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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1631 - (show annotations)
Wed Feb 10 19:13:17 2016 UTC (2 days, 1 hour ago) by ph10
File size: 287861 byte(s)
Error occurred while calculating annotation data.
Fix workspace overflow for (*ACCEPT) with deeply nested parentheses.
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. Migrating appropriate PCRE2 JIT improvements to PCRE.
48
49 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
50 assertion, caused pcretest to generate incorrect output, and also to read
51 uninitialized memory (detected by ASAN or valgrind).
52
53 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
54 nested set of parentheses of sufficient size caused an overflow of the
55 compiling workspace (which was diagnosed, but of course is not desirable).
56
57
58 Version 8.38 23-November-2015
59 -----------------------------
60
61 1. If a group that contained a recursive back reference also contained a
62 forward reference subroutine call followed by a non-forward-reference
63 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
64 compile correct code, leading to undefined behaviour or an internally
65 detected error. This bug was discovered by the LLVM fuzzer.
66
67 2. Quantification of certain items (e.g. atomic back references) could cause
68 incorrect code to be compiled when recursive forward references were
69 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
70 This bug was discovered by the LLVM fuzzer.
71
72 3. A repeated conditional group whose condition was a reference by name caused
73 a buffer overflow if there was more than one group with the given name.
74 This bug was discovered by the LLVM fuzzer.
75
76 4. A recursive back reference by name within a group that had the same name as
77 another group caused a buffer overflow. For example:
78 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
79
80 5. A forward reference by name to a group whose number is the same as the
81 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
82 a buffer overflow at compile time. This bug was discovered by the LLVM
83 fuzzer.
84
85 6. A lookbehind assertion within a set of mutually recursive subpatterns could
86 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
87
88 7. Another buffer overflow bug involved duplicate named groups with a
89 reference between their definition, with a group that reset capture
90 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
91 fixed by always allowing for more memory, even if not needed. (A proper fix
92 is implemented in PCRE2, but it involves more refactoring.)
93
94 8. There was no check for integer overflow in subroutine calls such as (?123).
95
96 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
97 being treated as a literal 'l' instead of causing an error.
98
99 10. There was a buffer overflow if pcre_exec() was called with an ovector of
100 size 1. This bug was found by american fuzzy lop.
101
102 11. If a non-capturing group containing a conditional group that could match
103 an empty string was repeated, it was not identified as matching an empty
104 string itself. For example: /^(?:(?(1)x|)+)+$()/.
105
106 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
107 \a and \e in test subject lines.
108
109 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
110 instead of the EBCDIC value.
111
112 14. The handling of \c in an EBCDIC environment has been revised so that it is
113 now compatible with the specification in Perl's perlebcdic page.
114
115 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
116 ASCII/Unicode. This has now been added to the list of characters that are
117 recognized as white space in EBCDIC.
118
119 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
120 error (correctly) when used outside a class, but did not give an error
121 within a class.
122
123 17. \h within a class was incorrectly compiled in EBCDIC environments.
124
125 18. A pattern with an unmatched closing parenthesis that contained a backward
126 assertion which itself contained a forward reference caused buffer
127 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
128
129 19. JIT should return with error when the compiled pattern requires more stack
130 space than the maximum.
131
132 20. A possessively repeated conditional group that could match an empty string,
133 for example, /(?(R))*+/, was incorrectly compiled.
134
135 21. Fix infinite recursion in the JIT compiler when certain patterns such as
136 /(?:|a|){100}x/ are analysed.
137
138 22. Some patterns with character classes involving [: and \\ were incorrectly
139 compiled and could cause reading from uninitialized memory or an incorrect
140 error diagnosis.
141
142 23. Pathological patterns containing many nested occurrences of [: caused
143 pcre_compile() to run for a very long time.
144
145 24. A conditional group with only one branch has an implicit empty alternative
146 branch and must therefore be treated as potentially matching an empty
147 string.
148
149 25. If (?R was followed by - or + incorrect behaviour happened instead of a
150 diagnostic.
151
152 26. Arrange to give up on finding the minimum matching length for overly
153 complex patterns.
154
155 27. Similar to (4) above: in a pattern with duplicated named groups and an
156 occurrence of (?| it is possible for an apparently non-recursive back
157 reference to become recursive if a later named group with the relevant
158 number is encountered. This could lead to a buffer overflow. Wen Guanxing
159 from Venustech ADLAB discovered this bug.
160
161 28. If pcregrep was given the -q option with -c or -l, or when handling a
162 binary file, it incorrectly wrote output to stdout.
163
164 29. The JIT compiler did not restore the control verb head in case of *THEN
165 control verbs. This issue was found by Karl Skomski with a custom LLVM
166 fuzzer.
167
168 30. Error messages for syntax errors following \g and \k were giving inaccurate
169 offsets in the pattern.
170
171 31. Added a check for integer overflow in conditions (?(<digits>) and
172 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
173 fuzzer.
174
175 32. Handling recursive references such as (?2) when the reference is to a group
176 later in the pattern uses code that is very hacked about and error-prone.
177 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
178 give an internal error if it is obvious that compiling has gone wrong.
179
180 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
181 This issue was found by Karl Skomski with a custom LLVM fuzzer.
182
183 34. The JIT compiler should restore the control chain for empty possessive
184 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
185
186 35. Match limit check added to JIT recursion. This issue was found by Karl
187 Skomski with a custom LLVM fuzzer.
188
189 36. Yet another case similar to 27 above has been circumvented by an
190 unconditional allocation of extra memory. This issue is fixed "properly" in
191 PCRE2 by refactoring the way references are handled. Wen Guanxing
192 from Venustech ADLAB discovered this bug.
193
194 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
195 with a custom LLVM fuzzer.
196
197 38. Fixed a corner case of range optimization in JIT.
198
199 39. An incorrect error "overran compiling workspace" was given if there were
200 exactly enough group forward references such that the last one extended
201 into the workspace safety margin. The next one would have expanded the
202 workspace. The test for overflow was not including the safety margin.
203
204 40. A match limit issue is fixed in JIT which was found by Karl Skomski
205 with a custom LLVM fuzzer.
206
207 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
208 work under Windows. (Why has it taken so long for anyone to notice?)
209
210 42. In a character class such as [\W\p{Any}] where both a negative-type escape
211 ("not a word character") and a property escape were present, the property
212 escape was being ignored.
213
214 43. Fix crash caused by very long (*MARK) or (*THEN) names.
215
216 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
217 by a single ASCII character in a class item, was incorrectly compiled in
218 UCP mode. The POSIX class got lost, but only if the single character
219 followed it.
220
221 45. [:punct:] in UCP mode was matching some characters in the range 128-255
222 that should not have been matched.
223
224 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
225 class, all characters with code points greater than 255 are in the class.
226 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
227 such as \w are turned into Unicode properties), wide characters were not
228 correctly handled, and could fail to match.
229
230
231 Version 8.37 28-April-2015
232 --------------------------
233
234 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
235 for those parentheses to be closed with whatever has been captured so far.
236 However, it was failing to mark any other groups between the hightest
237 capture so far and the currrent group as "unset". Thus, the ovector for
238 those groups contained whatever was previously there. An example is the
239 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
240
241 2. If an assertion condition was quantified with a minimum of zero (an odd
242 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
243
244 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
245 unrecognized modifier, a crash could occur.
246
247 4. An attempt to do global matching in pcretest with a zero-length ovector
248 caused a crash.
249
250 5. Fixed a memory leak during matching that could occur for a subpattern
251 subroutine call (recursive or otherwise) if the number of captured groups
252 that had to be saved was greater than ten.
253
254 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
255 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
256 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
257
258 7. A UTF pattern containing a "not" match of a non-ASCII character and a
259 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
260
261 8. When a pattern is compiled, it remembers the highest back reference so that
262 when matching, if the ovector is too small, extra memory can be obtained to
263 use instead. A conditional subpattern whose condition is a check on a
264 capture having happened, such as, for example in the pattern
265 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
266 setting the highest backreference number. This mattered only if pcre_exec()
267 was called with an ovector that was too small to hold the capture, and there
268 was no other kind of back reference (a situation which is probably quite
269 rare). The effect of the bug was that the condition was always treated as
270 FALSE when the capture could not be consulted, leading to a incorrect
271 behaviour by pcre_exec(). This bug has been fixed.
272
273 9. A reference to a duplicated named group (either a back reference or a test
274 for being set in a conditional) that occurred in a part of the pattern where
275 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
276 to be incorrectly calculated, leading to overwriting.
277
278 10. A mutually recursive set of back references such as (\2)(\1) caused a
279 segfault at study time (while trying to find the minimum matching length).
280 The infinite loop is now broken (with the minimum length unset, that is,
281 zero).
282
283 11. If an assertion that was used as a condition was quantified with a minimum
284 of zero, matching went wrong. In particular, if the whole group had
285 unlimited repetition and could match an empty string, a segfault was
286 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
287 assertions to be quantified, but not if they are being used as conditions,
288 so the above pattern is faulted by Perl. PCRE has now been changed so that
289 it also rejects such patterns.
290
291 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
292 failed to allow the zero-repeat case if pcre2_exec() was called with an
293 ovector too small to capture the group.
294
295 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
296 Red Hat Product Security:
297
298 (a) A crash if /K and /F were both set with the option to save the compiled
299 pattern.
300
301 (b) Another crash if the option to print captured substrings in a callout
302 was combined with setting a null ovector, for example \O\C+ as a subject
303 string.
304
305 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
306 forward reference repeated a large (but limited) number of times within a
307 repeated outer group that has a zero minimum quantifier, caused incorrect
308 code to be compiled, leading to the error "internal error:
309 previously-checked referenced subpattern not found" when an incorrect
310 memory address was read. This bug was reported as "heap overflow",
311 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
312 CVE-2015-2325.
313
314 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
315 call within a group that also contained a recursive back reference caused
316 incorrect code to be compiled. This bug was reported as "heap overflow",
317 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
318 number CVE-2015-2326.
319
320 24. Computing the size of the JIT read-only data in advance has been a source
321 of various issues, and new ones are still appear unfortunately. To fix
322 existing and future issues, size computation is eliminated from the code,
323 and replaced by on-demand memory allocation.
324
325 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
326 adjacent to the end of the range, and the range contained characters with
327 more than one other case, caused incorrect behaviour when compiled in UTF
328 mode. In that example, the range a-j was left out of the class.
329
330 26. Fix JIT compilation of conditional blocks, which assertion
331 is converted to (*FAIL). E.g: /(?(?!))/.
332
333 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
334 discovered by the LLVM fuzzer.
335
336 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
337 when this assertion was used as a condition, for example (?(?!)a|b). In
338 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
339 error about an unsupported item.
340
341 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
342 possessification code could take exponential time to complete. A recursion
343 depth limit of 1000 has been imposed to limit the resources used by this
344 optimization.
345
346 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
347 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
348 because \S ensures they are all in the class. The code for doing this was
349 interacting badly with the code for computing the amount of space needed to
350 compile the pattern, leading to a buffer overflow. This bug was discovered
351 by the LLVM fuzzer.
352
353 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
354 other kinds of group caused stack overflow at compile time. This bug was
355 discovered by the LLVM fuzzer.
356
357 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
358 between a subroutine call and its quantifier was incorrectly compiled,
359 leading to buffer overflow or other errors. This bug was discovered by the
360 LLVM fuzzer.
361
362 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
363 assertion after (?(. The code was failing to check the character after
364 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
365 was discovered by the LLVM fuzzer.
366
367 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
368 a fixed maximum following a group that contains a subroutine reference was
369 incorrectly compiled and could trigger buffer overflow. This bug was
370 discovered by the LLVM fuzzer.
371
372 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
373 caused a stack overflow instead of the diagnosis of a non-fixed length
374 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
375
376 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
377 (e.g. /(?<=\Ka)/) could make pcregrep loop.
378
379 37. There was a similar problem to 36 in pcretest for global matches.
380
381 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
382 and a subsequent item in the pattern caused a non-match, backtracking over
383 the repeated \X did not stop, but carried on past the start of the subject,
384 causing reference to random memory and/or a segfault. There were also some
385 other cases where backtracking after \C could crash. This set of bugs was
386 discovered by the LLVM fuzzer.
387
388 39. The function for finding the minimum length of a matching string could take
389 a very long time if mutual recursion was present many times in a pattern,
390 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
391 method has been implemented. This infelicity was discovered by the LLVM
392 fuzzer.
393
394 40. Static linking against the PCRE library using the pkg-config module was
395 failing on missing pthread symbols.
396
397
398 Version 8.36 26-September-2014
399 ------------------------------
400
401 1. Got rid of some compiler warnings in the C++ modules that were shown up by
402 -Wmissing-field-initializers and -Wunused-parameter.
403
404 2. The tests for quantifiers being too big (greater than 65535) were being
405 applied after reading the number, and stupidly assuming that integer
406 overflow would give a negative number. The tests are now applied as the
407 numbers are read.
408
409 3. Tidy code in pcre_exec.c where two branches that used to be different are
410 now the same.
411
412 4. The JIT compiler did not generate match limit checks for certain
413 bracketed expressions with quantifiers. This may lead to exponential
414 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
415 issue should be resolved now.
416
417 5. Fixed an issue, which occures when nested alternatives are optimized
418 with table jumps.
419
420 6. Inserted two casts and changed some ints to size_t in the light of some
421 reported 64-bit compiler warnings (Bugzilla 1477).
422
423 7. Fixed a bug concerned with zero-minimum possessive groups that could match
424 an empty string, which sometimes were behaving incorrectly in the
425 interpreter (though correctly in the JIT matcher). This pcretest input is
426 an example:
427
428 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
429 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
430
431 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
432 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
433 for an empty string was breaking the inner loop and carrying on at a lower
434 level, when possessive repeated groups should always return to a higher
435 level as they have no backtrack points in them. The empty string test now
436 occurs at the outer level.
437
438 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
439 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
440
441 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
442 doesn't).
443
444 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
445 to hold a pattern to be incorrectly computed (too small) when there were
446 named back references to duplicated names. This could cause "internal
447 error: code overflow" or "double free or corruption" or other memory
448 handling errors.
449
450 11. When named subpatterns had the same prefixes, back references could be
451 confused. For example, in this pattern:
452
453 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
454
455 the reference to 'Name' was incorrectly treated as a reference to a
456 duplicate name.
457
458 12. A pattern such as /^s?c/mi8 where the optional character has more than
459 one "other case" was incorrectly compiled such that it would only try to
460 match starting at "c".
461
462 13. When a pattern starting with \s was studied, VT was not included in the
463 list of possible starting characters; this should have been part of the
464 8.34/18 patch.
465
466 14. If a character class started [\Qx]... where x is any character, the class
467 was incorrectly terminated at the ].
468
469 15. If a pattern that started with a caseless match for a character with more
470 than one "other case" was studied, PCRE did not set up the starting code
471 unit bit map for the list of possible characters. Now it does. This is an
472 optimization improvement, not a bug fix.
473
474 16. The Unicode data tables have been updated to Unicode 7.0.0.
475
476 17. Fixed a number of memory leaks in pcregrep.
477
478 18. Avoid a compiler warning (from some compilers) for a function call with
479 a cast that removes "const" from an lvalue by using an intermediate
480 variable (to which the compiler does not object).
481
482 19. Incorrect code was compiled if a group that contained an internal recursive
483 back reference was optional (had quantifier with a minimum of zero). This
484 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
485 caused segmentation faults because of stack overflows at compile time.
486
487 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
488 group that is quantified with an indefinite repeat, caused a compile-time
489 loop which used up all the system stack and provoked a segmentation fault.
490 This was not the same bug as 19 above.
491
492 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
493 Patch by Mike Frysinger.
494
495
496 Version 8.35 04-April-2014
497 --------------------------
498
499 1. A new flag is set, when property checks are present in an XCLASS.
500 When this flag is not set, PCRE can perform certain optimizations
501 such as studying these XCLASS-es.
502
503 2. The auto-possessification of character sets were improved: a normal
504 and an extended character set can be compared now. Furthermore
505 the JIT compiler optimizes more character set checks.
506
507 3. Got rid of some compiler warnings for potentially uninitialized variables
508 that show up only when compiled with -O2.
509
510 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
511 of a match later then the end of the match. The pcretest program was not
512 handling the case sensibly - it was outputting from the start to the next
513 binary zero. It now reports this situation in a message, and outputs the
514 text from the end to the start.
515
516 5. Fast forward search is improved in JIT. Instead of the first three
517 characters, any three characters with fixed position can be searched.
518 Search order: first, last, middle.
519
520 6. Improve character range checks in JIT. Characters are read by an inprecise
521 function now, which returns with an unknown value if the character code is
522 above a certain threshold (e.g: 256). The only limitation is that the value
523 must be bigger than the threshold as well. This function is useful when
524 the characters above the threshold are handled in the same way.
525
526 7. The macros whose names start with RAWUCHAR are placeholders for a future
527 mode in which only the bottom 21 bits of 32-bit data items are used. To
528 make this more memorable for those maintaining the code, the names have
529 been changed to start with UCHAR21, and an extensive comment has been added
530 to their definition.
531
532 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
533 to the export list in Makefile.am (they were accidentally omitted from the
534 8.34 tarball).
535
536 9. The informational output from pcretest used the phrase "starting byte set"
537 which is inappropriate for the 16-bit and 32-bit libraries. As the output
538 for "first char" and "need char" really means "non-UTF-char", I've changed
539 "byte" to "char", and slightly reworded the output. The documentation about
540 these values has also been (I hope) clarified.
541
542 10. Another JIT related optimization: use table jumps for selecting the correct
543 backtracking path, when more than four alternatives are present inside a
544 bracket.
545
546 11. Empty match is not possible, when the minimum length is greater than zero,
547 and there is no \K in the pattern. JIT should avoid empty match checks in
548 such cases.
549
550 12. In a caseless character class with UCP support, when a character with more
551 than one alternative case was not the first character of a range, not all
552 the alternative cases were added to the class. For example, s and \x{17f}
553 are both alternative cases for S: the class [RST] was handled correctly,
554 but [R-T] was not.
555
556 13. The configure.ac file always checked for pthread support when JIT was
557 enabled. This is not used in Windows, so I have put this test inside a
558 check for the presence of windows.h (which was already tested for).
559
560 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
561 The algorithm provides a way to skip certain starting offsets, and usually
562 faster than linear prefix searches.
563
564 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
565 as for 'fr_FR' and 'french'. For some reason, however, it then used the
566 Windows-specific input and output files, which have 'french' screwed in.
567 So this could never have worked. One of the problems with locales is that
568 they aren't always the same. I have now updated RunTest so that it checks
569 the output of the locale test (test 3) against three different output
570 files, and it allows the test to pass if any one of them matches. With luck
571 this should make the test pass on some versions of Solaris where it was
572 failing. Because of the uncertainty, the script did not used to stop if
573 test 3 failed; it now does. If further versions of a French locale ever
574 come to light, they can now easily be added.
575
576 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
577 there was a message during ./configure, but it did not stop. This now
578 provokes an error. The invalid example in README has been corrected.
579 If a value less than the minimum is given, the minimum value has always
580 been used, but now a warning is given.
581
582 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
583 was a bug in the test system, which is now fixed. Also, the list of various
584 configurations that are tested for each release did not have one with both
585 16/32 bits and --enable-bar-anycrlf. It now does.
586
587 18. pcretest was missing "-C bsr" for displaying the \R default setting.
588
589 19. Little endian PowerPC systems are supported now by the JIT compiler.
590
591 20. The fast forward newline mechanism could enter to an infinite loop on
592 certain invalid UTF-8 input. Although we don't support these cases
593 this issue can be fixed by a performance optimization.
594
595 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
596 not take account if existing stack usage. There is now a new global
597 variable called pcre_stack_guard that can be set to point to an external
598 function to check stack availability. It is called at the start of
599 processing every parenthesized group.
600
601 22. A typo in the code meant that in ungreedy mode the max/min qualifier
602 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
603 not match "ab".
604
605 23. When UTF was disabled, the JIT program reported some incorrect compile
606 errors. These messages are silenced now.
607
608 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
609 compiler.
610
611 25. Change all the temporary files used in RunGrepTest to be different to those
612 used by RunTest so that the tests can be run simultaneously, for example by
613 "make -j check".
614
615
616 Version 8.34 15-December-2013
617 -----------------------------
618
619 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
620 executable memory. Patch inspired by Carsten Klein.
621
622 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
623 this macro is never tested and has no effect, because the work to support
624 coverage involves only compiling and linking options and special targets in
625 the Makefile. The comment in config.h implied that defining the macro would
626 enable coverage support, which is totally false. There was also support for
627 setting this macro in the CMake files (my fault, I just copied it from
628 configure). SUPPORT_GCOV has now been removed.
629
630 3. Make a small performance improvement in strlen16() and strlen32() in
631 pcretest.
632
633 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
634 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
635 these cases). There was also one in pcretest.c.
636
637 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
638
639 6. In UTF mode, the code for checking whether a group could match an empty
640 string (which is used for indefinitely repeated groups to allow for
641 breaking an infinite loop) was broken when the group contained a repeated
642 negated single-character class with a character that occupied more than one
643 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
644 UTF-8 mode). The effect was undefined: the group might or might not be
645 deemed as matching an empty string, or the program might have crashed.
646
647 7. The code for checking whether a group could match an empty string was not
648 recognizing that \h, \H, \v, \V, and \R must match a character.
649
650 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
651 an empty string. If it can, pcretest shows this in its information output.
652
653 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
654 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
655 matched by pcre_exec() without using JIT:
656
657 (a) If the rest of the pattern did not match after a maximal run of
658 grapheme clusters, the code for backing up to try with fewer of them
659 did not always back up over a full grapheme when characters that do not
660 have the modifier quality were involved, e.g. Hangul syllables.
661
662 (b) If the match point in a subject started with modifier character, and
663 there was no match, the code could incorrectly back up beyond the match
664 point, and potentially beyond the first character in the subject,
665 leading to a segfault or an incorrect match result.
666
667 10. A conditional group with an assertion condition could lead to PCRE
668 recording an incorrect first data item for a match if no other first data
669 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
670 first data item, and therefore matched "ca" after "c" instead of at the
671 start.
672
673 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
674 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
675
676 12. The source of pcregrep now includes z/OS-specific code so that it can be
677 compiled for z/OS as part of the special z/OS distribution.
678
679 13. Added the -T and -TM options to pcretest.
680
681 14. The code in pcre_compile.c for creating the table of named capturing groups
682 has been refactored. Instead of creating the table dynamically during the
683 actual compiling pass, the information is remembered during the pre-compile
684 pass (on the stack unless there are more than 20 named groups, in which
685 case malloc() is used) and the whole table is created before the actual
686 compile happens. This has simplified the code (it is now nearly 150 lines
687 shorter) and prepared the way for better handling of references to groups
688 with duplicate names.
689
690 15. A back reference to a named subpattern when there is more than one of the
691 same name now checks them in the order in which they appear in the pattern.
692 The first one that is set is used for the reference. Previously only the
693 first one was inspected. This change makes PCRE more compatible with Perl.
694
695 16. Unicode character properties were updated from Unicode 6.3.0.
696
697 17. The compile-time code for auto-possessification has been refactored, based
698 on a patch by Zoltan Herczeg. It now happens after instead of during
699 compilation. The code is cleaner, and more cases are handled. The option
700 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
701 options in pcretest are provided to set it. It can also be set by
702 (*NO_AUTO_POSSESS) at the start of a pattern.
703
704 18. The character VT has been added to the default ("C" locale) set of
705 characters that match \s and are generally treated as white space,
706 following this same change in Perl 5.18. There is now no difference between
707 "Perl space" and "POSIX space". Whether VT is treated as white space in
708 other locales depends on the locale.
709
710 19. The code for checking named groups as conditions, either for being set or
711 for being recursed, has been refactored (this is related to 14 and 15
712 above). Processing unduplicated named groups should now be as fast at
713 numerical groups, and processing duplicated groups should be faster than
714 before.
715
716 20. Two patches to the CMake build system, by Alexander Barkov:
717
718 (1) Replace the "source" command by "." in CMakeLists.txt because
719 "source" is a bash-ism.
720
721 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
722 without these the CMake build does not work on Solaris.
723
724 21. Perl has changed its handling of \8 and \9. If there is no previously
725 encountered capturing group of those numbers, they are treated as the
726 literal characters 8 and 9 instead of a binary zero followed by the
727 literals. PCRE now does the same.
728
729 22. Following Perl, added \o{} to specify codepoints in octal, making it
730 possible to specify values greater than 0777 and also making them
731 unambiguous.
732
733 23. Perl now gives an error for missing closing braces after \x{... instead of
734 treating the string as literal. PCRE now does the same.
735
736 24. RunTest used to grumble if an inappropriate test was selected explicitly,
737 but just skip it when running all tests. This make it awkward to run ranges
738 of tests when one of them was inappropriate. Now it just skips any
739 inappropriate tests, as it always did when running all tests.
740
741 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
742 character types such as \d or \w, too many callouts were inserted, and the
743 data that they returned was rubbish.
744
745 26. In UCP mode, \s was not matching two of the characters that Perl matches,
746 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
747 were matched by \h. The code has now been refactored so that the lists of
748 the horizontal and vertical whitespace characters used for \h and \v (which
749 are defined only in one place) are now also used for \s.
750
751 27. Add JIT support for the 64 bit TileGX architecture.
752 Patch by Jiong Wang (Tilera Corporation).
753
754 28. Possessive quantifiers for classes (both explicit and automatically
755 generated) now use special opcodes instead of wrapping in ONCE brackets.
756
757 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
758 (because it's meaningless), this was not happening when PCRE_CASELESS was
759 set. Not wrong, but inefficient.
760
761 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
762 properties for \w, \d, etc) is present in a test regex. Otherwise if the
763 test contains no characters greater than 255, Perl doesn't realise it
764 should be using Unicode semantics.
765
766 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
767 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
768 does in Unicode mode.
769
770 32. Added the "forbid" facility to pcretest so that putting tests into the
771 wrong test files can sometimes be quickly detected.
772
773 33. There is now a limit (default 250) on the depth of nesting of parentheses.
774 This limit is imposed to control the amount of system stack used at compile
775 time. It can be changed at build time by --with-parens-nest-limit=xxx or
776 the equivalent in CMake.
777
778 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
779 errors. Perl warns for these when in warning mode, but PCRE has no facility
780 for giving warnings.
781
782 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
783 However, this was not working for (?!) because it is optimized to (*FAIL),
784 for which PCRE does not allow quantifiers. The optimization is now disabled
785 when a quantifier follows (?!). I can't see any use for this, but it makes
786 things uniform.
787
788 36. Perl no longer allows group names to start with digits, so I have made this
789 change also in PCRE. It simplifies the code a bit.
790
791 37. In extended mode, Perl ignores spaces before a + that indicates a
792 possessive quantifier. PCRE allowed a space before the quantifier, but not
793 before the possessive +. It now does.
794
795 38. The use of \K (reset reported match start) within a repeated possessive
796 group such as (a\Kb)*+ was not working.
797
798 40. Document that the same character tables must be used at compile time and
799 run time, and that the facility to pass tables to pcre_exec() and
800 pcre_dfa_exec() is for use only with saved/restored patterns.
801
802 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
803 features for Builds with MSVC:
804
805 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
806 builds). The .rc files can be used to set FileDescription and many other
807 attributes.
808
809 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
810 This allows higher-level build scripts which want .pdb files to avoid
811 hard-coding the exact files needed."
812
813 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
814 mean "start of word" and "end of word", respectively, as a transition aid.
815
816 43. A minimizing repeat of a class containing codepoints greater than 255 in
817 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
818 compiled to use the heap for recursion.
819
820 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
821 is configured.
822
823
824 Version 8.33 28-May-2013
825 ------------------------
826
827 1. Added 'U' to some constants that are compared to unsigned integers, to
828 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
829 variables that are added to signed variables, to ensure the result is
830 signed and can be negated.
831
832 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
833 CMake config files.
834
835 3. Revise the creation of config.h.generic so that all boolean macros are
836 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
837 overriding via -D on the command line possible.
838
839 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
840 to unsigned int is reported to make a quite noticeable speed difference in
841 a specific Windows environment. Testing on Linux did also appear to show
842 some benefit (and it is clearly not harmful). Also fixed the definition of
843 Xop which should be unsigned.
844
845 5. Related to (4), changing the definition of the intermediate variable cc
846 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
847 improvements.
848
849 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
850 unnecessary spaces.
851
852 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
853 and later.
854
855 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
856
857 9. Optimizing fast_forward_start_bits in JIT.
858
859 10. Adding support for callouts in JIT, and fixing some issues revealed
860 during this work. Namely:
861
862 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
863
864 (b) Minimum length was not checked before the matching is started.
865
866 11. The value of capture_last that is passed to callouts was incorrect in some
867 cases when there was a capture on one path that was subsequently abandoned
868 after a backtrack. Also, the capture_last value is now reset after a
869 recursion, since all captures are also reset in this case.
870
871 12. The interpreter no longer returns the "too many substrings" error in the
872 case when an overflowing capture is in a branch that is subsequently
873 abandoned after a backtrack.
874
875 13. In the pathological case when an offset vector of size 2 is used, pcretest
876 now prints out the matched string after a yield of 0 or 1.
877
878 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
879 Only supported by the JIT compiler at the moment.
880
881 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
882
883 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
884 offset of the starting point of the matching process, provided the offsets
885 vector is large enough.
886
887 17. The \A escape now records a lookbehind value of 1, though its execution
888 does not actually inspect the previous character. This is to ensure that,
889 in partial multi-segment matching, at least one character from the old
890 segment is retained when a new segment is processed. Otherwise, if there
891 are no lookbehinds in the pattern, \A might match incorrectly at the start
892 of a new segment.
893
894 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
895
896 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
897 gives some modest performance improvement in 8-bit mode.
898
899 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
900 be expressed in certain programming languages using Universal Character
901 Names.
902
903 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
904 which points out that "non characters" are not "characters that may not
905 appear in Unicode strings" but rather "characters that are reserved for
906 internal use and have only local meaning".
907
908 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
909 there was a conditional group that depended on an assertion, if the
910 assertion was false, the callout that immediately followed the alternation
911 in the condition was skipped when pcre_exec() was used for matching.
912
913 23. Allow an explicit callout to be inserted before an assertion that is the
914 condition for a conditional group, for compatibility with automatic
915 callouts, which always insert a callout at this point.
916
917 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
918 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
919
920 25. (*PRUNE) is now supported by the JIT compiler.
921
922 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
923
924 27. Fix the case where there are two or more SKIPs with arguments that may be
925 ignored.
926
927 28. (*SKIP) is now supported by the JIT compiler.
928
929 29. (*THEN) is now supported by the JIT compiler.
930
931 30. Update RunTest with additional test selector options.
932
933 31. The way PCRE handles backtracking verbs has been changed in two ways.
934
935 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
936 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
937 backtracking. In some cases this makes it more Perl-compatible, but Perl's
938 rather obscure rules do not always do the same thing.
939
940 (2) Previously, backtracking verbs were confined within assertions. This is
941 no longer the case for positive assertions, except for (*ACCEPT). Again,
942 this sometimes improves Perl compatibility, and sometimes does not.
943
944 32. A number of tests that were in test 2 because Perl did things differently
945 have been moved to test 1, because either Perl or PCRE has changed, and
946 these tests are now compatible.
947
948 32. Backtracking control verbs are now handled in the same way in JIT and
949 interpreter.
950
951 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
952 contained a forward subroutine reference caused a compile error.
953
954 34. Auto-detect and optimize limited repetitions in JIT.
955
956 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
957 blocking (*UTF) etc.
958
959 36. In the interpreter, maximizing pattern repetitions for characters and
960 character types now use tail recursion, which reduces stack usage.
961
962 37. The value of the max lookbehind was not correctly preserved if a compiled
963 and saved regex was reloaded on a host of different endianness.
964
965 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
966 of the compiled pattern block, expand the flags field from 16 to 32 bits
967 because it was almost full.
968
969 39. Try madvise first before posix_madvise.
970
971 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
972 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
973 original change locked out all matches of empty strings. This has been
974 changed so that one match of an empty string per line is recognized.
975 Subsequent searches on the same line (for colouring or for --only-matching,
976 for example) do not recognize empty strings.
977
978 41. Applied a user patch to fix a number of spelling mistakes in comments.
979
980 42. Data lines longer than 65536 caused pcretest to crash.
981
982 43. Clarified the data type for length and startoffset arguments for pcre_exec
983 and pcre_dfa_exec in the function-specific man pages, where they were
984 explicitly stated to be in bytes, never having been updated. I also added
985 some clarification to the pcreapi man page.
986
987 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
988 a segmentation fault.
989
990
991 Version 8.32 30-November-2012
992 -----------------------------
993
994 1. Improved JIT compiler optimizations for first character search and single
995 character iterators.
996
997 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
998 Patch by Daniel Richard G.
999
1000 3. Single character iterator optimizations in the JIT compiler.
1001
1002 4. Improved JIT compiler optimizations for character ranges.
1003
1004 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1005 Reported by Giuseppe D'Angelo.
1006
1007 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1008 of a line, was being set incorrectly in cases where .* appeared inside
1009 atomic brackets at the start of a pattern, or where there was a subsequent
1010 *PRUNE or *SKIP.
1011
1012 7. Improved instruction cache flush for POWER/PowerPC.
1013 Patch by Daniel Richard G.
1014
1015 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1016 grep:
1017
1018 (a) There is now no limit to the number of patterns to be matched.
1019
1020 (b) An error is given if a pattern is too long.
1021
1022 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1023 are now supported.
1024
1025 (d) --exclude-from and --include-from (multiple use) have been added.
1026
1027 (e) Exclusions and inclusions now apply to all files and directories, not
1028 just to those obtained from scanning a directory recursively.
1029
1030 (f) Multiple uses of -f and --file-list are now supported.
1031
1032 (g) In a Windows environment, the default for -d has been changed from
1033 "read" (the GNU grep default) to "skip", because otherwise the presence
1034 of a directory in the file list provokes an error.
1035
1036 (h) The documentation has been revised and clarified in places.
1037
1038 9. Improve the matching speed of capturing brackets.
1039
1040 10. Changed the meaning of \X so that it now matches a Unicode extended
1041 grapheme cluster.
1042
1043 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1044 out POSIX threads when JIT support is configured.
1045
1046 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1047
1048 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1049 structure could go wrong in environments where size_t is not the same size
1050 as int.
1051
1052 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1053 set.
1054
1055 15. The EBCDIC support had decayed; later updates to the code had included
1056 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1057 general tidy up of EBCDIC-related issues, and the documentation was also
1058 not quite right. There is now a test that can be run on ASCII systems to
1059 check some of the EBCDIC-related things (but is it not a full test).
1060
1061 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1062 in a small tidy to the code.
1063
1064 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1065
1066 18. If the --only-matching (-o) option in pcregrep is specified multiple
1067 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1068 substrings matched by the 1st and 2nd capturing parentheses. A separating
1069 string can be specified by --om-separator (default empty).
1070
1071 19. Improving the first n character searches.
1072
1073 20. Turn case lists for horizontal and vertical white space into macros so that
1074 they are defined only once.
1075
1076 21. This set of changes together give more compatible Unicode case-folding
1077 behaviour for characters that have more than one other case when UCP
1078 support is available.
1079
1080 (a) The Unicode property table now has offsets into a new table of sets of
1081 three or more characters that are case-equivalent. The MultiStage2.py
1082 script that generates these tables (the pcre_ucd.c file) now scans
1083 CaseFolding.txt instead of UnicodeData.txt for character case
1084 information.
1085
1086 (b) The code for adding characters or ranges of characters to a character
1087 class has been abstracted into a generalized function that also handles
1088 case-independence. In UTF-mode with UCP support, this uses the new data
1089 to handle characters with more than one other case.
1090
1091 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1092 whose other case is greater than 256 are now correctly matched
1093 caselessly. Previously, the high codepoint matched the low one, but not
1094 vice versa.
1095
1096 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1097 of the new class addition function, using character lists defined as
1098 macros alongside the case definitions of 20 above.
1099
1100 (e) Caseless back references now work with characters that have more than
1101 one other case.
1102
1103 (f) General caseless matching of characters with more than one other case
1104 is supported.
1105
1106 22. Unicode character properties were updated from Unicode 6.2.0
1107
1108 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1109
1110 24. Add support for 32-bit character strings, and UTF-32
1111
1112 25. Major JIT compiler update (code refactoring and bugfixing).
1113 Experimental Sparc 32 support is added.
1114
1115 26. Applied a modified version of Daniel Richard G's patch to create
1116 pcre.h.generic and config.h.generic by "make" instead of in the
1117 PrepareRelease script.
1118
1119 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1120 pcre_compile.c when checking for a zero character.
1121
1122 28. Introducing a native interface for JIT. Through this interface, the compiled
1123 machine code can be directly executed. The purpose of this interface is to
1124 provide fast pattern matching, so several sanity checks are not performed.
1125 However, feature tests are still performed. The new interface provides
1126 1.4x speedup compared to the old one.
1127
1128 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1129 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1130 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1131
1132 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1133 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1134 the "old" RFC 2279). Instead, it ended up passing a negative length to
1135 pcre_exec().
1136
1137 31. Add support for GCC's visibility feature to hide internal functions.
1138
1139 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1140 "unknown -C option" after outputting 0 or 1.
1141
1142 33. There is now support for generating a code coverage report for the test
1143 suite in environments where gcc is the compiler and lcov is installed. This
1144 is mainly for the benefit of the developers.
1145
1146 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1147 unaddressable using valgrind annotations, allowing valgrind to detect
1148 invalid memory accesses. This is mainly for the benefit of the developers.
1149
1150 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1151
1152 26. Give configure error if --enable-cpp but no C++ compiler found.
1153
1154
1155 Version 8.31 06-July-2012
1156 -------------------------
1157
1158 1. Fixing a wrong JIT test case and some compiler warnings.
1159
1160 2. Removed a bashism from the RunTest script.
1161
1162 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1163 to unsigned type, result still unsigned" that was given by an MS compiler
1164 on encountering the code "-sizeof(xxx)".
1165
1166 4. Partial matching support is added to the JIT compiler.
1167
1168 5. Fixed several bugs concerned with partial matching of items that consist
1169 of more than one character:
1170
1171 (a) /^(..)\1/ did not partially match "aba" because checking references was
1172 done on an "all or nothing" basis. This also applied to repeated
1173 references.
1174
1175 (b) \R did not give a hard partial match if \r was found at the end of the
1176 subject.
1177
1178 (c) \X did not give a hard partial match after matching one or more
1179 characters at the end of the subject.
1180
1181 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1182 a partial match for the string "\r".
1183
1184 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1185 a partial match for a CR character at the end of the subject string.
1186
1187 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1188 running pcretest, the text "(JIT)" added to the output whenever JIT is
1189 actually used to run the match.
1190
1191 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1192 or /S+[+] with a digit between 1 and 7.
1193
1194 8. OP_NOT now supports any UTF character not just single-byte ones.
1195
1196 9. (*MARK) control verb is now supported by the JIT compiler.
1197
1198 10. The command "./RunTest list" lists the available tests without actually
1199 running any of them. (Because I keep forgetting what they all are.)
1200
1201 11. Add PCRE_INFO_MAXLOOKBEHIND.
1202
1203 12. Applied a (slightly modified) user-supplied patch that improves performance
1204 when the heap is used for recursion (compiled with --disable-stack-for-
1205 recursion). Instead of malloc and free for each heap frame each time a
1206 logical recursion happens, frames are retained on a chain and re-used where
1207 possible. This sometimes gives as much as 30% improvement.
1208
1209 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1210 call.
1211
1212 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1213
1214 15. It is now possible to link pcretest with libedit as an alternative to
1215 libreadline.
1216
1217 16. (*COMMIT) control verb is now supported by the JIT compiler.
1218
1219 17. The Unicode data tables have been updated to Unicode 6.1.0.
1220
1221 18. Added --file-list option to pcregrep.
1222
1223 19. Added binary file support to pcregrep, including the -a, --binary-files,
1224 -I, and --text options.
1225
1226 20. The madvise function is renamed for posix_madvise for QNX compatibility
1227 reasons. Fixed by Giuseppe D'Angelo.
1228
1229 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1230
1231 22. Changed the option for creating symbolic links for 16-bit man pages from
1232 -s to -sf so that re-installing does not cause issues.
1233
1234 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1235
1236 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1237 matches in certain environments (the workspace was not being correctly
1238 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1239 some of the workspace data at the beginning of a restart.
1240
1241 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1242 was not doing so when it should - probably a typo introduced by SVN 528
1243 (change 8.10/14).
1244
1245 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1246 \w+ when the character tables indicated that \x{c4} was a word character.
1247 There were several related cases, all because the tests for doing a table
1248 lookup were testing for characters less than 127 instead of 255.
1249
1250 27. If a pattern contains capturing parentheses that are not used in a match,
1251 their slots in the ovector are set to -1. For those that are higher than
1252 any matched groups, this happens at the end of processing. In the case when
1253 there were back references that the ovector was too small to contain
1254 (causing temporary malloc'd memory to be used during matching), and the
1255 highest capturing number was not used, memory off the end of the ovector
1256 was incorrectly being set to -1. (It was using the size of the temporary
1257 memory instead of the true size.)
1258
1259 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1260 ovector size, it uses memory at the end of the block that it has got.
1261
1262 29. Check for an overlong MARK name and give an error at compile time. The
1263 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1264
1265 30. JIT compiler update.
1266
1267 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1268 Rill for the patch.
1269
1270 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1271
1272 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1273
1274 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1275 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1276 was enabled.
1277
1278 35. Improve JIT code generation for greedy plus quantifier.
1279
1280 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1281 1 to "aa" instead of to an empty string. The bug affected repeated groups
1282 that could potentially match an empty string.
1283
1284 37. Optimizing single character iterators in JIT.
1285
1286 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1287 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1288 codepoints that are too big for the mode are faulted, and in a UTF mode,
1289 disallowed codepoints are also faulted.
1290
1291 39. If PCRE was compiled with UTF support, in three places in the DFA
1292 matcher there was code that should only have been obeyed in UTF mode, but
1293 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1294 processing when bytes with values greater than 127 were present. In 16-bit
1295 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1296 both cases the values are those that cannot be the first data item in a UTF
1297 character. The three items that might have provoked this were recursions,
1298 possessively repeated groups, and atomic groups.
1299
1300 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1301 and pcregrep, because some OS require shared objects to be explicitly
1302 passed to ld, causing the link step to fail if they are not.
1303
1304 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1305 mode, patterns that started with \h* or \R* might be incorrectly matched.
1306
1307
1308 Version 8.30 04-February-2012
1309 -----------------------------
1310
1311 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1312 name is defined in ctype.h.
1313
1314 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1315 only in quite long subpatterns.
1316
1317 3. Removed the function pcre_info(), which has been obsolete and deprecated
1318 since it was replaced by pcre_fullinfo() in February 2000.
1319
1320 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1321 match a (*MARK), and the match failed at the start of the subject, a
1322 reference to memory before the start of the subject could occur. This bug
1323 was introduced by fix 17 of release 8.21.
1324
1325 5. A reference to an unset group with zero minimum repetition was giving
1326 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1327 /(another)?(\1?)test/ matched against "hello world test". This bug was
1328 introduced in release 8.13.
1329
1330 6. Add support for 16-bit character strings (a large amount of work involving
1331 many changes and refactorings).
1332
1333 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1334 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1335 from a file.
1336
1337 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1338 rounding is not applied in this particular case).
1339
1340 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1341 if they appear, or are escaped, in patterns.
1342
1343 10. Get rid of a number of -Wunused-but-set-variable warnings.
1344
1345 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1346 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1347 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1348 also returns the mark "x". This bug applied to capturing parentheses,
1349 non-capturing parentheses, and atomic parentheses. It also applied to some
1350 assertions.
1351
1352 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1353 information out of configure.ac instead of relying on pcre.h.generic, which
1354 is not stored in the repository.
1355
1356 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1357 -lreadline.
1358
1359 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1360
1361 15. Applied Graycode's patch to put the top-level frame on the stack rather
1362 than the heap when not using the stack for recursion. This gives a
1363 performance improvement in many cases when recursion is not deep.
1364
1365 16. Experimental code added to "pcretest -C" to output the stack frame size.
1366
1367
1368 Version 8.21 12-Dec-2011
1369 ------------------------
1370
1371 1. Updating the JIT compiler.
1372
1373 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1374 are added as well.
1375
1376 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1377 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1378 calling _pcre_jit_exec. Some extra comments are added.
1379
1380 4. (*MARK) settings inside atomic groups that do not contain any capturing
1381 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1382 was introduced by change 18 for 8.20.
1383
1384 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1385 ECMA-262 standard.
1386
1387 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1388 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1389 This bug was probably introduced by change 9 of 8.13.
1390
1391 7. While fixing 6 above, I noticed that a number of other items were being
1392 incorrectly rejected as "not fixed length". This arose partly because newer
1393 opcodes had not been added to the fixed-length checking code. I have (a)
1394 corrected the bug and added tests for these items, and (b) arranged for an
1395 error to occur if an unknown opcode is encountered while checking for fixed
1396 length instead of just assuming "not fixed length". The items that were
1397 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1398 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1399 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1400
1401 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1402 being incorrectly compiled and would have given unpredicatble results.
1403
1404 9. A possessively repeated subpattern with minimum repeat count greater than
1405 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1406 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1407 the first (A) could occur when it should not.
1408
1409 10. Add a cast and remove a redundant test from the code.
1410
1411 11. JIT should use pcre_malloc/pcre_free for allocation.
1412
1413 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1414 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1415 is anything other than /usr, -L is still shown).
1416
1417 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1418
1419 14. Perl does not support \N without a following name in a [] class; PCRE now
1420 also gives an error.
1421
1422 15. If a forward reference was repeated with an upper limit of around 2000,
1423 it caused the error "internal error: overran compiling workspace". The
1424 maximum number of forward references (including repeats) was limited by the
1425 internal workspace, and dependent on the LINK_SIZE. The code has been
1426 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1427 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1428 of around 200,000 forward references. While doing this, I also speeded up
1429 the filling in of repeated forward references.
1430
1431 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1432 incorrectly expecting the subject to contain another "a" after the start.
1433
1434 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1435 in the match, the SKIP should be ignored. This was not happening; instead
1436 the SKIP was being treated as NOMATCH. For patterns such as
1437 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1438 tested.
1439
1440 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1441 now much more compatible with Perl, in particular in cases where the result
1442 is a non-match for a non-anchored pattern. For example, if
1443 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1444 "m", where previously it did not return a name. A side effect of this
1445 change is that for partial matches, the last encountered mark name is
1446 returned, as for non matches. A number of tests that were previously not
1447 Perl-compatible have been moved into the Perl-compatible test files. The
1448 refactoring has had the pleasing side effect of removing one argument from
1449 the match() function, thus reducing its stack requirements.
1450
1451 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1452 subsequent uses of /S (without +) incorrectly behaved like /S+.
1453
1454 21. Retrieve executable code size support for the JIT compiler and fixing
1455 some warnings.
1456
1457 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1458 not work when the shorter character appeared right at the end of the
1459 subject string.
1460
1461 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1462 systems.
1463
1464 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1465 output it when the /M option is used in pcretest.
1466
1467 25. The CheckMan script was not being included in the distribution. Also, added
1468 an explicit "perl" to run Perl scripts from the PrepareRelease script
1469 because this is reportedly needed in Windows.
1470
1471 26. If study data was being save in a file and studying had not found a set of
1472 "starts with" bytes for the pattern, the data written to the file (though
1473 never used) was taken from uninitialized memory and so caused valgrind to
1474 complain.
1475
1476 27. Updated RunTest.bat as provided by Sheri Pierce.
1477
1478 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1479
1480 29. Computation of memory usage for the table of capturing group names was
1481 giving an unnecessarily large value.
1482
1483
1484 Version 8.20 21-Oct-2011
1485 ------------------------
1486
1487 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1488 a POSIX class. After further experiments with Perl, which convinced me that
1489 Perl has bugs and confusions, a closing square bracket is no longer allowed
1490 in a POSIX name. This bug also affected patterns with classes that started
1491 with full stops.
1492
1493 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1494 captured substring, but while checking the failing first alternative,
1495 substring 1 is temporarily captured. If the output vector supplied to
1496 pcre_exec() was not big enough for this capture, the yield of the function
1497 was still zero ("insufficient space for captured substrings"). This cannot
1498 be totally fixed without adding another stack variable, which seems a lot
1499 of expense for a edge case. However, I have improved the situation in cases
1500 such as /(a)(b)x|abc/ matched against "abc", where the return code
1501 indicates that fewer than the maximum number of slots in the ovector have
1502 been set.
1503
1504 3. Related to (2) above: when there are more back references in a pattern than
1505 slots in the output vector, pcre_exec() uses temporary memory during
1506 matching, and copies in the captures as far as possible afterwards. It was
1507 using the entire output vector, but this conflicts with the specification
1508 that only 2/3 is used for passing back captured substrings. Now it uses
1509 only the first 2/3, for compatibility. This is, of course, another edge
1510 case.
1511
1512 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1513 main code base, and can be used by building with --enable-jit. When this is
1514 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1515 runtime --no-jit option is given.
1516
1517 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1518 ovector, the return from the function was zero, implying that there were
1519 other matches that did not fit. The correct "exactly full" value is now
1520 returned.
1521
1522 6. If a subpattern that was called recursively or as a subroutine contained
1523 (*PRUNE) or any other control that caused it to give a non-standard return,
1524 invalid errors such as "Error -26 (nested recursion at the same subject
1525 position)" or even infinite loops could occur.
1526
1527 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1528 computing the minimum length on reaching *ACCEPT, and so ended up with the
1529 wrong value of 1 rather than 0. Further investigation indicates that
1530 computing a minimum subject length in the presence of *ACCEPT is difficult
1531 (think back references, subroutine calls), and so I have changed the code
1532 so that no minimum is registered for a pattern that contains *ACCEPT.
1533
1534 8. If (*THEN) was present in the first (true) branch of a conditional group,
1535 it was not handled as intended. [But see 16 below.]
1536
1537 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1538 Sheri Pierce.
1539
1540 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1541 the first byte in a match must be "a".
1542
1543 11. Change 17 for 8.13 increased the recursion depth for patterns like
1544 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1545 pattern contains any instances of (*THEN). If it does not, the old
1546 optimizations are restored. It would be nice to do this on a per-group
1547 basis, but at the moment that is not feasible.
1548
1549 12. In some environments, the output of pcretest -C is CRLF terminated. This
1550 broke RunTest's code that checks for the link size. A single white space
1551 character after the value is now allowed for.
1552
1553 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1554 For "fr", it uses the Windows-specific input and output files.
1555
1556 14. If (*THEN) appeared in a group that was called recursively or as a
1557 subroutine, it did not work as intended. [But see next item.]
1558
1559 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1560 pattern fragments (but not containing any | characters). If A and B are
1561 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1562 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1563 D. In other words, Perl considers parentheses that do not contain any |
1564 characters to be part of a surrounding alternative, whereas PCRE was
1565 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1566 differently. PCRE now behaves in the same way as Perl, except in the case
1567 of subroutine/recursion calls such as (?1) which have in any case always
1568 been different (but PCRE had them first :-).
1569
1570 16. Related to 15 above: Perl does not treat the | in a conditional group as
1571 creating alternatives. Such a group is treated in the same way as an
1572 ordinary group without any | characters when processing (*THEN). PCRE has
1573 been changed to match Perl's behaviour.
1574
1575 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1576 RunGrepTest script failed.
1577
1578 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1579 inevitable for groups that contain captures, but it can lead to a lot of
1580 stack use in large patterns. The old behaviour has been restored for atomic
1581 groups that do not contain any capturing parentheses.
1582
1583 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1584 suppress the check for a minimum subject length at run time. (If it was
1585 given to pcre_exec() or pcre_dfa_exec() it did work.)
1586
1587 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1588 fail to work when decoding hex characters in data strings in EBCDIC
1589 environments.
1590
1591 21. It appears that in at least one Mac OS environment, the isxdigit() function
1592 is implemented as a macro that evaluates to its argument more than once,
1593 contravening the C 90 Standard (I haven't checked a later standard). There
1594 was an instance in pcretest which caused it to go wrong when processing
1595 \x{...} escapes in subject strings. The has been rewritten to avoid using
1596 things like p++ in the argument of isxdigit().
1597
1598
1599 Version 8.13 16-Aug-2011
1600 ------------------------
1601
1602 1. The Unicode data tables have been updated to Unicode 6.0.0.
1603
1604 2. Two minor typos in pcre_internal.h have been fixed.
1605
1606 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1607 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1608 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1609
1610 4. There were a number of related bugs in the code for matching backrefences
1611 caselessly in UTF-8 mode when codes for the characters concerned were
1612 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1613 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1614 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1615 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1616 2-byte code at the end of the subject (it thought there wasn't enough data
1617 left).
1618
1619 5. Comprehensive information about what went wrong is now returned by
1620 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1621 as the output vector has at least 2 elements. The offset of the start of
1622 the failing character and a reason code are placed in the vector.
1623
1624 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1625 now returned is for the first byte of the failing character, instead of the
1626 last byte inspected. This is an incompatible change, but I hope it is small
1627 enough not to be a problem. It makes the returned offset consistent with
1628 pcre_exec() and pcre_dfa_exec().
1629
1630 7. pcretest now gives a text phrase as well as the error number when
1631 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1632 failure, the offset and reason code are output.
1633
1634 8. When \R was used with a maximizing quantifier it failed to skip backwards
1635 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1636 back over a single character (\n). This seems wrong (because it treated the
1637 two characters as a single entity when going forwards), conflicts with the
1638 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1639 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1640 has been changed.
1641
1642 9. Some internal refactoring has changed the processing so that the handling
1643 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1644 time (the PCRE_DOTALL option was changed this way some time ago: version
1645 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1646 which was always a bit of a fudge. It also means that there is one less
1647 argument for the match() function, which reduces its stack requirements
1648 slightly. This change also fixes an incompatibility with Perl: the pattern
1649 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1650
1651 10. More internal refactoring has drastically reduced the number of recursive
1652 calls to match() for possessively repeated groups such as (abc)++ when
1653 using pcre_exec().
1654
1655 11. While implementing 10, a number of bugs in the handling of groups were
1656 discovered and fixed:
1657
1658 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1659 (a|)*(?1) gave a compile-time internal error.
1660 ((a|)+)+ did not notice that the outer group could match an empty string.
1661 (^a|^)+ was not marked as anchored.
1662 (.*a|.*)+ was not marked as matching at start or after a newline.
1663
1664 12. Yet more internal refactoring has removed another argument from the match()
1665 function. Special calls to this function are now indicated by setting a
1666 value in a variable in the "match data" data block.
1667
1668 13. Be more explicit in pcre_study() instead of relying on "default" for
1669 opcodes that mean there is no starting character; this means that when new
1670 ones are added and accidentally left out of pcre_study(), testing should
1671 pick them up.
1672
1673 14. The -s option of pcretest has been documented for ages as being an old
1674 synonym of -m (show memory usage). I have changed it to mean "force study
1675 for every regex", that is, assume /S for every regex. This is similar to -i
1676 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1677 using it. It makes it easier to run collections of tests with and without
1678 study enabled, and thereby test pcre_study() more easily. All the standard
1679 tests are now run with and without -s (but some patterns can be marked as
1680 "never study" - see 20 below).
1681
1682 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1683 restoration of the capturing data to the outer values was not happening
1684 correctly.
1685
1686 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1687 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1688 pattern had matched an empty string, and so incorrectly returned a no
1689 match.
1690
1691 17. There was optimizing code for the last branch of non-capturing parentheses,
1692 and also for the obeyed branch of a conditional subexpression, which used
1693 tail recursion to cut down on stack usage. Unfortunately, now that there is
1694 the possibility of (*THEN) occurring in these branches, tail recursion is
1695 no longer possible because the return has to be checked for (*THEN). These
1696 two optimizations have therefore been removed. [But see 8.20/11 above.]
1697
1698 18. If a pattern containing \R was studied, it was assumed that \R always
1699 matched two bytes, thus causing the minimum subject length to be
1700 incorrectly computed because \R can also match just one byte.
1701
1702 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1703 was incorrectly computed.
1704
1705 20. If /S is present twice on a test pattern in pcretest input, it now
1706 *disables* studying, thereby overriding the use of -s on the command line
1707 (see 14 above). This is necessary for one or two tests to keep the output
1708 identical in both cases.
1709
1710 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1711 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1712
1713 22. When an atomic group that contained a capturing parenthesis was
1714 successfully matched, but the branch in which it appeared failed, the
1715 capturing was not being forgotten if a higher numbered group was later
1716 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1717 group 1 to "a", when in fact it should be unset. This applied to multi-
1718 branched capturing and non-capturing groups, repeated or not, and also to
1719 positive assertions (capturing in negative assertions does not happen
1720 in PCRE) and also to nested atomic groups.
1721
1722 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1723 subject after a captured substring, to make it easier to tell which of a
1724 number of identical substrings has been captured.
1725
1726 24. The way atomic groups are processed by pcre_exec() has been changed so that
1727 if they are repeated, backtracking one repetition now resets captured
1728 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1729 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1730 "aaa". Previously, it would have been "a". As part of this code
1731 refactoring, the way recursive calls are handled has also been changed.
1732
1733 25. If an assertion condition captured any substrings, they were not passed
1734 back unless some other capturing happened later. For example, if
1735 (?(?=(a))a) was matched against "a", no capturing was returned.
1736
1737 26. When studying a pattern that contained subroutine calls or assertions,
1738 the code for finding the minimum length of a possible match was handling
1739 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1740 group 1 called group 2 while simultaneously a separate group 2 called group
1741 1). A stack overflow occurred in this case. I have fixed this by limiting
1742 the recursion depth to 10.
1743
1744 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1745 Fortmann. This supports explicit test numbers on the command line, and has
1746 argument validation and error reporting.
1747
1748 28. An instance of \X with an unlimited repeat could fail if at any point the
1749 first character it looked at was a mark character.
1750
1751 29. Some minor code refactoring concerning Unicode properties and scripts
1752 should reduce the stack requirement of match() slightly.
1753
1754 30. Added the '=' option to pcretest to check the setting of unused capturing
1755 slots at the end of the pattern, which are documented as being -1, but are
1756 not included in the return count.
1757
1758 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1759 compiled something random. Now it gives a compile-time error (as does
1760 Perl).
1761
1762 32. A *MARK encountered during the processing of a positive assertion is now
1763 recorded and passed back (compatible with Perl).
1764
1765 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1766 had alternative anchored branches, the search for a second match in a line
1767 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1768 matched the line "0102" twice. The same bug affected patterns that started
1769 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1770 twice.
1771
1772 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1773 does, and because of capturing effects, quantifying parenthesized
1774 assertions may at times be useful. Quantifiers are now allowed for
1775 parenthesized assertions.
1776
1777 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1778
1779 36. \g was being checked for fancy things in a character class, when it should
1780 just be a literal "g".
1781
1782 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1783 appearance of a nested POSIX class supersedes an apparent external class.
1784 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1785 unescaped square brackets may also appear as part of class names. For
1786 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1787 more like Perl. (But see 8.20/1 above.)
1788
1789 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1790 was because it thought it was \N{name}, which is not supported).
1791
1792 39. Add minix to OS list not supporting the -S option in pcretest.
1793
1794 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1795 cannot analyze patterns in sufficient detail to catch mutual recursions
1796 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1797 subgroup is called recursively as a subpattern for a second time at the
1798 same position in the subject string. In previous releases this might have
1799 been caught by the recursion limit, or it might have run out of stack.
1800
1801 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1802 happen only once. PCRE was, however incorrectly giving a compile time error
1803 "recursive call could loop indefinitely" because it cannot analyze the
1804 pattern in sufficient detail. The compile time test no longer happens when
1805 PCRE is compiling a conditional subpattern, but actual runaway loops are
1806 now caught at runtime (see 40 above).
1807
1808 42. It seems that Perl allows any characters other than a closing parenthesis
1809 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1810 has been changed to be the same.
1811
1812 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1813 as not to get warnings when autogen.sh is called. Also changed
1814 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1815
1816 44. To help people who use pcregrep to scan files containing exceedingly long
1817 lines, the following changes have been made:
1818
1819 (a) The default value of the buffer size parameter has been increased from
1820 8K to 20K. (The actual buffer used is three times this size.)
1821
1822 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1823 PCRE is built.
1824
1825 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1826 to be set at run time.
1827
1828 (d) Numerical values in pcregrep options can be followed by K or M, for
1829 example --buffer-size=50K.
1830
1831 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1832 given and the return code is set to 2.
1833
1834 45. Add a pointer to the latest mark to the callout data block.
1835
1836 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1837 partial match of an empty string instead of no match. This was specific to
1838 the use of ".".
1839
1840 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1841 complete match instead of a partial match. This bug was dependent on both
1842 the PCRE_UTF8 and PCRE_DOTALL options being set.
1843
1844 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1845 starting byte set, because \b was not being ignored.
1846
1847
1848 Version 8.12 15-Jan-2011
1849 ------------------------
1850
1851 1. Fixed some typos in the markup of the man pages, and wrote a script that
1852 checks for such things as part of the documentation building process.
1853
1854 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1855 --match-limit and --recursion-limit options (added for 8.11). In
1856 particular, this made one of the standard tests fail. (The integer value
1857 went into the wrong half of a long int.)
1858
1859 3. If the --colour option was given to pcregrep with -v (invert match), it
1860 did strange things, either producing crazy output, or crashing. It should,
1861 of course, ignore a request for colour when reporting lines that do not
1862 match.
1863
1864 4. Another pcregrep bug caused similar problems if --colour was specified with
1865 -M (multiline) and the pattern match finished with a line ending.
1866
1867 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1868 matched in multiline mode, the following line was shown as part of the
1869 match. This seems wrong, so I have changed it.
1870
1871 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1872 the check for further matches in the same line (so they could be coloured)
1873 to overrun the end of the current line. If another match was found, it was
1874 incorrectly shown (and then shown again when found in the next line).
1875
1876 7. If pcregrep was compiled under Windows, there was a reference to the
1877 function pcregrep_exit() before it was defined. I am assuming this was
1878 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1879 reported by a user. I've moved the definition above the reference.
1880
1881
1882 Version 8.11 10-Dec-2010
1883 ------------------------
1884
1885 1. (*THEN) was not working properly if there were untried alternatives prior
1886 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1887 backtracked to try for "b" instead of moving to the next alternative branch
1888 at the same level (in this case, to look for "c"). The Perl documentation
1889 is clear that when (*THEN) is backtracked onto, it goes to the "next
1890 alternative in the innermost enclosing group".
1891
1892 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1893 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1894 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1895 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1896 (*THEN).
1897
1898 3. If \s appeared in a character class, it removed the VT character from
1899 the class, even if it had been included by some previous item, for example
1900 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1901 of \s, but is part of the POSIX "space" class.)
1902
1903 4. A partial match never returns an empty string (because you can always
1904 match an empty string at the end of the subject); however the checking for
1905 an empty string was starting at the "start of match" point. This has been
1906 changed to the "earliest inspected character" point, because the returned
1907 data for a partial match starts at this character. This means that, for
1908 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1909 (previously it gave "no match").
1910
1911 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1912 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1913 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1914 has an implication that the given string is incomplete (because a partial
1915 match is preferred over a full match). For this reason, these items now
1916 give a partial match in this situation. [Aside: previously, the one case
1917 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1918 match rather than a full match, which was wrong by the old rules, but is
1919 now correct.]
1920
1921 6. There was a bug in the handling of #-introduced comments, recognized when
1922 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1923 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1924 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1925 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1926 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1927 places in pcre_compile().
1928
1929 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1930 comments when looking ahead for named forward references to subpatterns,
1931 the only newline sequence it recognized was NL. It now handles newlines
1932 according to the set newline convention.
1933
1934 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1935 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1936 cater for a lack of strerror(). These oversights have been fixed.
1937
1938 9. Added --match-limit and --recursion-limit to pcregrep.
1939
1940 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1941
1942 11. When the -o option was used, pcregrep was setting a return code of 1, even
1943 when matches were found, and --line-buffered was not being honoured.
1944
1945 12. Added an optional parentheses number to the -o and --only-matching options
1946 of pcregrep.
1947
1948 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1949 can match an empty string. The code to do it in pcretest and pcredemo
1950 needed fixing:
1951
1952 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1953 only one byte after an empty string match just before CRLF (this case
1954 just got forgotten; "any" and "anycrlf" were OK).
1955
1956 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1957 mode when an empty string match preceded an ASCII character followed by
1958 a non-ASCII character. (The code for advancing by one character rather
1959 than one byte was nonsense.)
1960
1961 (c) The pcredemo.c sample program did not have any code at all to handle
1962 the cases when CRLF is a valid newline sequence.
1963
1964 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1965 as a starting offset was within the subject string. There is now a new
1966 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1967 negative or greater than the length of the string. In order to test this,
1968 pcretest is extended to allow the setting of negative starting offsets.
1969
1970 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1971 starting offset points to the beginning of a UTF-8 character was
1972 unnecessarily clumsy. I tidied it up.
1973
1974 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1975 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1976
1977 17. Nobody had reported that the --include_dir option, which was added in
1978 release 7.7 should have been called --include-dir (hyphen, not underscore)
1979 for compatibility with GNU grep. I have changed it to --include-dir, but
1980 left --include_dir as an undocumented synonym, and the same for
1981 --exclude-dir, though that is not available in GNU grep, at least as of
1982 release 2.5.4.
1983
1984 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1985 characters from a string of bytes) have been redefined so as not to use
1986 loops, in order to improve performance in some environments. At the same
1987 time, I abstracted some of the common code into auxiliary macros to save
1988 repetition (this should not affect the compiled code).
1989
1990 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1991 compile-time error is now given if \c is not followed by an ASCII
1992 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1993 different, and any byte value is allowed.)
1994
1995 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1996 START_OPTIMIZE option, which is now allowed at compile time - but just
1997 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1998 to pcregrep and other applications that have no direct access to PCRE
1999 options. The new /Y option in pcretest sets this option when calling
2000 pcre_compile().
2001
2002 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2003 back references. Groups containing recursive back references were forced to
2004 be atomic by that change, but in the case of named groups, the amount of
2005 memory required was incorrectly computed, leading to "Failed: internal
2006 error: code overflow". This has been fixed.
2007
2008 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2009 pcretest.c, to avoid build problems in some Borland environments.
2010
2011
2012 Version 8.10 25-Jun-2010
2013 ------------------------
2014
2015 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2016 THEN.
2017
2018 2. (*ACCEPT) was not working when inside an atomic group.
2019
2020 3. Inside a character class, \B is treated as a literal by default, but
2021 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2022 causes the error). The code is unchanged, but I tidied the documentation.
2023
2024 4. Inside a character class, PCRE always treated \R and \X as literals,
2025 whereas Perl faults them if its -w option is set. I have changed PCRE so
2026 that it faults them when PCRE_EXTRA is set.
2027
2028 5. Added support for \N, which always matches any character other than
2029 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2030
2031 6. When compiling pcregrep with newer versions of gcc which may have
2032 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2033 declared with attribute warn_unused_result" were given. Just casting the
2034 result to (void) does not stop the warnings; a more elaborate fudge is
2035 needed. I've used a macro to implement this.
2036
2037 7. Minor change to pcretest.c to avoid a compiler warning.
2038
2039 8. Added four artifical Unicode properties to help with an option to make
2040 \s etc use properties (see next item). The new properties are: Xan
2041 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2042
2043 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2044 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2045 this option. Modified pcretest to add /W to test this facility. Added
2046 REG_UCP to make it available via the POSIX interface.
2047
2048 10. Added --line-buffered to pcregrep.
2049
2050 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2051 studied, and the match started with a letter with a code point greater than
2052 127 whose first byte was different to the first byte of the other case of
2053 the letter, the other case of this starting letter was not recognized
2054 (#976).
2055
2056 12. If a pattern that was studied started with a repeated Unicode property
2057 test, for example, \p{Nd}+, there was the theoretical possibility of
2058 setting up an incorrect bitmap of starting bytes, but fortunately it could
2059 not have actually happened in practice until change 8 above was made (it
2060 added property types that matched character-matching opcodes).
2061
2062 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2063 possible starting bytes for non-anchored patterns.
2064
2065 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2066 \R, and also a number of cases that involve Unicode properties, both
2067 explicit and implicit when PCRE_UCP is set.
2068
2069 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2070 input, it could crash or give wrong results if characters with values
2071 greater than 0xc0 were present in the subject string. (Detail: it assumed
2072 UTF-8 input when processing these items.)
2073
2074 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2075 size_t is 64-bit (#991).
2076
2077 17. Added a check for running out of memory when PCRE is compiled with
2078 --disable-stack-for-recursion (#990).
2079
2080 18. If the last data line in a file for pcretest does not have a newline on
2081 the end, a newline was missing in the output.
2082
2083 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2084 less than 128) in its various bitmaps. However, there is a facility for
2085 generating tables according to the current locale when PCRE is compiled. It
2086 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2087 characters, are recognized by isspace() and therefore were getting set in
2088 these tables, and indeed these tables seem to approximate to ISO 8859. This
2089 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2090 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2091 which of course cannot start UTF-8 characters. I have changed the code so
2092 that only real ASCII characters (less than 128) and the correct starting
2093 bytes for UTF-8 encodings are set for characters greater than 127 when in
2094 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2095 altogether.)
2096
2097 20. Added the /T option to pcretest so as to be able to run tests with non-
2098 standard character tables, thus making it possible to include the tests
2099 used for 19 above in the standard set of tests.
2100
2101 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2102 reference to a subpattern the other side of a comment that contains an
2103 opening parenthesis caused either an internal compiling error, or a
2104 reference to the wrong subpattern.
2105
2106
2107 Version 8.02 19-Mar-2010
2108 ------------------------
2109
2110 1. The Unicode data tables have been updated to Unicode 5.2.0.
2111
2112 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2113 configured.
2114
2115 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2116 original author of that file, following a query about its status.
2117
2118 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2119 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2120
2121 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2122 quantifier applied to a forward-referencing subroutine call, could compile
2123 incorrect code or give the error "internal error: previously-checked
2124 referenced subpattern not found".
2125
2126 6. Both MS Visual Studio and Symbian OS have problems with initializing
2127 variables to point to external functions. For these systems, therefore,
2128 pcre_malloc etc. are now initialized to local functions that call the
2129 relevant global functions.
2130
2131 7. There were two entries missing in the vectors called coptable and poptable
2132 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2133 I've fixed the data, and added a kludgy way of testing at compile time that
2134 the lengths are correct (equal to the number of opcodes).
2135
2136 8. Following on from 7, I added a similar kludge to check the length of the
2137 eint vector in pcreposix.c.
2138
2139 9. Error texts for pcre_compile() are held as one long string to avoid too
2140 much relocation at load time. To find a text, the string is searched,
2141 counting zeros. There was no check for running off the end of the string,
2142 which could happen if a new error number was added without updating the
2143 string.
2144
2145 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2146
2147 11. \K was not working if it appeared in an atomic group or in a group that
2148 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2149 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2150 the assertion is positive, but not if it is negative.
2151
2152 12. Change 11 fortuitously reduced the size of the stack frame used in the
2153 "match()" function of pcre_exec.c by one pointer. Forthcoming
2154 implementation of support for (*MARK) will need an extra pointer on the
2155 stack; I have reserved it now, so that the stack frame size does not
2156 decrease.
2157
2158 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2159 item in branch that calls a recursion is a subroutine call - as in the
2160 second branch in the above example - was incorrectly given the compile-
2161 time error "recursive call could loop indefinitely" because pcre_compile()
2162 was not correctly checking the subroutine for matching a non-empty string.
2163
2164 14. The checks for overrunning compiling workspace could trigger after an
2165 overrun had occurred. This is a "should never occur" error, but it can be
2166 triggered by pathological patterns such as hundreds of nested parentheses.
2167 The checks now trigger 100 bytes before the end of the workspace.
2168
2169 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2170
2171
2172 Version 8.01 19-Jan-2010
2173 ------------------------
2174
2175 1. If a pattern contained a conditional subpattern with only one branch (in
2176 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2177 computed the wrong minimum data length (which is of course zero for such
2178 subpatterns). This could cause incorrect "no match" results.
2179
2180 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2181 the pattern is reset in the first branch, pcre_compile() failed with
2182 "internal error: code overflow at offset...". This happened only when
2183 the reset was to the original external option setting. (An optimization
2184 abstracts leading options settings into an external setting, which was the
2185 cause of this.)
2186
2187 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2188 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2189 assertion pattern did not match (meaning that the assertion was true), it
2190 was incorrectly treated as false if the SKIP had been reached during the
2191 matching. This also applied to assertions used as conditions.
2192
2193 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2194 assertion subpattern, including such a pattern used as a condition,
2195 unpredictable results occurred, instead of the error return
2196 PCRE_ERROR_DFA_UITEM.
2197
2198 5. The C++ GlobalReplace function was not working like Perl for the special
2199 situation when an empty string is matched. It now does the fancy magic
2200 stuff that is necessary.
2201
2202 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2203 removed. (These were left over from very, very early versions of PCRE.)
2204
2205 7. Some cosmetic changes to the code to make life easier when compiling it
2206 as part of something else:
2207
2208 (a) Change DEBUG to PCRE_DEBUG.
2209
2210 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2211 called "current" as "current_branch", to prevent a collision with the
2212 Linux macro when compiled as a kernel module.
2213
2214 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2215 prevent a collision with the Linux macro when compiled as a kernel
2216 module.
2217
2218 8. In pcre_compile() there are some checks for integer overflows that used to
2219 cast potentially large values to (double). This has been changed to that
2220 when building, a check for int64_t is made, and if it is found, it is used
2221 instead, thus avoiding the use of floating point arithmetic. (There is no
2222 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2223 double.
2224
2225 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2226 2005 (difference between two addresses compared to an unsigned value).
2227
2228 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2229 custom one, because of the following reported problem in Windows:
2230
2231 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2232 under Win32.
2233 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2234 therefore missing the function definition.
2235 - The compiler thus generates a "C" signature for the test function.
2236 - The linker fails to find the "C" function.
2237 - PCRE fails to configure if asked to do so against libbz2.
2238
2239 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2240 messages were output:
2241
2242 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2243 rerunning libtoolize, to keep the correct libtool macros in-tree.
2244 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2245
2246 I have done both of these things.
2247
2248 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2249 most of the time, it *can* run out if it is given a pattern that contains a
2250 runaway infinite recursion. I updated the discussion in the pcrestack man
2251 page.
2252
2253 13. Now that we have gone to the x.xx style of version numbers, the minor
2254 version may start with zero. Using 08 or 09 is a bad idea because users
2255 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2256 interpreted as invalid octal numbers. I've updated the previous comment in
2257 configure.ac, and also added a check that gives an error if 08 or 09 are
2258 used.
2259
2260 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2261 causing partial matching to fail when the end of the subject matched \W
2262 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2263
2264 15. There were some discrepancies between the declarations in pcre_internal.h
2265 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2266 their definitions. The declarations used "const uschar *" and the
2267 definitions used USPTR. Even though USPTR is normally defined as "const
2268 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2269 reported that: "This difference in casting confuses some C++ compilers, for
2270 example, SunCC recognizes above declarations as different functions and
2271 generates broken code for hbpcre." I have changed the declarations to use
2272 USPTR.
2273
2274 16. GNU libtool is named differently on some systems. The autogen.sh script now
2275 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2276 (FreeBSD).
2277
2278 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2279 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2280 comment: "Figure out how to create a longlong from a string: strtoll and
2281 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2282 instance, but it only takes 2 args instead of 3!"
2283
2284 18. A subtle bug concerned with back references has been fixed by a change of
2285 specification, with a corresponding code fix. A pattern such as
2286 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2287 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2288 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2289 same bug. Such groups have to be quantified to be useful, or contained
2290 inside another quantified group. (If there's no repetition, the reference
2291 can never match.) The problem arises because, having left the group and
2292 moved on to the rest of the pattern, a later failure that backtracks into
2293 the group uses the captured value from the final iteration of the group
2294 rather than the correct earlier one. I have fixed this in PCRE by forcing
2295 any group that contains a reference to itself to be an atomic group; that
2296 is, there cannot be any backtracking into it once it has completed. This is
2297 similar to recursive and subroutine calls.
2298
2299
2300 Version 8.00 19-Oct-09
2301 ----------------------
2302
2303 1. The table for translating pcre_compile() error codes into POSIX error codes
2304 was out-of-date, and there was no check on the pcre_compile() error code
2305 being within the table. This could lead to an OK return being given in
2306 error.
2307
2308 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2309 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2310 in a Windows environment.
2311
2312 3. The pcregrep --count option prints the count for each file even when it is
2313 zero, as does GNU grep. However, pcregrep was also printing all files when
2314 --files-with-matches was added. Now, when both options are given, it prints
2315 counts only for those files that have at least one match. (GNU grep just
2316 prints the file name in this circumstance, but including the count seems
2317 more useful - otherwise, why use --count?) Also ensured that the
2318 combination -clh just lists non-zero counts, with no names.
2319
2320 4. The long form of the pcregrep -F option was incorrectly implemented as
2321 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2322 but it seems right to fix it, and I didn't think it was worth preserving
2323 the old behaviour.
2324
2325 5. The command line items --regex=pattern and --regexp=pattern were not
2326 recognized by pcregrep, which required --regex pattern or --regexp pattern
2327 (with a space rather than an '='). The man page documented the '=' forms,
2328 which are compatible with GNU grep; these now work.
2329
2330 6. No libpcreposix.pc file was created for pkg-config; there was just
2331 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2332
2333 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2334 when UCP support is not needed, by modifying the Python script that
2335 generates it from Unicode data files. This should not matter if the module
2336 is correctly used as a library, but I received one complaint about 50K of
2337 unwanted data. My guess is that the person linked everything into his
2338 program rather than using a library. Anyway, it does no harm.
2339
2340 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2341 was a minimum greater than 1 for a wide character in a possessive
2342 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2343 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2344 character. Chaos in the form of incorrect output or a compiling loop could
2345 result.
2346
2347 9. The restrictions on what a pattern can contain when partial matching is
2348 requested for pcre_exec() have been removed. All patterns can now be
2349 partially matched by this function. In addition, if there are at least two
2350 slots in the offset vector, the offset of the earliest inspected character
2351 for the match and the offset of the end of the subject are set in them when
2352 PCRE_ERROR_PARTIAL is returned.
2353
2354 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2355 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2356 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2357 and may be more useful for multi-segment matching.
2358
2359 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2360 used to be given if ever the end of the subject was reached; now it is
2361 given only if matching could not proceed because another character was
2362 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2363 string "Z", which now yields "no match" instead of "partial match". In the
2364 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2365 final character ended with (*FAIL).
2366
2367 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2368 if the pattern had a "must contain" character that was already found in the
2369 earlier partial match, unless partial matching was again requested. For
2370 example, with the pattern /dog.(body)?/, the "must contain" character is
2371 "g". If the first part-match was for the string "dog", restarting with
2372 "sbody" failed. This bug has been fixed.
2373
2374 13. The string returned by pcre_dfa_exec() after a partial match has been
2375 changed so that it starts at the first inspected character rather than the
2376 first character of the match. This makes a difference only if the pattern
2377 starts with a lookbehind assertion or \b or \B (\K is not supported by
2378 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2379 matching functions compatible, and I think it's the right thing to do.
2380
2381 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2382 so that the demonstration program is easily available in environments where
2383 PCRE has not been installed from source.
2384
2385 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2386 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2387 library.
2388
2389 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2390 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2391 is not the first non-POSIX option to be added. Clearly some people find
2392 these options useful.
2393
2394 17. If a caller to the POSIX matching function regexec() passes a non-zero
2395 value for nmatch with a NULL value for pmatch, the value of
2396 nmatch is forced to zero.
2397
2398 18. RunGrepTest did not have a test for the availability of the -u option of
2399 the diff command, as RunTest does. It now checks in the same way as
2400 RunTest, and also checks for the -b option.
2401
2402 19. If an odd number of negated classes containing just a single character
2403 interposed, within parentheses, between a forward reference to a named
2404 subpattern and the definition of the subpattern, compilation crashed with
2405 an internal error, complaining that it could not find the referenced
2406 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2407 [The bug was that it was starting one character too far in when skipping
2408 over the character class, thus treating the ] as data rather than
2409 terminating the class. This meant it could skip too much.]
2410
2411 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2412 /g option in pcretest when the pattern contains \K, which makes it possible
2413 to have an empty string match not at the start, even when the pattern is
2414 anchored. Updated pcretest and pcredemo to use this option.
2415
2416 21. If the maximum number of capturing subpatterns in a recursion was greater
2417 than the maximum at the outer level, the higher number was returned, but
2418 with unset values at the outer level. The correct (outer level) value is
2419 now given.
2420
2421 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2422 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2423 make it do so. The string so far is captured, making this feature
2424 compatible with Perl.
2425
2426 23. The tests have been re-organized, adding tests 11 and 12, to make it
2427 possible to check the Perl 5.10 features against Perl 5.10.
2428
2429 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2430 pattern matches a fixed length string. PCRE did not allow this; now it
2431 does. Neither allows recursion.
2432
2433 25. I finally figured out how to implement a request to provide the minimum
2434 length of subject string that was needed in order to match a given pattern.
2435 (It was back references and recursion that I had previously got hung up
2436 on.) This code has now been added to pcre_study(); it finds a lower bound
2437 to the length of subject needed. It is not necessarily the greatest lower
2438 bound, but using it to avoid searching strings that are too short does give
2439 some useful speed-ups. The value is available to calling programs via
2440 pcre_fullinfo().
2441
2442 26. While implementing 25, I discovered to my embarrassment that pcretest had
2443 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2444 study optimizations had never been tested with that matching function.
2445 Oops. What is worse, even when it was passed study data, there was a bug in
2446 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2447 were also very few tests of studied patterns with pcre_dfa_exec().
2448
2449 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2450 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2451 on the other side of the coin, they are no longer allowed to have different
2452 names, because these cannot be distinguished in PCRE, and this has caused
2453 confusion. (This is a difference from Perl.)
2454
2455 28. When duplicate subpattern names are present (necessarily with different
2456 numbers, as required by 27 above), and a test is made by name in a
2457 conditional pattern, either for a subpattern having been matched, or for
2458 recursion in such a pattern, all the associated numbered subpatterns are
2459 tested, and the overall condition is true if the condition is true for any
2460 one of them. This is the way Perl works, and is also more like the way
2461 testing by number works.
2462
2463
2464 Version 7.9 11-Apr-09
2465 ---------------------
2466
2467 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2468 (pcretest), all targets were linked against these libraries. This included
2469 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2470 libraries. This caused unwanted dependencies to be created. This problem
2471 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2472 pcretest is linked with readline.
2473
2474 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2475 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2476 moved outside it again, because FALSE and TRUE are already defined in AIX,
2477 but BOOL is not.
2478
2479 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2480 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2481
2482 4. The pcregrep documentation said spaces were inserted as well as colons (or
2483 hyphens) following file names and line numbers when outputting matching
2484 lines. This is not true; no spaces are inserted. I have also clarified the
2485 wording for the --colour (or --color) option.
2486
2487 5. In pcregrep, when --colour was used with -o, the list of matching strings
2488 was not coloured; this is different to GNU grep, so I have changed it to be
2489 the same.
2490
2491 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2492 each matching line was coloured. Now it goes on to look for further matches
2493 of any of the test patterns, which is the same behaviour as GNU grep.
2494
2495 7. A pattern that could match an empty string could cause pcregrep to loop; it
2496 doesn't make sense to accept an empty string match in pcregrep, so I have
2497 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2498 seems to be how GNU grep behaves. [But see later change 40 for release
2499 8.33.]
2500
2501 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2502 start or after a newline", because the conditional assertion was not being
2503 correctly handled. The rule now is that both the assertion and what follows
2504 in the first alternative must satisfy the test.
2505
2506 9. If auto-callout was enabled in a pattern with a conditional group whose
2507 condition was an assertion, PCRE could crash during matching, both with
2508 pcre_exec() and pcre_dfa_exec().
2509
2510 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2511 used for matching.
2512
2513 11. Unicode property support in character classes was not working for
2514 characters (bytes) greater than 127 when not in UTF-8 mode.
2515
2516 12. Added the -M command line option to pcretest.
2517
2518 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2519
2520 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2521
2522 16. Added comments and documentation about mis-use of no_arg in the C++
2523 wrapper.
2524
2525 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2526 from Martin Jerabek that uses macro names for all relevant character and
2527 string constants.
2528
2529 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2530 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2531 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2532 these, but not everybody uses configure.
2533
2534 19. A conditional group that had only one branch was not being correctly
2535 recognized as an item that could match an empty string. This meant that an
2536 enclosing group might also not be so recognized, causing infinite looping
2537 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2538 with the subject "ab", where knowledge that the repeated group can match
2539 nothing is needed in order to break the loop.
2540
2541 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2542 exec(), but without supplying a callout function, matching went wrong.
2543
2544 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2545 leak if the size of the offset vector was greater than 30. When the vector
2546 is smaller, the saved offsets during recursion go onto a local stack
2547 vector, but for larger vectors malloc() is used. It was failing to free
2548 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2549 error, in fact).
2550
2551 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2552 heapframe that is used only when UTF-8 support is enabled. This caused no
2553 problem, but was untidy.
2554
2555 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2556 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2557 included within another project.
2558
2559 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2560 slightly modified by me:
2561
2562 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2563 not building pcregrep.
2564
2565 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2566 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2567
2568 25. Forward references, both numeric and by name, in patterns that made use of
2569 duplicate group numbers, could behave incorrectly or give incorrect errors,
2570 because when scanning forward to find the reference group, PCRE was not
2571 taking into account the duplicate group numbers. A pattern such as
2572 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2573
2574 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2575 the feature of a custom pointer more persuasive (as requested by a user).
2576
2577 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2578 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2579 pre-defined.
2580
2581 28. Added support for (*UTF8) at the start of a pattern.
2582
2583 29. Arrange for flags added by the "release type" setting in CMake to be shown
2584 in the configuration summary.
2585
2586
2587 Version 7.8 05-Sep-08
2588 ---------------------
2589
2590 1. Replaced UCP searching code with optimized version as implemented for Ad
2591 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2592 stage table and inline lookup instead of a function, giving speed ups of 2
2593 to 5 times on some simple patterns that I tested. Permission was given to
2594 distribute the MultiStage2.py script that generates the tables (it's not in
2595 the tarball, but is in the Subversion repository).
2596
2597 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2598 scripts.
2599
2600 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2601 a group with a zero qualifier. The result of the study could be incorrect,
2602 or the function might crash, depending on the pattern.
2603
2604 4. Caseless matching was not working for non-ASCII characters in back
2605 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2606 It now works when Unicode Property Support is available.
2607
2608 5. In pcretest, an escape such as \x{de} in the data was always generating
2609 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2610 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2611 truncation.
2612
2613 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2614
2615 7. Added two (int) casts to pcregrep when printing the difference of two
2616 pointers, in case they are 64-bit values.
2617
2618 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2619 test 2 if it fails.
2620
2621 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2622 and a #define of that name to empty if it is not externally set. This is to
2623 allow users of MSVC to set it if necessary.
2624
2625 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2626 the convenience functions in the pcre_get.c source file.
2627
2628 11. An option change at the start of a pattern that had top-level alternatives
2629 could cause overwriting and/or a crash. This command provoked a crash in
2630 some environments:
2631
2632 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2633
2634 This potential security problem was recorded as CVE-2008-2371.
2635
2636 12. For a pattern where the match had to start at the beginning or immediately
2637 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2638 pcre_dfa_exec() could read past the end of the passed subject if there was
2639 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2640 pcretest so that it places the subject at the end of its malloc-ed buffer.
2641
2642 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2643 exec() might read past the end of the data buffer in UTF-8 mode.
2644
2645 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2646 the data contained the byte 0x85 as part of a UTF-8 character within its
2647 first line. This applied both to normal and DFA matching.
2648
2649 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2650 /^[^d]*?$/8 failed to match "abc".
2651
2652 16. Added a missing copyright notice to pcrecpp_internal.h.
2653
2654 17. Make it more clear in the documentation that values returned from
2655 pcre_exec() in ovector are byte offsets, not character counts.
2656
2657 18. Tidied a few places to stop certain compilers from issuing warnings.
2658
2659 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2660 supplied by Stefan Weber. I made a further small update for 7.8 because
2661 there is a change of source arrangements: the pcre_searchfuncs.c module is
2662 replaced by pcre_ucd.c.
2663
2664
2665 Version 7.7 07-May-08
2666 ---------------------
2667
2668 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2669 a string to a long long, pretend we don't even have a long long." This is
2670 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2671
2672 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2673 pre-7.6 versions, which defined a global no_arg variable instead of putting
2674 it in the RE class. (See also #8 below.)
2675
2676 3. Remove a line of dead code, identified by coverity and reported by Nuno
2677 Lopes.
2678
2679 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2680
2681 (1) The include/exclude patterns were being applied to the whole pathnames
2682 of files, instead of just to the final components.
2683
2684 (2) If there was more than one level of directory, the subdirectories were
2685 skipped unless they satisfied the include/exclude conditions. This is
2686 inconsistent with GNU grep (and could even be seen as contrary to the
2687 pcregrep specification - which I improved to make it absolutely clear).
2688 The action now is always to scan all levels of directory, and just
2689 apply the include/exclude patterns to regular files.
2690
2691 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2692 --exclude_dir in the tests to avoid scanning .svn directories.
2693
2694 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2695 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2696 doesn't support NULs in patterns.
2697
2698 7. Added some missing "const"s to declarations of static tables in
2699 pcre_compile.c and pcre_dfa_exec.c.
2700
2701 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2702 caused by fix #2 above. (Subsequently also a second patch to fix the
2703 first patch. And a third patch - this was a messy problem.)
2704
2705 9. Applied Craig's patch to remove the use of push_back().
2706
2707 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2708 matching function regexec().
2709
2710 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2711 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2712 references. PCRE supports relative numbers with this syntax (I don't think
2713 Oniguruma does).
2714
2715 12. Previously, a group with a zero repeat such as (...){0} was completely
2716 omitted from the compiled regex. However, this means that if the group
2717 was called as a subroutine from elsewhere in the pattern, things went wrong
2718 (an internal error was given). Such groups are now left in the compiled
2719 pattern, with a new opcode that causes them to be skipped at execution
2720 time.
2721
2722 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2723 to the way PCRE behaves:
2724
2725 (a) A lone ] character is dis-allowed (Perl treats it as data).
2726
2727 (b) A back reference to an unmatched subpattern matches an empty string
2728 (Perl fails the current match path).
2729
2730 (c) A data ] in a character class must be notated as \] because if the
2731 first data character in a class is ], it defines an empty class. (In
2732 Perl it is not possible to have an empty class.) The empty class []
2733 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2734 The negative empty class [^] matches any one character, independently
2735 of the DOTALL setting.
2736
2737 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2738 non-existent subpattern following a character class starting with ']' and
2739 containing () gave an internal compiling error instead of "reference to
2740 non-existent subpattern". Fortunately, when the pattern did exist, the
2741 compiled code was correct. (When scanning forwards to check for the
2742 existence of the subpattern, it was treating the data ']' as terminating
2743 the class, so got the count wrong. When actually compiling, the reference
2744 was subsequently set up correctly.)
2745
2746 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2747 it was being rejected as not supported by pcre_dfa_exec(), even though
2748 other assertions are supported. I have made pcre_dfa_exec() support
2749 (*FAIL).
2750
2751 16. The implementation of 13c above involved the invention of a new opcode,
2752 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2753 cannot be changed at match time, I realized I could make a small
2754 improvement to matching performance by compiling OP_ALLANY instead of
2755 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2756 on the OP_ANY path.
2757
2758 17. Compiling pcretest on Windows with readline support failed without the
2759 following two fixes: (1) Make the unistd.h include conditional on
2760 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2761
2762 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2763 ncurses library to be included for pcretest when ReadLine support is
2764 requested, but also to allow for it to be overridden. This patch came from
2765 Daniel Bergström.
2766
2767 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2768 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2769 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2770 spotting this.
2771
2772
2773 Version 7.6 28-Jan-08
2774 ---------------------
2775
2776 1. A character class containing a very large number of characters with
2777 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2778 overflow.
2779
2780 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2781 HAVE_LONG_LONG is not defined.
2782
2783 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2784 bring them up to date and include new features. This patch includes:
2785
2786 - Fixed PH's badly added libz and libbz2 support.
2787 - Fixed a problem with static linking.
2788 - Added pcredemo. [But later removed - see 7 below.]
2789 - Fixed dftables problem and added an option.
2790 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2791 HAVE_LONG_LONG.
2792 - Added readline support for pcretest.
2793 - Added an listing of the option settings after cmake has run.
2794
2795 4. A user submitted a patch to Makefile that makes it easy to create
2796 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2797 Makefile.am that cause it to include this special target, without
2798 affecting anything else. Note that the same mingw target plus all
2799 the other distribution libraries and programs are now supported
2800 when configuring with CMake (see 6 below) instead of with
2801 Configure/Make.
2802
2803 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2804 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2805 exported in the Windows port". It has not yet been confirmed that the patch
2806 solves the problem, but it does no harm.
2807
2808 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2809 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2810 with CMake, and also correct the comment about stack recursion.
2811
2812 7. Remove the automatic building of pcredemo from the ./configure system and
2813 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2814 of a program that users should build themselves after PCRE is installed, so
2815 building it automatically is not really right. What is more, it gave
2816 trouble in some build environments.
2817
2818 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2819
2820
2821 Version 7.5 10-Jan-08
2822 ---------------------
2823
2824 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2825 values in parens when parsing an RE using the C++ wrapper."
2826
2827 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2828 Characters greater than 255 were excluded from the class instead of being
2829 included.
2830
2831 3. The same bug as (2) above applied to negated POSIX classes such as
2832 [:^space:].
2833
2834 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2835 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2836 I have changed it.
2837
2838 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2839 first named subpattern) and a construct such as (?&a) would reference the
2840 first named subpattern whose name started with "a" (in other words, the
2841 length check was missing). Both these problems are fixed. "Subpattern name
2842 expected" is now given for (?&) (a zero-length name), and this patch also
2843 makes it give the same error for \k'' (previously it complained that that
2844 was a reference to a non-existent subpattern).
2845
2846 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2847 this is right because (?- can be followed by option settings as well as by
2848 digits. I have, however, made the messages clearer.
2849
2850 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2851 than the number used in the conditional) now cause a compile-time error.
2852 This is actually not compatible with Perl, which accepts such patterns, but
2853 treats the conditional as always being FALSE (as PCRE used to), but it
2854 seems to me that giving a diagnostic is better.
2855
2856 8. Change "alphameric" to the more common word "alphanumeric" in comments
2857 and messages.
2858
2859 9. Fix two occurrences of "backslash" in comments that should have been
2860 "backspace".
2861
2862 10. Remove two redundant lines of code that can never be obeyed (their function
2863 was moved elsewhere).
2864
2865 11. The program that makes PCRE's Unicode character property table had a bug
2866 which caused it to generate incorrect table entries for sequences of
2867 characters that have the same character type, but are in different scripts.
2868 It amalgamated them into a single range, with the script of the first of
2869 them. In other words, some characters were in the wrong script. There were
2870 thirteen such cases, affecting characters in the following ranges:
2871
2872 U+002b0 - U+002c1
2873 U+0060c - U+0060d
2874 U+0061e - U+00612
2875 U+0064b - U+0065e
2876 U+0074d - U+0076d
2877 U+01800 - U+01805
2878 U+01d00 - U+01d77
2879 U+01d9b - U+01dbf
2880 U+0200b - U+0200f
2881 U+030fc - U+030fe
2882 U+03260 - U+0327f
2883 U+0fb46 - U+0fbb1
2884 U+10450 - U+1049d
2885
2886 12. The -o option (show only the matching part of a line) for pcregrep was not
2887 compatible with GNU grep in that, if there was more than one match in a
2888 line, it showed only the first of them. It now behaves in the same way as
2889 GNU grep.
2890
2891 13. If the -o and -v options were combined for pcregrep, it printed a blank
2892 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2893 does the same. The return code can be used to tell if there were any
2894 non-matching lines.
2895
2896 14. Added --file-offsets and --line-offsets to pcregrep.
2897
2898 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2899 infinitely looping recursion. The bug was that positive lookaheads were not
2900 being skipped when checking for a possible empty match (negative lookaheads
2901 and both kinds of lookbehind were skipped).
2902
2903 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2904 inclusion of <windows.h> to before rather than after the definition of
2905 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2906
2907 17. Specifying a possessive quantifier with a specific limit for a Unicode
2908 character property caused pcre_compile() to compile bad code, which led at
2909 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2910 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2911 caused the error; without that there was no problem.
2912
2913 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2914
2915 19. Added --enable-pcretest-libreadline.
2916
2917 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2918 RE::GlobalReplace(). As a result, the number of replacements returned was
2919 double what it should be. I removed one of the increments, but Craig sent a
2920 later patch that removed the other one (the right fix) and added unit tests
2921 that check the return values (which was not done before).
2922
2923 21. Several CMake things:
2924
2925 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2926 the names libpcre and libpcreposix, not just pcre and pcreposix.
2927
2928 (2) The above change means that pcretest and pcregrep are now correctly
2929 linked with the newly-built libraries, not previously installed ones.
2930
2931 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2932
2933 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2934 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2935 UTF-8 newline character). The key issue is that the pattern starts .*;
2936 this means that the match must be either at the beginning, or after a
2937 newline. The bug was in the code for advancing after a failed match and
2938 checking that the new position followed a newline. It was not taking
2939 account of UTF-8 characters correctly.
2940
2941 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2942 character classes. PCRE was not treating the sequence [:...:] as a
2943 character class unless the ... were all letters. Perl, however, seems to
2944 allow any characters between [: and :], though of course it rejects as
2945 unknown any "names" that contain non-letters, because all the known class
2946 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2947 for example, whereas PCRE did not - it did not recognize a POSIX character
2948 class. This seemed a bit dangerous, so the code has been changed to be
2949 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2950 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2951 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2952 Perl does, and where it didn't before.
2953
2954 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2955 Windows environments %n is disabled by default.
2956
2957
2958 Version 7.4 21-Sep-07
2959 ---------------------
2960
2961 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2962 means that a class such as [\s] counted as "explicit reference to CR or
2963 LF". That isn't really right - the whole point of the change was to try to
2964 help when there was an actual mention of one of the two characters. So now
2965 the change happens only if \r or \n (or a literal CR or LF) character is
2966 encountered.
2967
2968 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2969 of both had grown to the point where there were only 3 bits left.
2970 Fortunately, there was spare space in the data structure, and so I have
2971 moved the internal flags into a new 16-bit field to free up more option
2972 bits.
2973
2974 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2975 but did not set the internal JCHANGED flag - either of these is enough to
2976 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2977 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2978 start sets both bits.
2979
2980 4. Added options (at build time, compile time, exec time) to change \R from
2981 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2982
2983 5. doc/pcresyntax.html was missing from the distribution.
2984
2985 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2986 compatibility, even though it is no longer used.
2987
2988 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2989 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2990 windows.h file is present (where different names are used). [This was
2991 reversed later after testing - see 16 below.]
2992
2993 8. Changed all #include <config.h> to #include "config.h". There were also
2994 some further <pcre.h> cases that I changed to "pcre.h".
2995
2996 9. When pcregrep was used with the --colour option, it missed the line ending
2997 sequence off the lines that it output.
2998
2999 10. It was pointed out to me that arrays of string pointers cause lots of
3000 relocations when a shared library is dynamically loaded. A technique of
3001 using a single long string with a table of offsets can drastically reduce
3002 these. I have refactored PCRE in four places to do this. The result is
3003 dramatic:
3004
3005 Originally: 290
3006 After changing UCP table: 187
3007 After changing error message table: 43
3008 After changing table of "verbs" 36
3009 After changing table of Posix names 22
3010
3011 Thanks to the folks working on Gregex for glib for this insight.
3012
3013 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3014 unicode-properties was also set.
3015
3016 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3017
3018 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3019 checked only for CRLF.
3020
3021 14. Added casts to pcretest.c to avoid compiler warnings.
3022
3023 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3024
3025 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3026 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3027 entirely. This removes changes made in 7 above.
3028
3029 17. The CMake files have been updated, and there is now more information about
3030 building with CMake in the NON-UNIX-USE document.
3031
3032
3033 Version 7.3 28-Aug-07
3034 ---------------------
3035
3036 1. In the rejigging of the build system that eventually resulted in 7.1, the
3037 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3038 brackets there is not right, since it causes compilers to look for an
3039 installed pcre.h, not the version that is in the source that is being
3040 compiled (which of course may be different). I have changed it back to:
3041
3042 #include "pcre.h"
3043
3044 I have a vague recollection that the change was concerned with compiling in
3045 different directories, but in the new build system, that is taken care of
3046 by the VPATH setting the Makefile.
3047
3048 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3049 when the subject happened to end in the byte 0x85 (e.g. if the last
3050 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3051 characters but of course it shouldn't be taken as a newline when it is part
3052 of another character. The bug was that, for an unlimited repeat of . in
3053 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3054 characters when looking for a newline.
3055
3056 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3057
3058 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3059 in debug output.
3060
3061 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3062 long printing in the pcrecpp unittest when running under MinGW.
3063
3064 6. ESC_K was left out of the EBCDIC table.
3065
3066 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3067 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3068 limit also applies to "virtual nesting" when a pattern is recursive, and in
3069 this case 1000 isn't so big. I have been able to remove this limit at the
3070 expense of backing off one optimization in certain circumstances. Normally,
3071 when pcre_exec() would call its internal match() function recursively and
3072 immediately return the result unconditionally, it uses a "tail recursion"
3073 feature to save stack. However, when a subpattern that can match an empty
3074 string has an unlimited repetition quantifier, it no longer makes this
3075 optimization. That gives it a stack frame in which to save the data for
3076 checking that an empty string has been matched. Previously this was taken
3077 from the 1000-entry workspace that had been reserved. So now there is no
3078 explicit limit, but more stack is used.
3079
3080 8. Applied Daniel's patches to solve problems with the import/export magic
3081 syntax that is required for Windows, and which was going wrong for the
3082 pcreposix and pcrecpp parts of the library. These were overlooked when this
3083 problem was solved for the main library.
3084
3085 9. There were some crude static tests to avoid integer overflow when computing
3086 the size of patterns that contain repeated groups with explicit upper
3087 limits. As the maximum quantifier is 65535, the maximum group length was
3088 set at 30,000 so that the product of these two numbers did not overflow a
3089 32-bit integer. However, it turns out that people want to use groups that
3090 are longer than 30,000 bytes (though not repeat them that many times).
3091 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3092 made it possible to implement the integer overflow checks in a much more
3093 dynamic way, which I have now done. The artificial limitation on group
3094 length has been removed - we now have only the limit on the total length of
3095 the compiled pattern, which depends on the LINK_SIZE setting.
3096
3097 10. Fixed a bug in the documentation for get/copy named substring when
3098 duplicate names are permitted. If none of the named substrings are set, the
3099 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3100 empty string.
3101
3102 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3103 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3104 because the ] is interpreted as the first data character and the
3105 terminating ] is not found. PCRE has been made compatible with Perl in this
3106 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3107 cause memory overwriting.
3108
3109 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3110 string has been matched (to stop an infinite loop). It was not recognizing
3111 a conditional subpattern that could match an empty string if that
3112 subpattern was within another subpattern. For example, it looped when
3113 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3114 condition was not nested. This bug has been fixed.
3115
3116 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3117 past the start of the subject in the presence of bytes with the top bit
3118 set, for example "\x8aBCD".
3119
3120 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3121 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3122
3123 14. Optimized (?!) to (*FAIL).
3124
3125 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3126 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3127 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3128 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3129 does: it's just the validity check that is more restrictive.
3130
3131 16. Inserted checks for integer overflows during escape sequence (backslash)
3132 processing, and also fixed erroneous offset values for syntax errors during
3133 backslash processing.
3134
3135 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3136 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3137
3138 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3139 caused an overrun.
3140
3141 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3142 something other than just ASCII characters) inside a group that had an
3143 unlimited repeat caused a loop at compile time (while checking to see
3144 whether the group could match an empty string).
3145
3146 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3147 [\P{Any}] did so. (Error in the code for printing property names.)
3148
3149 21. An orphan \E inside a character class could cause a crash.
3150
3151 22. A repeated capturing bracket such as (A)? could cause a wild memory
3152 reference during compilation.
3153
3154 23. There are several functions in pcre_compile() that scan along a compiled
3155 expression for various reasons (e.g. to see if it's fixed length for look
3156 behind). There were bugs in these functions when a repeated \p or \P was
3157 present in the pattern. These operators have additional parameters compared
3158 with \d, etc, and these were not being taken into account when moving along
3159 the compiled data. Specifically:
3160
3161 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3162 length.
3163
3164 (b) An item such as \pL+ within a repeated group could cause crashes or
3165 loops.
3166
3167 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3168 "reference to non-existent subpattern" error.
3169
3170 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3171
3172 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3173 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3174
3175 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3176
3177 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3178 character were causing crashes (broken optimization).
3179
3180 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3181 \p or \P) caused a compile-time loop.
3182
3183 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3184 break. For example, the unstudied pattern [\r\n]A does not match the string
3185 "\r\nA" because change 7.0/46 below moves the current point on by two
3186 characters after failing to match at the start. However, the pattern \nA
3187 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3188 the same is true. There doesn't seem any very clean way out of this, but
3189 what I have chosen to do makes the common cases work: PCRE now takes note
3190 of whether there can be an explicit match for \r or \n anywhere in the
3191 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3192 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3193 pattern has explicit CR or LF references.
3194
3195 29. Added (*CR) etc for changing newline setting at start of pattern.
3196
3197
3198 Version 7.2 19-Jun-07
3199 ---------------------
3200
3201 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3202 which is apparently normally available under Windows.
3203
3204 2. Re-jig the pcregrep tests with different newline settings in an attempt
3205 to make them independent of the local environment's newline setting.
3206
3207 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3208
3209 4. Some of the "internals" tests were previously cut out when the link size
3210 was not 2, because the output contained actual offsets. The recent new
3211 "Z" feature of pcretest means that these can be cut out, making the tests
3212 usable with all link sizes.
3213
3214 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3215 stack recursion. This gives a massive performance boost under BSD, but just
3216 a small improvement under Linux. However, it saves one field in the frame
3217 in all cases.
3218
3219 6. Added more features from the forthcoming Perl 5.10:
3220
3221 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3222 recursion call. It refers to the nth most recently opened parentheses.
3223
3224 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3225 to be opened parentheses.
3226
3227 (c) Conditions that refer to capturing parentheses can be specified
3228 relatively, for example, (?(-2)... or (?(+3)...
3229
3230 (d) \K resets the start of the current match so that everything before
3231 is not part of it.
3232
3233 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3234
3235 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3236 reference syntax.
3237
3238 (g) (?| introduces a group in which the numbering of parentheses in each
3239 alternative starts with the same number.
3240
3241 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3242
3243 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3244 PCRE_INFO_JCHANGED.
3245
3246 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3247 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3248 for detecting groups that can match an empty string.
3249
3250 9. A pattern with a very large number of alternatives (more than several
3251 hundred) was running out of internal workspace during the pre-compile
3252 phase, where pcre_compile() figures out how much memory will be needed. A
3253 bit of new cunning has reduced the workspace needed for groups with
3254 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3255 workspace instead of running out of the 4096 that are available.
3256
3257 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3258
3259 11. Applied patch from Google to remove an optimization that didn't quite work.
3260 The report of the bug said:
3261
3262 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3263 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3264 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3265
3266 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3267 it matched the wrong number of bytes.
3268
3269
3270 Version 7.1 24-Apr-07
3271 ---------------------
3272
3273 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3274 that is more "standard", making use of automake and other Autotools. There
3275 is some re-arrangement of the files and adjustment of comments consequent
3276 on this.
3277
3278 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3279 for recursive directory scanning broke on some systems because the files
3280 are not scanned in any specific order and on different systems the order
3281 was different. A call to "sort" has been inserted into RunGrepTest for the
3282 approprate test as a short-term fix. In the longer term there may be an
3283 alternative.
3284
3285 3. I had an email from Eric Raymond about problems translating some of PCRE's
3286 man pages to HTML (despite the fact that I distribute HTML pages, some
3287 people do their own conversions for various reasons). The problems
3288 concerned the use of low-level troff macros .br and .in. I have therefore
3289 removed all such uses from the man pages (some were redundant, some could
3290 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3291 HTML has been updated to handle .nf/.fi and to complain if it encounters
3292 .br or .in.
3293
3294 4. Updated comments in configure.ac that get placed in config.h.in and also
3295 arranged for config.h to be included in the distribution, with the name
3296 config.h.generic, for the benefit of those who have to compile without
3297 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3298
3299 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3300 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3301 makevp.bat for the current PCRE, using the additional files
3302 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3303
3304 6. A Windows user reported a minor discrepancy with test 2, which turned out
3305 to be caused by a trailing space on an input line that had got lost in his
3306 copy. The trailing space was an accident, so I've just removed it.
3307
3308 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3309 that is needed.
3310
3311 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3312 as "const" (a) because they are and (b) because it helps the PHP
3313 maintainers who have recently made a script to detect big data structures
3314 in the php code that should be moved to the .rodata section. I remembered
3315 to update Builducptable as well, so it won't revert if ucptable.h is ever
3316 re-created.
3317
3318 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3319 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3320 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3321 support is not required. This saves 1.5-2K of code, which is important in
3322 some applications.
3323
3324 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3325 so as not to refer to the tables, even though these functions will never be
3326 called when UTF-8 support is disabled. Otherwise there are problems with a
3327 shared library.
3328
3329 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3330
3331 (a) It was defining its arguments as char * instead of void *.
3332
3333 (b) It was assuming that all moves were upwards in memory; this was true
3334 a long time ago when I wrote it, but is no longer the case.
3335
3336 The emulated memove() is provided for those environments that have neither
3337 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3338 is clearly not the case, as these two bugs were recently reported.
3339
3340 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3341 and Detrail to create the HTML documentation, the .txt form of the man
3342 pages, and it removes trailing spaces from listed files. It also creates
3343 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3344 case, it wraps all the #defines with #ifndefs. This script should be run
3345 before "make dist".
3346
3347 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3348 with Unicode property support.
3349
3350 (a) For a maximizing quantifier, if the two different cases of the
3351 character were of different lengths in their UTF-8 codings (there are
3352 some cases like this - I found 11), and the matching function had to
3353 back up over a mixture of the two cases, it incorrectly assumed they
3354 were both the same length.
3355
3356 (b) When PCRE was configured to use the heap rather than the stack for
3357 recursion during matching, it was not correctly preserving the data for
3358 the other case of a UTF-8 character when checking ahead for a match
3359 while processing a minimizing repeat. If the check also involved
3360 matching a wide character, but failed, corruption could cause an
3361 erroneous result when trying to check for a repeat of the original
3362 character.
3363
3364 13. Some tidying changes to the testing mechanism:
3365
3366 (a) The RunTest script now detects the internal link size and whether there
3367 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3368 values substituted by "configure". (The RunGrepTest script already did
3369 this for UTF-8.) The configure.ac script no longer substitutes the
3370 relevant variables.
3371
3372 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3373 with length and offset values. This means that the output is different
3374 for different internal link sizes. Test 2 is skipped for link sizes
3375 other than 2 because of this, bypassing the problem. Unfortunately,
3376 there was also a test in test 3 (the locale tests) that used /B and
3377 failed for link sizes other than 2. Rather than cut the whole test out,
3378 I have added a new /Z option to pcretest that replaces the length and
3379 offset values with spaces. This is now used to make test 3 independent
3380 of link size. (Test 2 will be tidied up later.)
3381
3382 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3383 segmentation fault instead of returning the appropriate error message.
3384
3385 15. In multiline mode when the newline sequence was set to "any", the pattern
3386 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3387 This doesn't seem right; it now treats the CRLF combination as the line
3388 ending, and so does not match in that case. It's only a pattern such as ^$
3389 that would hit this one: something like ^ABC$ would have failed after \r
3390 and then tried again after \r\n.
3391
3392 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3393 in an attempt to make files that differ only in their line terminators
3394 compare equal. This works on Linux.
3395
3396 17. Under certain error circumstances pcregrep might try to free random memory
3397 as it exited. This is now fixed, thanks to valgrind.
3398
3399 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3400 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3401 was because its rules for how to advance for /g after matching an empty
3402 string at the end of a line did not allow for this case. They now check for
3403 it specially.
3404
3405 20. pcretest is supposed to handle patterns and data of any length, by
3406 extending its buffers when necessary. It was getting this wrong when the
3407 buffer for a data line had to be extended.
3408
3409 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3410 CRLF as a newline sequence.
3411
3412 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3413 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3414 I have nevertheless tidied it up.
3415
3416 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3417
3418 24. Added a man page for pcre-config.
3419
3420
3421 Version 7.0 19-Dec-06
3422 ---------------------
3423
3424 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3425 moving to gcc 4.1.1.
3426
3427 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3428 sys/time.h, which is documented as needed for this function. It doesn't
3429 seem to matter on Linux, but it showed up on some releases of OS X.
3430
3431 3. It seems that there are systems where bytes whose values are greater than
3432 127 match isprint() in the "C" locale. The "C" locale should be the
3433 default when a C program starts up. In most systems, only ASCII printing
3434 characters match isprint(). This difference caused the output from pcretest
3435 to vary, making some of the tests fail. I have changed pcretest so that:
3436
3437 (a) When it is outputting text in the compiled version of a pattern, bytes
3438 other than 32-126 are always shown as hex escapes.
3439
3440 (b) When it is outputting text that is a matched part of a subject string,
3441 it does the same, unless a different locale has been set for the match
3442 (using the /L modifier). In this case, it uses isprint() to decide.
3443
3444 4. Fixed a major bug that caused incorrect computation of the amount of memory
3445 required for a compiled pattern when options that changed within the
3446 pattern affected the logic of the preliminary scan that determines the
3447 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3448 that the computed length was too small. The symptoms of this bug were
3449 either the PCRE error "internal error: code overflow" from pcre_compile(),
3450 or a glibc crash with a message such as "pcretest: free(): invalid next
3451 size (fast)". Examples of patterns that provoked this bug (shown in
3452 pcretest format) are:
3453
3454 /(?-x: )/x
3455 /(?x)(?-x: \s*#\s*)/
3456 /((?i)[\x{c0}])/8
3457 /(?i:[\x{c0}])/8
3458
3459 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3460 is now done differently.
3461
3462 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3463 wrapper classes; (b) implement a new function in the C++ scanner that is
3464 more efficient than the old way of doing things because it avoids levels of
3465 recursion in the regex matching; (c) add a paragraph to the documentation
3466 for the FullMatch() function.
3467
3468 6. The escape sequence \n was being treated as whatever was defined as
3469 "newline". Not only was this contrary to the documentation, which states
3470 that \n is character 10 (hex 0A), but it also went horribly wrong when
3471 "newline" was defined as CRLF. This has been fixed.
3472
3473 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3474 was being set to -1 for the "end of line" case (supposedly a value that no
3475 character can have). Though this value is never used (the check for end of
3476 line is "zero bytes in current character"), it caused compiler complaints.
3477 I've changed it to 0xffffffff.
3478
3479 8. In pcre_version.c, the version string was being built by a sequence of
3480 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3481 string (as it is for production releases) called a macro with an empty
3482 argument. The C standard says the result of this is undefined. The gcc
3483 compiler treats it as an empty string (which was what was wanted) but it is
3484 reported that Visual C gives an error. The source has been hacked around to
3485 avoid this problem.
3486
3487 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3488 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3489 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3490 of them did).
3491
3492 10. Originally, pcretest opened its input and output without "b"; then I was
3493 told that "b" was needed in some environments, so it was added for release
3494 5.0 to both the input and output. (It makes no difference on Unix-like
3495 systems.) Later I was told that it is wrong for the input on Windows. I've
3496 now abstracted the modes into two macros, to make it easier to fiddle with
3497 them, and removed "b" from the input mode under Windows.
3498
3499 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3500
3501 12. Added -help and --help to pcretest as an official way of being reminded
3502 of the options.
3503
3504 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3505 and pcrecpp.cc because they annoy compilers at high warning levels.
3506
3507 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3508
3509 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3510 scripts are not C programs :-) and which was not noticed because it works
3511 on Linux.
3512
3513 16. pcretest is supposed to handle any length of pattern and data line (as one
3514 line or as a continued sequence of lines) by extending its input buffer if
3515 necessary. This feature was broken for very long pattern lines, leading to
3516 a string of junk being passed to pcre_compile() if the pattern was longer
3517 than about 50K.
3518
3519 17. I have done a major re-factoring of the way pcre_compile() computes the
3520 amount of memory needed for a compiled pattern. Previously, there was code
3521 that made a preliminary scan of the pattern in order to do this. That was
3522 OK when PCRE was new, but as the facilities have expanded, it has become
3523 harder and harder to keep it in step with the real compile phase, and there
3524 have been a number of bugs (see for example, 4 above). I have now found a
3525 cunning way of running the real compile function in a "fake" mode that
3526 enables it to compute how much memory it would need, while actually only
3527 ever using a few hundred bytes of working memory and without too many
3528 tests of the mode. This should make future maintenance and development
3529 easier. A side effect of this work is that the limit of 200 on the nesting
3530 depth of parentheses has been removed (though this was never a serious
3531 limitation, I suspect). However, there is a downside: pcre_compile() now
3532 runs more slowly than before (30% or more, depending on the pattern). I
3533 hope this isn't a big issue. There is no effect on runtime performance.
3534
3535 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3536 newline (only possible for the last line of a file) and it was a
3537 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3538
3539 19. Added additional timing features to pcretest. (1) The -tm option now times
3540 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3541 separate command line item, by a number that specifies the number of
3542 repeats to use when timing. The default is 50000; this gives better
3543 precision, but takes uncomfortably long for very large patterns.
3544
3545 20. Extended pcre_study() to be more clever in cases where a branch of a
3546 subpattern has no definite first character. For example, (a*|b*)[cd] would
3547 previously give no result from pcre_study(). Now it recognizes that the
3548 first character must be a, b, c, or d.
3549
3550 21. There was an incorrect error "recursive call could loop indefinitely" if
3551 a subpattern (or the entire pattern) that was being tested for matching an
3552 empty string contained only one non-empty item after a nested subpattern.
3553 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3554 incorrectly, because the \d was being skipped in the check.
3555
3556 22. The pcretest program now has a new pattern option /B and a command line
3557 option -b, which is equivalent to adding /B to every pattern. This causes
3558 it to show the compiled bytecode, without the additional information that
3559 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3560 is the same as /B/I).
3561
3562 23. A new optimization is now able automatically to treat some sequences such
3563 as a*b as a*+b. More specifically, if something simple (such as a character
3564 or a simple class like \d) has an unlimited quantifier, and is followed by
3565 something that cannot possibly match the quantified thing, the quantifier
3566 is automatically "possessified".
3567
3568 24. A recursive reference to a subpattern whose number was greater than 39
3569 went wrong under certain circumstances in UTF-8 mode. This bug could also
3570 have affected the operation of pcre_study().
3571
3572 25. Realized that a little bit of performance could be had by replacing
3573 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3574
3575 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3576
3577 27. Possessive quantifiers such as a++ were previously implemented by turning
3578 them into atomic groups such as ($>a+). Now they have their own opcodes,
3579 which improves performance. This includes the automatically created ones
3580 from 23 above.
3581
3582 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3583 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3584 the first matched character to be a colon. This applied both to named and
3585 numbered groups.
3586
3587 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3588
3589 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3590 building PCRE on Windows easier, so I have included it in the distribution.
3591
3592 31. There is now a check in pcretest against a ridiculously large number being
3593 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3594 loop, the loop is abandoned.
3595
3596 32. Forward references to subpatterns in conditions such as (?(2)...) where
3597 subpattern 2 is defined later cause pcre_compile() to search forwards in
3598 the pattern for the relevant set of parentheses. This search went wrong
3599 when there were unescaped parentheses in a character class, parentheses
3600 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3601
3602 33. "Subroutine" calls and backreferences were previously restricted to
3603 referencing subpatterns earlier in the regex. This restriction has now
3604 been removed.
3605
3606 34. Added a number of extra features that are going to be in Perl 5.10. On the
3607 whole, these are just syntactic alternatives for features that PCRE had
3608 previously implemented using the Python syntax or my own invention. The
3609 other formats are all retained for compatibility.
3610
3611 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3612 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3613 also .NET compatible.
3614
3615 (b) A recursion or subroutine call to a named group can now be defined as
3616 (?&name) as well as (?P>name).
3617
3618 (c) A backreference to a named group can now be defined as \k<name> or
3619 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3620 5.10, are also .NET compatible.
3621
3622 (d) A conditional reference to a named group can now use the syntax
3623 (?(<name>) or (?('name') as well as (?(name).
3624
3625 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3626 groups (named and numbered) that are never evaluated inline, but can be
3627 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3628 is always false. There may be only one alternative in such a group.
3629
3630 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3631 as the simple (?(R). The condition is true only if the most recent
3632 recursion is that of the given number or name. It does not search out
3633 through the entire recursion stack.
3634
3635 (g) The escape \gN or \g{N} has been added, where N is a positive or
3636 negative number, specifying an absolute or relative reference.
3637
3638 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3639 some "unreachable code" warnings.
3640
3641 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3642 things, this adds five new scripts.
3643
3644 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3645 There were also incompatibilities regarding the handling of \Q..\E inside
3646 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3647 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3648
3649 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3650 matches an empty string, and forcibly breaks the loop. There were bugs in
3651 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3652 against aaaa the result was just "a" rather than "aaaa", for example. Two
3653 separate and independent bugs (that affected different cases) have been
3654 fixed.
3655
3656 39. Refactored the code to abolish the use of different opcodes for small
3657 capturing bracket numbers. This is a tidy that I avoided doing when I
3658 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3659 The new approach is not only tidier, it makes it possible to reduce the
3660 memory needed to fix the previous bug (38).
3661
3662 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3663 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3664 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3665 mode.
3666
3667 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3668 report.
3669
3670 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3671 copy construction and assignment in the C++ wrapper.
3672
3673 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3674 couple of bugs that could have given wrong results in the "--newline=crlf"
3675 case.
3676
3677 44. Added a number of casts and did some reorganization of signed/unsigned int
3678 variables following suggestions from Dair Grant. Also renamed the variable
3679 "this" as "item" because it is a C++ keyword.
3680
3681 45. Arranged for dftables to add
3682
3683 #include "pcre_internal.h"
3684
3685 to pcre_chartables.c because without it, gcc 4.x may remove the array
3686 definition from the final binary if PCRE is built into a static library and
3687 dead code stripping is activated.
3688
3689 46. For an unanchored pattern, if a match attempt fails at the start of a
3690 newline sequence, and the newline setting is CRLF or ANY, and the next two
3691 characters are CRLF, advance by two characters instead of one.
3692
3693
3694 Version 6.7 04-Jul-06
3695 ---------------------
3696
3697 1. In order to handle tests when input lines are enormously long, pcretest has
3698 been re-factored so that it automatically extends its buffers when
3699 necessary. The code is crude, but this _is_ just a test program. The
3700 default size has been increased from 32K to 50K.
3701
3702 2. The code in pcre_study() was using the value of the re argument before
3703 testing it for NULL. (Of course, in any sensible call of the function, it
3704 won't be NULL.)
3705
3706 3. The memmove() emulation function in pcre_internal.h, which is used on
3707 systems that lack both memmove() and bcopy() - that is, hardly ever -
3708 was missing a "static" storage class specifier.
3709
3710 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3711 containing an extended class (one that cannot be represented by a bitmap
3712 because it contains high-valued characters or Unicode property items, e.g.
3713 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3714 pattern, but PCRE should not loop if you do not (it no longer does).
3715 [Detail: two cases were found: (a) a repeated subpattern containing an
3716 extended class; (b) a recursive reference to a subpattern that followed a
3717 previous extended class. It wasn't skipping over the extended class
3718 correctly when UTF-8 mode was not set.]
3719
3720 5. A negated single-character class was not being recognized as fixed-length
3721 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3722 compile error "lookbehind assertion is not fixed length".
3723
3724 6. The RunPerlTest auxiliary script was showing an unexpected difference
3725 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3726 write a Perl script that can interpret lines of an input file either as
3727 byte characters or as UTF-8, which is what "perltest" was being required to
3728 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3729 can't do is switch easily at run time between having the "use utf8;" pragma
3730 or not. In the end, I fudged it by using the RunPerlTest script to insert
3731 "use utf8;" explicitly for the UTF-8 tests.
3732
3733 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3734 the end of the subject string, contrary to the documentation and to what
3735 Perl does. This was true of both matching functions. Now it matches only at
3736 the start of the subject and immediately after *internal* newlines.
3737
3738 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3739 a pointer to an int instead of a pointer to an unsigned long int. This
3740 caused problems on 64-bit systems.
3741
3742 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3743 instance of the 'standard' template library not being so standard".
3744
3745 10. There was no check on the number of named subpatterns nor the maximum
3746 length of a subpattern name. The product of these values is used to compute
3747 the size of the memory block for a compiled pattern. By supplying a very
3748 long subpattern name and a large number of named subpatterns, the size
3749 computation could be caused to overflow. This is now prevented by limiting
3750 the length of names to 32 characters, and the number of named subpatterns
3751 to 10,000.
3752
3753 11. Subpatterns that are repeated with specific counts have to be replicated in
3754 the compiled pattern. The size of memory for this was computed from the
3755 length of the subpattern and the repeat count. The latter is limited to
3756 65535, but there was no limit on the former, meaning that integer overflow
3757 could in principle occur. The compiled length of a repeated subpattern is
3758 now limited to 30,000 bytes in order to prevent this.
3759
3760 12. Added the optional facility to have named substrings with the same name.
3761
3762 13. Added the ability to use a named substring as a condition, using the
3763 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3764 are numbers (not recommended). Forward references are permitted.
3765
3766 14. Added forward references in named backreferences (if you see what I mean).
3767
3768 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3769 pattern could run off the end of the subject. For example, the pattern
3770 "(?s)(.{1,5})"8 did this with the subject "ab".
3771
3772 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3773 PCRE_CASELESS was set when matching characters that were quantified with ?
3774 or *.
3775
3776 17. A character class other than a single negated character that had a minimum
3777 but no maximum quantifier - for example [ab]{6,} - was not handled
3778 correctly by pce_dfa_exec(). It would match only one character.
3779
3780 18. A valid (though odd) pattern that looked like a POSIX character
3781 class but used an invalid character after [ (for example [[,abc,]]) caused
3782 pcre_compile() to give the error "Failed: internal error: code overflow" or
3783 in some cases to crash with a glibc free() error. This could even happen if
3784 the pattern terminated after [[ but there just happened to be a sequence of
3785 letters, a binary zero, and a closing ] in the memory that followed.
3786
3787 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3788 over the years. Originally (before any Unicode support), just the bottom 8
3789 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3790 output from "man perlunicode" includes this:
3791
3792 The regular expression compiler produces polymorphic opcodes. That
3793 is, the pattern adapts to the data and automatically switches to
3794 the Unicode character scheme when presented with Unicode data--or
3795 instead uses a traditional byte scheme when presented with byte
3796 data.
3797
3798 Sadly, a wide octal escape does not cause a switch, and in a string with
3799 no other multibyte characters, these octal escapes are treated as before.
3800 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3801 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3802 Unicode string.
3803
3804 I have not perpetrated such confusion in PCRE. Up till now, it took just
3805 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3806 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3807 translate to the appropriate multibyte character.
3808
3809 29. Applied some refactoring to reduce the number of warnings from Microsoft
3810 and Borland compilers. This has included removing the fudge introduced
3811 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3812 a warning about an unused variable.
3813
3814 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3815 characters since release 4.0, because Perl (from release 5.004) does not.
3816 [Or at least, is documented not to: some releases seem to be in conflict
3817 with the documentation.] However, when a pattern was studied with
3818 pcre_study() and all its branches started with \s, PCRE still included VT
3819 as a possible starting character. Of course, this did no harm; it just
3820 caused an unnecessary match attempt.
3821
3822 22. Removed a now-redundant internal flag bit that recorded the fact that case
3823 dependency changed within the pattern. This was once needed for "required
3824 byte" processing, but is no longer used. This recovers a now-scarce options
3825 bit. Also moved the least significant internal flag bit to the most-
3826 significant bit of the word, which was not previously used (hangover from
3827 the days when it was an int rather than a uint) to free up another bit for
3828 the future.
3829
3830 23. Added support for CRLF line endings as well as CR and LF. As well as the
3831 default being selectable at build time, it can now be changed at runtime
3832 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3833 specify that it is scanning data with non-default line endings.
3834
3835 24. Changed the definition of CXXLINK to make it agree with the definition of
3836 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3837
3838 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3839 recursions. This makes a big different to stack usage for some patterns.
3840
3841 26. If a subpattern containing a named recursion or subroutine reference such
3842 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3843 the space required for the compiled pattern went wrong and gave too small a
3844 value. Depending on the environment, this could lead to "Failed: internal
3845 error: code overflow at offset 49" or "glibc detected double free or
3846 corruption" errors.
3847
3848 27. Applied patches from Google (a) to support the new newline modes and (b) to
3849 advance over multibyte UTF-8 characters in GlobalReplace.
3850
3851 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3852 difference for some implementation of PCRE in some Windows version.
3853
3854 29. Added some extra testing facilities to pcretest:
3855
3856 \q<number> in a data line sets the "match limit" value
3857 \Q<number> in a data line sets the "match recursion limt" value
3858 -S <number> sets the stack size, where <number> is in megabytes
3859
3860 The -S option isn't available for Windows.
3861
3862
3863 Version 6.6 06-Feb-06
3864 ---------------------
3865
3866 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3867 in pcreposix.h. I have copied the definition from pcre.h.
3868
3869 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3870 because pcre.h is no longer a built file.
3871
3872 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3873 not normally included in the compiled code.
3874
3875
3876 Version 6.5 01-Feb-06
3877 ---------------------
3878
3879 1. When using the partial match feature with pcre_dfa_exec(), it was not
3880 anchoring the second and subsequent partial matches at the new starting
3881 point. This could lead to incorrect results. For example, with the pattern
3882 /1234/, partially matching against "123" and then "a4" gave a match.
3883
3884 2. Changes to pcregrep:
3885
3886 (a) All non-match returns from pcre_exec() were being treated as failures
3887 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3888 error message is output. Some extra information is given for the
3889 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3890 probably the only errors that are likely to be caused by users (by
3891 specifying a regex that has nested indefinite repeats, for instance).
3892 If there are more than 20 of these errors, pcregrep is abandoned.
3893
3894 (b) A binary zero was treated as data while matching, but terminated the
3895 output line if it was written out. This has been fixed: binary zeroes
3896 are now no different to any other data bytes.
3897
3898 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3899 used to set a locale for matching. The --locale=xxxx long option has
3900 been added (no short equivalent) to specify a locale explicitly on the
3901 pcregrep command, overriding the environment variables.
3902
3903 (d) When -B was used with -n, some line numbers in the output were one less
3904 than they should have been.
3905
3906 (e) Added the -o (--only-matching) option.
3907
3908 (f) If -A or -C was used with -c (count only), some lines of context were
3909 accidentally printed for the final match.
3910
3911 (g) Added the -H (--with-filename) option.
3912
3913 (h) The combination of options -rh failed to suppress file names for files
3914 that were found from directory arguments.
3915
3916 (i) Added the -D (--devices) and -d (--directories) options.
3917
3918 (j) Added the -F (--fixed-strings) option.
3919
3920 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3921
3922 (l) Added the --colo(u)r option.
3923
3924 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3925 is not present by default.
3926
3927 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3928 items such as (?R) or (?1), when the recursion could match a number of
3929 alternatives. If it matched one of the alternatives, but subsequently,
3930 outside the recursion, there was a failure, the code tried to back up into
3931 the recursion. However, because of the way PCRE is implemented, this is not
3932 possible, and the result was an incorrect result from the match.
3933
3934 In order to prevent this happening, the specification of recursion has
3935 been changed so that all such subpatterns are automatically treated as
3936 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3937
3938 4. I had overlooked the fact that, in some locales, there are characters for
3939 which isalpha() is true but neither isupper() nor islower() are true. In
3940 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3941 and ordfeminine) are like this. This affected the treatment of \w and \W
3942 when they appeared in character classes, but not when they appeared outside
3943 a character class. The bit map for "word" characters is now created
3944 separately from the results of isalnum() instead of just taking it from the
3945 upper, lower, and digit maps. (Plus the underscore character, of course.)
3946
3947 5. The above bug also affected the handling of POSIX character classes such as
3948 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3949 permanent tables. Instead, the bit maps for such a class were previously
3950 created as the appropriate unions of the upper, lower, and digit bitmaps.
3951 Now they are created by subtraction from the [[:word:]] class, which has
3952 its own bitmap.
3953
3954 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3955 It is created by subtracting the vertical space characters (\x09, \x0a,
3956 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3957 subtraction was done in the overall bitmap for a character class, meaning
3958 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3959 be recognized. This bug has been fixed.
3960
3961 7. Patches from the folks at Google:
3962
3963 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3964 real life, but is still worth protecting against".
3965
3966 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3967 regular expressions".
3968
3969 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3970 have it.
3971
3972 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3973 "configure" and the latter not, in order to fix a problem somebody had
3974 with compiling the Arg class on HP-UX.
3975
3976 (e) Improve the error-handling of the C++ wrapper a little bit.
3977
3978 (f) New tests for checking recursion limiting.
3979
3980 8. The pcre_memmove() function, which is used only if the environment does not
3981 have a standard memmove() function (and is therefore rarely compiled),
3982 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3983 returning a result (though PCRE never actually uses the result).
3984
3985 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3986 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3987 returned instead of calling malloc() with an overflowing number that would
3988 most likely cause subsequent chaos.
3989
3990 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3991
3992 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3993 with this option is matched, the nmatch and pmatch options of regexec() are
3994 ignored.
3995
3996 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3997 provided in case anyone wants to the the POSIX interface with UTF-8
3998 strings.
3999
4000 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4001 C++ linking (needed for some HP-UX environments).
4002
4003 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4004 (unused parameter) and in the pcre_printint() function (omitted "default"
4005 switch label when the default is to do nothing).
4006
4007 15. Added some code to make it possible, when PCRE is compiled as a C++
4008 library, to replace subject pointers for pcre_exec() with a smart pointer
4009 class, thus making it possible to process discontinuous strings.
4010
4011 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4012 much the same function. They were added by different people who were trying
4013 to make PCRE easy to compile on non-Unix systems. It has been suggested
4014 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4015 for compiling on Windows systems. I have therefore replaced it with
4016 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4017 defaults to "extern" for C or "extern C" for C++, which works fine on
4018 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4019 SCOPE with something explicit in config.h. In addition:
4020
4021 (a) pcreposix.h still had just "extern" instead of either of these macros;
4022 I have replaced it with PCRE_DATA_SCOPE.
4023
4024 (b) Functions such as _pcre_xclass(), which are internal to the library,
4025 but external in the C sense, all had PCRE_EXPORT in their definitions.
4026 This is apparently wrong for the Windows case, so I have removed it.
4027 (It makes no difference on Unix-like systems.)
4028
4029 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4030 of recursive calls to match(). This is different to MATCH_LIMIT because
4031 that limits the total number of calls to match(), not all of which increase
4032 the depth of recursion. Limiting the recursion depth limits the amount of
4033 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4034 when PCRE is compiled, and changed at run time. A patch from Google adds
4035 this functionality to the C++ interface.
4036
4037 18. Changes to the handling of Unicode character properties:
4038
4039 (a) Updated the table to Unicode 4.1.0.
4040
4041 (b) Recognize characters that are not in the table as "Cn" (undefined).
4042
4043 (c) I revised the way the table is implemented to a much improved format
4044 which includes recognition of ranges. It now supports the ranges that
4045 are defined in UnicodeData.txt, and it also amalgamates other
4046 characters into ranges. This has reduced the number of entries in the
4047 table from around 16,000 to around 3,000, thus reducing its size
4048 considerably. I realized I did not need to use a tree structure after
4049 all - a binary chop search is just as efficient. Having reduced the
4050 number of entries, I extended their size from 6 bytes to 8 bytes to
4051 allow for more data.
4052
4053 (d) Added support for Unicode script names via properties such as \p{Han}.
4054
4055 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4056 matching that character.
4057
4058 20. When matching a repeated Unicode property with a minimum greater than zero,
4059 (for example \pL{2,}), PCRE could look past the end of the subject if it
4060 reached it while seeking the minimum number of characters. This could
4061 happen only if some of the characters were more than one byte long, because
4062 there is a check for at least the minimum number of bytes.
4063
4064 21. Refactored the implementation of \p and \P so as to be more general, to
4065 allow for more different types of property in future. This has changed the
4066 compiled form incompatibly. Anybody with saved compiled patterns that use
4067 \p or \P will have to recompile them.
4068
4069 22. Added "Any" and "L&" to the supported property types.
4070
4071 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4072 but give a compile time error if the value is greater than 0xff.
4073
4074 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4075 accidentally not being installed or uninstalled.
4076
4077 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4078 made were to insert the current release number. This seemed silly, because
4079 it made things harder for people building PCRE on systems that don't run
4080 "configure". I have turned pcre.h into a distributed file, no longer built
4081 by "configure", with the version identification directly included. There is
4082 no longer a pcre.h.in file.
4083
4084 However, this change necessitated a change to the pcre-config script as
4085 well. It is built from pcre-config.in, and one of the substitutions was the
4086 release number. I have updated configure.ac so that ./configure now finds
4087 the release number by grepping pcre.h.
4088
4089 26. Added the ability to run the tests under valgrind.
4090
4091
4092 Version 6.4 05-Sep-05
4093 ---------------------
4094
4095 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4096 "--" to be printed when multiple files were scanned, even when none of the
4097 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4098 consider it to be a bug, and have restored the previous behaviour.
4099
4100 2. A couple of code tidies to get rid of compiler warnings.
4101
4102 3. The pcretest program used to cheat by referring to symbols in the library
4103 whose names begin with _pcre_. These are internal symbols that are not
4104 really supposed to be visible externally, and in some environments it is
4105 possible to suppress them. The cheating is now confined to including
4106 certain files from the library's source, which is a bit cleaner.
4107
4108 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4109 file's purpose clearer.
4110
4111 5. Reorganized pcre_ucp_findchar().
4112
4113
4114 Version 6.3 15-Aug-05
4115 ---------------------
4116
4117 1. The file libpcre.pc.in did not have general read permission in the tarball.
4118
4119 2. There were some problems when building without C++ support:
4120
4121 (a) If C++ support was not built, "make install" and "make test" still
4122 tried to test it.
4123
4124 (b) There were problems when the value of CXX was explicitly set. Some
4125 changes have been made to try to fix these, and ...
4126
4127 (c) --disable-cpp can now be used to explicitly disable C++ support.
4128
4129 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4130 backslash in a target when C++ was disabled. This confuses some
4131 versions of "make", apparently. Using an intermediate variable solves
4132 this. (Same for CPP_LOBJ.)
4133
4134 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4135 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4136 necessary on certain architectures.
4137
4138 4. Added a setting of -export-symbols-regex to the link command to remove
4139 those symbols that are exported in the C sense, but actually are local
4140 within the library, and not documented. Their names all begin with
4141 "_pcre_". This is not a perfect job, because (a) we have to except some
4142 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4143 available (and never for static libraries). I have made a note to try to
4144 find a way round (a) in the future.
4145
4146
4147 Version 6.2 01-Aug-05
4148 ---------------------
4149
4150 1. There was no test for integer overflow of quantifier values. A construction
4151 such as {1111111111111111} would give undefined results. What is worse, if
4152 a minimum quantifier for a parenthesized subpattern overflowed and became
4153 negative, the calculation of the memory size went wrong. This could have
4154 led to memory overwriting.
4155
4156 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4157
4158 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4159 operating environments where this matters.
4160
4161 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4162 PCRE options from within the C++ wrapper.
4163
4164 5. Named capturing subpatterns were not being correctly counted when a pattern
4165 was compiled. This caused two problems: (a) If there were more than 100
4166 such subpatterns, the calculation of the memory needed for the whole
4167 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4168 back references of the form \12, where the number was greater than 9, were
4169 not recognized as back references, even though there were sufficient
4170 previous subpatterns.
4171
4172 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4173 versions of gcc, e.g. 2.95.4.
4174
4175
4176 Version 6.1 21-Jun-05
4177 ---------------------
4178
4179 1. There was one reference to the variable "posix" in pcretest.c that was not
4180 surrounded by "#if !defined NOPOSIX".
4181
4182 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4183 the cross-check on the old pcre_info() function, for the benefit of the
4184 cut-down version of PCRE that is currently imported into Exim.
4185
4186 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4187 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4188 patterns that start (?i) but I don't think that matters. The use of (?i) is
4189 just an example; this all applies to the other options as well.
4190
4191 4. Since libtool seems to echo the compile commands it is issuing, the output
4192 from "make" can be reduced a bit by putting "@" in front of each libtool
4193 compile command.
4194
4195 5. Patch from the folks at Google for configure.in to be a bit more thorough
4196 in checking for a suitable C++ installation before trying to compile the
4197 C++ stuff. This should fix a reported problem when a compiler was present,
4198 but no suitable headers.
4199
4200 6. The man pages all had just "PCRE" as their title. I have changed them to
4201 be the relevant file name. I have also arranged that these names are
4202 retained in the file doc/pcre.txt, which is a concatenation in text format
4203 of all the man pages except the little individual ones for each function.
4204
4205 7. The NON-UNIX-USE file had not been updated for the different set of source
4206 files that come with release 6. I also added a few comments about the C++
4207 wrapper.
4208
4209
4210 Version 6.0 07-Jun-05
4211 ---------------------
4212
4213 1. Some minor internal re-organization to help with my DFA experiments.
4214
4215 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4216 didn't matter for the library itself when fully configured, but did matter
4217 when compiling without UCP support, or within Exim, where the ucp files are
4218 not imported.
4219
4220 3. Refactoring of the library code to split up the various functions into
4221 different source modules. The addition of the new DFA matching code (see
4222 below) to a single monolithic source would have made it really too
4223 unwieldy, quite apart from causing all the code to be include in a
4224 statically linked application, when only some functions are used. This is
4225 relevant even without the DFA addition now that patterns can be compiled in
4226 one application and matched in another.
4227
4228 The downside of splitting up is that there have to be some external
4229 functions and data tables that are used internally in different modules of
4230 the library but which are not part of the API. These have all had their
4231 names changed to start with "_pcre_" so that they are unlikely to clash
4232 with other external names.
4233
4234 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4235 a different (DFA) algorithm. Although it is slower than the original
4236 function, it does have some advantages for certain types of matching
4237 problem.
4238
4239 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4240 including restarting after a partial match.
4241
4242 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4243 defined when compiling for Windows was sent to me. I have put it into the
4244 code, though I have no means of testing or verifying it.
4245
4246 7. Added the pcre_refcount() auxiliary function.
4247
4248 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4249 match before or at the first newline in the subject string. In pcretest,
4250 the /f option on a pattern can be used to set this.
4251
4252 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4253 would behave wrongly. This has been present in PCRE since release 4.0.
4254
4255 10. A number of changes to the pcregrep command:
4256
4257 (a) Refactored how -x works; insert ^(...)$ instead of setting
4258 PCRE_ANCHORED and checking the length, in preparation for adding
4259 something similar for -w.
4260
4261 (b) Added the -w (match as a word) option.
4262
4263 (c) Refactored the way lines are read and buffered so as to have more
4264 than one at a time available.
4265
4266 (d) Implemented a pcregrep test script.
4267
4268 (e) Added the -M (multiline match) option. This allows patterns to match
4269 over several lines of the subject. The buffering ensures that at least
4270 8K, or the rest of the document (whichever is the shorter) is available
4271 for matching (and similarly the previous 8K for lookbehind assertions).
4272
4273 (f) Changed the --help output so that it now says
4274
4275 -w, --word-regex(p)
4276
4277 instead of two lines, one with "regex" and the other with "regexp"
4278 because that confused at least one person since the short forms are the
4279 same. (This required a bit of code, as the output is generated
4280 automatically from a table. It wasn't just a text change.)
4281
4282 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4283 option but starts with a hyphen. Could be a pattern or a path name
4284 starting with a hyphen, for instance.
4285
4286 (h) "-" can be given as a file name to represent stdin.
4287
4288 (i) When file names are being printed, "(standard input)" is used for
4289 the standard input, for compatibility with GNU grep. Previously
4290 "<stdin>" was used.
4291
4292 (j) The option --label=xxx can be used to supply a name to be used for
4293 stdin when file names are being printed. There is no short form.
4294
4295 (k) Re-factored the options decoding logic because we are going to add
4296 two more options that take data. Such options can now be given in four
4297 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4298
4299 (l) Added the -A, -B, and -C options for requesting that lines of context
4300 around matches be printed.
4301
4302 (m) Added the -L option to print the names of files that do not contain
4303 any matching lines, that is, the complement of -l.
4304
4305 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4306 continue to scan other files.
4307
4308 (o) The -s option was incorrectly implemented. For compatibility with other
4309 greps, it now suppresses the error message for a non-existent or non-
4310 accessible file (but not the return code). There is a new option called
4311 -q that suppresses the output of matching lines, which was what -s was
4312 previously doing.
4313
4314 (p) Added --include and --exclude options to specify files for inclusion
4315 and exclusion when recursing.
4316
4317 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4318 Hopefully, it now does.
4319
4320 12. Missing cast in pcre_study().
4321
4322 13. Added an "uninstall" target to the makefile.
4323
4324 14. Replaced "extern" in the function prototypes in Makefile.in with
4325 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4326 world, but is set differently for Windows.
4327
4328 15. Added a second compiling function called pcre_compile2(). The only
4329 difference is that it has an extra argument, which is a pointer to an
4330 integer error code. When there is a compile-time failure, this is set
4331 non-zero, in addition to the error test pointer being set to point to an
4332 error message. The new argument may be NULL if no error number is required
4333 (but then you may as well call pcre_compile(), which is now just a
4334 wrapper). This facility is provided because some applications need a
4335 numeric error indication, but it has also enabled me to tidy up the way
4336 compile-time errors are handled in the POSIX wrapper.
4337
4338 16. Added VPATH=.libs to the makefile; this should help when building with one
4339 prefix path and installing with another. (Or so I'm told by someone who
4340 knows more about this stuff than I do.)
4341
4342 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4343 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4344 match everything, including newlines. This is not POSIX-compatible, but
4345 somebody wanted the feature. From pcretest it can be activated by using
4346 both the P and the s flags.
4347
4348 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4349
4350 19. libpcre.pc was being incorrectly installed as executable.
4351
4352 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4353 it now also looks for '\r' so that it will work unmodified on Windows.
4354
4355 21. Added Google's contributed C++ wrapper to the distribution.
4356
4357 22. Added some untidy missing memory free() calls in pcretest, to keep
4358 Electric Fence happy when testing.
4359
4360
4361
4362 Version 5.0 13-Sep-04
4363 ---------------------
4364
4365 1. Internal change: literal characters are no longer packed up into items
4366 containing multiple characters in a single byte-string. Each character
4367 is now matched using a separate opcode. However, there may be more than one
4368 byte in the character in UTF-8 mode.
4369
4370 2. The pcre_callout_block structure has two new fields: pattern_position and
4371 next_item_length. These contain the offset in the pattern to the next match
4372 item, and its length, respectively.
4373
4374 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4375 insertion of callouts before each pattern item. Added the /C option to
4376 pcretest to make use of this.
4377
4378 4. On the advice of a Windows user, the lines
4379
4380 #if defined(_WIN32) || defined(WIN32)
4381 _setmode( _fileno( stdout ), 0x8000 );
4382 #endif /* defined(_WIN32) || defined(WIN32) */
4383
4384 have been added to the source of pcretest. This apparently does useful
4385 magic in relation to line terminators.
4386
4387 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4388 for the benefit of those environments where the "b" makes a difference.
4389
4390 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4391 to know about it. I have put a hack into configure.in that adds in code
4392 to set GCC=yes if CC=icc. This seems to end up at a point in the
4393 generated configure script that is early enough to affect the setting of
4394 compiler options, which is what is needed, but I have no means of testing
4395 whether it really works. (The user who reported this had patched the
4396 generated configure script, which of course I cannot do.)
4397
4398 LATER: After change 22 below (new libtool files), the configure script
4399 seems to know about icc (and also ecc). Therefore, I have commented out
4400 this hack in configure.in.
4401
4402 7. Added support for pkg-config (2 patches were sent in).
4403
4404 8. Negated POSIX character classes that used a combination of internal tables
4405 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4406 [[:^ascii]]. Typically, they would match almost any characters. The other
4407 POSIX classes were not broken in this way.
4408
4409 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4410 to find the match, as PCRE was deluded into thinking that the match had to
4411 start at the start point or following a newline. The same bug applied to
4412 patterns with negative forward assertions or any backward assertions
4413 preceding ".*" at the start, unless the pattern required a fixed first
4414 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4415
4416 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4417 starting at the last subject character, bytes beyond the end of the subject
4418 string were read.
4419
4420 11. Renamed the variable "class" as "classbits" to make life easier for C++
4421 users. (Previously there was a macro definition, but it apparently wasn't
4422 enough.)
4423
4424 12. Added the new field "tables" to the extra data so that tables can be passed
4425 in at exec time, or the internal tables can be re-selected. This allows
4426 a compiled regex to be saved and re-used at a later time by a different
4427 program that might have everything at different addresses.
4428
4429 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4430 -R library as well as a -L library.
4431
4432 14. The debugging options of pcretest (-d on the command line or D on a
4433 pattern) showed incorrect output for anything following an extended class
4434 that contained multibyte characters and which was followed by a quantifier.
4435
4436 15. Added optional support for general category Unicode character properties
4437 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4438 support. It adds about 90K to the size of the library. The meanings of the
4439 inbuilt class escapes such as \d and \s have NOT been changed.
4440
4441 16. Updated pcredemo.c to include calls to free() to release the memory for the
4442 compiled pattern.
4443
4444 17. The generated file chartables.c was being created in the source directory
4445 instead of in the building directory. This caused the build to fail if the
4446 source directory was different from the building directory, and was
4447 read-only.
4448
4449 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4450 file. No doubt somebody will tell me if they don't make sense... Also added
4451 Dan Mooney's comments about building on OpenVMS.
4452
4453 19. Added support for partial matching via the PCRE_PARTIAL option for
4454 pcre_exec() and the \P data escape in pcretest.
4455
4456 20. Extended pcretest with 3 new pattern features:
4457
4458 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4459 write the compiled pattern to the file whose name is "rest-of-line".
4460 This is a straight binary dump of the data, with the saved pointer to
4461 the character tables forced to be NULL. The study data, if any, is
4462 written too. After writing, pcretest reads a new pattern.
4463
4464 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4465 compiled pattern from the given file. There must not be any
4466 occurrences of "<" in the file name (pretty unlikely); if there are,
4467 pcretest will instead treat the initial "<" as a pattern delimiter.
4468 After reading in the pattern, pcretest goes on to read data lines as
4469 usual.
4470
4471 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4472 and 16-bit fields in a compiled pattern, to simulate a pattern that
4473 was compiled on a host of opposite endianness.
4474
4475 21. The pcre-exec() function can now cope with patterns that were compiled on
4476 hosts of opposite endianness, with this restriction:
4477
4478 As for any compiled expression that is saved and used later, the tables
4479 pointer field cannot be preserved; the extra_data field in the arguments
4480 to pcre_exec() should be used to pass in a tables address if a value
4481 other than the default internal tables were used at compile time.
4482
4483 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4484 now diagnosed as an error. Previously, most of the time, a negative number
4485 would have been treated as zero, but if in addition "ovector" was passed as
4486 NULL, a crash could occur.
4487
4488 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4489 new versions from the libtool 1.5 distribution (the last one is a copy of
4490 a file called libtool.m4). This seems to have fixed the need to patch
4491 "configure" to support Darwin 1.3 (which I used to do). However, I still
4492 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4493 workstation).
4494
4495 24. Changed the PCRE licence to be the more standard "BSD" licence.
4496
4497
4498 Version 4.5 01-Dec-03
4499 ---------------------
4500
4501 1. There has been some re-arrangement of the code for the match() function so
4502 that it can be compiled in a version that does not call itself recursively.
4503 Instead, it keeps those local variables that need separate instances for
4504 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4505 needs to "recurse". Keeping track of where control must go is done by means
4506 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4507 hide most of the details from the main code, and operates only if
4508 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4509 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4510 operating.
4511
4512 To make it easier for callers to provide specially tailored get/free
4513 functions for this usage, two new functions, pcre_stack_malloc, and
4514 pcre_stack_free, are used. They are always called in strict stacking order,
4515 and the size of block requested is always the same.
4516
4517 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4518 PCRE has been compiled to use the stack or the heap for recursion. The
4519 -C option of pcretest uses this to show which version is compiled.
4520
4521 A new data escape \S, is added to pcretest; it causes the amounts of store
4522 obtained and freed by both kinds of malloc/free at match time to be added
4523 to the output.
4524
4525 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4526 what's available on my current Linux desktop machine.
4527
4528 3. When matching a UTF-8 string, the test for a valid string at the start has
4529 been extended. If start_offset is not zero, PCRE now checks that it points
4530 to a byte that is the start of a UTF-8 character. If not, it returns
4531 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4532 this is necessary because there may be backward assertions in the pattern.
4533 When matching the same subject several times, it may save resources to use
4534 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4535
4536 4. The code for checking the validity of UTF-8 strings has been tightened so
4537 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4538 containing "overlong sequences".
4539
4540 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4541 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4542 should have been "&", but it just so happened that all the cases this let
4543 through by mistake were picked up later in the function.
4544
4545 6. I had used a variable called "isblank" - this is a C99 function, causing
4546 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4547
4548 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4549 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4550 so the version is shown; (c) stop "make test" echoing "./RunTest".
4551
4552 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4553
4554 9. The prototype for memmove() for systems that don't have it was using
4555 size_t, but the inclusion of the header that defines size_t was later. I've
4556 moved the #includes for the C headers earlier to avoid this.
4557
4558 10. Added some adjustments to the code to make it easier to compiler on certain
4559 special systems:
4560
4561 (a) Some "const" qualifiers were missing.
4562 (b) Added the macro EXPORT before all exported functions; by default this
4563 is defined to be empty.
4564 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4565 that it reads its output file name as an argument instead of writing
4566 to the standard output and assuming this can be redirected.
4567
4568 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4569 class containing characters with values greater than 255, PCRE compilation
4570 went into a loop.
4571
4572 12. A recursive reference to a subpattern that was within another subpattern
4573 that had a minimum quantifier of zero caused PCRE to crash. For example,
4574 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4575 recursion. If the recursively-called subpattern itself had a zero repeat,
4576 that was OK.
4577
4578 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4579 buffer into which it was copied (for escape processing) was still set at
4580 1024, so long lines caused crashes.
4581
4582 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4583 "internal error: code overflow...". This applied to any character class
4584 that was followed by a possessive quantifier.
4585
4586 15. Modified the Makefile to add libpcre.la as a prerequisite for
4587 libpcreposix.la because I was told this is needed for a parallel build to
4588 work.
4589
4590 16. If a pattern that contained .* following optional items at the start was
4591 studied, the wrong optimizing data was generated, leading to matching
4592 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4593 matching string must start with a or b or c. The correct conclusion for
4594 this pattern is that a match can start with any character.
4595
4596
4597 Version 4.4 13-Aug-03
4598 ---------------------
4599
4600 1. In UTF-8 mode, a character class containing characters with values between
4601 127 and 255 was not handled correctly if the compiled pattern was studied.
4602 In fixing this, I have also improved the studying algorithm for such
4603 classes (slightly).
4604
4605 2. Three internal functions had redundant arguments passed to them. Removal
4606 might give a very teeny performance improvement.
4607
4608 3. Documentation bug: the value of the capture_top field in a callout is *one
4609 more than* the number of the hightest numbered captured substring.
4610
4611 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4612 in incorrectly linking with a previously installed version. They now link
4613 explicitly with libpcre.la.
4614
4615 5. configure.in no longer needs to recognize Cygwin specially.
4616
4617 6. A problem in pcre.in for Windows platforms is fixed.
4618
4619 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4620 pcretest, it used to include the size of the study block as part of its
4621 output. Unfortunately, the structure contains a field that has a different
4622 size on different hardware architectures. This meant that the tests that
4623 showed this size failed. As the block is currently always of a fixed size,
4624 this information isn't actually particularly useful in pcretest output, so
4625 I have just removed it.
4626
4627 8. Three pre-processor statements accidentally did not start in column 1.
4628 Sadly, there are *still* compilers around that complain, even though
4629 standard C has not required this for well over a decade. Sigh.
4630
4631 9. In pcretest, the code for checking callouts passed small integers in the
4632 callout_data field, which is a void * field. However, some picky compilers
4633 complained about the casts involved for this on 64-bit systems. Now
4634 pcretest passes the address of the small integer instead, which should get
4635 rid of the warnings.
4636
4637 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4638 both compile and run time, and gives an error if an invalid UTF-8 sequence
4639 is found. There is a option for disabling this check in cases where the
4640 string is known to be correct and/or the maximum performance is wanted.
4641
4642 11. In response to a bug report, I changed one line in Makefile.in from
4643
4644 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4645 to
4646 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4647
4648 to look similar to other lines, but I have no way of telling whether this
4649 is the right thing to do, as I do not use Windows. No doubt I'll get told
4650 if it's wrong...
4651
4652
4653 Version 4.3 21-May-03
4654 ---------------------
4655
4656 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4657 Makefile.
4658
4659 2. Some refactoring to improve the quality of the code:
4660
4661 (i) The utf8_table... variables are now declared "const".
4662
4663 (ii) The code for \cx, which used the "case flipping" table to upper case
4664 lower case letters, now just substracts 32. This is ASCII-specific,
4665 but the whole concept of \cx is ASCII-specific, so it seems
4666 reasonable.
4667
4668 (iii) PCRE was using its character types table to recognize decimal and
4669 hexadecimal digits in the pattern. This is silly, because it handles
4670 only 0-9, a-f, and A-F, but the character types table is locale-
4671 specific, which means strange things might happen. A private
4672 table is now used for this - though it costs 256 bytes, a table is
4673 much faster than multiple explicit tests. Of course, the standard
4674 character types table is still used for matching digits in subject
4675 strings against \d.
4676
4677 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4678 ending in _t are). So I've renamed it as ESC_tee.
4679
4680 3. The first argument for regexec() in the POSIX wrapper should have been
4681 defined as "const".
4682
4683 4. Changed pcretest to use malloc() for its buffers so that they can be
4684 Electric Fenced for debugging.
4685
4686 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4687 to read one or more bytes before the start of the subject string. Often this
4688 had no effect on PCRE's behaviour, but in some circumstances it could
4689 provoke a segmentation fault.
4690
4691 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4692 to try to read one or more bytes before the start of the subject string.
4693
4694 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4695 UTF-8 support could misbehave in various ways if the subject string
4696 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4697 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4698 back over UTF-8 characters.)
4699
4700
4701 Version 4.2 14-Apr-03
4702 ---------------------
4703
4704 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4705
4706 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4707 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4708 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4709 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4710 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4711 and BUILD_EXEEXT
4712 Note: automatic setting of the BUILD variables is not yet working
4713 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4714 compile-time but not at link-time
4715 [LINK]: use for linking executables only
4716 make different versions for Windows and non-Windows
4717 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4718 libraries
4719 [LINK_FOR_BUILD]: new variable
4720 [OBJEXT]: use throughout
4721 [EXEEXT]: use throughout
4722 <winshared>: new target
4723 <wininstall>: new target
4724 <dftables.o>: use native compiler
4725 <dftables>: use native linker
4726 <install>: handle Windows platform correctly
4727 <clean>: ditto
4728 <check>: ditto
4729 copy DLL to top builddir before testing
4730
4731 As part of these changes, -no-undefined was removed again. This was reported
4732 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4733 in any case.
4734
4735 3. Some tidies to get rid of compiler warnings:
4736
4737 . In the match_data structure, match_limit was an unsigned long int, whereas
4738 match_call_count was an int. I've made them both unsigned long ints.
4739
4740 . In pcretest the fact that a const uschar * doesn't automatically cast to
4741 a void * provoked a warning.
4742
4743 . Turning on some more compiler warnings threw up some "shadow" variables
4744 and a few more missing casts.
4745
4746 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4747 option, a class that contained a single character with a value between 128
4748 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4749
4750 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4751 option, a class that contained several characters, but with at least one
4752 whose value was between 128 and 255 caused PCRE to crash.
4753
4754
4755 Version 4.1 12-Mar-03
4756 ---------------------
4757
4758 1. Compiling with gcc -pedantic found a couple of places where casts were
4759 needed, and a string in dftables.c that was longer than standard compilers are
4760 required to support.
4761
4762 2. Compiling with Sun's compiler found a few more places where the code could
4763 be tidied up in order to avoid warnings.
4764
4765 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4766 first of these names is deprecated in the latest Autoconf in favour of the name
4767 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4768 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4769 analogy I have changed it to CFLAGS_FOR_BUILD.
4770
4771 4. Added -no-undefined to the linking command in the Makefile, because this is
4772 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4773 linking step for the pcreposix library.
4774
4775 5. PCRE was failing to diagnose the case of two named groups with the same
4776 name.
4777
4778 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4779 literal character that is needed in the subject for a match, and scans along to
4780 ensure that it is present before embarking on the full matching process. This
4781 saves time in cases of nested unlimited repeats that are never going to match.
4782 Problem: the scan can take a lot of time if the subject is very long (e.g.
4783 megabytes), thus penalizing straightforward matches. It is now done only if the
4784 amount of subject to be scanned is less than 1000 bytes.
4785
4786 7. A lesser problem with the same optimization is that it was recording the
4787 first character of an anchored pattern as "needed", thus provoking a search
4788 right along the subject, even when the first match of the pattern was going to
4789 fail. The "needed" character is now not set for anchored patterns, unless it
4790 follows something in the pattern that is of non-fixed length. Thus, it still
4791 fulfils its original purpose of finding quick non-matches in cases of nested
4792 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4793
4794
4795 Version 4.0 17-Feb-03
4796 ---------------------
4797
4798 1. If a comment in an extended regex that started immediately after a meta-item
4799 extended to the end of string, PCRE compiled incorrect data. This could lead to
4800 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4801
4802 2. Moved to autoconf 2.53 and libtool 1.4.2.
4803
4804 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4805 the special perltest8 script is no longer needed - all the tests can be run
4806 from a single perltest script.
4807
4808 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4809 by \s. It has now been removed in PCRE. This means it isn't recognized as
4810 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4811 class [:space:] *does* include VT, thereby creating a mess.
4812
4813 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4814 space and tab.
4815
4816 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4817 its new features into the main test script, reducing the number of scripts.
4818
4819 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4820 were backward compatible, and made the (?i) apply to the whole pattern, as if
4821 /i were given. Now it behaves more logically, and applies the option setting
4822 only to what follows. PCRE has been changed to follow suit. However, if it
4823 finds options settings right at the start of the pattern, it extracts them into
4824 the global options, as before. Thus, they show up in the info data.
4825
4826 8. Added support for the \Q...\E escape sequence. Characters in between are
4827 treated as literals. This is slightly different from Perl in that $ and @ are
4828 also handled as literals inside the quotes. In Perl, they will cause variable
4829 interpolation. Note the following examples:
4830
4831 Pattern PCRE matches Perl matches
4832
4833 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4834 \Qabc\$xyz\E abc\$xyz abc\$xyz
4835 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4836
4837 For compatibility with Perl, \Q...\E sequences are recognized inside character
4838 classes as well as outside them.
4839
4840 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4841 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4842 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4843 signed/unsigned warnings.
4844
4845 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4846 option for pcretest, so I've replaced it by a simple function that does just
4847 that job.
4848
4849 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4850 "pcregrep -".
4851
4852 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4853 Java package. This provides some syntactic sugar for simple cases of what my
4854 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4855 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4856 item, you can use this simplified notation. Note that only makes sense with
4857 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4858 greediness, whatever the setting of the PCRE_UNGREEDY option.
4859
4860 13. A change of greediness default within a pattern was not taking effect at
4861 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4862 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4863 was abstracted outside.
4864
4865 14. PCRE now supports the \G assertion. It is true when the current matching
4866 position is at the start point of the match. This differs from \A when the
4867 starting offset is non-zero. Used with the /g option of pcretest (or similar
4868 code), it works in the same way as it does for Perl's /g option. If all
4869 alternatives of a regex begin with \G, the expression is anchored to the start
4870 match position, and the "anchored" flag is set in the compiled expression.
4871
4872 15. Some bugs concerning the handling of certain option changes within patterns
4873 have been fixed. These applied to options other than (?ims). For example,
4874 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4875 been the other way round. Some of this was related to change 7 above.
4876
4877 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4878 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4879 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4880 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4881
4882 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4883 mode. Unlike ".", it always matches newline, whatever the setting of
4884 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4885 assertions. Perl allows it, but it doesn't (in general) work because it can't
4886 calculate the length of the lookbehind. At least, that's the case for Perl
4887 5.8.0 - I've been told they are going to document that it doesn't work in
4888 future.
4889
4890 18. Added an error diagnosis for escapes that PCRE does not support: these are
4891 \L, \l, \N, \P, \p, \U, \u, and \X.
4892
4893 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4894 reading past the end of the pattern in cases such as /[abcd/.
4895
4896 20. PCRE was getting more memory than necessary for patterns with classes that
4897 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4898
4899 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4900 compiling PCRE for use with Virtual Pascal.
4901
4902 22. Small fix to the Makefile to make it work properly if the build is done
4903 outside the source tree.
4904
4905 23. Added a new extension: a condition to go with recursion. If a conditional
4906 subpattern starts with (?(R) the "true" branch is used if recursion has
4907 happened, whereas the "false" branch is used only at the top level.
4908
4909 24. When there was a very long string of literal characters (over 255 bytes
4910 without UTF support, over 250 bytes with UTF support), the computation of how
4911 much memory was required could be incorrect, leading to segfaults or other
4912 strange effects.
4913
4914 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4915 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4916 there was a subsequent back reference to those brackets. This meant that, for
4917 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4918 possible to check for precisely this case. All we can do is abandon the
4919 optimization if .* occurs inside capturing brackets when there are any back
4920 references whatsoever. (See below for a better fix that came later.)
4921
4922 26. The handling of the optimization for finding the first character of a
4923 non-anchored pattern, and for finding a character that is required later in the
4924 match were failing in some cases. This didn't break the matching; it just
4925 failed to optimize when it could. The way this is done has been re-implemented.
4926
4927 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4928
4929 28. Added a new feature that provides some of the functionality that Perl
4930 provides with (?{...}). The facility is termed a "callout". The way it is done
4931 in PCRE is for the caller to provide an optional function, by setting
4932 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4933 global variable. By default it is unset, which disables all calling out. To get
4934 the function called, the regex must include (?C) at appropriate points. This
4935 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4936 This provides a means of identifying different callout points. When PCRE
4937 reaches such a point in the regex, if pcre_callout has been set, the external
4938 function is called. It is provided with data in a structure called
4939 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4940 matching continues; if it returns a non-zero value, the match at the current
4941 point fails. However, backtracking will occur if possible. [This was changed
4942 later and other features added - see item 49 below.]
4943
4944 29. pcretest is upgraded to test the callout functionality. It provides a
4945 callout function that displays information. By default, it shows the start of
4946 the match and the current position in the text. There are some new data escapes
4947 to vary what happens:
4948
4949 \C+ in addition, show current contents of captured substrings
4950 \C- do not supply a callout function
4951 \C!n return 1 when callout number n is reached
4952 \C!n!m return 1 when callout number n is reached for the mth time
4953
4954 30. If pcregrep was called with the -l option and just a single file name, it
4955 output "<stdin>" if a match was found, instead of the file name.
4956
4957 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4958 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4959 pcre_exec(). This saves a malloc/free per call. The default value of
4960 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4961 when configuring.
4962
4963 32. The default maximum size of a compiled pattern is 64K. There have been a
4964 few cases of people hitting this limit. The code now uses macros to handle the
4965 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4966 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4967 configuring. Tests 2 and 5 work only with 2-byte links because they output
4968 debugging information about compiled patterns.
4969
4970 33. Internal code re-arrangements:
4971
4972 (a) Moved the debugging function for printing out a compiled regex into
4973 its own source file (printint.c) and used #include to pull it into
4974 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4975 separate copies.
4976
4977 (b) Defined the list of op-code names for debugging as a macro in
4978 internal.h so that it is next to the definition of the opcodes.
4979
4980 (c) Defined a table of op-code lengths for simpler skipping along compiled
4981 code. This is again a macro in internal.h so that it is next to the
4982 definition of the opcodes.
4983
4984 34. Added support for recursive calls to individual subpatterns, along the
4985 lines of Robin Houston's patch (but implemented somewhat differently).
4986
4987 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4988 allow it to read and process whole directories in Win32. This code was
4989 contributed by Lionel Fourquaux; it has not been tested by me.
4990
4991 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4992 used to name a group. Names consist of alphanumerics and underscores, and must
4993 be unique. Back references use the syntax (?P=name) and recursive calls use
4994 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4995 numbers. The function pcre_fullinfo() can be used after compilation to extract
4996 a name/number map. There are three relevant calls:
4997
4998 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4999 PCRE_INFO_NAMECOUNT yields the number of entries
5000 PCRE_INFO_NAMETABLE yields a pointer to the map.
5001
5002 The map is a vector of fixed-size entries. The size of each entry depends on
5003 the length of the longest name used. The first two bytes of each entry are the
5004 group number, most significant byte first. There follows the corresponding
5005 name, zero terminated. The names are in alphabetical order.
5006
5007 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5008 case instead of 255. Making it the same both with and without UTF-8 support
5009 means that the same test output works with both.
5010
5011 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5012 calling malloc() with a zero argument.
5013
5014 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5015 optimization. I've improved things by keeping a bitmap of backreferences with
5016 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5017 fact referenced, the optimization can be applied. It is unlikely that a
5018 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5019 the match to follow \n) will appear inside brackets with a number greater than
5020 31, but if it does, any back reference > 31 suppresses the optimization.
5021
5022 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5023 of disabling numbered capturing parentheses. Any opening parenthesis that is
5024 not followed by ? behaves as if it were followed by ?: but named parentheses
5025 can still be used for capturing (and they will acquire numbers in the usual
5026 way).
5027
5028 41. Redesigned the return codes from the match() function into yes/no/error so
5029 that errors can be passed back from deep inside the nested calls. A malloc
5030 failure while inside a recursive subpattern call now causes the
5031 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5032
5033 42. It is now possible to set a limit on the number of times the match()
5034 function is called in a call to pcre_exec(). This facility makes it possible to
5035 limit the amount of recursion and backtracking, though not in a directly
5036 obvious way, because the match() function is used in a number of different
5037 circumstances. The count starts from zero for each position in the subject
5038 string (for non-anchored patterns). The default limit is, for compatibility, a
5039 large number, namely 10 000 000. You can change this in two ways:
5040
5041 (a) When configuring PCRE before making, you can use --with-match-limit=n
5042 to set a default value for the compiled library.
5043
5044 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5045 a different value is set. See 45 below.
5046
5047 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5048
5049 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5050 of things that can be changed at compile time. The first argument specifies
5051 what is wanted and the second points to where the information is to be placed.
5052 The current list of available information is:
5053
5054 PCRE_CONFIG_UTF8
5055
5056 The output is an integer that is set to one if UTF-8 support is available;
5057 otherwise it is set to zero.
5058
5059 PCRE_CONFIG_NEWLINE
5060
5061 The output is an integer that it set to the value of the code that is used for
5062 newline. It is either LF (10) or CR (13).
5063
5064 PCRE_CONFIG_LINK_SIZE
5065
5066 The output is an integer that contains the number of bytes used for internal
5067 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5068
5069 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5070
5071 The output is an integer that contains the threshold above which the POSIX
5072 interface uses malloc() for output vectors. See item 31 above.
5073
5074 PCRE_CONFIG_MATCH_LIMIT
5075
5076 The output is an unsigned integer that contains the default limit of the number
5077 of match() calls in a pcre_exec() execution. See 42 above.
5078
5079 44. pcretest has been upgraded by the addition of the -C option. This causes it
5080 to extract all the available output from the new pcre_config() function, and to
5081 output it. The program then exits immediately.
5082
5083 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5084 order to support additional features. One way would have been to define
5085 pcre_exec2() (for example) with extra arguments, but this would not have been
5086 extensible, and would also have required all calls to the original function to
5087 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5088 is used for passing in "extra" data from pcre_study().
5089
5090 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5091 contains the following fields:
5092
5093 flags a bitmap indicating which of the following fields are set
5094 study_data opaque data from pcre_study()
5095 match_limit a way of specifying a limit on match() calls for a specific
5096 call to pcre_exec()
5097 callout_data data for callouts (see 49 below)
5098
5099 The flag bits are also defined in pcre.h, and are
5100
5101 PCRE_EXTRA_STUDY_DATA
5102 PCRE_EXTRA_MATCH_LIMIT
5103 PCRE_EXTRA_CALLOUT_DATA
5104
5105 The pcre_study() function now returns one of these new pcre_extra blocks, with
5106 the actual study data pointed to by the study_data field, and the
5107 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5108 before. That is, this change is entirely upwards-compatible and requires no
5109 change to existing code.
5110
5111 If you want to pass in additional data to pcre_exec(), you can either place it
5112 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5113 block.
5114
5115 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5116 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5117 times with different match limits, until it finds the minimum value needed for
5118 pcre_exec() to complete. The value is then output. This can be instructive; for
5119 most simple matches the number is quite small, but for pathological cases it
5120 gets very large very quickly.
5121
5122 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5123 returns the size of the data block pointed to by the study_data field in a
5124 pcre_extra block, that is, the value that was passed as the argument to
5125 pcre_malloc() when PCRE was getting memory in which to place the information
5126 created by pcre_study(). The fourth argument should point to a size_t variable.
5127 pcretest has been extended so that this information is shown after a successful
5128 pcre_study() call when information about the compiled regex is being displayed.
5129
5130 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5131 because what follows is always an absolute path. (Later: it turns out that this
5132 is more than cosmetic for MinGW, because it doesn't like empty path
5133 components.)
5134
5135 49. Some changes have been made to the callout feature (see 28 above):
5136
5137 (i) A callout function now has three choices for what it returns:
5138
5139 0 => success, carry on matching
5140 > 0 => failure at this point, but backtrack if possible
5141 < 0 => serious error, return this value from pcre_exec()
5142
5143 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5144 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5145 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5146 use by callout functions. It will never be used by PCRE itself.
5147
5148 (ii) The pcre_extra structure (see 45 above) has a void * field called
5149 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5150 pcre_callout_block structure has a field of the same name. The contents of
5151 the field passed in the pcre_extra structure are passed to the callout
5152 function in the corresponding field in the callout block. This makes it
5153 easier to use the same callout-containing regex from multiple threads. For
5154 testing, the pcretest program has a new data escape
5155
5156 \C*n pass the number n (may be negative) as callout_data
5157
5158 If the callout function in pcretest receives a non-zero value as
5159 callout_data, it returns that value.
5160
5161 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5162 there were some redundant $(CFLAGS) in commands that are now specified as
5163 $(LINK), which already includes $(CFLAGS).
5164
5165 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5166 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5167 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5168 one-byte characters throughout. Note that case-insensitive matching applies
5169 only to characters whose values are less than 256. PCRE doesn't support the
5170 notion of cases for higher-valued characters.
5171
5172 (i) A character class whose characters are all within 0-255 is handled as
5173 a bit map, and the map is inverted for negative classes. Previously, a
5174 character > 255 always failed to match such a class; however it should
5175 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5176
5177 (ii) A negated character class with a single character < 255 is coded as
5178 "not this character" (OP_NOT). This wasn't working properly when the test
5179 character was multibyte, either singly or repeated.
5180
5181 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5182 mode, for example: \x{100}{2,3}.
5183
5184 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5185 singly or repeated) now correctly test multibyte characters. However,
5186 PCRE doesn't recognize any characters with values greater than 255 as
5187 digits, spaces, or word characters. Such characters always match \D, \S,
5188 and \W, and never match \d, \s, or \w.
5189
5190 (v) Classes may now contain characters and character ranges with values
5191 greater than 255. For example: [ab\x{100}-\x{400}].
5192
5193 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5194 PCRE in UTF-8 mode.
5195
5196 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5197 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5198 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5199 value.)
5200
5201 53. The single man page has become too large. I have therefore split it up into
5202 a number of separate man pages. These also give rise to individual HTML pages;
5203 these are now put in a separate directory, and there is an index.html page that
5204 lists them all. Some hyperlinking between the pages has been installed.
5205
5206 54. Added convenience functions for handling named capturing parentheses.
5207
5208 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5209 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5210 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5211 are faulted.
5212
5213 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5214 calling configure. These values are used when compiling the dftables.c program
5215 which is run to generate the source of the default character tables. They
5216 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5217 you will need to set these values.
5218
5219 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5220
5221
5222 Version 3.9 02-Jan-02
5223 ---------------------
5224
5225 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5226
5227 2. If --disable-static was given, the building process failed when trying to
5228 build pcretest and pcregrep. (For some reason it was using libtool to compile
5229 them, which is not right, as they aren't part of the library.)
5230
5231
5232 Version 3.8 18-Dec-01
5233 ---------------------
5234
5235 1. The experimental UTF-8 code was completely screwed up. It was packing the
5236 bytes in the wrong order. How dumb can you get?
5237
5238
5239 Version 3.7 29-Oct-01
5240 ---------------------
5241
5242 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5243 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5244 this didn't happen under Solaris 8, where I normally test things.
5245
5246 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5247 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5248 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5249 this fix an hour or so after the initial 3.7 release.)
5250
5251
5252 Version 3.6 23-Oct-01
5253 ---------------------
5254
5255 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5256 offsets passed as NULL with zero offset count.
5257
5258 2. The config.guess and config.sub files had not been updated when I moved to
5259 the latest autoconf.
5260
5261
5262 Version 3.5 15-Aug-01
5263 ---------------------
5264
5265 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5266 had been forgotten.
5267
5268 2. By using declared but undefined structures, we can avoid using "void"
5269 definitions in pcre.h while keeping the internal definitions of the structures
5270 private.
5271
5272 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5273 user point of view, this means that both static and shared libraries are built
5274 by default, but this can be individually controlled. More of the work of
5275 handling this static/shared cases is now inside libtool instead of PCRE's make
5276 file.
5277
5278 4. The pcretest utility is now installed along with pcregrep because it is
5279 useful for users (to test regexs) and by doing this, it automatically gets
5280 relinked by libtool. The documentation has been turned into a man page, so
5281 there are now .1, .txt, and .html versions in /doc.
5282
5283 5. Upgrades to pcregrep:
5284 (i) Added long-form option names like gnu grep.
5285 (ii) Added --help to list all options with an explanatory phrase.
5286 (iii) Added -r, --recursive to recurse into sub-directories.
5287 (iv) Added -f, --file to read patterns from a file.
5288
5289 6. pcre_exec() was referring to its "code" argument before testing that
5290 argument for NULL (and giving an error if it was NULL).
5291
5292 7. Upgraded Makefile.in to allow for compiling in a different directory from
5293 the source directory.
5294
5295 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5296 options bits, the pointer it was passed was to an int instead of to an unsigned
5297 long int. This mattered only on 64-bit systems.
5298
5299 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5300 generated) instead of pcre.in, which it its source. Also made the same change
5301 in several of the .c files.
5302
5303 10. A new release of gcc defines printf() as a macro, which broke pcretest
5304 because it had an ifdef in the middle of a string argument for printf(). Fixed
5305 by using separate calls to printf().
5306
5307 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5308 script, to force use of CR or LF instead of \n in the source. On non-Unix
5309 systems, the value can be set in config.h.
5310
5311 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5312 absolute limit. Changed the text of the error message to make this clear, and
5313 likewise updated the man page.
5314
5315 13. The limit of 99 on the number of capturing subpatterns has been removed.
5316 The new limit is 65535, which I hope will not be a "real" limit.
5317
5318
5319 Version 3.4 22-Aug-00
5320 ---------------------
5321
5322 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5323
5324 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5325
5326
5327 Version 3.3 01-Aug-00
5328 ---------------------
5329
5330 1. If an octal character was given, but the value was greater than \377, it
5331 was not getting masked to the least significant bits, as documented. This could
5332 lead to crashes in some systems.
5333
5334 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5335 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5336
5337 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5338 These just pass their arguments on to (pcre_free)(), but they are provided
5339 because some uses of PCRE bind it to non-C systems that can call its functions,
5340 but cannot call free() or pcre_free() directly.
5341
5342 4. Add "make test" as a synonym for "make check". Corrected some comments in
5343 the Makefile.
5344
5345 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5346 Makefile.
5347
5348 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5349 command called pgrep for grepping around the active processes.
5350
5351 7. Added the beginnings of support for UTF-8 character strings.
5352
5353 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5354 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5355 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5356 out for the ar command.)
5357
5358
5359 Version 3.2 12-May-00
5360 ---------------------
5361
5362 This is purely a bug fixing release.
5363
5364 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5365 of ZA. This was just one example of several cases that could provoke this bug,
5366 which was introduced by change 9 of version 2.00. The code for breaking
5367 infinite loops after an iteration that matches an empty string was't working
5368 correctly.
5369
5370 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5371 when matched against abbab (for example). After matching an empty string, it
5372 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5373 caused it to match further down the string than it should.
5374
5375 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5376 was there because it doesn't seem to be needed, and it causes trouble on some
5377 systems, as it is not a Standard C header. It has been removed.
5378
5379 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5380 were reported on the Macintosh. The changes were from
5381
5382 while ((c = *(++ptr)) != 0 && c != '\n');
5383 to
5384 while ((c = *(++ptr)) != 0 && c != '\n') ;
5385
5386 Totally extraordinary, but if that's what it takes...
5387
5388 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5389 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5390 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5391 assumes the way PCRE uses memmove() (always moving upwards).
5392
5393 6. PCRE is being used in one environment where strchr() is not available. There
5394 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5395 faster code anyway.
5396
5397
5398 Version 3.1 09-Feb-00
5399 ---------------------
5400
5401 The only change in this release is the fixing of some bugs in Makefile.in for
5402 the "install" target:
5403
5404 (1) It was failing to install pcreposix.h.
5405
5406 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5407
5408
5409 Version 3.0 01-Feb-00
5410 ---------------------
5411
5412 1. Add support for the /+ modifier to perltest (to output $` like it does in
5413 pcretest).
5414
5415 2. Add support for the /g modifier to perltest.
5416
5417 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5418 matches null strings.
5419
5420 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5421 pattern. Perl treats empty patterns specially - it reuses the most recent
5422 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5423 effect.
5424
5425 5. The POSIX interface was broken in that it was just handing over the POSIX
5426 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5427 required a bigger vector, with some working space on the end. This means that
5428 the POSIX wrapper now has to get and free some memory, and copy the results.
5429
5430 6. Added some simple autoconf support, placing the test data and the
5431 documentation in separate directories, re-organizing some of the
5432 information files, and making it build pcre-config (a GNU standard). Also added
5433 libtool support for building PCRE as a shared library, which is now the
5434 default.
5435
5436 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5437 09 are not valid octal constants. Single digits will be used for minor values
5438 less than 10.
5439
5440 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5441 existing programs that set these in the POSIX interface can use PCRE without
5442 modification.
5443
5444 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5445 return all that pcre_info() returns, plus additional data. The pcre_info()
5446 function is retained for compatibility, but is considered to be obsolete.
5447
5448 10. Added experimental recursion feature (?R) to handle one common case that
5449 Perl 5.6 will be able to do with (?p{...}).
5450
5451 11. Added support for POSIX character classes like [:alpha:], which Perl is
5452 adopting.
5453
5454
5455 Version 2.08 31-Aug-99
5456 ----------------------
5457
5458 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5459 trying to match at the startoffset position, but instead was moving forward to
5460 the next newline as if a previous match had failed.
5461
5462 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5463 and could get into a loop if a null string was matched other than at the start
5464 of the subject.
5465
5466 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5467 be distinguished at compile time, and for completeness also added PCRE_DATE.
5468
5469 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5470 in GnuWin32 environments.
5471
5472
5473 Version 2.07 29-Jul-99
5474 ----------------------
5475
5476 1. The documentation is now supplied in plain text form and HTML as well as in
5477 the form of man page sources.
5478
5479 2. C++ compilers don't like assigning (void *) values to other pointer types.
5480 In particular this affects malloc(). Although there is no problem in Standard
5481 C, I've put in casts to keep C++ compilers happy.
5482
5483 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5484 should be (const char *).
5485
5486 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5487 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5488 However, I haven't made this a standard facility. The documentation doesn't
5489 mention it, and the Makefile doesn't support it.
5490
5491 5. The Makefile now contains an "install" target, with editable destinations at
5492 the top of the file. The pcretest program is not installed.
5493
5494 6. pgrep -V now gives the PCRE version number and date.
5495
5496 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5497 causing the entire string to be ignored, instead of just the last character.
5498
5499 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5500 non-matching string, it can take a very, very long time, even for strings of
5501 quite modest length, because of the nested recursion. PCRE now does better in
5502 some of these cases. It does this by remembering the last required literal
5503 character in the pattern, and pre-searching the subject to ensure it is present
5504 before running the real match. In other words, it applies a heuristic to detect
5505 some types of certain failure quickly, and in the above example, if presented
5506 with a string that has no trailing " it gives "no match" very quickly.
5507
5508 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5509 other alternatives are tried instead.
5510
5511
5512 Version 2.06 09-Jun-99
5513 ----------------------
5514
5515 1. Change pcretest's output for amount of store used to show just the code
5516 space, because the remainder (the data block) varies in size between 32-bit and
5517 64-bit systems.
5518
5519 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5520 start matching at. This allows lookbehinds to work when searching for multiple
5521 occurrences in a string.
5522
5523 3. Added additional options to pcretest for testing multiple occurrences:
5524
5525 /+ outputs the rest of the string that follows a match
5526 /g loops for multiple occurrences, using the new startoffset argument
5527 /G loops for multiple occurrences by passing an incremented pointer
5528
5529 4. PCRE wasn't doing the "first character" optimization for patterns starting
5530 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5531 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5532 the letter 'x'. On long subject strings, this gives a significant speed-up.
5533
5534
5535 Version 2.05 21-Apr-99
5536 ----------------------
5537
5538 1. Changed the type of magic_number from int to long int so that it works
5539 properly on 16-bit systems.
5540
5541 2. Fixed a bug which caused patterns starting with .* not to work correctly
5542 when the subject string contained newline characters. PCRE was assuming
5543 anchoring for such patterns in all cases, which is not correct because .* will
5544 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5545 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5546 must be retried after every newline in the subject.
5547
5548
5549 Version 2.04 18-Feb-99
5550 ----------------------
5551
5552 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5553 computation of the store needed to hold the pattern was incorrect (too large).
5554 If such patterns were nested a few deep, this could multiply and become a real
5555 problem.
5556
5557 2. Added /M option to pcretest to show the memory requirement of a specific
5558 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5559
5560 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5561 compiled in such a way that the backtracking after subsequent failure was
5562 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5563 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5564
5565
5566 Version 2.03 02-Feb-99
5567 ----------------------
5568
5569 1. Fixed typo and small mistake in man page.
5570
5571 2. Added 4th condition (GPL supersedes if conflict) and created separate
5572 LICENCE file containing the conditions.
5573
5574 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5575 Perl, that is the internal \ allows the delimiter to be included in the
5576 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5577 the final delimiter, add \ to the end of the pattern (to test the error).
5578
5579 4. Added the convenience functions for extracting substrings after a successful
5580 match. Updated pcretest to make it able to test these functions.
5581
5582
5583 Version 2.02 14-Jan-99
5584 ----------------------
5585
5586 1. Initialized the working variables associated with each extraction so that
5587 their saving and restoring doesn't refer to uninitialized store.
5588
5589 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5590 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5591 fix the problem.
5592
5593 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5594 calls, and wasn't printing the correct value for compiling calls. Increased the
5595 default value of LOOPREPEAT, and the number of significant figures in the
5596 times.
5597
5598 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5599
5600 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5601 a building problem on Windows NT with a FAT file system.
5602
5603
5604 Version 2.01 21-Oct-98
5605 ----------------------
5606
5607 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5608 to character tables built by pcre_maketables() in the current locale. If NULL
5609 is passed, the default tables are used.
5610
5611
5612 Version 2.00 24-Sep-98
5613 ----------------------
5614
5615 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5616 it any more.
5617
5618 2. Allow quantification of (?>) groups, and make it work correctly.
5619
5620 3. The first character computation wasn't working for (?>) groups.
5621
5622 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5623 end of the subject) and add 5.005's \z, which really does match only at the
5624 very end of the subject.
5625
5626 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5627
5628 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5629 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5630 localized options. All options to pcre_study() were also removed.
5631
5632 7. Add other new features from 5.005:
5633
5634 $(?<= positive lookbehind
5635 $(?<! negative lookbehind
5636 (?imsx-imsx) added the unsetting capability
5637 such a setting is global if at outer level; local otherwise
5638 (?imsx-imsx:) non-capturing groups with option setting
5639 (?(cond)re|re) conditional pattern matching
5640
5641 A backreference to itself in a repeated group matches the previous
5642 captured string.
5643
5644 8. General tidying up of studying (both automatic and via "study")
5645 consequential on the addition of new assertions.
5646
5647 9. As in 5.005, unlimited repeated groups that could match an empty substring
5648 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5649 runtime if any iteration does actually match an empty substring.
5650
5651 10. Include the RunTest script in the distribution.
5652
5653 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5654 discrepancies, some of which were old and were also with respect to 5.004. They
5655 have now been fixed.
5656
5657
5658 Version 1.09 28-Apr-98
5659 ----------------------
5660
5661 1. A negated single character class followed by a quantifier with a minimum
5662 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5663 program crashes, or just wrong answers. This did not apply to negated classes
5664 containing more than one character, or to minima other than one.
5665
5666
5667 Version 1.08 27-Mar-98
5668 ----------------------
5669
5670 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5671
5672 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5673 latter must appear before anything that relies on it in the pattern.
5674
5675
5676 Version 1.07 16-Feb-98
5677 ----------------------
5678
5679 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5680 repeat of a potentially empty string).
5681
5682
5683 Version 1.06 23-Jan-98
5684 ----------------------
5685
5686 1. Added Markus Oberhumer's little patches for C++.
5687
5688 2. Literal strings longer than 255 characters were broken.
5689
5690
5691 Version 1.05 23-Dec-97
5692 ----------------------
5693
5694 1. Negated character classes containing more than one character were failing if
5695 PCRE_CASELESS was set at run time.
5696
5697
5698 Version 1.04 19-Dec-97
5699 ----------------------
5700
5701 1. Corrected the man page, where some "const" qualifiers had been omitted.
5702
5703 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5704 input syntax.
5705
5706 3. Fixed memory leak which occurred when a regex with back references was
5707 matched with an offsets vector that wasn't big enough. The temporary memory
5708 that is used in this case wasn't being freed if the match failed.
5709
5710 4. Tidied pcretest to ensure it frees memory that it gets.
5711
5712 5. Temporary memory was being obtained in the case where the passed offsets
5713 vector was exactly big enough.
5714
5715 6. Corrected definition of offsetof() from change 5 below.
5716
5717 7. I had screwed up change 6 below and broken the rules for the use of
5718 setjmp(). Now fixed.
5719
5720
5721 Version 1.03 18-Dec-97
5722 ----------------------
5723
5724 1. A erroneous regex with a missing opening parenthesis was correctly
5725 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5726 on some systems.
5727
5728 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5729 it was reported that one broken compiler failed on the former because "code" is
5730 also an independent variable.
5731
5732 3. The erroneous regex a[]b caused an array overrun reference.
5733
5734 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5735 fail on data ending with that character. (It was going on too far, and checking
5736 the next character, typically a binary zero.) This was specific to the
5737 optimized code for single-character negative classes.
5738
5739 5. Added a contributed patch from the TIN world which does the following:
5740
5741 + Add an undef for memmove, in case the the system defines a macro for it.
5742
5743 + Add a definition of offsetof(), in case there isn't one. (I don't know
5744 the reason behind this - offsetof() is part of the ANSI standard - but
5745 it does no harm).
5746
5747 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5748 most of the places where whitespace preceded '#'. I have given up and
5749 allowed the remaining 2 cases to be at the margin.
5750
5751 + Rename some variables in pcre to eliminate shadowing. This seems very
5752 pedantic, but does no harm, of course.
5753
5754 6. Moved the call to setjmp() into its own function, to get rid of warnings
5755 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5756
5757 7. Constructs such as \d{8,} were compiling into the equivalent of
5758 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5759 outcome, but in this particular case used more store than had been allocated,
5760 which caused the bug to be discovered because it threw up an internal error.
5761
5762 8. The debugging code in both pcre and pcretest for outputting the compiled
5763 form of a regex was going wrong in the case of back references followed by
5764 curly-bracketed repeats.
5765
5766
5767 Version 1.02 12-Dec-97
5768 ----------------------
5769
5770 1. Typos in pcre.3 and comments in the source fixed.
5771
5772 2. Applied a contributed patch to get rid of places where it used to remove
5773 'const' from variables, and fixed some signed/unsigned and uninitialized
5774 variable warnings.
5775
5776 3. Added the "runtest" target to Makefile.
5777
5778 4. Set default compiler flag to -O2 rather than just -O.
5779
5780
5781 Version 1.01 19-Nov-97
5782 ----------------------
5783
5784 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5785 like /([ab]*)*/, that is, for classes with more than one character in them.
5786
5787 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5788 as /((?>a*))*/ (a PCRE_EXTRA facility).
5789
5790
5791 Version 1.00 18-Nov-97
5792 ----------------------
5793
5794 1. Added compile-time macros to support systems such as SunOS4 which don't have
5795 memmove() or strerror() but have other things that can be used instead.
5796
5797 2. Arranged that "make clean" removes the executables.
5798
5799
5800 Version 0.99 27-Oct-97
5801 ----------------------
5802
5803 1. Fixed bug in code for optimizing classes with only one character. It was
5804 initializing a 32-byte map regardless, which could cause it to run off the end
5805 of the memory it had got.
5806
5807 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5808
5809
5810 Version 0.98 22-Oct-97
5811 ----------------------
5812
5813 1. Fixed bug in code for handling temporary memory usage when there are more
5814 back references than supplied space in the ovector. This could cause segfaults.
5815
5816
5817 Version 0.97 21-Oct-97
5818 ----------------------
5819
5820 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5821
5822 2. Optimized negated single characters not to use a bit map.
5823
5824 3. Brought error texts together as macro definitions; clarified some of them;
5825 fixed one that was wrong - it said "range out of order" when it meant "invalid
5826 escape sequence".
5827
5828 4. Changed some char * arguments to const char *.
5829
5830 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5831
5832 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5833 pcretest.
5834
5835
5836 Version 0.96 16-Oct-97
5837 ----------------------
5838
5839 1. Added a simple "pgrep" utility to the distribution.
5840
5841 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5842 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5843 where "ddd" means "one or more decimal digits".
5844
5845 3. Fixed serious bug. If a pattern had a back reference, but the call to
5846 pcre_exec() didn't supply a large enough ovector to record the related
5847 identifying subpattern, the match always failed. PCRE now remembers the number
5848 of the largest back reference, and gets some temporary memory in which to save
5849 the offsets during matching if necessary, in order to ensure that
5850 backreferences always work.
5851
5852 4. Increased the compatibility with Perl in a number of ways:
5853
5854 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5855 to request this handling. The option can be set at compile or exec time.
5856
5857 (b) $ matches before a terminating newline by default; an option
5858 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5859 mode). The option can be set at compile or exec time.
5860
5861 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5862 the same as Perl's. If the decimal number it represents is less than 10
5863 or there aren't that many previous left capturing parentheses, an octal
5864 escape is read. Inside a character class, it's always an octal escape,
5865 even if it is a single digit.
5866
5867 (d) An escaped but undefined alphabetic character is taken as a literal,
5868 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5869 escapes.
5870
5871 (e) {0} is now permitted. (The previous item is removed from the compiled
5872 pattern).
5873
5874 5. Changed all the names of code files so that the basic parts are no longer
5875 than 10 characters, and abolished the teeny "globals.c" file.
5876
5877 6. Changed the handling of character classes; they are now done with a 32-byte
5878 bit map always.
5879
5880 7. Added the -d and /D options to pcretest to make it possible to look at the
5881 internals of compilation without having to recompile pcre.
5882
5883
5884 Version 0.95 23-Sep-97
5885 ----------------------
5886
5887 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5888 \x20 at the start of a run of normal characters. These were being treated as
5889 real characters, instead of the source characters being re-checked.
5890
5891
5892 Version 0.94 18-Sep-97
5893 ----------------------
5894
5895 1. The functions are now thread-safe, with the caveat that the global variables
5896 containing pointers to malloc() and free() or alternative functions are the
5897 same for all threads.
5898
5899 2. Get pcre_study() to generate a bitmap of initial characters for non-
5900 anchored patterns when this is possible, and use it if passed to pcre_exec().
5901
5902
5903 Version 0.93 15-Sep-97
5904 ----------------------
5905
5906 1. /(b)|(:+)/ was computing an incorrect first character.
5907
5908 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5909 but not actually doing anything yet.
5910
5911 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5912 as Perl does (e.g. [-az] or [az-]).
5913
5914 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5915 all possible positions.
5916
5917 5. Split up into different modules to avoid including unneeded functions in a
5918 compiled binary. However, compile and exec are still in one module. The "study"
5919 function is split off.
5920
5921 6. The character tables are now in a separate module whose source is generated
5922 by an auxiliary program - but can then be edited by hand if required. There are
5923 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5924 toupper() in the code.
5925
5926 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5927 make them global. Abolish the function for setting them, as the caller can now
5928 set them directly.
5929
5930
5931 Version 0.92 11-Sep-97
5932 ----------------------
5933
5934 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5935 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5936
5937 2. Caseless matching was not working in character classes if the characters in
5938 the pattern were in upper case.
5939
5940 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5941
5942 4. Make PCRE_ANCHORED public and accept as a compile option.
5943
5944 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5945 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5946 pass them.
5947
5948 6. Give an error if bad option bits passed at compile or run time.
5949
5950 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5951 pcretest to cause it to pass that flag.
5952
5953 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5954 options, and the first character, if set.
5955
5956 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5957
5958
5959 Version 0.91 10-Sep-97
5960 ----------------------
5961
5962 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5963 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5964
5965 2. PCRE was looping on encountering an indefinitely repeated back reference to
5966 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5967 Perl does - treats the match as successful.
5968
5969 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5