showed incorrect output for anything following an extended class
4433 that contained multibyte characters and which was followed by a quantifier.
4434
4435 15. Added optional support for general category Unicode character properties
4436 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4437 support. It adds about 90K to the size of the library. The meanings of the
4438 inbuilt class escapes such as \d and \s have NOT been changed.
4439
4440 16. Updated pcredemo.c to include calls to free() to release the memory for the
4441 compiled pattern.
4442
4443 17. The generated file chartables.c was being created in the source directory
4444 instead of in the building directory. This caused the build to fail if the
4445 source directory was different from the building directory, and was
4446 read-only.
4447
4448 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4449 file. No doubt somebody will tell me if they don't make sense... Also added
4450 Dan Mooney's comments about building on OpenVMS.
4451
4452 19. Added support for partial matching via the PCRE_PARTIAL option for
4453 pcre_exec() and the \P data escape in pcretest.
4454
4455 20. Extended pcretest with 3 new pattern features:
4456
4457 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4458 write the compiled pattern to the file whose name is "rest-of-line".
4459 This is a straight binary dump of the data, with the saved pointer to
4460 the character tables forced to be NULL. The study data, if any, is
4461 written too. After writing, pcretest reads a new pattern.
4462
4463 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4464 compiled pattern from the given file. There must not be any
4465 occurrences of "<" in the file name (pretty unlikely); if there are,
4466 pcretest will instead treat the initial "<" as a pattern delimiter.
4467 After reading in the pattern, pcretest goes on to read data lines as
4468 usual.
4469
4470 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4471 and 16-bit fields in a compiled pattern, to simulate a pattern that
4472 was compiled on a host of opposite endianness.
4473
4474 21. The pcre-exec() function can now cope with patterns that were compiled on
4475 hosts of opposite endianness, with this restriction:
4476
4477 As for any compiled expression that is saved and used later, the tables
4478 pointer field cannot be preserved; the extra_data field in the arguments
4479 to pcre_exec() should be used to pass in a tables address if a value
4480 other than the default internal tables were used at compile time.
4481
4482 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4483 now diagnosed as an error. Previously, most of the time, a negative number
4484 would have been treated as zero, but if in addition "ovector" was passed as
4485 NULL, a crash could occur.
4486
4487 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4488 new versions from the libtool 1.5 distribution (the last one is a copy of
4489 a file called libtool.m4). This seems to have fixed the need to patch
4490 "configure" to support Darwin 1.3 (which I used to do). However, I still
4491 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4492 workstation).
4493
4494 24. Changed the PCRE licence to be the more standard "BSD" licence.
4495
4496
4497 Version 4.5 01-Dec-03
4498 ---------------------
4499
4500 1. There has been some re-arrangement of the code for the match() function so
4501 that it can be compiled in a version that does not call itself recursively.
4502 Instead, it keeps those local variables that need separate instances for
4503 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4504 needs to "recurse". Keeping track of where control must go is done by means
4505 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4506 hide most of the details from the main code, and operates only if
4507 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4508 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4509 operating.
4510
4511 To make it easier for callers to provide specially tailored get/free
4512 functions for this usage, two new functions, pcre_stack_malloc, and
4513 pcre_stack_free, are used. They are always called in strict stacking order,
4514 and the size of block requested is always the same.
4515
4516 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4517 PCRE has been compiled to use the stack or the heap for recursion. The
4518 -C option of pcretest uses this to show which version is compiled.
4519
4520 A new data escape \S, is added to pcretest; it causes the amounts of store
4521 obtained and freed by both kinds of malloc/free at match time to be added
4522 to the output.
4523
4524 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4525 what's available on my current Linux desktop machine.
4526
4527 3. When matching a UTF-8 string, the test for a valid string at the start has
4528 been extended. If start_offset is not zero, PCRE now checks that it points
4529 to a byte that is the start of a UTF-8 character. If not, it returns
4530 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4531 this is necessary because there may be backward assertions in the pattern.
4532 When matching the same subject several times, it may save resources to use
4533 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4534
4535 4. The code for checking the validity of UTF-8 strings has been tightened so
4536 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4537 containing "overlong sequences".
4538
4539 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4540 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4541 should have been "&", but it just so happened that all the cases this let
4542 through by mistake were picked up later in the function.
4543
4544 6. I had used a variable called "isblank" - this is a C99 function, causing
4545 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4546
4547 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4548 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4549 so the version is shown; (c) stop "make test" echoing "./RunTest".
4550
4551 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4552
4553 9. The prototype for memmove() for systems that don't have it was using
4554 size_t, but the inclusion of the header that defines size_t was later. I've
4555 moved the #includes for the C headers earlier to avoid this.
4556
4557 10. Added some adjustments to the code to make it easier to compiler on certain
4558 special systems:
4559
4560 (a) Some "const" qualifiers were missing.
4561 (b) Added the macro EXPORT before all exported functions; by default this
4562 is defined to be empty.
4563 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4564 that it reads its output file name as an argument instead of writing
4565 to the standard output and assuming this can be redirected.
4566
4567 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4568 class containing characters with values greater than 255, PCRE compilation
4569 went into a loop.
4570
4571 12. A recursive reference to a subpattern that was within another subpattern
4572 that had a minimum quantifier of zero caused PCRE to crash. For example,
4573 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4574 recursion. If the recursively-called subpattern itself had a zero repeat,
4575 that was OK.
4576
4577 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4578 buffer into which it was copied (for escape processing) was still set at
4579 1024, so long lines caused crashes.
4580
4581 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4582 "internal error: code overflow...". This applied to any character class
4583 that was followed by a possessive quantifier.
4584
4585 15. Modified the Makefile to add libpcre.la as a prerequisite for
4586 libpcreposix.la because I was told this is needed for a parallel build to
4587 work.
4588
4589 16. If a pattern that contained .* following optional items at the start was
4590 studied, the wrong optimizing data was generated, leading to matching
4591 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4592 matching string must start with a or b or c. The correct conclusion for
4593 this pattern is that a match can start with any character.
4594
4595
4596 Version 4.4 13-Aug-03
4597 ---------------------
4598
4599 1. In UTF-8 mode, a character class containing characters with values between
4600 127 and 255 was not handled correctly if the compiled pattern was studied.
4601 In fixing this, I have also improved the studying algorithm for such
4602 classes (slightly).
4603
4604 2. Three internal functions had redundant arguments passed to them. Removal
4605 might give a very teeny performance improvement.
4606
4607 3. Documentation bug: the value of the capture_top field in a callout is *one
4608 more than* the number of the hightest numbered captured substring.
4609
4610 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4611 in incorrectly linking with a previously installed version. They now link
4612 explicitly with libpcre.la.
4613
4614 5. configure.in no longer needs to recognize Cygwin specially.
4615
4616 6. A problem in pcre.in for Windows platforms is fixed.
4617
4618 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4619 pcretest, it used to include the size of the study block as part of its
4620 output. Unfortunately, the structure contains a field that has a different
4621 size on different hardware architectures. This meant that the tests that
4622 showed this size failed. As the block is currently always of a fixed size,
4623 this information isn't actually particularly useful in pcretest output, so
4624 I have just removed it.
4625
4626 8. Three pre-processor statements accidentally did not start in column 1.
4627 Sadly, there are *still* compilers around that complain, even though
4628 standard C has not required this for well over a decade. Sigh.
4629
4630 9. In pcretest, the code for checking callouts passed small integers in the
4631 callout_data field, which is a void * field. However, some picky compilers
4632 complained about the casts involved for this on 64-bit systems. Now
4633 pcretest passes the address of the small integer instead, which should get
4634 rid of the warnings.
4635
4636 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4637 both compile and run time, and gives an error if an invalid UTF-8 sequence
4638 is found. There is a option for disabling this check in cases where the
4639 string is known to be correct and/or the maximum performance is wanted.
4640
4641 11. In response to a bug report, I changed one line in Makefile.in from
4642
4643 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4644 to
4645 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4646
4647 to look similar to other lines, but I have no way of telling whether this
4648 is the right thing to do, as I do not use Windows. No doubt I'll get told
4649 if it's wrong...
4650
4651
4652 Version 4.3 21-May-03
4653 ---------------------
4654
4655 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4656 Makefile.
4657
4658 2. Some refactoring to improve the quality of the code:
4659
4660 (i) The utf8_table... variables are now declared "const".
4661
4662 (ii) The code for \cx, which used the "case flipping" table to upper case
4663 lower case letters, now just substracts 32. This is ASCII-specific,
4664 but the whole concept of \cx is ASCII-specific, so it seems
4665 reasonable.
4666
4667 (iii) PCRE was using its character types table to recognize decimal and
4668 hexadecimal digits in the pattern. This is silly, because it handles
4669 only 0-9, a-f, and A-F, but the character types table is locale-
4670 specific, which means strange things might happen. A private
4671 table is now used for this - though it costs 256 bytes, a table is
4672 much faster than multiple explicit tests. Of course, the standard
4673 character types table is still used for matching digits in subject
4674 strings against \d.
4675
4676 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4677 ending in _t are). So I've renamed it as ESC_tee.
4678
4679 3. The first argument for regexec() in the POSIX wrapper should have been
4680 defined as "const".
4681
4682 4. Changed pcretest to use malloc() for its buffers so that they can be
4683 Electric Fenced for debugging.
4684
4685 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4686 to read one or more bytes before the start of the subject string. Often this
4687 had no effect on PCRE's behaviour, but in some circumstances it could
4688 provoke a segmentation fault.
4689
4690 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4691 to try to read one or more bytes before the start of the subject string.
4692
4693 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4694 UTF-8 support could misbehave in various ways if the subject string
4695 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4696 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4697 back over UTF-8 characters.)
4698
4699
4700 Version 4.2 14-Apr-03
4701 ---------------------
4702
4703 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4704
4705 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4706 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4707 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4708 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4709 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4710 and BUILD_EXEEXT
4711 Note: automatic setting of the BUILD variables is not yet working
4712 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4713 compile-time but not at link-time
4714 [LINK]: use for linking executables only
4715 make different versions for Windows and non-Windows
4716 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4717 libraries
4718 [LINK_FOR_BUILD]: new variable
4719 [OBJEXT]: use throughout
4720 [EXEEXT]: use throughout
4721 <winshared>: new target
4722 <wininstall>: new target
4723 <dftables.o>: use native compiler
4724 <dftables>: use native linker
4725 <install>: handle Windows platform correctly
4726 <clean>: ditto
4727 <check>: ditto
4728 copy DLL to top builddir before testing
4729
4730 As part of these changes, -no-undefined was removed again. This was reported
4731 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4732 in any case.
4733
4734 3. Some tidies to get rid of compiler warnings:
4735
4736 . In the match_data structure, match_limit was an unsigned long int, whereas
4737 match_call_count was an int. I've made them both unsigned long ints.
4738
4739 . In pcretest the fact that a const uschar * doesn't automatically cast to
4740 a void * provoked a warning.
4741
4742 . Turning on some more compiler warnings threw up some "shadow" variables
4743 and a few more missing casts.
4744
4745 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4746 option, a class that contained a single character with a value between 128
4747 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4748
4749 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4750 option, a class that contained several characters, but with at least one
4751 whose value was between 128 and 255 caused PCRE to crash.
4752
4753
4754 Version 4.1 12-Mar-03
4755 ---------------------
4756
4757 1. Compiling with gcc -pedantic found a couple of places where casts were
4758 needed, and a string in dftables.c that was longer than standard compilers are
4759 required to support.
4760
4761 2. Compiling with Sun's compiler found a few more places where the code could
4762 be tidied up in order to avoid warnings.
4763
4764 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4765 first of these names is deprecated in the latest Autoconf in favour of the name
4766 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4767 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4768 analogy I have changed it to CFLAGS_FOR_BUILD.
4769
4770 4. Added -no-undefined to the linking command in the Makefile, because this is
4771 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4772 linking step for the pcreposix library.
4773
4774 5. PCRE was failing to diagnose the case of two named groups with the same
4775 name.
4776
4777 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4778 literal character that is needed in the subject for a match, and scans along to
4779 ensure that it is present before embarking on the full matching process. This
4780 saves time in cases of nested unlimited repeats that are never going to match.
4781 Problem: the scan can take a lot of time if the subject is very long (e.g.
4782 megabytes), thus penalizing straightforward matches. It is now done only if the
4783 amount of subject to be scanned is less than 1000 bytes.
4784
4785 7. A lesser problem with the same optimization is that it was recording the
4786 first character of an anchored pattern as "needed", thus provoking a search
4787 right along the subject, even when the first match of the pattern was going to
4788 fail. The "needed" character is now not set for anchored patterns, unless it
4789 follows something in the pattern that is of non-fixed length. Thus, it still
4790 fulfils its original purpose of finding quick non-matches in cases of nested
4791 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4792
4793
4794 Version 4.0 17-Feb-03
4795 ---------------------
4796
4797 1. If a comment in an extended regex that started immediately after a meta-item
4798 extended to the end of string, PCRE compiled incorrect data. This could lead to
4799 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4800
4801 2. Moved to autoconf 2.53 and libtool 1.4.2.
4802
4803 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4804 the special perltest8 script is no longer needed - all the tests can be run
4805 from a single perltest script.
4806
4807 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4808 by \s. It has now been removed in PCRE. This means it isn't recognized as
4809 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4810 class [:space:] *does* include VT, thereby creating a mess.
4811
4812 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4813 space and tab.
4814
4815 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4816 its new features into the main test script, reducing the number of scripts.
4817
4818 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4819 were backward compatible, and made the (?i) apply to the whole pattern, as if
4820 /i were given. Now it behaves more logically, and applies the option setting
4821 only to what follows. PCRE has been changed to follow suit. However, if it
4822 finds options settings right at the start of the pattern, it extracts them into
4823 the global options, as before. Thus, they show up in the info data.
4824
4825 8. Added support for the \Q...\E escape sequence. Characters in between are
4826 treated as literals. This is slightly different from Perl in that $ and @ are
4827 also handled as literals inside the quotes. In Perl, they will cause variable
4828 interpolation. Note the following examples:
4829
4830 Pattern PCRE matches Perl matches
4831
4832 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4833 \Qabc\$xyz\E abc\$xyz abc\$xyz
4834 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4835
4836 For compatibility with Perl, \Q...\E sequences are recognized inside character
4837 classes as well as outside them.
4838
4839 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4840 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4841 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4842 signed/unsigned warnings.
4843
4844 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4845 option for pcretest, so I've replaced it by a simple function that does just
4846 that job.
4847
4848 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4849 "pcregrep -".
4850
4851 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4852 Java package. This provides some syntactic sugar for simple cases of what my
4853 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4854 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4855 item, you can use this simplified notation. Note that only makes sense with
4856 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4857 greediness, whatever the setting of the PCRE_UNGREEDY option.
4858
4859 13. A change of greediness default within a pattern was not taking effect at
4860 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4861 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4862 was abstracted outside.
4863
4864 14. PCRE now supports the \G assertion. It is true when the current matching
4865 position is at the start point of the match. This differs from \A when the
4866 starting offset is non-zero. Used with the /g option of pcretest (or similar
4867 code), it works in the same way as it does for Perl's /g option. If all
4868 alternatives of a regex begin with \G, the expression is anchored to the start
4869 match position, and the "anchored" flag is set in the compiled expression.
4870
4871 15. Some bugs concerning the handling of certain option changes within patterns
4872 have been fixed. These applied to options other than (?ims). For example,
4873 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4874 been the other way round. Some of this was related to change 7 above.
4875
4876 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4877 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4878 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4879 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4880
4881 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4882 mode. Unlike ".", it always matches newline, whatever the setting of
4883 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4884 assertions. Perl allows it, but it doesn't (in general) work because it can't
4885 calculate the length of the lookbehind. At least, that's the case for Perl
4886 5.8.0 - I've been told they are going to document that it doesn't work in
4887 future.
4888
4889 18. Added an error diagnosis for escapes that PCRE does not support: these are
4890 \L, \l, \N, \P, \p, \U, \u, and \X.
4891
4892 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4893 reading past the end of the pattern in cases such as /[abcd/.
4894
4895 20. PCRE was getting more memory than necessary for patterns with classes that
4896 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4897
4898 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4899 compiling PCRE for use with Virtual Pascal.
4900
4901 22. Small fix to the Makefile to make it work properly if the build is done
4902 outside the source tree.
4903
4904 23. Added a new extension: a condition to go with recursion. If a conditional
4905 subpattern starts with (?(R) the "true" branch is used if recursion has
4906 happened, whereas the "false" branch is used only at the top level.
4907
4908 24. When there was a very long string of literal characters (over 255 bytes
4909 without UTF support, over 250 bytes with UTF support), the computation of how
4910 much memory was required could be incorrect, leading to segfaults or other
4911 strange effects.
4912
4913 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4914 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4915 there was a subsequent back reference to those brackets. This meant that, for
4916 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4917 possible to check for precisely this case. All we can do is abandon the
4918 optimization if .* occurs inside capturing brackets when there are any back
4919 references whatsoever. (See below for a better fix that came later.)
4920
4921 26. The handling of the optimization for finding the first character of a
4922 non-anchored pattern, and for finding a character that is required later in the
4923 match were failing in some cases. This didn't break the matching; it just
4924 failed to optimize when it could. The way this is done has been re-implemented.
4925
4926 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4927
4928 28. Added a new feature that provides some of the functionality that Perl
4929 provides with (?{...}). The facility is termed a "callout". The way it is done
4930 in PCRE is for the caller to provide an optional function, by setting
4931 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4932 global variable. By default it is unset, which disables all calling out. To get
4933 the function called, the regex must include (?C) at appropriate points. This
4934 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4935 This provides a means of identifying different callout points. When PCRE
4936 reaches such a point in the regex, if pcre_callout has been set, the external
4937 function is called. It is provided with data in a structure called
4938 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4939 matching continues; if it returns a non-zero value, the match at the current
4940 point fails. However, backtracking will occur if possible. [This was changed
4941 later and other features added - see item 49 below.]
4942
4943 29. pcretest is upgraded to test the callout functionality. It provides a
4944 callout function that displays information. By default, it shows the start of
4945 the match and the current position in the text. There are some new data escapes
4946 to vary what happens:
4947
4948 \C+ in addition, show current contents of captured substrings
4949 \C- do not supply a callout function
4950 \C!n return 1 when callout number n is reached
4951 \C!n!m return 1 when callout number n is reached for the mth time
4952
4953 30. If pcregrep was called with the -l option and just a single file name, it
4954 output "<stdin>" if a match was found, instead of the file name.
4955
4956 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4957 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4958 pcre_exec(). This saves a malloc/free per call. The default value of
4959 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4960 when configuring.
4961
4962 32. The default maximum size of a compiled pattern is 64K. There have been a
4963 few cases of people hitting this limit. The code now uses macros to handle the
4964 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4965 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4966 configuring. Tests 2 and 5 work only with 2-byte links because they output
4967 debugging information about compiled patterns.
4968
4969 33. Internal code re-arrangements:
4970
4971 (a) Moved the debugging function for printing out a compiled regex into
4972 its own source file (printint.c) and used #include to pull it into
4973 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4974 separate copies.
4975
4976 (b) Defined the list of op-code names for debugging as a macro in
4977 internal.h so that it is next to the definition of the opcodes.
4978
4979 (c) Defined a table of op-code lengths for simpler skipping along compiled
4980 code. This is again a macro in internal.h so that it is next to the
4981 definition of the opcodes.
4982
4983 34. Added support for recursive calls to individual subpatterns, along the
4984 lines of Robin Houston's patch (but implemented somewhat differently).
4985
4986 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4987 allow it to read and process whole directories in Win32. This code was
4988 contributed by Lionel Fourquaux; it has not been tested by me.
4989
4990 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4991 used to name a group. Names consist of alphanumerics and underscores, and must
4992 be unique. Back references use the syntax (?P=name) and recursive calls use
4993 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4994 numbers. The function pcre_fullinfo() can be used after compilation to extract
4995 a name/number map. There are three relevant calls:
4996
4997 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4998 PCRE_INFO_NAMECOUNT yields the number of entries
4999 PCRE_INFO_NAMETABLE yields a pointer to the map.
5000
5001 The map is a vector of fixed-size entries. The size of each entry depends on
5002 the length of the longest name used. The first two bytes of each entry are the
5003 group number, most significant byte first. There follows the corresponding
5004 name, zero terminated. The names are in alphabetical order.
5005
5006 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5007 case instead of 255. Making it the same both with and without UTF-8 support
5008 means that the same test output works with both.
5009
5010 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5011 calling malloc() with a zero argument.
5012
5013 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5014 optimization. I've improved things by keeping a bitmap of backreferences with
5015 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5016 fact referenced, the optimization can be applied. It is unlikely that a
5017 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5018 the match to follow \n) will appear inside brackets with a number greater than
5019 31, but if it does, any back reference > 31 suppresses the optimization.
5020
5021 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5022 of disabling numbered capturing parentheses. Any opening parenthesis that is
5023 not followed by ? behaves as if it were followed by ?: but named parentheses
5024 can still be used for capturing (and they will acquire numbers in the usual
5025 way).
5026
5027 41. Redesigned the return codes from the match() function into yes/no/error so
5028 that errors can be passed back from deep inside the nested calls. A malloc
5029 failure while inside a recursive subpattern call now causes the
5030 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5031
5032 42. It is now possible to set a limit on the number of times the match()
5033 function is called in a call to pcre_exec(). This facility makes it possible to
5034 limit the amount of recursion and backtracking, though not in a directly
5035 obvious way, because the match() function is used in a number of different
5036 circumstances. The count starts from zero for each position in the subject
5037 string (for non-anchored patterns). The default limit is, for compatibility, a
5038 large number, namely 10 000 000. You can change this in two ways:
5039
5040 (a) When configuring PCRE before making, you can use --with-match-limit=n
5041 to set a default value for the compiled library.
5042
5043 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5044 a different value is set. See 45 below.
5045
5046 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5047
5048 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5049 of things that can be changed at compile time. The first argument specifies
5050 what is wanted and the second points to where the information is to be placed.
5051 The current list of available information is:
5052
5053 PCRE_CONFIG_UTF8
5054
5055 The output is an integer that is set to one if UTF-8 support is available;
5056 otherwise it is set to zero.
5057
5058 PCRE_CONFIG_NEWLINE
5059
5060 The output is an integer that it set to the value of the code that is used for
5061 newline. It is either LF (10) or CR (13).
5062
5063 PCRE_CONFIG_LINK_SIZE
5064
5065 The output is an integer that contains the number of bytes used for internal
5066 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5067
5068 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5069
5070 The output is an integer that contains the threshold above which the POSIX
5071 interface uses malloc() for output vectors. See item 31 above.
5072
5073 PCRE_CONFIG_MATCH_LIMIT
5074
5075 The output is an unsigned integer that contains the default limit of the number
5076 of match() calls in a pcre_exec() execution. See 42 above.
5077
5078 44. pcretest has been upgraded by the addition of the -C option. This causes it
5079 to extract all the available output from the new pcre_config() function, and to
5080 output it. The program then exits immediately.
5081
5082 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5083 order to support additional features. One way would have been to define
5084 pcre_exec2() (for example) with extra arguments, but this would not have been
5085 extensible, and would also have required all calls to the original function to
5086 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5087 is used for passing in "extra" data from pcre_study().
5088
5089 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5090 contains the following fields:
5091
5092 flags a bitmap indicating which of the following fields are set
5093 study_data opaque data from pcre_study()
5094 match_limit a way of specifying a limit on match() calls for a specific
5095 call to pcre_exec()
5096 callout_data data for callouts (see 49 below)
5097
5098 The flag bits are also defined in pcre.h, and are
5099
5100 PCRE_EXTRA_STUDY_DATA
5101 PCRE_EXTRA_MATCH_LIMIT
5102 PCRE_EXTRA_CALLOUT_DATA
5103
5104 The pcre_study() function now returns one of these new pcre_extra blocks, with
5105 the actual study data pointed to by the study_data field, and the
5106 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5107 before. That is, this change is entirely upwards-compatible and requires no
5108 change to existing code.
5109
5110 If you want to pass in additional data to pcre_exec(), you can either place it
5111 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5112 block.
5113
5114 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5115 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5116 times with different match limits, until it finds the minimum value needed for
5117 pcre_exec() to complete. The value is then output. This can be instructive; for
5118 most simple matches the number is quite small, but for pathological cases it
5119 gets very large very quickly.
5120
5121 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5122 returns the size of the data block pointed to by the study_data field in a
5123 pcre_extra block, that is, the value that was passed as the argument to
5124 pcre_malloc() when PCRE was getting memory in which to place the information
5125 created by pcre_study(). The fourth argument should point to a size_t variable.
5126 pcretest has been extended so that this information is shown after a successful
5127 pcre_study() call when information about the compiled regex is being displayed.
5128
5129 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5130 because what follows is always an absolute path. (Later: it turns out that this
5131 is more than cosmetic for MinGW, because it doesn't like empty path
5132 components.)
5133
5134 49. Some changes have been made to the callout feature (see 28 above):
5135
5136 (i) A callout function now has three choices for what it returns:
5137
5138 0 => success, carry on matching
5139 > 0 => failure at this point, but backtrack if possible
5140 < 0 => serious error, return this value from pcre_exec()
5141
5142 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5143 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5144 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5145 use by callout functions. It will never be used by PCRE itself.
5146
5147 (ii) The pcre_extra structure (see 45 above) has a void * field called
5148 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5149 pcre_callout_block structure has a field of the same name. The contents of
5150 the field passed in the pcre_extra structure are passed to the callout
5151 function in the corresponding field in the callout block. This makes it
5152 easier to use the same callout-containing regex from multiple threads. For
5153 testing, the pcretest program has a new data escape
5154
5155 \C*n pass the number n (may be negative) as callout_data
5156
5157 If the callout function in pcretest receives a non-zero value as
5158 callout_data, it returns that value.
5159
5160 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5161 there were some redundant $(CFLAGS) in commands that are now specified as
5162 $(LINK), which already includes $(CFLAGS).
5163
5164 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5165 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5166 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5167 one-byte characters throughout. Note that case-insensitive matching applies
5168 only to characters whose values are less than 256. PCRE doesn't support the
5169 notion of cases for higher-valued characters.
5170
5171 (i) A character class whose characters are all within 0-255 is handled as
5172 a bit map, and the map is inverted for negative classes. Previously, a
5173 character > 255 always failed to match such a class; however it should
5174 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5175
5176 (ii) A negated character class with a single character < 255 is coded as
5177 "not this character" (OP_NOT). This wasn't working properly when the test
5178 character was multibyte, either singly or repeated.
5179
5180 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5181 mode, for example: \x{100}{2,3}.
5182
5183 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5184 singly or repeated) now correctly test multibyte characters. However,
5185 PCRE doesn't recognize any characters with values greater than 255 as
5186 digits, spaces, or word characters. Such characters always match \D, \S,
5187 and \W, and never match \d, \s, or \w.
5188
5189 (v) Classes may now contain characters and character ranges with values
5190 greater than 255. For example: [ab\x{100}-\x{400}].
5191
5192 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5193 PCRE in UTF-8 mode.
5194
5195 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5196 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5197 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5198 value.)
5199
5200 53. The single man page has become too large. I have therefore split it up into
5201 a number of separate man pages. These also give rise to individual HTML pages;
5202 these are now put in a separate directory, and there is an index.html page that
5203 lists them all. Some hyperlinking between the pages has been installed.
5204
5205 54. Added convenience functions for handling named capturing parentheses.
5206
5207 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5208 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5209 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5210 are faulted.
5211
5212 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5213 calling configure. These values are used when compiling the dftables.c program
5214 which is run to generate the source of the default character tables. They
5215 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5216 you will need to set these values.
5217
5218 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5219
5220
5221 Version 3.9 02-Jan-02
5222 ---------------------
5223
5224 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5225
5226 2. If --disable-static was given, the building process failed when trying to
5227 build pcretest and pcregrep. (For some reason it was using libtool to compile
5228 them, which is not right, as they aren't part of the library.)
5229
5230
5231 Version 3.8 18-Dec-01
5232 ---------------------
5233
5234 1. The experimental UTF-8 code was completely screwed up. It was packing the
5235 bytes in the wrong order. How dumb can you get?
5236
5237
5238 Version 3.7 29-Oct-01
5239 ---------------------
5240
5241 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5242 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5243 this didn't happen under Solaris 8, where I normally test things.
5244
5245 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5246 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5247 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5248 this fix an hour or so after the initial 3.7 release.)
5249
5250
5251 Version 3.6 23-Oct-01
5252 ---------------------
5253
5254 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5255 offsets passed as NULL with zero offset count.
5256
5257 2. The config.guess and config.sub files had not been updated when I moved to
5258 the latest autoconf.
5259
5260
5261 Version 3.5 15-Aug-01
5262 ---------------------
5263
5264 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5265 had been forgotten.
5266
5267 2. By using declared but undefined structures, we can avoid using "void"
5268 definitions in pcre.h while keeping the internal definitions of the structures
5269 private.
5270
5271 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5272 user point of view, this means that both static and shared libraries are built
5273 by default, but this can be individually controlled. More of the work of
5274 handling this static/shared cases is now inside libtool instead of PCRE's make
5275 file.
5276
5277 4. The pcretest utility is now installed along with pcregrep because it is
5278 useful for users (to test regexs) and by doing this, it automatically gets
5279 relinked by libtool. The documentation has been turned into a man page, so
5280 there are now .1, .txt, and .html versions in /doc.
5281
5282 5. Upgrades to pcregrep:
5283 (i) Added long-form option names like gnu grep.
5284 (ii) Added --help to list all options with an explanatory phrase.
5285 (iii) Added -r, --recursive to recurse into sub-directories.
5286 (iv) Added -f, --file to read patterns from a file.
5287
5288 6. pcre_exec() was referring to its "code" argument before testing that
5289 argument for NULL (and giving an error if it was NULL).
5290
5291 7. Upgraded Makefile.in to allow for compiling in a different directory from
5292 the source directory.
5293
5294 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5295 options bits, the pointer it was passed was to an int instead of to an unsigned
5296 long int. This mattered only on 64-bit systems.
5297
5298 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5299 generated) instead of pcre.in, which it its source. Also made the same change
5300 in several of the .c files.
5301
5302 10. A new release of gcc defines printf() as a macro, which broke pcretest
5303 because it had an ifdef in the middle of a string argument for printf(). Fixed
5304 by using separate calls to printf().
5305
5306 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5307 script, to force use of CR or LF instead of \n in the source. On non-Unix
5308 systems, the value can be set in config.h.
5309
5310 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5311 absolute limit. Changed the text of the error message to make this clear, and
5312 likewise updated the man page.
5313
5314 13. The limit of 99 on the number of capturing subpatterns has been removed.
5315 The new limit is 65535, which I hope will not be a "real" limit.
5316
5317
5318 Version 3.4 22-Aug-00
5319 ---------------------
5320
5321 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5322
5323 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5324
5325
5326 Version 3.3 01-Aug-00
5327 ---------------------
5328
5329 1. If an octal character was given, but the value was greater than \377, it
5330 was not getting masked to the least significant bits, as documented. This could
5331 lead to crashes in some systems.
5332
5333 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5334 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5335
5336 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5337 These just pass their arguments on to (pcre_free)(), but they are provided
5338 because some uses of PCRE bind it to non-C systems that can call its functions,
5339 but cannot call free() or pcre_free() directly.
5340
5341 4. Add "make test" as a synonym for "make check". Corrected some comments in
5342 the Makefile.
5343
5344 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5345 Makefile.
5346
5347 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5348 command called pgrep for grepping around the active processes.
5349
5350 7. Added the beginnings of support for UTF-8 character strings.
5351
5352 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5353 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5354 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5355 out for the ar command.)
5356
5357
5358 Version 3.2 12-May-00
5359 ---------------------
5360
5361 This is purely a bug fixing release.
5362
5363 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5364 of ZA. This was just one example of several cases that could provoke this bug,
5365 which was introduced by change 9 of version 2.00. The code for breaking
5366 infinite loops after an iteration that matches an empty string was't working
5367 correctly.
5368
5369 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5370 when matched against abbab (for example). After matching an empty string, it
5371 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5372 caused it to match further down the string than it should.
5373
5374 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5375 was there because it doesn't seem to be needed, and it causes trouble on some
5376 systems, as it is not a Standard C header. It has been removed.
5377
5378 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5379 were reported on the Macintosh. The changes were from
5380
5381 while ((c = *(++ptr)) != 0 && c != '\n');
5382 to
5383 while ((c = *(++ptr)) != 0 && c != '\n') ;
5384
5385 Totally extraordinary, but if that's what it takes...
5386
5387 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5388 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5389 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5390 assumes the way PCRE uses memmove() (always moving upwards).
5391
5392 6. PCRE is being used in one environment where strchr() is not available. There
5393 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5394 faster code anyway.
5395
5396
5397 Version 3.1 09-Feb-00
5398 ---------------------
5399
5400 The only change in this release is the fixing of some bugs in Makefile.in for
5401 the "install" target:
5402
5403 (1) It was failing to install pcreposix.h.
5404
5405 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5406
5407
5408 Version 3.0 01-Feb-00
5409 ---------------------
5410
5411 1. Add support for the /+ modifier to perltest (to output $` like it does in
5412 pcretest).
5413
5414 2. Add support for the /g modifier to perltest.
5415
5416 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5417 matches null strings.
5418
5419 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5420 pattern. Perl treats empty patterns specially - it reuses the most recent
5421 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5422 effect.
5423
5424 5. The POSIX interface was broken in that it was just handing over the POSIX
5425 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5426 required a bigger vector, with some working space on the end. This means that
5427 the POSIX wrapper now has to get and free some memory, and copy the results.
5428
5429 6. Added some simple autoconf support, placing the test data and the
5430 documentation in separate directories, re-organizing some of the
5431 information files, and making it build pcre-config (a GNU standard). Also added
5432 libtool support for building PCRE as a shared library, which is now the
5433 default.
5434
5435 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5436 09 are not valid octal constants. Single digits will be used for minor values
5437 less than 10.
5438
5439 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5440 existing programs that set these in the POSIX interface can use PCRE without
5441 modification.
5442
5443 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5444 return all that pcre_info() returns, plus additional data. The pcre_info()
5445 function is retained for compatibility, but is considered to be obsolete.
5446
5447 10. Added experimental recursion feature (?R) to handle one common case that
5448 Perl 5.6 will be able to do with (?p{...}).
5449
5450 11. Added support for POSIX character classes like [:alpha:], which Perl is
5451 adopting.
5452
5453
5454 Version 2.08 31-Aug-99
5455 ----------------------
5456
5457 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5458 trying to match at the startoffset position, but instead was moving forward to
5459 the next newline as if a previous match had failed.
5460
5461 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5462 and could get into a loop if a null string was matched other than at the start
5463 of the subject.
5464
5465 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5466 be distinguished at compile time, and for completeness also added PCRE_DATE.
5467
5468 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5469 in GnuWin32 environments.
5470
5471
5472 Version 2.07 29-Jul-99
5473 ----------------------
5474
5475 1. The documentation is now supplied in plain text form and HTML as well as in
5476 the form of man page sources.
5477
5478 2. C++ compilers don't like assigning (void *) values to other pointer types.
5479 In particular this affects malloc(). Although there is no problem in Standard
5480 C, I've put in casts to keep C++ compilers happy.
5481
5482 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5483 should be (const char *).
5484
5485 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5486 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5487 However, I haven't made this a standard facility. The documentation doesn't
5488 mention it, and the Makefile doesn't support it.
5489
5490 5. The Makefile now contains an "install" target, with editable destinations at
5491 the top of the file. The pcretest program is not installed.
5492
5493 6. pgrep -V now gives the PCRE version number and date.
5494
5495 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5496 causing the entire string to be ignored, instead of just the last character.
5497
5498 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5499 non-matching string, it can take a very, very long time, even for strings of
5500 quite modest length, because of the nested recursion. PCRE now does better in
5501 some of these cases. It does this by remembering the last required literal
5502 character in the pattern, and pre-searching the subject to ensure it is present
5503 before running the real match. In other words, it applies a heuristic to detect
5504 some types of certain failure quickly, and in the above example, if presented
5505 with a string that has no trailing " it gives "no match" very quickly.
5506
5507 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5508 other alternatives are tried instead.
5509
5510
5511 Version 2.06 09-Jun-99
5512 ----------------------
5513
5514 1. Change pcretest's output for amount of store used to show just the code
5515 space, because the remainder (the data block) varies in size between 32-bit and
5516 64-bit systems.
5517
5518 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5519 start matching at. This allows lookbehinds to work when searching for multiple
5520 occurrences in a string.
5521
5522 3. Added additional options to pcretest for testing multiple occurrences:
5523
5524 /+ outputs the rest of the string that follows a match
5525 /g loops for multiple occurrences, using the new startoffset argument
5526 /G loops for multiple occurrences by passing an incremented pointer
5527
5528 4. PCRE wasn't doing the "first character" optimization for patterns starting
5529 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5530 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5531 the letter 'x'. On long subject strings, this gives a significant speed-up.
5532
5533
5534 Version 2.05 21-Apr-99
5535 ----------------------
5536
5537 1. Changed the type of magic_number from int to long int so that it works
5538 properly on 16-bit systems.
5539
5540 2. Fixed a bug which caused patterns starting with .* not to work correctly
5541 when the subject string contained newline characters. PCRE was assuming
5542 anchoring for such patterns in all cases, which is not correct because .* will
5543 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5544 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5545 must be retried after every newline in the subject.
5546
5547
5548 Version 2.04 18-Feb-99
5549 ----------------------
5550
5551 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5552 computation of the store needed to hold the pattern was incorrect (too large).
5553 If such patterns were nested a few deep, this could multiply and become a real
5554 problem.
5555
5556 2. Added /M option to pcretest to show the memory requirement of a specific
5557 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5558
5559 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5560 compiled in such a way that the backtracking after subsequent failure was
5561 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5562 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5563
5564
5565 Version 2.03 02-Feb-99
5566 ----------------------
5567
5568 1. Fixed typo and small mistake in man page.
5569
5570 2. Added 4th condition (GPL supersedes if conflict) and created separate
5571 LICENCE file containing the conditions.
5572
5573 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5574 Perl, that is the internal \ allows the delimiter to be included in the
5575 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5576 the final delimiter, add \ to the end of the pattern (to test the error).
5577
5578 4. Added the convenience functions for extracting substrings after a successful
5579 match. Updated pcretest to make it able to test these functions.
5580
5581
5582 Version 2.02 14-Jan-99
5583 ----------------------
5584
5585 1. Initialized the working variables associated with each extraction so that
5586 their saving and restoring doesn't refer to uninitialized store.
5587
5588 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5589 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5590 fix the problem.
5591
5592 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5593 calls, and wasn't printing the correct value for compiling calls. Increased the
5594 default value of LOOPREPEAT, and the number of significant figures in the
5595 times.
5596
5597 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5598
5599 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5600 a building problem on Windows NT with a FAT file system.
5601
5602
5603 Version 2.01 21-Oct-98
5604 ----------------------
5605
5606 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5607 to character tables built by pcre_maketables() in the current locale. If NULL
5608 is passed, the default tables are used.
5609
5610
5611 Version 2.00 24-Sep-98
5612 ----------------------
5613
5614 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5615 it any more.
5616
5617 2. Allow quantification of (?>) groups, and make it work correctly.
5618
5619 3. The first character computation wasn't working for (?>) groups.
5620
5621 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5622 end of the subject) and add 5.005's \z, which really does match only at the
5623 very end of the subject.
5624
5625 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5626
5627 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5628 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5629 localized options. All options to pcre_study() were also removed.
5630
5631 7. Add other new features from 5.005:
5632
5633 $(?<= positive lookbehind
5634 $(?<! negative lookbehind
5635 (?imsx-imsx) added the unsetting capability
5636 such a setting is global if at outer level; local otherwise
5637 (?imsx-imsx:) non-capturing groups with option setting
5638 (?(cond)re|re) conditional pattern matching
5639
5640 A backreference to itself in a repeated group matches the previous
5641 captured string.
5642
5643 8. General tidying up of studying (both automatic and via "study")
5644 consequential on the addition of new assertions.
5645
5646 9. As in 5.005, unlimited repeated groups that could match an empty substring
5647 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5648 runtime if any iteration does actually match an empty substring.
5649
5650 10. Include the RunTest script in the distribution.
5651
5652 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5653 discrepancies, some of which were old and were also with respect to 5.004. They
5654 have now been fixed.
5655
5656
5657 Version 1.09 28-Apr-98
5658 ----------------------
5659
5660 1. A negated single character class followed by a quantifier with a minimum
5661 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5662 program crashes, or just wrong answers. This did not apply to negated classes
5663 containing more than one character, or to minima other than one.
5664
5665
5666 Version 1.08 27-Mar-98
5667 ----------------------
5668
5669 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5670
5671 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5672 latter must appear before anything that relies on it in the pattern.
5673
5674
5675 Version 1.07 16-Feb-98
5676 ----------------------
5677
5678 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5679 repeat of a potentially empty string).
5680
5681
5682 Version 1.06 23-Jan-98
5683 ----------------------
5684
5685 1. Added Markus Oberhumer's little patches for C++.
5686
5687 2. Literal strings longer than 255 characters were broken.
5688
5689
5690 Version 1.05 23-Dec-97
5691 ----------------------
5692
5693 1. Negated character classes containing more than one character were failing if
5694 PCRE_CASELESS was set at run time.
5695
5696
5697 Version 1.04 19-Dec-97
5698 ----------------------
5699
5700 1. Corrected the man page, where some "const" qualifiers had been omitted.
5701
5702 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5703 input syntax.
5704
5705 3. Fixed memory leak which occurred when a regex with back references was
5706 matched with an offsets vector that wasn't big enough. The temporary memory
5707 that is used in this case wasn't being freed if the match failed.
5708
5709 4. Tidied pcretest to ensure it frees memory that it gets.
5710
5711 5. Temporary memory was being obtained in the case where the passed offsets
5712 vector was exactly big enough.
5713
5714 6. Corrected definition of offsetof() from change 5 below.
5715
5716 7. I had screwed up change 6 below and broken the rules for the use of
5717 setjmp(). Now fixed.
5718
5719
5720 Version 1.03 18-Dec-97
5721 ----------------------
5722
5723 1. A erroneous regex with a missing opening parenthesis was correctly
5724 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5725 on some systems.
5726
5727 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5728 it was reported that one broken compiler failed on the former because "code" is
5729 also an independent variable.
5730
5731 3. The erroneous regex a[]b caused an array overrun reference.
5732
5733 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5734 fail on data ending with that character. (It was going on too far, and checking
5735 the next character, typically a binary zero.) This was specific to the
5736 optimized code for single-character negative classes.
5737
5738 5. Added a contributed patch from the TIN world which does the following:
5739
5740 + Add an undef for memmove, in case the the system defines a macro for it.
5741
5742 + Add a definition of offsetof(), in case there isn't one. (I don't know
5743 the reason behind this - offsetof() is part of the ANSI standard - but
5744 it does no harm).
5745
5746 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5747 most of the places where whitespace preceded '#'. I have given up and
5748 allowed the remaining 2 cases to be at the margin.
5749
5750 + Rename some variables in pcre to eliminate shadowing. This seems very
5751 pedantic, but does no harm, of course.
5752
5753 6. Moved the call to setjmp() into its own function, to get rid of warnings
5754 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5755
5756 7. Constructs such as \d{8,} were compiling into the equivalent of
5757 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5758 outcome, but in this particular case used more store than had been allocated,
5759 which caused the bug to be discovered because it threw up an internal error.
5760
5761 8. The debugging code in both pcre and pcretest for outputting the compiled
5762 form of a regex was going wrong in the case of back references followed by
5763 curly-bracketed repeats.
5764
5765
5766 Version 1.02 12-Dec-97
5767 ----------------------
5768
5769 1. Typos in pcre.3 and comments in the source fixed.
5770
5771 2. Applied a contributed patch to get rid of places where it used to remove
5772 'const' from variables, and fixed some signed/unsigned and uninitialized
5773 variable warnings.
5774
5775 3. Added the "runtest" target to Makefile.
5776
5777 4. Set default compiler flag to -O2 rather than just -O.
5778
5779
5780 Version 1.01 19-Nov-97
5781 ----------------------
5782
5783 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5784 like /([ab]*)*/, that is, for classes with more than one character in them.
5785
5786 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5787 as /((?>a*))*/ (a PCRE_EXTRA facility).
5788
5789
5790 Version 1.00 18-Nov-97
5791 ----------------------
5792
5793 1. Added compile-time macros to support systems such as SunOS4 which don't have
5794 memmove() or strerror() but have other things that can be used instead.
5795
5796 2. Arranged that "make clean" removes the executables.
5797
5798
5799 Version 0.99 27-Oct-97
5800 ----------------------
5801
5802 1. Fixed bug in code for optimizing classes with only one character. It was
5803 initializing a 32-byte map regardless, which could cause it to run off the end
5804 of the memory it had got.
5805
5806 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5807
5808
5809 Version 0.98 22-Oct-97
5810 ----------------------
5811
5812 1. Fixed bug in code for handling temporary memory usage when there are more
5813 back references than supplied space in the ovector. This could cause segfaults.
5814
5815
5816 Version 0.97 21-Oct-97
5817 ----------------------
5818
5819 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5820
5821 2. Optimized negated single characters not to use a bit map.
5822
5823 3. Brought error texts together as macro definitions; clarified some of them;
5824 fixed one that was wrong - it said "range out of order" when it meant "invalid
5825 escape sequence".
5826
5827 4. Changed some char * arguments to const char *.
5828
5829 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5830
5831 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5832 pcretest.
5833
5834
5835 Version 0.96 16-Oct-97
5836 ----------------------
5837
5838 1. Added a simple "pgrep" utility to the distribution.
5839
5840 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5841 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5842 where "ddd" means "one or more decimal digits".
5843
5844 3. Fixed serious bug. If a pattern had a back reference, but the call to
5845 pcre_exec() didn't supply a large enough ovector to record the related
5846 identifying subpattern, the match always failed. PCRE now remembers the number
5847 of the largest back reference, and gets some temporary memory in which to save
5848 the offsets during matching if necessary, in order to ensure that
5849 backreferences always work.
5850
5851 4. Increased the compatibility with Perl in a number of ways:
5852
5853 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5854 to request this handling. The option can be set at compile or exec time.
5855
5856 (b) $ matches before a terminating newline by default; an option
5857 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5858 mode). The option can be set at compile or exec time.
5859
5860 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5861 the same as Perl's. If the decimal number it represents is less than 10
5862 or there aren't that many previous left capturing parentheses, an octal
5863 escape is read. Inside a character class, it's always an octal escape,
5864 even if it is a single digit.
5865
5866 (d) An escaped but undefined alphabetic character is taken as a literal,
5867 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5868 escapes.
5869
5870 (e) {0} is now permitted. (The previous item is removed from the compiled
5871 pattern).
5872
5873 5. Changed all the names of code files so that the basic parts are no longer
5874 than 10 characters, and abolished the teeny "globals.c" file.
5875
5876 6. Changed the handling of character classes; they are now done with a 32-byte
5877 bit map always.
5878
5879 7. Added the -d and /D options to pcretest to make it possible to look at the
5880 internals of compilation without having to recompile pcre.
5881
5882
5883 Version 0.95 23-Sep-97
5884 ----------------------
5885
5886 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5887 \x20 at the start of a run of normal characters. These were being treated as
5888 real characters, instead of the source characters being re-checked.
5889
5890
5891 Version 0.94 18-Sep-97
5892 ----------------------
5893
5894 1. The functions are now thread-safe, with the caveat that the global variables
5895 containing pointers to malloc() and free() or alternative functions are the
5896 same for all threads.
5897
5898 2. Get pcre_study() to generate a bitmap of initial characters for non-
5899 anchored patterns when this is possible, and use it if passed to pcre_exec().
5900
5901
5902 Version 0.93 15-Sep-97
5903 ----------------------
5904
5905 1. /(b)|(:+)/ was computing an incorrect first character.
5906
5907 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5908 but not actually doing anything yet.
5909
5910 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5911 as Perl does (e.g. [-az] or [az-]).
5912
5913 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5914 all possible positions.
5915
5916 5. Split up into different modules to avoid including unneeded functions in a
5917 compiled binary. However, compile and exec are still in one module. The "study"
5918 function is split off.
5919
5920 6. The character tables are now in a separate module whose source is generated
5921 by an auxiliary program - but can then be edited by hand if required. There are
5922 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5923 toupper() in the code.
5924
5925 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5926 make them global. Abolish the function for setting them, as the caller can now
5927 set them directly.
5928
5929
5930 Version 0.92 11-Sep-97
5931 ----------------------
5932
5933 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5934 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5935
5936 2. Caseless matching was not working in character classes if the characters in
5937 the pattern were in upper case.
5938
5939 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5940
5941 4. Make PCRE_ANCHORED public and accept as a compile option.
5942
5943 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5944 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5945 pass them.
5946
5947 6. Give an error if bad option bits passed at compile or run time.
5948
5949 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5950 pcretest to cause it to pass that flag.
5951
5952 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5953 options, and the first character, if set.
5954
5955 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5956
5957
5958 Version 0.91 10-Sep-97
5959 ----------------------
5960
5961 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5962 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5963
5964 2. PCRE was looping on encountering an indefinitely repeated back reference to
5965 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5966 Perl does - treats the match as successful.
5967
5968 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5