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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1651 - (show annotations)
Mon May 23 15:06:35 2016 UTC (3 years, 4 months ago) by ph10
File size: 289326 byte(s)
File tidies for 8.39-RC1.
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 21-May-2016
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 14. And yet another buffer overflow bug involving duplicate named groups, this
58 time nested, with a nested back reference. Yet again, I have just allowed
59 for more memory, because anything more needs all the refactoring that has
60 been done for PCRE2. An example pattern that provoked this bug is:
61 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
62 registered as CVE-2016-1283.
63
64 15. pcretest went into a loop if global matching was requested with an ovector
65 size less than 2. It now gives an error message. This bug was found by
66 afl-fuzz.
67
68 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
69 ("assertion expected") when (?(?C) was not followed by an opening
70 parenthesis.
71
72 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
73 actually affect anything, by sheer luck.
74
75 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
76 static compilation.
77
78 19. Modified the RunTest script to incorporate a valgrind suppressions file so
79 that certain errors, provoked by the SSE2 instruction set when JIT is used,
80 are ignored.
81
82 20. A racing condition is fixed in JIT reported by Mozilla.
83
84 21. Minor code refactor to avoid "array subscript is below array bounds"
85 compiler warning.
86
87 22. Minor code refactor to avoid "left shift of negative number" warning.
88
89 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
90 UCP support.
91
92
93 Version 8.38 23-November-2015
94 -----------------------------
95
96 1. If a group that contained a recursive back reference also contained a
97 forward reference subroutine call followed by a non-forward-reference
98 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
99 compile correct code, leading to undefined behaviour or an internally
100 detected error. This bug was discovered by the LLVM fuzzer.
101
102 2. Quantification of certain items (e.g. atomic back references) could cause
103 incorrect code to be compiled when recursive forward references were
104 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
105 This bug was discovered by the LLVM fuzzer.
106
107 3. A repeated conditional group whose condition was a reference by name caused
108 a buffer overflow if there was more than one group with the given name.
109 This bug was discovered by the LLVM fuzzer.
110
111 4. A recursive back reference by name within a group that had the same name as
112 another group caused a buffer overflow. For example:
113 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
114
115 5. A forward reference by name to a group whose number is the same as the
116 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
117 a buffer overflow at compile time. This bug was discovered by the LLVM
118 fuzzer.
119
120 6. A lookbehind assertion within a set of mutually recursive subpatterns could
121 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
122
123 7. Another buffer overflow bug involved duplicate named groups with a
124 reference between their definition, with a group that reset capture
125 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
126 fixed by always allowing for more memory, even if not needed. (A proper fix
127 is implemented in PCRE2, but it involves more refactoring.)
128
129 8. There was no check for integer overflow in subroutine calls such as (?123).
130
131 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
132 being treated as a literal 'l' instead of causing an error.
133
134 10. There was a buffer overflow if pcre_exec() was called with an ovector of
135 size 1. This bug was found by american fuzzy lop.
136
137 11. If a non-capturing group containing a conditional group that could match
138 an empty string was repeated, it was not identified as matching an empty
139 string itself. For example: /^(?:(?(1)x|)+)+$()/.
140
141 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
142 \a and \e in test subject lines.
143
144 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
145 instead of the EBCDIC value.
146
147 14. The handling of \c in an EBCDIC environment has been revised so that it is
148 now compatible with the specification in Perl's perlebcdic page.
149
150 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
151 ASCII/Unicode. This has now been added to the list of characters that are
152 recognized as white space in EBCDIC.
153
154 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
155 error (correctly) when used outside a class, but did not give an error
156 within a class.
157
158 17. \h within a class was incorrectly compiled in EBCDIC environments.
159
160 18. A pattern with an unmatched closing parenthesis that contained a backward
161 assertion which itself contained a forward reference caused buffer
162 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
163
164 19. JIT should return with error when the compiled pattern requires more stack
165 space than the maximum.
166
167 20. A possessively repeated conditional group that could match an empty string,
168 for example, /(?(R))*+/, was incorrectly compiled.
169
170 21. Fix infinite recursion in the JIT compiler when certain patterns such as
171 /(?:|a|){100}x/ are analysed.
172
173 22. Some patterns with character classes involving [: and \\ were incorrectly
174 compiled and could cause reading from uninitialized memory or an incorrect
175 error diagnosis.
176
177 23. Pathological patterns containing many nested occurrences of [: caused
178 pcre_compile() to run for a very long time.
179
180 24. A conditional group with only one branch has an implicit empty alternative
181 branch and must therefore be treated as potentially matching an empty
182 string.
183
184 25. If (?R was followed by - or + incorrect behaviour happened instead of a
185 diagnostic.
186
187 26. Arrange to give up on finding the minimum matching length for overly
188 complex patterns.
189
190 27. Similar to (4) above: in a pattern with duplicated named groups and an
191 occurrence of (?| it is possible for an apparently non-recursive back
192 reference to become recursive if a later named group with the relevant
193 number is encountered. This could lead to a buffer overflow. Wen Guanxing
194 from Venustech ADLAB discovered this bug.
195
196 28. If pcregrep was given the -q option with -c or -l, or when handling a
197 binary file, it incorrectly wrote output to stdout.
198
199 29. The JIT compiler did not restore the control verb head in case of *THEN
200 control verbs. This issue was found by Karl Skomski with a custom LLVM
201 fuzzer.
202
203 30. Error messages for syntax errors following \g and \k were giving inaccurate
204 offsets in the pattern.
205
206 31. Added a check for integer overflow in conditions (?(<digits>) and
207 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
208 fuzzer.
209
210 32. Handling recursive references such as (?2) when the reference is to a group
211 later in the pattern uses code that is very hacked about and error-prone.
212 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
213 give an internal error if it is obvious that compiling has gone wrong.
214
215 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
216 This issue was found by Karl Skomski with a custom LLVM fuzzer.
217
218 34. The JIT compiler should restore the control chain for empty possessive
219 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
220
221 35. Match limit check added to JIT recursion. This issue was found by Karl
222 Skomski with a custom LLVM fuzzer.
223
224 36. Yet another case similar to 27 above has been circumvented by an
225 unconditional allocation of extra memory. This issue is fixed "properly" in
226 PCRE2 by refactoring the way references are handled. Wen Guanxing
227 from Venustech ADLAB discovered this bug.
228
229 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
230 with a custom LLVM fuzzer.
231
232 38. Fixed a corner case of range optimization in JIT.
233
234 39. An incorrect error "overran compiling workspace" was given if there were
235 exactly enough group forward references such that the last one extended
236 into the workspace safety margin. The next one would have expanded the
237 workspace. The test for overflow was not including the safety margin.
238
239 40. A match limit issue is fixed in JIT which was found by Karl Skomski
240 with a custom LLVM fuzzer.
241
242 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
243 work under Windows. (Why has it taken so long for anyone to notice?)
244
245 42. In a character class such as [\W\p{Any}] where both a negative-type escape
246 ("not a word character") and a property escape were present, the property
247 escape was being ignored.
248
249 43. Fix crash caused by very long (*MARK) or (*THEN) names.
250
251 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
252 by a single ASCII character in a class item, was incorrectly compiled in
253 UCP mode. The POSIX class got lost, but only if the single character
254 followed it.
255
256 45. [:punct:] in UCP mode was matching some characters in the range 128-255
257 that should not have been matched.
258
259 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
260 class, all characters with code points greater than 255 are in the class.
261 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
262 such as \w are turned into Unicode properties), wide characters were not
263 correctly handled, and could fail to match.
264
265
266 Version 8.37 28-April-2015
267 --------------------------
268
269 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
270 for those parentheses to be closed with whatever has been captured so far.
271 However, it was failing to mark any other groups between the hightest
272 capture so far and the currrent group as "unset". Thus, the ovector for
273 those groups contained whatever was previously there. An example is the
274 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
275
276 2. If an assertion condition was quantified with a minimum of zero (an odd
277 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
278
279 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
280 unrecognized modifier, a crash could occur.
281
282 4. An attempt to do global matching in pcretest with a zero-length ovector
283 caused a crash.
284
285 5. Fixed a memory leak during matching that could occur for a subpattern
286 subroutine call (recursive or otherwise) if the number of captured groups
287 that had to be saved was greater than ten.
288
289 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
290 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
291 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
292
293 7. A UTF pattern containing a "not" match of a non-ASCII character and a
294 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
295
296 8. When a pattern is compiled, it remembers the highest back reference so that
297 when matching, if the ovector is too small, extra memory can be obtained to
298 use instead. A conditional subpattern whose condition is a check on a
299 capture having happened, such as, for example in the pattern
300 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
301 setting the highest backreference number. This mattered only if pcre_exec()
302 was called with an ovector that was too small to hold the capture, and there
303 was no other kind of back reference (a situation which is probably quite
304 rare). The effect of the bug was that the condition was always treated as
305 FALSE when the capture could not be consulted, leading to a incorrect
306 behaviour by pcre_exec(). This bug has been fixed.
307
308 9. A reference to a duplicated named group (either a back reference or a test
309 for being set in a conditional) that occurred in a part of the pattern where
310 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
311 to be incorrectly calculated, leading to overwriting.
312
313 10. A mutually recursive set of back references such as (\2)(\1) caused a
314 segfault at study time (while trying to find the minimum matching length).
315 The infinite loop is now broken (with the minimum length unset, that is,
316 zero).
317
318 11. If an assertion that was used as a condition was quantified with a minimum
319 of zero, matching went wrong. In particular, if the whole group had
320 unlimited repetition and could match an empty string, a segfault was
321 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
322 assertions to be quantified, but not if they are being used as conditions,
323 so the above pattern is faulted by Perl. PCRE has now been changed so that
324 it also rejects such patterns.
325
326 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
327 failed to allow the zero-repeat case if pcre2_exec() was called with an
328 ovector too small to capture the group.
329
330 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
331 Red Hat Product Security:
332
333 (a) A crash if /K and /F were both set with the option to save the compiled
334 pattern.
335
336 (b) Another crash if the option to print captured substrings in a callout
337 was combined with setting a null ovector, for example \O\C+ as a subject
338 string.
339
340 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
341 forward reference repeated a large (but limited) number of times within a
342 repeated outer group that has a zero minimum quantifier, caused incorrect
343 code to be compiled, leading to the error "internal error:
344 previously-checked referenced subpattern not found" when an incorrect
345 memory address was read. This bug was reported as "heap overflow",
346 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
347 CVE-2015-2325.
348
349 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
350 call within a group that also contained a recursive back reference caused
351 incorrect code to be compiled. This bug was reported as "heap overflow",
352 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
353 number CVE-2015-2326.
354
355 24. Computing the size of the JIT read-only data in advance has been a source
356 of various issues, and new ones are still appear unfortunately. To fix
357 existing and future issues, size computation is eliminated from the code,
358 and replaced by on-demand memory allocation.
359
360 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
361 adjacent to the end of the range, and the range contained characters with
362 more than one other case, caused incorrect behaviour when compiled in UTF
363 mode. In that example, the range a-j was left out of the class.
364
365 26. Fix JIT compilation of conditional blocks, which assertion
366 is converted to (*FAIL). E.g: /(?(?!))/.
367
368 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
369 discovered by the LLVM fuzzer.
370
371 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
372 when this assertion was used as a condition, for example (?(?!)a|b). In
373 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
374 error about an unsupported item.
375
376 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
377 possessification code could take exponential time to complete. A recursion
378 depth limit of 1000 has been imposed to limit the resources used by this
379 optimization.
380
381 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
382 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
383 because \S ensures they are all in the class. The code for doing this was
384 interacting badly with the code for computing the amount of space needed to
385 compile the pattern, leading to a buffer overflow. This bug was discovered
386 by the LLVM fuzzer.
387
388 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
389 other kinds of group caused stack overflow at compile time. This bug was
390 discovered by the LLVM fuzzer.
391
392 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
393 between a subroutine call and its quantifier was incorrectly compiled,
394 leading to buffer overflow or other errors. This bug was discovered by the
395 LLVM fuzzer.
396
397 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
398 assertion after (?(. The code was failing to check the character after
399 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
400 was discovered by the LLVM fuzzer.
401
402 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
403 a fixed maximum following a group that contains a subroutine reference was
404 incorrectly compiled and could trigger buffer overflow. This bug was
405 discovered by the LLVM fuzzer.
406
407 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
408 caused a stack overflow instead of the diagnosis of a non-fixed length
409 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
410
411 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
412 (e.g. /(?<=\Ka)/) could make pcregrep loop.
413
414 37. There was a similar problem to 36 in pcretest for global matches.
415
416 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
417 and a subsequent item in the pattern caused a non-match, backtracking over
418 the repeated \X did not stop, but carried on past the start of the subject,
419 causing reference to random memory and/or a segfault. There were also some
420 other cases where backtracking after \C could crash. This set of bugs was
421 discovered by the LLVM fuzzer.
422
423 39. The function for finding the minimum length of a matching string could take
424 a very long time if mutual recursion was present many times in a pattern,
425 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
426 method has been implemented. This infelicity was discovered by the LLVM
427 fuzzer.
428
429 40. Static linking against the PCRE library using the pkg-config module was
430 failing on missing pthread symbols.
431
432
433 Version 8.36 26-September-2014
434 ------------------------------
435
436 1. Got rid of some compiler warnings in the C++ modules that were shown up by
437 -Wmissing-field-initializers and -Wunused-parameter.
438
439 2. The tests for quantifiers being too big (greater than 65535) were being
440 applied after reading the number, and stupidly assuming that integer
441 overflow would give a negative number. The tests are now applied as the
442 numbers are read.
443
444 3. Tidy code in pcre_exec.c where two branches that used to be different are
445 now the same.
446
447 4. The JIT compiler did not generate match limit checks for certain
448 bracketed expressions with quantifiers. This may lead to exponential
449 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
450 issue should be resolved now.
451
452 5. Fixed an issue, which occures when nested alternatives are optimized
453 with table jumps.
454
455 6. Inserted two casts and changed some ints to size_t in the light of some
456 reported 64-bit compiler warnings (Bugzilla 1477).
457
458 7. Fixed a bug concerned with zero-minimum possessive groups that could match
459 an empty string, which sometimes were behaving incorrectly in the
460 interpreter (though correctly in the JIT matcher). This pcretest input is
461 an example:
462
463 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
464 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
465
466 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
467 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
468 for an empty string was breaking the inner loop and carrying on at a lower
469 level, when possessive repeated groups should always return to a higher
470 level as they have no backtrack points in them. The empty string test now
471 occurs at the outer level.
472
473 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
474 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
475
476 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
477 doesn't).
478
479 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
480 to hold a pattern to be incorrectly computed (too small) when there were
481 named back references to duplicated names. This could cause "internal
482 error: code overflow" or "double free or corruption" or other memory
483 handling errors.
484
485 11. When named subpatterns had the same prefixes, back references could be
486 confused. For example, in this pattern:
487
488 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
489
490 the reference to 'Name' was incorrectly treated as a reference to a
491 duplicate name.
492
493 12. A pattern such as /^s?c/mi8 where the optional character has more than
494 one "other case" was incorrectly compiled such that it would only try to
495 match starting at "c".
496
497 13. When a pattern starting with \s was studied, VT was not included in the
498 list of possible starting characters; this should have been part of the
499 8.34/18 patch.
500
501 14. If a character class started [\Qx]... where x is any character, the class
502 was incorrectly terminated at the ].
503
504 15. If a pattern that started with a caseless match for a character with more
505 than one "other case" was studied, PCRE did not set up the starting code
506 unit bit map for the list of possible characters. Now it does. This is an
507 optimization improvement, not a bug fix.
508
509 16. The Unicode data tables have been updated to Unicode 7.0.0.
510
511 17. Fixed a number of memory leaks in pcregrep.
512
513 18. Avoid a compiler warning (from some compilers) for a function call with
514 a cast that removes "const" from an lvalue by using an intermediate
515 variable (to which the compiler does not object).
516
517 19. Incorrect code was compiled if a group that contained an internal recursive
518 back reference was optional (had quantifier with a minimum of zero). This
519 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
520 caused segmentation faults because of stack overflows at compile time.
521
522 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
523 group that is quantified with an indefinite repeat, caused a compile-time
524 loop which used up all the system stack and provoked a segmentation fault.
525 This was not the same bug as 19 above.
526
527 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
528 Patch by Mike Frysinger.
529
530
531 Version 8.35 04-April-2014
532 --------------------------
533
534 1. A new flag is set, when property checks are present in an XCLASS.
535 When this flag is not set, PCRE can perform certain optimizations
536 such as studying these XCLASS-es.
537
538 2. The auto-possessification of character sets were improved: a normal
539 and an extended character set can be compared now. Furthermore
540 the JIT compiler optimizes more character set checks.
541
542 3. Got rid of some compiler warnings for potentially uninitialized variables
543 that show up only when compiled with -O2.
544
545 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
546 of a match later then the end of the match. The pcretest program was not
547 handling the case sensibly - it was outputting from the start to the next
548 binary zero. It now reports this situation in a message, and outputs the
549 text from the end to the start.
550
551 5. Fast forward search is improved in JIT. Instead of the first three
552 characters, any three characters with fixed position can be searched.
553 Search order: first, last, middle.
554
555 6. Improve character range checks in JIT. Characters are read by an inprecise
556 function now, which returns with an unknown value if the character code is
557 above a certain threshold (e.g: 256). The only limitation is that the value
558 must be bigger than the threshold as well. This function is useful when
559 the characters above the threshold are handled in the same way.
560
561 7. The macros whose names start with RAWUCHAR are placeholders for a future
562 mode in which only the bottom 21 bits of 32-bit data items are used. To
563 make this more memorable for those maintaining the code, the names have
564 been changed to start with UCHAR21, and an extensive comment has been added
565 to their definition.
566
567 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
568 to the export list in Makefile.am (they were accidentally omitted from the
569 8.34 tarball).
570
571 9. The informational output from pcretest used the phrase "starting byte set"
572 which is inappropriate for the 16-bit and 32-bit libraries. As the output
573 for "first char" and "need char" really means "non-UTF-char", I've changed
574 "byte" to "char", and slightly reworded the output. The documentation about
575 these values has also been (I hope) clarified.
576
577 10. Another JIT related optimization: use table jumps for selecting the correct
578 backtracking path, when more than four alternatives are present inside a
579 bracket.
580
581 11. Empty match is not possible, when the minimum length is greater than zero,
582 and there is no \K in the pattern. JIT should avoid empty match checks in
583 such cases.
584
585 12. In a caseless character class with UCP support, when a character with more
586 than one alternative case was not the first character of a range, not all
587 the alternative cases were added to the class. For example, s and \x{17f}
588 are both alternative cases for S: the class [RST] was handled correctly,
589 but [R-T] was not.
590
591 13. The configure.ac file always checked for pthread support when JIT was
592 enabled. This is not used in Windows, so I have put this test inside a
593 check for the presence of windows.h (which was already tested for).
594
595 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
596 The algorithm provides a way to skip certain starting offsets, and usually
597 faster than linear prefix searches.
598
599 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
600 as for 'fr_FR' and 'french'. For some reason, however, it then used the
601 Windows-specific input and output files, which have 'french' screwed in.
602 So this could never have worked. One of the problems with locales is that
603 they aren't always the same. I have now updated RunTest so that it checks
604 the output of the locale test (test 3) against three different output
605 files, and it allows the test to pass if any one of them matches. With luck
606 this should make the test pass on some versions of Solaris where it was
607 failing. Because of the uncertainty, the script did not used to stop if
608 test 3 failed; it now does. If further versions of a French locale ever
609 come to light, they can now easily be added.
610
611 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
612 there was a message during ./configure, but it did not stop. This now
613 provokes an error. The invalid example in README has been corrected.
614 If a value less than the minimum is given, the minimum value has always
615 been used, but now a warning is given.
616
617 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
618 was a bug in the test system, which is now fixed. Also, the list of various
619 configurations that are tested for each release did not have one with both
620 16/32 bits and --enable-bar-anycrlf. It now does.
621
622 18. pcretest was missing "-C bsr" for displaying the \R default setting.
623
624 19. Little endian PowerPC systems are supported now by the JIT compiler.
625
626 20. The fast forward newline mechanism could enter to an infinite loop on
627 certain invalid UTF-8 input. Although we don't support these cases
628 this issue can be fixed by a performance optimization.
629
630 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
631 not take account if existing stack usage. There is now a new global
632 variable called pcre_stack_guard that can be set to point to an external
633 function to check stack availability. It is called at the start of
634 processing every parenthesized group.
635
636 22. A typo in the code meant that in ungreedy mode the max/min qualifier
637 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
638 not match "ab".
639
640 23. When UTF was disabled, the JIT program reported some incorrect compile
641 errors. These messages are silenced now.
642
643 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
644 compiler.
645
646 25. Change all the temporary files used in RunGrepTest to be different to those
647 used by RunTest so that the tests can be run simultaneously, for example by
648 "make -j check".
649
650
651 Version 8.34 15-December-2013
652 -----------------------------
653
654 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
655 executable memory. Patch inspired by Carsten Klein.
656
657 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
658 this macro is never tested and has no effect, because the work to support
659 coverage involves only compiling and linking options and special targets in
660 the Makefile. The comment in config.h implied that defining the macro would
661 enable coverage support, which is totally false. There was also support for
662 setting this macro in the CMake files (my fault, I just copied it from
663 configure). SUPPORT_GCOV has now been removed.
664
665 3. Make a small performance improvement in strlen16() and strlen32() in
666 pcretest.
667
668 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
669 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
670 these cases). There was also one in pcretest.c.
671
672 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
673
674 6. In UTF mode, the code for checking whether a group could match an empty
675 string (which is used for indefinitely repeated groups to allow for
676 breaking an infinite loop) was broken when the group contained a repeated
677 negated single-character class with a character that occupied more than one
678 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
679 UTF-8 mode). The effect was undefined: the group might or might not be
680 deemed as matching an empty string, or the program might have crashed.
681
682 7. The code for checking whether a group could match an empty string was not
683 recognizing that \h, \H, \v, \V, and \R must match a character.
684
685 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
686 an empty string. If it can, pcretest shows this in its information output.
687
688 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
689 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
690 matched by pcre_exec() without using JIT:
691
692 (a) If the rest of the pattern did not match after a maximal run of
693 grapheme clusters, the code for backing up to try with fewer of them
694 did not always back up over a full grapheme when characters that do not
695 have the modifier quality were involved, e.g. Hangul syllables.
696
697 (b) If the match point in a subject started with modifier character, and
698 there was no match, the code could incorrectly back up beyond the match
699 point, and potentially beyond the first character in the subject,
700 leading to a segfault or an incorrect match result.
701
702 10. A conditional group with an assertion condition could lead to PCRE
703 recording an incorrect first data item for a match if no other first data
704 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
705 first data item, and therefore matched "ca" after "c" instead of at the
706 start.
707
708 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
709 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
710
711 12. The source of pcregrep now includes z/OS-specific code so that it can be
712 compiled for z/OS as part of the special z/OS distribution.
713
714 13. Added the -T and -TM options to pcretest.
715
716 14. The code in pcre_compile.c for creating the table of named capturing groups
717 has been refactored. Instead of creating the table dynamically during the
718 actual compiling pass, the information is remembered during the pre-compile
719 pass (on the stack unless there are more than 20 named groups, in which
720 case malloc() is used) and the whole table is created before the actual
721 compile happens. This has simplified the code (it is now nearly 150 lines
722 shorter) and prepared the way for better handling of references to groups
723 with duplicate names.
724
725 15. A back reference to a named subpattern when there is more than one of the
726 same name now checks them in the order in which they appear in the pattern.
727 The first one that is set is used for the reference. Previously only the
728 first one was inspected. This change makes PCRE more compatible with Perl.
729
730 16. Unicode character properties were updated from Unicode 6.3.0.
731
732 17. The compile-time code for auto-possessification has been refactored, based
733 on a patch by Zoltan Herczeg. It now happens after instead of during
734 compilation. The code is cleaner, and more cases are handled. The option
735 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
736 options in pcretest are provided to set it. It can also be set by
737 (*NO_AUTO_POSSESS) at the start of a pattern.
738
739 18. The character VT has been added to the default ("C" locale) set of
740 characters that match \s and are generally treated as white space,
741 following this same change in Perl 5.18. There is now no difference between
742 "Perl space" and "POSIX space". Whether VT is treated as white space in
743 other locales depends on the locale.
744
745 19. The code for checking named groups as conditions, either for being set or
746 for being recursed, has been refactored (this is related to 14 and 15
747 above). Processing unduplicated named groups should now be as fast at
748 numerical groups, and processing duplicated groups should be faster than
749 before.
750
751 20. Two patches to the CMake build system, by Alexander Barkov:
752
753 (1) Replace the "source" command by "." in CMakeLists.txt because
754 "source" is a bash-ism.
755
756 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
757 without these the CMake build does not work on Solaris.
758
759 21. Perl has changed its handling of \8 and \9. If there is no previously
760 encountered capturing group of those numbers, they are treated as the
761 literal characters 8 and 9 instead of a binary zero followed by the
762 literals. PCRE now does the same.
763
764 22. Following Perl, added \o{} to specify codepoints in octal, making it
765 possible to specify values greater than 0777 and also making them
766 unambiguous.
767
768 23. Perl now gives an error for missing closing braces after \x{... instead of
769 treating the string as literal. PCRE now does the same.
770
771 24. RunTest used to grumble if an inappropriate test was selected explicitly,
772 but just skip it when running all tests. This make it awkward to run ranges
773 of tests when one of them was inappropriate. Now it just skips any
774 inappropriate tests, as it always did when running all tests.
775
776 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
777 character types such as \d or \w, too many callouts were inserted, and the
778 data that they returned was rubbish.
779
780 26. In UCP mode, \s was not matching two of the characters that Perl matches,
781 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
782 were matched by \h. The code has now been refactored so that the lists of
783 the horizontal and vertical whitespace characters used for \h and \v (which
784 are defined only in one place) are now also used for \s.
785
786 27. Add JIT support for the 64 bit TileGX architecture.
787 Patch by Jiong Wang (Tilera Corporation).
788
789 28. Possessive quantifiers for classes (both explicit and automatically
790 generated) now use special opcodes instead of wrapping in ONCE brackets.
791
792 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
793 (because it's meaningless), this was not happening when PCRE_CASELESS was
794 set. Not wrong, but inefficient.
795
796 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
797 properties for \w, \d, etc) is present in a test regex. Otherwise if the
798 test contains no characters greater than 255, Perl doesn't realise it
799 should be using Unicode semantics.
800
801 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
802 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
803 does in Unicode mode.
804
805 32. Added the "forbid" facility to pcretest so that putting tests into the
806 wrong test files can sometimes be quickly detected.
807
808 33. There is now a limit (default 250) on the depth of nesting of parentheses.
809 This limit is imposed to control the amount of system stack used at compile
810 time. It can be changed at build time by --with-parens-nest-limit=xxx or
811 the equivalent in CMake.
812
813 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
814 errors. Perl warns for these when in warning mode, but PCRE has no facility
815 for giving warnings.
816
817 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
818 However, this was not working for (?!) because it is optimized to (*FAIL),
819 for which PCRE does not allow quantifiers. The optimization is now disabled
820 when a quantifier follows (?!). I can't see any use for this, but it makes
821 things uniform.
822
823 36. Perl no longer allows group names to start with digits, so I have made this
824 change also in PCRE. It simplifies the code a bit.
825
826 37. In extended mode, Perl ignores spaces before a + that indicates a
827 possessive quantifier. PCRE allowed a space before the quantifier, but not
828 before the possessive +. It now does.
829
830 38. The use of \K (reset reported match start) within a repeated possessive
831 group such as (a\Kb)*+ was not working.
832
833 40. Document that the same character tables must be used at compile time and
834 run time, and that the facility to pass tables to pcre_exec() and
835 pcre_dfa_exec() is for use only with saved/restored patterns.
836
837 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
838 features for Builds with MSVC:
839
840 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
841 builds). The .rc files can be used to set FileDescription and many other
842 attributes.
843
844 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
845 This allows higher-level build scripts which want .pdb files to avoid
846 hard-coding the exact files needed."
847
848 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
849 mean "start of word" and "end of word", respectively, as a transition aid.
850
851 43. A minimizing repeat of a class containing codepoints greater than 255 in
852 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
853 compiled to use the heap for recursion.
854
855 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
856 is configured.
857
858
859 Version 8.33 28-May-2013
860 ------------------------
861
862 1. Added 'U' to some constants that are compared to unsigned integers, to
863 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
864 variables that are added to signed variables, to ensure the result is
865 signed and can be negated.
866
867 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
868 CMake config files.
869
870 3. Revise the creation of config.h.generic so that all boolean macros are
871 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
872 overriding via -D on the command line possible.
873
874 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
875 to unsigned int is reported to make a quite noticeable speed difference in
876 a specific Windows environment. Testing on Linux did also appear to show
877 some benefit (and it is clearly not harmful). Also fixed the definition of
878 Xop which should be unsigned.
879
880 5. Related to (4), changing the definition of the intermediate variable cc
881 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
882 improvements.
883
884 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
885 unnecessary spaces.
886
887 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
888 and later.
889
890 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
891
892 9. Optimizing fast_forward_start_bits in JIT.
893
894 10. Adding support for callouts in JIT, and fixing some issues revealed
895 during this work. Namely:
896
897 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
898
899 (b) Minimum length was not checked before the matching is started.
900
901 11. The value of capture_last that is passed to callouts was incorrect in some
902 cases when there was a capture on one path that was subsequently abandoned
903 after a backtrack. Also, the capture_last value is now reset after a
904 recursion, since all captures are also reset in this case.
905
906 12. The interpreter no longer returns the "too many substrings" error in the
907 case when an overflowing capture is in a branch that is subsequently
908 abandoned after a backtrack.
909
910 13. In the pathological case when an offset vector of size 2 is used, pcretest
911 now prints out the matched string after a yield of 0 or 1.
912
913 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
914 Only supported by the JIT compiler at the moment.
915
916 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
917
918 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
919 offset of the starting point of the matching process, provided the offsets
920 vector is large enough.
921
922 17. The \A escape now records a lookbehind value of 1, though its execution
923 does not actually inspect the previous character. This is to ensure that,
924 in partial multi-segment matching, at least one character from the old
925 segment is retained when a new segment is processed. Otherwise, if there
926 are no lookbehinds in the pattern, \A might match incorrectly at the start
927 of a new segment.
928
929 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
930
931 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
932 gives some modest performance improvement in 8-bit mode.
933
934 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
935 be expressed in certain programming languages using Universal Character
936 Names.
937
938 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
939 which points out that "non characters" are not "characters that may not
940 appear in Unicode strings" but rather "characters that are reserved for
941 internal use and have only local meaning".
942
943 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
944 there was a conditional group that depended on an assertion, if the
945 assertion was false, the callout that immediately followed the alternation
946 in the condition was skipped when pcre_exec() was used for matching.
947
948 23. Allow an explicit callout to be inserted before an assertion that is the
949 condition for a conditional group, for compatibility with automatic
950 callouts, which always insert a callout at this point.
951
952 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
953 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
954
955 25. (*PRUNE) is now supported by the JIT compiler.
956
957 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
958
959 27. Fix the case where there are two or more SKIPs with arguments that may be
960 ignored.
961
962 28. (*SKIP) is now supported by the JIT compiler.
963
964 29. (*THEN) is now supported by the JIT compiler.
965
966 30. Update RunTest with additional test selector options.
967
968 31. The way PCRE handles backtracking verbs has been changed in two ways.
969
970 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
971 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
972 backtracking. In some cases this makes it more Perl-compatible, but Perl's
973 rather obscure rules do not always do the same thing.
974
975 (2) Previously, backtracking verbs were confined within assertions. This is
976 no longer the case for positive assertions, except for (*ACCEPT). Again,
977 this sometimes improves Perl compatibility, and sometimes does not.
978
979 32. A number of tests that were in test 2 because Perl did things differently
980 have been moved to test 1, because either Perl or PCRE has changed, and
981 these tests are now compatible.
982
983 32. Backtracking control verbs are now handled in the same way in JIT and
984 interpreter.
985
986 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
987 contained a forward subroutine reference caused a compile error.
988
989 34. Auto-detect and optimize limited repetitions in JIT.
990
991 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
992 blocking (*UTF) etc.
993
994 36. In the interpreter, maximizing pattern repetitions for characters and
995 character types now use tail recursion, which reduces stack usage.
996
997 37. The value of the max lookbehind was not correctly preserved if a compiled
998 and saved regex was reloaded on a host of different endianness.
999
1000 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1001 of the compiled pattern block, expand the flags field from 16 to 32 bits
1002 because it was almost full.
1003
1004 39. Try madvise first before posix_madvise.
1005
1006 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1007 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1008 original change locked out all matches of empty strings. This has been
1009 changed so that one match of an empty string per line is recognized.
1010 Subsequent searches on the same line (for colouring or for --only-matching,
1011 for example) do not recognize empty strings.
1012
1013 41. Applied a user patch to fix a number of spelling mistakes in comments.
1014
1015 42. Data lines longer than 65536 caused pcretest to crash.
1016
1017 43. Clarified the data type for length and startoffset arguments for pcre_exec
1018 and pcre_dfa_exec in the function-specific man pages, where they were
1019 explicitly stated to be in bytes, never having been updated. I also added
1020 some clarification to the pcreapi man page.
1021
1022 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1023 a segmentation fault.
1024
1025
1026 Version 8.32 30-November-2012
1027 -----------------------------
1028
1029 1. Improved JIT compiler optimizations for first character search and single
1030 character iterators.
1031
1032 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1033 Patch by Daniel Richard G.
1034
1035 3. Single character iterator optimizations in the JIT compiler.
1036
1037 4. Improved JIT compiler optimizations for character ranges.
1038
1039 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1040 Reported by Giuseppe D'Angelo.
1041
1042 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1043 of a line, was being set incorrectly in cases where .* appeared inside
1044 atomic brackets at the start of a pattern, or where there was a subsequent
1045 *PRUNE or *SKIP.
1046
1047 7. Improved instruction cache flush for POWER/PowerPC.
1048 Patch by Daniel Richard G.
1049
1050 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1051 grep:
1052
1053 (a) There is now no limit to the number of patterns to be matched.
1054
1055 (b) An error is given if a pattern is too long.
1056
1057 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1058 are now supported.
1059
1060 (d) --exclude-from and --include-from (multiple use) have been added.
1061
1062 (e) Exclusions and inclusions now apply to all files and directories, not
1063 just to those obtained from scanning a directory recursively.
1064
1065 (f) Multiple uses of -f and --file-list are now supported.
1066
1067 (g) In a Windows environment, the default for -d has been changed from
1068 "read" (the GNU grep default) to "skip", because otherwise the presence
1069 of a directory in the file list provokes an error.
1070
1071 (h) The documentation has been revised and clarified in places.
1072
1073 9. Improve the matching speed of capturing brackets.
1074
1075 10. Changed the meaning of \X so that it now matches a Unicode extended
1076 grapheme cluster.
1077
1078 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1079 out POSIX threads when JIT support is configured.
1080
1081 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1082
1083 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1084 structure could go wrong in environments where size_t is not the same size
1085 as int.
1086
1087 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1088 set.
1089
1090 15. The EBCDIC support had decayed; later updates to the code had included
1091 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1092 general tidy up of EBCDIC-related issues, and the documentation was also
1093 not quite right. There is now a test that can be run on ASCII systems to
1094 check some of the EBCDIC-related things (but is it not a full test).
1095
1096 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1097 in a small tidy to the code.
1098
1099 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1100
1101 18. If the --only-matching (-o) option in pcregrep is specified multiple
1102 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1103 substrings matched by the 1st and 2nd capturing parentheses. A separating
1104 string can be specified by --om-separator (default empty).
1105
1106 19. Improving the first n character searches.
1107
1108 20. Turn case lists for horizontal and vertical white space into macros so that
1109 they are defined only once.
1110
1111 21. This set of changes together give more compatible Unicode case-folding
1112 behaviour for characters that have more than one other case when UCP
1113 support is available.
1114
1115 (a) The Unicode property table now has offsets into a new table of sets of
1116 three or more characters that are case-equivalent. The MultiStage2.py
1117 script that generates these tables (the pcre_ucd.c file) now scans
1118 CaseFolding.txt instead of UnicodeData.txt for character case
1119 information.
1120
1121 (b) The code for adding characters or ranges of characters to a character
1122 class has been abstracted into a generalized function that also handles
1123 case-independence. In UTF-mode with UCP support, this uses the new data
1124 to handle characters with more than one other case.
1125
1126 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1127 whose other case is greater than 256 are now correctly matched
1128 caselessly. Previously, the high codepoint matched the low one, but not
1129 vice versa.
1130
1131 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1132 of the new class addition function, using character lists defined as
1133 macros alongside the case definitions of 20 above.
1134
1135 (e) Caseless back references now work with characters that have more than
1136 one other case.
1137
1138 (f) General caseless matching of characters with more than one other case
1139 is supported.
1140
1141 22. Unicode character properties were updated from Unicode 6.2.0
1142
1143 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1144
1145 24. Add support for 32-bit character strings, and UTF-32
1146
1147 25. Major JIT compiler update (code refactoring and bugfixing).
1148 Experimental Sparc 32 support is added.
1149
1150 26. Applied a modified version of Daniel Richard G's patch to create
1151 pcre.h.generic and config.h.generic by "make" instead of in the
1152 PrepareRelease script.
1153
1154 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1155 pcre_compile.c when checking for a zero character.
1156
1157 28. Introducing a native interface for JIT. Through this interface, the compiled
1158 machine code can be directly executed. The purpose of this interface is to
1159 provide fast pattern matching, so several sanity checks are not performed.
1160 However, feature tests are still performed. The new interface provides
1161 1.4x speedup compared to the old one.
1162
1163 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1164 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1165 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1166
1167 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1168 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1169 the "old" RFC 2279). Instead, it ended up passing a negative length to
1170 pcre_exec().
1171
1172 31. Add support for GCC's visibility feature to hide internal functions.
1173
1174 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1175 "unknown -C option" after outputting 0 or 1.
1176
1177 33. There is now support for generating a code coverage report for the test
1178 suite in environments where gcc is the compiler and lcov is installed. This
1179 is mainly for the benefit of the developers.
1180
1181 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1182 unaddressable using valgrind annotations, allowing valgrind to detect
1183 invalid memory accesses. This is mainly for the benefit of the developers.
1184
1185 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1186
1187 26. Give configure error if --enable-cpp but no C++ compiler found.
1188
1189
1190 Version 8.31 06-July-2012
1191 -------------------------
1192
1193 1. Fixing a wrong JIT test case and some compiler warnings.
1194
1195 2. Removed a bashism from the RunTest script.
1196
1197 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1198 to unsigned type, result still unsigned" that was given by an MS compiler
1199 on encountering the code "-sizeof(xxx)".
1200
1201 4. Partial matching support is added to the JIT compiler.
1202
1203 5. Fixed several bugs concerned with partial matching of items that consist
1204 of more than one character:
1205
1206 (a) /^(..)\1/ did not partially match "aba" because checking references was
1207 done on an "all or nothing" basis. This also applied to repeated
1208 references.
1209
1210 (b) \R did not give a hard partial match if \r was found at the end of the
1211 subject.
1212
1213 (c) \X did not give a hard partial match after matching one or more
1214 characters at the end of the subject.
1215
1216 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1217 a partial match for the string "\r".
1218
1219 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1220 a partial match for a CR character at the end of the subject string.
1221
1222 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1223 running pcretest, the text "(JIT)" added to the output whenever JIT is
1224 actually used to run the match.
1225
1226 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1227 or /S+[+] with a digit between 1 and 7.
1228
1229 8. OP_NOT now supports any UTF character not just single-byte ones.
1230
1231 9. (*MARK) control verb is now supported by the JIT compiler.
1232
1233 10. The command "./RunTest list" lists the available tests without actually
1234 running any of them. (Because I keep forgetting what they all are.)
1235
1236 11. Add PCRE_INFO_MAXLOOKBEHIND.
1237
1238 12. Applied a (slightly modified) user-supplied patch that improves performance
1239 when the heap is used for recursion (compiled with --disable-stack-for-
1240 recursion). Instead of malloc and free for each heap frame each time a
1241 logical recursion happens, frames are retained on a chain and re-used where
1242 possible. This sometimes gives as much as 30% improvement.
1243
1244 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1245 call.
1246
1247 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1248
1249 15. It is now possible to link pcretest with libedit as an alternative to
1250 libreadline.
1251
1252 16. (*COMMIT) control verb is now supported by the JIT compiler.
1253
1254 17. The Unicode data tables have been updated to Unicode 6.1.0.
1255
1256 18. Added --file-list option to pcregrep.
1257
1258 19. Added binary file support to pcregrep, including the -a, --binary-files,
1259 -I, and --text options.
1260
1261 20. The madvise function is renamed for posix_madvise for QNX compatibility
1262 reasons. Fixed by Giuseppe D'Angelo.
1263
1264 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1265
1266 22. Changed the option for creating symbolic links for 16-bit man pages from
1267 -s to -sf so that re-installing does not cause issues.
1268
1269 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1270
1271 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1272 matches in certain environments (the workspace was not being correctly
1273 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1274 some of the workspace data at the beginning of a restart.
1275
1276 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1277 was not doing so when it should - probably a typo introduced by SVN 528
1278 (change 8.10/14).
1279
1280 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1281 \w+ when the character tables indicated that \x{c4} was a word character.
1282 There were several related cases, all because the tests for doing a table
1283 lookup were testing for characters less than 127 instead of 255.
1284
1285 27. If a pattern contains capturing parentheses that are not used in a match,
1286 their slots in the ovector are set to -1. For those that are higher than
1287 any matched groups, this happens at the end of processing. In the case when
1288 there were back references that the ovector was too small to contain
1289 (causing temporary malloc'd memory to be used during matching), and the
1290 highest capturing number was not used, memory off the end of the ovector
1291 was incorrectly being set to -1. (It was using the size of the temporary
1292 memory instead of the true size.)
1293
1294 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1295 ovector size, it uses memory at the end of the block that it has got.
1296
1297 29. Check for an overlong MARK name and give an error at compile time. The
1298 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1299
1300 30. JIT compiler update.
1301
1302 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1303 Rill for the patch.
1304
1305 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1306
1307 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1308
1309 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1310 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1311 was enabled.
1312
1313 35. Improve JIT code generation for greedy plus quantifier.
1314
1315 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1316 1 to "aa" instead of to an empty string. The bug affected repeated groups
1317 that could potentially match an empty string.
1318
1319 37. Optimizing single character iterators in JIT.
1320
1321 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1322 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1323 codepoints that are too big for the mode are faulted, and in a UTF mode,
1324 disallowed codepoints are also faulted.
1325
1326 39. If PCRE was compiled with UTF support, in three places in the DFA
1327 matcher there was code that should only have been obeyed in UTF mode, but
1328 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1329 processing when bytes with values greater than 127 were present. In 16-bit
1330 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1331 both cases the values are those that cannot be the first data item in a UTF
1332 character. The three items that might have provoked this were recursions,
1333 possessively repeated groups, and atomic groups.
1334
1335 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1336 and pcregrep, because some OS require shared objects to be explicitly
1337 passed to ld, causing the link step to fail if they are not.
1338
1339 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1340 mode, patterns that started with \h* or \R* might be incorrectly matched.
1341
1342
1343 Version 8.30 04-February-2012
1344 -----------------------------
1345
1346 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1347 name is defined in ctype.h.
1348
1349 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1350 only in quite long subpatterns.
1351
1352 3. Removed the function pcre_info(), which has been obsolete and deprecated
1353 since it was replaced by pcre_fullinfo() in February 2000.
1354
1355 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1356 match a (*MARK), and the match failed at the start of the subject, a
1357 reference to memory before the start of the subject could occur. This bug
1358 was introduced by fix 17 of release 8.21.
1359
1360 5. A reference to an unset group with zero minimum repetition was giving
1361 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1362 /(another)?(\1?)test/ matched against "hello world test". This bug was
1363 introduced in release 8.13.
1364
1365 6. Add support for 16-bit character strings (a large amount of work involving
1366 many changes and refactorings).
1367
1368 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1369 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1370 from a file.
1371
1372 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1373 rounding is not applied in this particular case).
1374
1375 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1376 if they appear, or are escaped, in patterns.
1377
1378 10. Get rid of a number of -Wunused-but-set-variable warnings.
1379
1380 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1381 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1382 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1383 also returns the mark "x". This bug applied to capturing parentheses,
1384 non-capturing parentheses, and atomic parentheses. It also applied to some
1385 assertions.
1386
1387 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1388 information out of configure.ac instead of relying on pcre.h.generic, which
1389 is not stored in the repository.
1390
1391 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1392 -lreadline.
1393
1394 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1395
1396 15. Applied Graycode's patch to put the top-level frame on the stack rather
1397 than the heap when not using the stack for recursion. This gives a
1398 performance improvement in many cases when recursion is not deep.
1399
1400 16. Experimental code added to "pcretest -C" to output the stack frame size.
1401
1402
1403 Version 8.21 12-Dec-2011
1404 ------------------------
1405
1406 1. Updating the JIT compiler.
1407
1408 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1409 are added as well.
1410
1411 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1412 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1413 calling _pcre_jit_exec. Some extra comments are added.
1414
1415 4. (*MARK) settings inside atomic groups that do not contain any capturing
1416 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1417 was introduced by change 18 for 8.20.
1418
1419 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1420 ECMA-262 standard.
1421
1422 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1423 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1424 This bug was probably introduced by change 9 of 8.13.
1425
1426 7. While fixing 6 above, I noticed that a number of other items were being
1427 incorrectly rejected as "not fixed length". This arose partly because newer
1428 opcodes had not been added to the fixed-length checking code. I have (a)
1429 corrected the bug and added tests for these items, and (b) arranged for an
1430 error to occur if an unknown opcode is encountered while checking for fixed
1431 length instead of just assuming "not fixed length". The items that were
1432 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1433 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1434 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1435
1436 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1437 being incorrectly compiled and would have given unpredicatble results.
1438
1439 9. A possessively repeated subpattern with minimum repeat count greater than
1440 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1441 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1442 the first (A) could occur when it should not.
1443
1444 10. Add a cast and remove a redundant test from the code.
1445
1446 11. JIT should use pcre_malloc/pcre_free for allocation.
1447
1448 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1449 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1450 is anything other than /usr, -L is still shown).
1451
1452 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1453
1454 14. Perl does not support \N without a following name in a [] class; PCRE now
1455 also gives an error.
1456
1457 15. If a forward reference was repeated with an upper limit of around 2000,
1458 it caused the error "internal error: overran compiling workspace". The
1459 maximum number of forward references (including repeats) was limited by the
1460 internal workspace, and dependent on the LINK_SIZE. The code has been
1461 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1462 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1463 of around 200,000 forward references. While doing this, I also speeded up
1464 the filling in of repeated forward references.
1465
1466 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1467 incorrectly expecting the subject to contain another "a" after the start.
1468
1469 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1470 in the match, the SKIP should be ignored. This was not happening; instead
1471 the SKIP was being treated as NOMATCH. For patterns such as
1472 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1473 tested.
1474
1475 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1476 now much more compatible with Perl, in particular in cases where the result
1477 is a non-match for a non-anchored pattern. For example, if
1478 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1479 "m", where previously it did not return a name. A side effect of this
1480 change is that for partial matches, the last encountered mark name is
1481 returned, as for non matches. A number of tests that were previously not
1482 Perl-compatible have been moved into the Perl-compatible test files. The
1483 refactoring has had the pleasing side effect of removing one argument from
1484 the match() function, thus reducing its stack requirements.
1485
1486 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1487 subsequent uses of /S (without +) incorrectly behaved like /S+.
1488
1489 21. Retrieve executable code size support for the JIT compiler and fixing
1490 some warnings.
1491
1492 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1493 not work when the shorter character appeared right at the end of the
1494 subject string.
1495
1496 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1497 systems.
1498
1499 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1500 output it when the /M option is used in pcretest.
1501
1502 25. The CheckMan script was not being included in the distribution. Also, added
1503 an explicit "perl" to run Perl scripts from the PrepareRelease script
1504 because this is reportedly needed in Windows.
1505
1506 26. If study data was being save in a file and studying had not found a set of
1507 "starts with" bytes for the pattern, the data written to the file (though
1508 never used) was taken from uninitialized memory and so caused valgrind to
1509 complain.
1510
1511 27. Updated RunTest.bat as provided by Sheri Pierce.
1512
1513 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1514
1515 29. Computation of memory usage for the table of capturing group names was
1516 giving an unnecessarily large value.
1517
1518
1519 Version 8.20 21-Oct-2011
1520 ------------------------
1521
1522 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1523 a POSIX class. After further experiments with Perl, which convinced me that
1524 Perl has bugs and confusions, a closing square bracket is no longer allowed
1525 in a POSIX name. This bug also affected patterns with classes that started
1526 with full stops.
1527
1528 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1529 captured substring, but while checking the failing first alternative,
1530 substring 1 is temporarily captured. If the output vector supplied to
1531 pcre_exec() was not big enough for this capture, the yield of the function
1532 was still zero ("insufficient space for captured substrings"). This cannot
1533 be totally fixed without adding another stack variable, which seems a lot
1534 of expense for a edge case. However, I have improved the situation in cases
1535 such as /(a)(b)x|abc/ matched against "abc", where the return code
1536 indicates that fewer than the maximum number of slots in the ovector have
1537 been set.
1538
1539 3. Related to (2) above: when there are more back references in a pattern than
1540 slots in the output vector, pcre_exec() uses temporary memory during
1541 matching, and copies in the captures as far as possible afterwards. It was
1542 using the entire output vector, but this conflicts with the specification
1543 that only 2/3 is used for passing back captured substrings. Now it uses
1544 only the first 2/3, for compatibility. This is, of course, another edge
1545 case.
1546
1547 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1548 main code base, and can be used by building with --enable-jit. When this is
1549 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1550 runtime --no-jit option is given.
1551
1552 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1553 ovector, the return from the function was zero, implying that there were
1554 other matches that did not fit. The correct "exactly full" value is now
1555 returned.
1556
1557 6. If a subpattern that was called recursively or as a subroutine contained
1558 (*PRUNE) or any other control that caused it to give a non-standard return,
1559 invalid errors such as "Error -26 (nested recursion at the same subject
1560 position)" or even infinite loops could occur.
1561
1562 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1563 computing the minimum length on reaching *ACCEPT, and so ended up with the
1564 wrong value of 1 rather than 0. Further investigation indicates that
1565 computing a minimum subject length in the presence of *ACCEPT is difficult
1566 (think back references, subroutine calls), and so I have changed the code
1567 so that no minimum is registered for a pattern that contains *ACCEPT.
1568
1569 8. If (*THEN) was present in the first (true) branch of a conditional group,
1570 it was not handled as intended. [But see 16 below.]
1571
1572 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1573 Sheri Pierce.
1574
1575 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1576 the first byte in a match must be "a".
1577
1578 11. Change 17 for 8.13 increased the recursion depth for patterns like
1579 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1580 pattern contains any instances of (*THEN). If it does not, the old
1581 optimizations are restored. It would be nice to do this on a per-group
1582 basis, but at the moment that is not feasible.
1583
1584 12. In some environments, the output of pcretest -C is CRLF terminated. This
1585 broke RunTest's code that checks for the link size. A single white space
1586 character after the value is now allowed for.
1587
1588 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1589 For "fr", it uses the Windows-specific input and output files.
1590
1591 14. If (*THEN) appeared in a group that was called recursively or as a
1592 subroutine, it did not work as intended. [But see next item.]
1593
1594 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1595 pattern fragments (but not containing any | characters). If A and B are
1596 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1597 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1598 D. In other words, Perl considers parentheses that do not contain any |
1599 characters to be part of a surrounding alternative, whereas PCRE was
1600 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1601 differently. PCRE now behaves in the same way as Perl, except in the case
1602 of subroutine/recursion calls such as (?1) which have in any case always
1603 been different (but PCRE had them first :-).
1604
1605 16. Related to 15 above: Perl does not treat the | in a conditional group as
1606 creating alternatives. Such a group is treated in the same way as an
1607 ordinary group without any | characters when processing (*THEN). PCRE has
1608 been changed to match Perl's behaviour.
1609
1610 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1611 RunGrepTest script failed.
1612
1613 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1614 inevitable for groups that contain captures, but it can lead to a lot of
1615 stack use in large patterns. The old behaviour has been restored for atomic
1616 groups that do not contain any capturing parentheses.
1617
1618 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1619 suppress the check for a minimum subject length at run time. (If it was
1620 given to pcre_exec() or pcre_dfa_exec() it did work.)
1621
1622 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1623 fail to work when decoding hex characters in data strings in EBCDIC
1624 environments.
1625
1626 21. It appears that in at least one Mac OS environment, the isxdigit() function
1627 is implemented as a macro that evaluates to its argument more than once,
1628 contravening the C 90 Standard (I haven't checked a later standard). There
1629 was an instance in pcretest which caused it to go wrong when processing
1630 \x{...} escapes in subject strings. The has been rewritten to avoid using
1631 things like p++ in the argument of isxdigit().
1632
1633
1634 Version 8.13 16-Aug-2011
1635 ------------------------
1636
1637 1. The Unicode data tables have been updated to Unicode 6.0.0.
1638
1639 2. Two minor typos in pcre_internal.h have been fixed.
1640
1641 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1642 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1643 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1644
1645 4. There were a number of related bugs in the code for matching backrefences
1646 caselessly in UTF-8 mode when codes for the characters concerned were
1647 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1648 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1649 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1650 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1651 2-byte code at the end of the subject (it thought there wasn't enough data
1652 left).
1653
1654 5. Comprehensive information about what went wrong is now returned by
1655 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1656 as the output vector has at least 2 elements. The offset of the start of
1657 the failing character and a reason code are placed in the vector.
1658
1659 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1660 now returned is for the first byte of the failing character, instead of the
1661 last byte inspected. This is an incompatible change, but I hope it is small
1662 enough not to be a problem. It makes the returned offset consistent with
1663 pcre_exec() and pcre_dfa_exec().
1664
1665 7. pcretest now gives a text phrase as well as the error number when
1666 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1667 failure, the offset and reason code are output.
1668
1669 8. When \R was used with a maximizing quantifier it failed to skip backwards
1670 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1671 back over a single character (\n). This seems wrong (because it treated the
1672 two characters as a single entity when going forwards), conflicts with the
1673 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1674 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1675 has been changed.
1676
1677 9. Some internal refactoring has changed the processing so that the handling
1678 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1679 time (the PCRE_DOTALL option was changed this way some time ago: version
1680 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1681 which was always a bit of a fudge. It also means that there is one less
1682 argument for the match() function, which reduces its stack requirements
1683 slightly. This change also fixes an incompatibility with Perl: the pattern
1684 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1685
1686 10. More internal refactoring has drastically reduced the number of recursive
1687 calls to match() for possessively repeated groups such as (abc)++ when
1688 using pcre_exec().
1689
1690 11. While implementing 10, a number of bugs in the handling of groups were
1691 discovered and fixed:
1692
1693 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1694 (a|)*(?1) gave a compile-time internal error.
1695 ((a|)+)+ did not notice that the outer group could match an empty string.
1696 (^a|^)+ was not marked as anchored.
1697 (.*a|.*)+ was not marked as matching at start or after a newline.
1698
1699 12. Yet more internal refactoring has removed another argument from the match()
1700 function. Special calls to this function are now indicated by setting a
1701 value in a variable in the "match data" data block.
1702
1703 13. Be more explicit in pcre_study() instead of relying on "default" for
1704 opcodes that mean there is no starting character; this means that when new
1705 ones are added and accidentally left out of pcre_study(), testing should
1706 pick them up.
1707
1708 14. The -s option of pcretest has been documented for ages as being an old
1709 synonym of -m (show memory usage). I have changed it to mean "force study
1710 for every regex", that is, assume /S for every regex. This is similar to -i
1711 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1712 using it. It makes it easier to run collections of tests with and without
1713 study enabled, and thereby test pcre_study() more easily. All the standard
1714 tests are now run with and without -s (but some patterns can be marked as
1715 "never study" - see 20 below).
1716
1717 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1718 restoration of the capturing data to the outer values was not happening
1719 correctly.
1720
1721 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1722 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1723 pattern had matched an empty string, and so incorrectly returned a no
1724 match.
1725
1726 17. There was optimizing code for the last branch of non-capturing parentheses,
1727 and also for the obeyed branch of a conditional subexpression, which used
1728 tail recursion to cut down on stack usage. Unfortunately, now that there is
1729 the possibility of (*THEN) occurring in these branches, tail recursion is
1730 no longer possible because the return has to be checked for (*THEN). These
1731 two optimizations have therefore been removed. [But see 8.20/11 above.]
1732
1733 18. If a pattern containing \R was studied, it was assumed that \R always
1734 matched two bytes, thus causing the minimum subject length to be
1735 incorrectly computed because \R can also match just one byte.
1736
1737 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1738 was incorrectly computed.
1739
1740 20. If /S is present twice on a test pattern in pcretest input, it now
1741 *disables* studying, thereby overriding the use of -s on the command line
1742 (see 14 above). This is necessary for one or two tests to keep the output
1743 identical in both cases.
1744
1745 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1746 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1747
1748 22. When an atomic group that contained a capturing parenthesis was
1749 successfully matched, but the branch in which it appeared failed, the
1750 capturing was not being forgotten if a higher numbered group was later
1751 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1752 group 1 to "a", when in fact it should be unset. This applied to multi-
1753 branched capturing and non-capturing groups, repeated or not, and also to
1754 positive assertions (capturing in negative assertions does not happen
1755 in PCRE) and also to nested atomic groups.
1756
1757 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1758 subject after a captured substring, to make it easier to tell which of a
1759 number of identical substrings has been captured.
1760
1761 24. The way atomic groups are processed by pcre_exec() has been changed so that
1762 if they are repeated, backtracking one repetition now resets captured
1763 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1764 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1765 "aaa". Previously, it would have been "a". As part of this code
1766 refactoring, the way recursive calls are handled has also been changed.
1767
1768 25. If an assertion condition captured any substrings, they were not passed
1769 back unless some other capturing happened later. For example, if
1770 (?(?=(a))a) was matched against "a", no capturing was returned.
1771
1772 26. When studying a pattern that contained subroutine calls or assertions,
1773 the code for finding the minimum length of a possible match was handling
1774 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1775 group 1 called group 2 while simultaneously a separate group 2 called group
1776 1). A stack overflow occurred in this case. I have fixed this by limiting
1777 the recursion depth to 10.
1778
1779 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1780 Fortmann. This supports explicit test numbers on the command line, and has
1781 argument validation and error reporting.
1782
1783 28. An instance of \X with an unlimited repeat could fail if at any point the
1784 first character it looked at was a mark character.
1785
1786 29. Some minor code refactoring concerning Unicode properties and scripts
1787 should reduce the stack requirement of match() slightly.
1788
1789 30. Added the '=' option to pcretest to check the setting of unused capturing
1790 slots at the end of the pattern, which are documented as being -1, but are
1791 not included in the return count.
1792
1793 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1794 compiled something random. Now it gives a compile-time error (as does
1795 Perl).
1796
1797 32. A *MARK encountered during the processing of a positive assertion is now
1798 recorded and passed back (compatible with Perl).
1799
1800 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1801 had alternative anchored branches, the search for a second match in a line
1802 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1803 matched the line "0102" twice. The same bug affected patterns that started
1804 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1805 twice.
1806
1807 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1808 does, and because of capturing effects, quantifying parenthesized
1809 assertions may at times be useful. Quantifiers are now allowed for
1810 parenthesized assertions.
1811
1812 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1813
1814 36. \g was being checked for fancy things in a character class, when it should
1815 just be a literal "g".
1816
1817 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1818 appearance of a nested POSIX class supersedes an apparent external class.
1819 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1820 unescaped square brackets may also appear as part of class names. For
1821 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1822 more like Perl. (But see 8.20/1 above.)
1823
1824 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1825 was because it thought it was \N{name}, which is not supported).
1826
1827 39. Add minix to OS list not supporting the -S option in pcretest.
1828
1829 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1830 cannot analyze patterns in sufficient detail to catch mutual recursions
1831 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1832 subgroup is called recursively as a subpattern for a second time at the
1833 same position in the subject string. In previous releases this might have
1834 been caught by the recursion limit, or it might have run out of stack.
1835
1836 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1837 happen only once. PCRE was, however incorrectly giving a compile time error
1838 "recursive call could loop indefinitely" because it cannot analyze the
1839 pattern in sufficient detail. The compile time test no longer happens when
1840 PCRE is compiling a conditional subpattern, but actual runaway loops are
1841 now caught at runtime (see 40 above).
1842
1843 42. It seems that Perl allows any characters other than a closing parenthesis
1844 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1845 has been changed to be the same.
1846
1847 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1848 as not to get warnings when autogen.sh is called. Also changed
1849 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1850
1851 44. To help people who use pcregrep to scan files containing exceedingly long
1852 lines, the following changes have been made:
1853
1854 (a) The default value of the buffer size parameter has been increased from
1855 8K to 20K. (The actual buffer used is three times this size.)
1856
1857 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1858 PCRE is built.
1859
1860 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1861 to be set at run time.
1862
1863 (d) Numerical values in pcregrep options can be followed by K or M, for
1864 example --buffer-size=50K.
1865
1866 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1867 given and the return code is set to 2.
1868
1869 45. Add a pointer to the latest mark to the callout data block.
1870
1871 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1872 partial match of an empty string instead of no match. This was specific to
1873 the use of ".".
1874
1875 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1876 complete match instead of a partial match. This bug was dependent on both
1877 the PCRE_UTF8 and PCRE_DOTALL options being set.
1878
1879 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1880 starting byte set, because \b was not being ignored.
1881
1882
1883 Version 8.12 15-Jan-2011
1884 ------------------------
1885
1886 1. Fixed some typos in the markup of the man pages, and wrote a script that
1887 checks for such things as part of the documentation building process.
1888
1889 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1890 --match-limit and --recursion-limit options (added for 8.11). In
1891 particular, this made one of the standard tests fail. (The integer value
1892 went into the wrong half of a long int.)
1893
1894 3. If the --colour option was given to pcregrep with -v (invert match), it
1895 did strange things, either producing crazy output, or crashing. It should,
1896 of course, ignore a request for colour when reporting lines that do not
1897 match.
1898
1899 4. Another pcregrep bug caused similar problems if --colour was specified with
1900 -M (multiline) and the pattern match finished with a line ending.
1901
1902 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1903 matched in multiline mode, the following line was shown as part of the
1904 match. This seems wrong, so I have changed it.
1905
1906 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1907 the check for further matches in the same line (so they could be coloured)
1908 to overrun the end of the current line. If another match was found, it was
1909 incorrectly shown (and then shown again when found in the next line).
1910
1911 7. If pcregrep was compiled under Windows, there was a reference to the
1912 function pcregrep_exit() before it was defined. I am assuming this was
1913 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1914 reported by a user. I've moved the definition above the reference.
1915
1916
1917 Version 8.11 10-Dec-2010
1918 ------------------------
1919
1920 1. (*THEN) was not working properly if there were untried alternatives prior
1921 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1922 backtracked to try for "b" instead of moving to the next alternative branch
1923 at the same level (in this case, to look for "c"). The Perl documentation
1924 is clear that when (*THEN) is backtracked onto, it goes to the "next
1925 alternative in the innermost enclosing group".
1926
1927 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1928 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1929 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1930 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1931 (*THEN).
1932
1933 3. If \s appeared in a character class, it removed the VT character from
1934 the class, even if it had been included by some previous item, for example
1935 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1936 of \s, but is part of the POSIX "space" class.)
1937
1938 4. A partial match never returns an empty string (because you can always
1939 match an empty string at the end of the subject); however the checking for
1940 an empty string was starting at the "start of match" point. This has been
1941 changed to the "earliest inspected character" point, because the returned
1942 data for a partial match starts at this character. This means that, for
1943 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1944 (previously it gave "no match").
1945
1946 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1947 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1948 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1949 has an implication that the given string is incomplete (because a partial
1950 match is preferred over a full match). For this reason, these items now
1951 give a partial match in this situation. [Aside: previously, the one case
1952 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1953 match rather than a full match, which was wrong by the old rules, but is
1954 now correct.]
1955
1956 6. There was a bug in the handling of #-introduced comments, recognized when
1957 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1958 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1959 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1960 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1961 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1962 places in pcre_compile().
1963
1964 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1965 comments when looking ahead for named forward references to subpatterns,
1966 the only newline sequence it recognized was NL. It now handles newlines
1967 according to the set newline convention.
1968
1969 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1970 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1971 cater for a lack of strerror(). These oversights have been fixed.
1972
1973 9. Added --match-limit and --recursion-limit to pcregrep.
1974
1975 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1976
1977 11. When the -o option was used, pcregrep was setting a return code of 1, even
1978 when matches were found, and --line-buffered was not being honoured.
1979
1980 12. Added an optional parentheses number to the -o and --only-matching options
1981 of pcregrep.
1982
1983 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1984 can match an empty string. The code to do it in pcretest and pcredemo
1985 needed fixing:
1986
1987 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1988 only one byte after an empty string match just before CRLF (this case
1989 just got forgotten; "any" and "anycrlf" were OK).
1990
1991 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1992 mode when an empty string match preceded an ASCII character followed by
1993 a non-ASCII character. (The code for advancing by one character rather
1994 than one byte was nonsense.)
1995
1996 (c) The pcredemo.c sample program did not have any code at all to handle
1997 the cases when CRLF is a valid newline sequence.
1998
1999 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2000 as a starting offset was within the subject string. There is now a new
2001 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2002 negative or greater than the length of the string. In order to test this,
2003 pcretest is extended to allow the setting of negative starting offsets.
2004
2005 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2006 starting offset points to the beginning of a UTF-8 character was
2007 unnecessarily clumsy. I tidied it up.
2008
2009 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2010 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2011
2012 17. Nobody had reported that the --include_dir option, which was added in
2013 release 7.7 should have been called --include-dir (hyphen, not underscore)
2014 for compatibility with GNU grep. I have changed it to --include-dir, but
2015 left --include_dir as an undocumented synonym, and the same for
2016 --exclude-dir, though that is not available in GNU grep, at least as of
2017 release 2.5.4.
2018
2019 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2020 characters from a string of bytes) have been redefined so as not to use
2021 loops, in order to improve performance in some environments. At the same
2022 time, I abstracted some of the common code into auxiliary macros to save
2023 repetition (this should not affect the compiled code).
2024
2025 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2026 compile-time error is now given if \c is not followed by an ASCII
2027 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2028 different, and any byte value is allowed.)
2029
2030 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2031 START_OPTIMIZE option, which is now allowed at compile time - but just
2032 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2033 to pcregrep and other applications that have no direct access to PCRE
2034 options. The new /Y option in pcretest sets this option when calling
2035 pcre_compile().
2036
2037 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2038 back references. Groups containing recursive back references were forced to
2039 be atomic by that change, but in the case of named groups, the amount of
2040 memory required was incorrectly computed, leading to "Failed: internal
2041 error: code overflow". This has been fixed.
2042
2043 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2044 pcretest.c, to avoid build problems in some Borland environments.
2045
2046
2047 Version 8.10 25-Jun-2010
2048 ------------------------
2049
2050 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2051 THEN.
2052
2053 2. (*ACCEPT) was not working when inside an atomic group.
2054
2055 3. Inside a character class, \B is treated as a literal by default, but
2056 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2057 causes the error). The code is unchanged, but I tidied the documentation.
2058
2059 4. Inside a character class, PCRE always treated \R and \X as literals,
2060 whereas Perl faults them if its -w option is set. I have changed PCRE so
2061 that it faults them when PCRE_EXTRA is set.
2062
2063 5. Added support for \N, which always matches any character other than
2064 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2065
2066 6. When compiling pcregrep with newer versions of gcc which may have
2067 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2068 declared with attribute warn_unused_result" were given. Just casting the
2069 result to (void) does not stop the warnings; a more elaborate fudge is
2070 needed. I've used a macro to implement this.
2071
2072 7. Minor change to pcretest.c to avoid a compiler warning.
2073
2074 8. Added four artifical Unicode properties to help with an option to make
2075 \s etc use properties (see next item). The new properties are: Xan
2076 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2077
2078 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2079 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2080 this option. Modified pcretest to add /W to test this facility. Added
2081 REG_UCP to make it available via the POSIX interface.
2082
2083 10. Added --line-buffered to pcregrep.
2084
2085 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2086 studied, and the match started with a letter with a code point greater than
2087 127 whose first byte was different to the first byte of the other case of
2088 the letter, the other case of this starting letter was not recognized
2089 (#976).
2090
2091 12. If a pattern that was studied started with a repeated Unicode property
2092 test, for example, \p{Nd}+, there was the theoretical possibility of
2093 setting up an incorrect bitmap of starting bytes, but fortunately it could
2094 not have actually happened in practice until change 8 above was made (it
2095 added property types that matched character-matching opcodes).
2096
2097 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2098 possible starting bytes for non-anchored patterns.
2099
2100 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2101 \R, and also a number of cases that involve Unicode properties, both
2102 explicit and implicit when PCRE_UCP is set.
2103
2104 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2105 input, it could crash or give wrong results if characters with values
2106 greater than 0xc0 were present in the subject string. (Detail: it assumed
2107 UTF-8 input when processing these items.)
2108
2109 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2110 size_t is 64-bit (#991).
2111
2112 17. Added a check for running out of memory when PCRE is compiled with
2113 --disable-stack-for-recursion (#990).
2114
2115 18. If the last data line in a file for pcretest does not have a newline on
2116 the end, a newline was missing in the output.
2117
2118 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2119 less than 128) in its various bitmaps. However, there is a facility for
2120 generating tables according to the current locale when PCRE is compiled. It
2121 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2122 characters, are recognized by isspace() and therefore were getting set in
2123 these tables, and indeed these tables seem to approximate to ISO 8859. This
2124 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2125 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2126 which of course cannot start UTF-8 characters. I have changed the code so
2127 that only real ASCII characters (less than 128) and the correct starting
2128 bytes for UTF-8 encodings are set for characters greater than 127 when in
2129 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2130 altogether.)
2131
2132 20. Added the /T option to pcretest so as to be able to run tests with non-
2133 standard character tables, thus making it possible to include the tests
2134 used for 19 above in the standard set of tests.
2135
2136 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2137 reference to a subpattern the other side of a comment that contains an
2138 opening parenthesis caused either an internal compiling error, or a
2139 reference to the wrong subpattern.
2140
2141
2142 Version 8.02 19-Mar-2010
2143 ------------------------
2144
2145 1. The Unicode data tables have been updated to Unicode 5.2.0.
2146
2147 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2148 configured.
2149
2150 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2151 original author of that file, following a query about its status.
2152
2153 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2154 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2155
2156 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2157 quantifier applied to a forward-referencing subroutine call, could compile
2158 incorrect code or give the error "internal error: previously-checked
2159 referenced subpattern not found".
2160
2161 6. Both MS Visual Studio and Symbian OS have problems with initializing
2162 variables to point to external functions. For these systems, therefore,
2163 pcre_malloc etc. are now initialized to local functions that call the
2164 relevant global functions.
2165
2166 7. There were two entries missing in the vectors called coptable and poptable
2167 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2168 I've fixed the data, and added a kludgy way of testing at compile time that
2169 the lengths are correct (equal to the number of opcodes).
2170
2171 8. Following on from 7, I added a similar kludge to check the length of the
2172 eint vector in pcreposix.c.
2173
2174 9. Error texts for pcre_compile() are held as one long string to avoid too
2175 much relocation at load time. To find a text, the string is searched,
2176 counting zeros. There was no check for running off the end of the string,
2177 which could happen if a new error number was added without updating the
2178 string.
2179
2180 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2181
2182 11. \K was not working if it appeared in an atomic group or in a group that
2183 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2184 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2185 the assertion is positive, but not if it is negative.
2186
2187 12. Change 11 fortuitously reduced the size of the stack frame used in the
2188 "match()" function of pcre_exec.c by one pointer. Forthcoming
2189 implementation of support for (*MARK) will need an extra pointer on the
2190 stack; I have reserved it now, so that the stack frame size does not
2191 decrease.
2192
2193 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2194 item in branch that calls a recursion is a subroutine call - as in the
2195 second branch in the above example - was incorrectly given the compile-
2196 time error "recursive call could loop indefinitely" because pcre_compile()
2197 was not correctly checking the subroutine for matching a non-empty string.
2198
2199 14. The checks for overrunning compiling workspace could trigger after an
2200 overrun had occurred. This is a "should never occur" error, but it can be
2201 triggered by pathological patterns such as hundreds of nested parentheses.
2202 The checks now trigger 100 bytes before the end of the workspace.
2203
2204 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2205
2206
2207 Version 8.01 19-Jan-2010
2208 ------------------------
2209
2210 1. If a pattern contained a conditional subpattern with only one branch (in
2211 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2212 computed the wrong minimum data length (which is of course zero for such
2213 subpatterns). This could cause incorrect "no match" results.
2214
2215 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2216 the pattern is reset in the first branch, pcre_compile() failed with
2217 "internal error: code overflow at offset...". This happened only when
2218 the reset was to the original external option setting. (An optimization
2219 abstracts leading options settings into an external setting, which was the
2220 cause of this.)
2221
2222 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2223 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2224 assertion pattern did not match (meaning that the assertion was true), it
2225 was incorrectly treated as false if the SKIP had been reached during the
2226 matching. This also applied to assertions used as conditions.
2227
2228 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2229 assertion subpattern, including such a pattern used as a condition,
2230 unpredictable results occurred, instead of the error return
2231 PCRE_ERROR_DFA_UITEM.
2232
2233 5. The C++ GlobalReplace function was not working like Perl for the special
2234 situation when an empty string is matched. It now does the fancy magic
2235 stuff that is necessary.
2236
2237 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2238 removed. (These were left over from very, very early versions of PCRE.)
2239
2240 7. Some cosmetic changes to the code to make life easier when compiling it
2241 as part of something else:
2242
2243 (a) Change DEBUG to PCRE_DEBUG.
2244
2245 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2246 called "current" as "current_branch", to prevent a collision with the
2247 Linux macro when compiled as a kernel module.
2248
2249 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2250 prevent a collision with the Linux macro when compiled as a kernel
2251 module.
2252
2253 8. In pcre_compile() there are some checks for integer overflows that used to
2254 cast potentially large values to (double). This has been changed to that
2255 when building, a check for int64_t is made, and if it is found, it is used
2256 instead, thus avoiding the use of floating point arithmetic. (There is no
2257 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2258 double.
2259
2260 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2261 2005 (difference between two addresses compared to an unsigned value).
2262
2263 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2264 custom one, because of the following reported problem in Windows:
2265
2266 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2267 under Win32.
2268 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2269 therefore missing the function definition.
2270 - The compiler thus generates a "C" signature for the test function.
2271 - The linker fails to find the "C" function.
2272 - PCRE fails to configure if asked to do so against libbz2.
2273
2274 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2275 messages were output:
2276
2277 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2278 rerunning libtoolize, to keep the correct libtool macros in-tree.
2279 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2280
2281 I have done both of these things.
2282
2283 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2284 most of the time, it *can* run out if it is given a pattern that contains a
2285 runaway infinite recursion. I updated the discussion in the pcrestack man
2286 page.
2287
2288 13. Now that we have gone to the x.xx style of version numbers, the minor
2289 version may start with zero. Using 08 or 09 is a bad idea because users
2290 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2291 interpreted as invalid octal numbers. I've updated the previous comment in
2292 configure.ac, and also added a check that gives an error if 08 or 09 are
2293 used.
2294
2295 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2296 causing partial matching to fail when the end of the subject matched \W
2297 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2298
2299 15. There were some discrepancies between the declarations in pcre_internal.h
2300 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2301 their definitions. The declarations used "const uschar *" and the
2302 definitions used USPTR. Even though USPTR is normally defined as "const
2303 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2304 reported that: "This difference in casting confuses some C++ compilers, for
2305 example, SunCC recognizes above declarations as different functions and
2306 generates broken code for hbpcre." I have changed the declarations to use
2307 USPTR.
2308
2309 16. GNU libtool is named differently on some systems. The autogen.sh script now
2310 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2311 (FreeBSD).
2312
2313 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2314 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2315 comment: "Figure out how to create a longlong from a string: strtoll and
2316 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2317 instance, but it only takes 2 args instead of 3!"
2318
2319 18. A subtle bug concerned with back references has been fixed by a change of
2320 specification, with a corresponding code fix. A pattern such as
2321 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2322 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2323 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2324 same bug. Such groups have to be quantified to be useful, or contained
2325 inside another quantified group. (If there's no repetition, the reference
2326 can never match.) The problem arises because, having left the group and
2327 moved on to the rest of the pattern, a later failure that backtracks into
2328 the group uses the captured value from the final iteration of the group
2329 rather than the correct earlier one. I have fixed this in PCRE by forcing
2330 any group that contains a reference to itself to be an atomic group; that
2331 is, there cannot be any backtracking into it once it has completed. This is
2332 similar to recursive and subroutine calls.
2333
2334
2335 Version 8.00 19-Oct-09
2336 ----------------------
2337
2338 1. The table for translating pcre_compile() error codes into POSIX error codes
2339 was out-of-date, and there was no check on the pcre_compile() error code
2340 being within the table. This could lead to an OK return being given in
2341 error.
2342
2343 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2344 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2345 in a Windows environment.
2346
2347 3. The pcregrep --count option prints the count for each file even when it is
2348 zero, as does GNU grep. However, pcregrep was also printing all files when
2349 --files-with-matches was added. Now, when both options are given, it prints
2350 counts only for those files that have at least one match. (GNU grep just
2351 prints the file name in this circumstance, but including the count seems
2352 more useful - otherwise, why use --count?) Also ensured that the
2353 combination -clh just lists non-zero counts, with no names.
2354
2355 4. The long form of the pcregrep -F option was incorrectly implemented as
2356 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2357 but it seems right to fix it, and I didn't think it was worth preserving
2358 the old behaviour.
2359
2360 5. The command line items --regex=pattern and --regexp=pattern were not
2361 recognized by pcregrep, which required --regex pattern or --regexp pattern
2362 (with a space rather than an '='). The man page documented the '=' forms,
2363 which are compatible with GNU grep; these now work.
2364
2365 6. No libpcreposix.pc file was created for pkg-config; there was just
2366 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2367
2368 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2369 when UCP support is not needed, by modifying the Python script that
2370 generates it from Unicode data files. This should not matter if the module
2371 is correctly used as a library, but I received one complaint about 50K of
2372 unwanted data. My guess is that the person linked everything into his
2373 program rather than using a library. Anyway, it does no harm.
2374
2375 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2376 was a minimum greater than 1 for a wide character in a possessive
2377 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2378 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2379 character. Chaos in the form of incorrect output or a compiling loop could
2380 result.
2381
2382 9. The restrictions on what a pattern can contain when partial matching is
2383 requested for pcre_exec() have been removed. All patterns can now be
2384 partially matched by this function. In addition, if there are at least two
2385 slots in the offset vector, the offset of the earliest inspected character
2386 for the match and the offset of the end of the subject are set in them when
2387 PCRE_ERROR_PARTIAL is returned.
2388
2389 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2390 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2391 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2392 and may be more useful for multi-segment matching.
2393
2394 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2395 used to be given if ever the end of the subject was reached; now it is
2396 given only if matching could not proceed because another character was
2397 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2398 string "Z", which now yields "no match" instead of "partial match". In the
2399 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2400 final character ended with (*FAIL).
2401
2402 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2403 if the pattern had a "must contain" character that was already found in the
2404 earlier partial match, unless partial matching was again requested. For
2405 example, with the pattern /dog.(body)?/, the "must contain" character is
2406 "g". If the first part-match was for the string "dog", restarting with
2407 "sbody" failed. This bug has been fixed.
2408
2409 13. The string returned by pcre_dfa_exec() after a partial match has been
2410 changed so that it starts at the first inspected character rather than the
2411 first character of the match. This makes a difference only if the pattern
2412 starts with a lookbehind assertion or \b or \B (\K is not supported by
2413 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2414 matching functions compatible, and I think it's the right thing to do.
2415
2416 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2417 so that the demonstration program is easily available in environments where
2418 PCRE has not been installed from source.
2419
2420 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2421 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2422 library.
2423
2424 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2425 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2426 is not the first non-POSIX option to be added. Clearly some people find
2427 these options useful.
2428
2429 17. If a caller to the POSIX matching function regexec() passes a non-zero
2430 value for nmatch with a NULL value for pmatch, the value of
2431 nmatch is forced to zero.
2432
2433 18. RunGrepTest did not have a test for the availability of the -u option of
2434 the diff command, as RunTest does. It now checks in the same way as
2435 RunTest, and also checks for the -b option.
2436
2437 19. If an odd number of negated classes containing just a single character
2438 interposed, within parentheses, between a forward reference to a named
2439 subpattern and the definition of the subpattern, compilation crashed with
2440 an internal error, complaining that it could not find the referenced
2441 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2442 [The bug was that it was starting one character too far in when skipping
2443 over the character class, thus treating the ] as data rather than
2444 terminating the class. This meant it could skip too much.]
2445
2446 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2447 /g option in pcretest when the pattern contains \K, which makes it possible
2448 to have an empty string match not at the start, even when the pattern is
2449 anchored. Updated pcretest and pcredemo to use this option.
2450
2451 21. If the maximum number of capturing subpatterns in a recursion was greater
2452 than the maximum at the outer level, the higher number was returned, but
2453 with unset values at the outer level. The correct (outer level) value is
2454 now given.
2455
2456 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2457 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2458 make it do so. The string so far is captured, making this feature
2459 compatible with Perl.
2460
2461 23. The tests have been re-organized, adding tests 11 and 12, to make it
2462 possible to check the Perl 5.10 features against Perl 5.10.
2463
2464 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2465 pattern matches a fixed length string. PCRE did not allow this; now it
2466 does. Neither allows recursion.
2467
2468 25. I finally figured out how to implement a request to provide the minimum
2469 length of subject string that was needed in order to match a given pattern.
2470 (It was back references and recursion that I had previously got hung up
2471 on.) This code has now been added to pcre_study(); it finds a lower bound
2472 to the length of subject needed. It is not necessarily the greatest lower
2473 bound, but using it to avoid searching strings that are too short does give
2474 some useful speed-ups. The value is available to calling programs via
2475 pcre_fullinfo().
2476
2477 26. While implementing 25, I discovered to my embarrassment that pcretest had
2478 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2479 study optimizations had never been tested with that matching function.
2480 Oops. What is worse, even when it was passed study data, there was a bug in
2481 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2482 were also very few tests of studied patterns with pcre_dfa_exec().
2483
2484 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2485 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2486 on the other side of the coin, they are no longer allowed to have different
2487 names, because these cannot be distinguished in PCRE, and this has caused
2488 confusion. (This is a difference from Perl.)
2489
2490 28. When duplicate subpattern names are present (necessarily with different
2491 numbers, as required by 27 above), and a test is made by name in a
2492 conditional pattern, either for a subpattern having been matched, or for
2493 recursion in such a pattern, all the associated numbered subpatterns are
2494 tested, and the overall condition is true if the condition is true for any
2495 one of them. This is the way Perl works, and is also more like the way
2496 testing by number works.
2497
2498
2499 Version 7.9 11-Apr-09
2500 ---------------------
2501
2502 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2503 (pcretest), all targets were linked against these libraries. This included
2504 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2505 libraries. This caused unwanted dependencies to be created. This problem
2506 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2507 pcretest is linked with readline.
2508
2509 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2510 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2511 moved outside it again, because FALSE and TRUE are already defined in AIX,
2512 but BOOL is not.
2513
2514 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2515 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2516
2517 4. The pcregrep documentation said spaces were inserted as well as colons (or
2518 hyphens) following file names and line numbers when outputting matching
2519 lines. This is not true; no spaces are inserted. I have also clarified the
2520 wording for the --colour (or --color) option.
2521
2522 5. In pcregrep, when --colour was used with -o, the list of matching strings
2523 was not coloured; this is different to GNU grep, so I have changed it to be
2524 the same.
2525
2526 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2527 each matching line was coloured. Now it goes on to look for further matches
2528 of any of the test patterns, which is the same behaviour as GNU grep.
2529
2530 7. A pattern that could match an empty string could cause pcregrep to loop; it
2531 doesn't make sense to accept an empty string match in pcregrep, so I have
2532 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2533 seems to be how GNU grep behaves. [But see later change 40 for release
2534 8.33.]
2535
2536 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2537 start or after a newline", because the conditional assertion was not being
2538 correctly handled. The rule now is that both the assertion and what follows
2539 in the first alternative must satisfy the test.
2540
2541 9. If auto-callout was enabled in a pattern with a conditional group whose
2542 condition was an assertion, PCRE could crash during matching, both with
2543 pcre_exec() and pcre_dfa_exec().
2544
2545 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2546 used for matching.
2547
2548 11. Unicode property support in character classes was not working for
2549 characters (bytes) greater than 127 when not in UTF-8 mode.
2550
2551 12. Added the -M command line option to pcretest.
2552
2553 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2554
2555 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2556
2557 16. Added comments and documentation about mis-use of no_arg in the C++
2558 wrapper.
2559
2560 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2561 from Martin Jerabek that uses macro names for all relevant character and
2562 string constants.
2563
2564 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2565 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2566 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2567 these, but not everybody uses configure.
2568
2569 19. A conditional group that had only one branch was not being correctly
2570 recognized as an item that could match an empty string. This meant that an
2571 enclosing group might also not be so recognized, causing infinite looping
2572 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2573 with the subject "ab", where knowledge that the repeated group can match
2574 nothing is needed in order to break the loop.
2575
2576 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2577 exec(), but without supplying a callout function, matching went wrong.
2578
2579 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2580 leak if the size of the offset vector was greater than 30. When the vector
2581 is smaller, the saved offsets during recursion go onto a local stack
2582 vector, but for larger vectors malloc() is used. It was failing to free
2583 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2584 error, in fact).
2585
2586 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2587 heapframe that is used only when UTF-8 support is enabled. This caused no
2588 problem, but was untidy.
2589
2590 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2591 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2592 included within another project.
2593
2594 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2595 slightly modified by me:
2596
2597 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2598 not building pcregrep.
2599
2600 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2601 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2602
2603 25. Forward references, both numeric and by name, in patterns that made use of
2604 duplicate group numbers, could behave incorrectly or give incorrect errors,
2605 because when scanning forward to find the reference group, PCRE was not
2606 taking into account the duplicate group numbers. A pattern such as
2607 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2608
2609 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2610 the feature of a custom pointer more persuasive (as requested by a user).
2611
2612 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2613 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2614 pre-defined.
2615
2616 28. Added support for (*UTF8) at the start of a pattern.
2617
2618 29. Arrange for flags added by the "release type" setting in CMake to be shown
2619 in the configuration summary.
2620
2621
2622 Version 7.8 05-Sep-08
2623 ---------------------
2624
2625 1. Replaced UCP searching code with optimized version as implemented for Ad
2626 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2627 stage table and inline lookup instead of a function, giving speed ups of 2
2628 to 5 times on some simple patterns that I tested. Permission was given to
2629 distribute the MultiStage2.py script that generates the tables (it's not in
2630 the tarball, but is in the Subversion repository).
2631
2632 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2633 scripts.
2634
2635 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2636 a group with a zero qualifier. The result of the study could be incorrect,
2637 or the function might crash, depending on the pattern.
2638
2639 4. Caseless matching was not working for non-ASCII characters in back
2640 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2641 It now works when Unicode Property Support is available.
2642
2643 5. In pcretest, an escape such as \x{de} in the data was always generating
2644 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2645 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2646 truncation.
2647
2648 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2649
2650 7. Added two (int) casts to pcregrep when printing the difference of two
2651 pointers, in case they are 64-bit values.
2652
2653 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2654 test 2 if it fails.
2655
2656 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2657 and a #define of that name to empty if it is not externally set. This is to
2658 allow users of MSVC to set it if necessary.
2659
2660 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2661 the convenience functions in the pcre_get.c source file.
2662
2663 11. An option change at the start of a pattern that had top-level alternatives
2664 could cause overwriting and/or a crash. This command provoked a crash in
2665 some environments:
2666
2667 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2668
2669 This potential security problem was recorded as CVE-2008-2371.
2670
2671 12. For a pattern where the match had to start at the beginning or immediately
2672 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2673 pcre_dfa_exec() could read past the end of the passed subject if there was
2674 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2675 pcretest so that it places the subject at the end of its malloc-ed buffer.
2676
2677 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2678 exec() might read past the end of the data buffer in UTF-8 mode.
2679
2680 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2681 the data contained the byte 0x85 as part of a UTF-8 character within its
2682 first line. This applied both to normal and DFA matching.
2683
2684 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2685 /^[^d]*?$/8 failed to match "abc".
2686
2687 16. Added a missing copyright notice to pcrecpp_internal.h.
2688
2689 17. Make it more clear in the documentation that values returned from
2690 pcre_exec() in ovector are byte offsets, not character counts.
2691
2692 18. Tidied a few places to stop certain compilers from issuing warnings.
2693
2694 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2695 supplied by Stefan Weber. I made a further small update for 7.8 because
2696 there is a change of source arrangements: the pcre_searchfuncs.c module is
2697 replaced by pcre_ucd.c.
2698
2699
2700 Version 7.7 07-May-08
2701 ---------------------
2702
2703 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2704 a string to a long long, pretend we don't even have a long long." This is
2705 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2706
2707 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2708 pre-7.6 versions, which defined a global no_arg variable instead of putting
2709 it in the RE class. (See also #8 below.)
2710
2711 3. Remove a line of dead code, identified by coverity and reported by Nuno
2712 Lopes.
2713
2714 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2715
2716 (1) The include/exclude patterns were being applied to the whole pathnames
2717 of files, instead of just to the final components.
2718
2719 (2) If there was more than one level of directory, the subdirectories were
2720 skipped unless they satisfied the include/exclude conditions. This is
2721 inconsistent with GNU grep (and could even be seen as contrary to the
2722 pcregrep specification - which I improved to make it absolutely clear).
2723 The action now is always to scan all levels of directory, and just
2724 apply the include/exclude patterns to regular files.
2725
2726 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2727 --exclude_dir in the tests to avoid scanning .svn directories.
2728
2729 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2730 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2731 doesn't support NULs in patterns.
2732
2733 7. Added some missing "const"s to declarations of static tables in
2734 pcre_compile.c and pcre_dfa_exec.c.
2735
2736 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2737 caused by fix #2 above. (Subsequently also a second patch to fix the
2738 first patch. And a third patch - this was a messy problem.)
2739
2740 9. Applied Craig's patch to remove the use of push_back().
2741
2742 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2743 matching function regexec().
2744
2745 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2746 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2747 references. PCRE supports relative numbers with this syntax (I don't think
2748 Oniguruma does).
2749
2750 12. Previously, a group with a zero repeat such as (...){0} was completely
2751 omitted from the compiled regex. However, this means that if the group
2752 was called as a subroutine from elsewhere in the pattern, things went wrong
2753 (an internal error was given). Such groups are now left in the compiled
2754 pattern, with a new opcode that causes them to be skipped at execution
2755 time.
2756
2757 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2758 to the way PCRE behaves:
2759
2760 (a) A lone ] character is dis-allowed (Perl treats it as data).
2761
2762 (b) A back reference to an unmatched subpattern matches an empty string
2763 (Perl fails the current match path).
2764
2765 (c) A data ] in a character class must be notated as \] because if the
2766 first data character in a class is ], it defines an empty class. (In
2767 Perl it is not possible to have an empty class.) The empty class []
2768 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2769 The negative empty class [^] matches any one character, independently
2770 of the DOTALL setting.
2771
2772 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2773 non-existent subpattern following a character class starting with ']' and
2774 containing () gave an internal compiling error instead of "reference to
2775 non-existent subpattern". Fortunately, when the pattern did exist, the
2776 compiled code was correct. (When scanning forwards to check for the
2777 existence of the subpattern, it was treating the data ']' as terminating
2778 the class, so got the count wrong. When actually compiling, the reference
2779 was subsequently set up correctly.)
2780
2781 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2782 it was being rejected as not supported by pcre_dfa_exec(), even though
2783 other assertions are supported. I have made pcre_dfa_exec() support
2784 (*FAIL).
2785
2786 16. The implementation of 13c above involved the invention of a new opcode,
2787 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2788 cannot be changed at match time, I realized I could make a small
2789 improvement to matching performance by compiling OP_ALLANY instead of
2790 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2791 on the OP_ANY path.
2792
2793 17. Compiling pcretest on Windows with readline support failed without the
2794 following two fixes: (1) Make the unistd.h include conditional on
2795 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2796
2797 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2798 ncurses library to be included for pcretest when ReadLine support is
2799 requested, but also to allow for it to be overridden. This patch came from
2800 Daniel Bergström.
2801
2802 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2803 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2804 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2805 spotting this.
2806
2807
2808 Version 7.6 28-Jan-08
2809 ---------------------
2810
2811 1. A character class containing a very large number of characters with
2812 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2813 overflow.
2814
2815 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2816 HAVE_LONG_LONG is not defined.
2817
2818 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2819 bring them up to date and include new features. This patch includes:
2820
2821 - Fixed PH's badly added libz and libbz2 support.
2822 - Fixed a problem with static linking.
2823 - Added pcredemo. [But later removed - see 7 below.]
2824 - Fixed dftables problem and added an option.
2825 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2826 HAVE_LONG_LONG.
2827 - Added readline support for pcretest.
2828 - Added an listing of the option settings after cmake has run.
2829
2830 4. A user submitted a patch to Makefile that makes it easy to create
2831 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2832 Makefile.am that cause it to include this special target, without
2833 affecting anything else. Note that the same mingw target plus all
2834 the other distribution libraries and programs are now supported
2835 when configuring with CMake (see 6 below) instead of with
2836 Configure/Make.
2837
2838 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2839 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2840 exported in the Windows port". It has not yet been confirmed that the patch
2841 solves the problem, but it does no harm.
2842
2843 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2844 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2845 with CMake, and also correct the comment about stack recursion.
2846
2847 7. Remove the automatic building of pcredemo from the ./configure system and
2848 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2849 of a program that users should build themselves after PCRE is installed, so
2850 building it automatically is not really right. What is more, it gave
2851 trouble in some build environments.
2852
2853 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2854
2855
2856 Version 7.5 10-Jan-08
2857 ---------------------
2858
2859 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2860 values in parens when parsing an RE using the C++ wrapper."
2861
2862 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2863 Characters greater than 255 were excluded from the class instead of being
2864 included.
2865
2866 3. The same bug as (2) above applied to negated POSIX classes such as
2867 [:^space:].
2868
2869 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2870 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2871 I have changed it.
2872
2873 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2874 first named subpattern) and a construct such as (?&a) would reference the
2875 first named subpattern whose name started with "a" (in other words, the
2876 length check was missing). Both these problems are fixed. "Subpattern name
2877 expected" is now given for (?&) (a zero-length name), and this patch also
2878 makes it give the same error for \k'' (previously it complained that that
2879 was a reference to a non-existent subpattern).
2880
2881 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2882 this is right because (?- can be followed by option settings as well as by
2883 digits. I have, however, made the messages clearer.
2884
2885 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2886 than the number used in the conditional) now cause a compile-time error.
2887 This is actually not compatible with Perl, which accepts such patterns, but
2888 treats the conditional as always being FALSE (as PCRE used to), but it
2889 seems to me that giving a diagnostic is better.
2890
2891 8. Change "alphameric" to the more common word "alphanumeric" in comments
2892 and messages.
2893
2894 9. Fix two occurrences of "backslash" in comments that should have been
2895 "backspace".
2896
2897 10. Remove two redundant lines of code that can never be obeyed (their function
2898 was moved elsewhere).
2899
2900 11. The program that makes PCRE's Unicode character property table had a bug
2901 which caused it to generate incorrect table entries for sequences of
2902 characters that have the same character type, but are in different scripts.
2903 It amalgamated them into a single range, with the script of the first of
2904 them. In other words, some characters were in the wrong script. There were
2905 thirteen such cases, affecting characters in the following ranges:
2906
2907 U+002b0 - U+002c1
2908 U+0060c - U+0060d
2909 U+0061e - U+00612
2910 U+0064b - U+0065e
2911 U+0074d - U+0076d
2912 U+01800 - U+01805
2913 U+01d00 - U+01d77
2914 U+01d9b - U+01dbf
2915 U+0200b - U+0200f
2916 U+030fc - U+030fe
2917 U+03260 - U+0327f
2918 U+0fb46 - U+0fbb1
2919 U+10450 - U+1049d
2920
2921 12. The -o option (show only the matching part of a line) for pcregrep was not
2922 compatible with GNU grep in that, if there was more than one match in a
2923 line, it showed only the first of them. It now behaves in the same way as
2924 GNU grep.
2925
2926 13. If the -o and -v options were combined for pcregrep, it printed a blank
2927 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2928 does the same. The return code can be used to tell if there were any
2929 non-matching lines.
2930
2931 14. Added --file-offsets and --line-offsets to pcregrep.
2932
2933 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2934 infinitely looping recursion. The bug was that positive lookaheads were not
2935 being skipped when checking for a possible empty match (negative lookaheads
2936 and both kinds of lookbehind were skipped).
2937
2938 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2939 inclusion of <windows.h> to before rather than after the definition of
2940 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2941
2942 17. Specifying a possessive quantifier with a specific limit for a Unicode
2943 character property caused pcre_compile() to compile bad code, which led at
2944 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2945 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2946 caused the error; without that there was no problem.
2947
2948 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2949
2950 19. Added --enable-pcretest-libreadline.
2951
2952 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2953 RE::GlobalReplace(). As a result, the number of replacements returned was
2954 double what it should be. I removed one of the increments, but Craig sent a
2955 later patch that removed the other one (the right fix) and added unit tests
2956 that check the return values (which was not done before).
2957
2958 21. Several CMake things:
2959
2960 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2961 the names libpcre and libpcreposix, not just pcre and pcreposix.
2962
2963 (2) The above change means that pcretest and pcregrep are now correctly
2964 linked with the newly-built libraries, not previously installed ones.
2965
2966 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2967
2968 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2969 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2970 UTF-8 newline character). The key issue is that the pattern starts .*;
2971 this means that the match must be either at the beginning, or after a
2972 newline. The bug was in the code for advancing after a failed match and
2973 checking that the new position followed a newline. It was not taking
2974 account of UTF-8 characters correctly.
2975
2976 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2977 character classes. PCRE was not treating the sequence [:...:] as a
2978 character class unless the ... were all letters. Perl, however, seems to
2979 allow any characters between [: and :], though of course it rejects as
2980 unknown any "names" that contain non-letters, because all the known class
2981 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2982 for example, whereas PCRE did not - it did not recognize a POSIX character
2983 class. This seemed a bit dangerous, so the code has been changed to be
2984 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2985 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2986 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2987 Perl does, and where it didn't before.
2988
2989 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2990 Windows environments %n is disabled by default.
2991
2992
2993 Version 7.4 21-Sep-07
2994 ---------------------
2995
2996 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2997 means that a class such as [\s] counted as "explicit reference to CR or
2998 LF". That isn't really right - the whole point of the change was to try to
2999 help when there was an actual mention of one of the two characters. So now
3000 the change happens only if \r or \n (or a literal CR or LF) character is
3001 encountered.
3002
3003 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3004 of both had grown to the point where there were only 3 bits left.
3005 Fortunately, there was spare space in the data structure, and so I have
3006 moved the internal flags into a new 16-bit field to free up more option
3007 bits.
3008
3009 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3010 but did not set the internal JCHANGED flag - either of these is enough to
3011 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3012 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3013 start sets both bits.
3014
3015 4. Added options (at build time, compile time, exec time) to change \R from
3016 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3017
3018 5. doc/pcresyntax.html was missing from the distribution.
3019
3020 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3021 compatibility, even though it is no longer used.
3022
3023 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3024 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3025 windows.h file is present (where different names are used). [This was
3026 reversed later after testing - see 16 below.]
3027
3028 8. Changed all #include <config.h> to #include "config.h". There were also
3029 some further <pcre.h> cases that I changed to "pcre.h".
3030
3031 9. When pcregrep was used with the --colour option, it missed the line ending
3032 sequence off the lines that it output.
3033
3034 10. It was pointed out to me that arrays of string pointers cause lots of
3035 relocations when a shared library is dynamically loaded. A technique of
3036 using a single long string with a table of offsets can drastically reduce
3037 these. I have refactored PCRE in four places to do this. The result is
3038 dramatic:
3039
3040 Originally: 290
3041 After changing UCP table: 187
3042 After changing error message table: 43
3043 After changing table of "verbs" 36
3044 After changing table of Posix names 22
3045
3046 Thanks to the folks working on Gregex for glib for this insight.
3047
3048 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3049 unicode-properties was also set.
3050
3051 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3052
3053 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3054 checked only for CRLF.
3055
3056 14. Added casts to pcretest.c to avoid compiler warnings.
3057
3058 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3059
3060 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3061 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3062 entirely. This removes changes made in 7 above.
3063
3064 17. The CMake files have been updated, and there is now more information about
3065 building with CMake in the NON-UNIX-USE document.
3066
3067
3068 Version 7.3 28-Aug-07
3069 ---------------------
3070
3071 1. In the rejigging of the build system that eventually resulted in 7.1, the
3072 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3073 brackets there is not right, since it causes compilers to look for an
3074 installed pcre.h, not the version that is in the source that is being
3075 compiled (which of course may be different). I have changed it back to:
3076
3077 #include "pcre.h"
3078
3079 I have a vague recollection that the change was concerned with compiling in
3080 different directories, but in the new build system, that is taken care of
3081 by the VPATH setting the Makefile.
3082
3083 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3084 when the subject happened to end in the byte 0x85 (e.g. if the last
3085 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3086 characters but of course it shouldn't be taken as a newline when it is part
3087 of another character. The bug was that, for an unlimited repeat of . in
3088 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3089 characters when looking for a newline.
3090
3091 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3092
3093 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3094 in debug output.
3095
3096 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3097 long printing in the pcrecpp unittest when running under MinGW.
3098
3099 6. ESC_K was left out of the EBCDIC table.
3100
3101 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3102 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3103 limit also applies to "virtual nesting" when a pattern is recursive, and in
3104 this case 1000 isn't so big. I have been able to remove this limit at the
3105 expense of backing off one optimization in certain circumstances. Normally,
3106 when pcre_exec() would call its internal match() function recursively and
3107 immediately return the result unconditionally, it uses a "tail recursion"
3108 feature to save stack. However, when a subpattern that can match an empty
3109 string has an unlimited repetition quantifier, it no longer makes this
3110 optimization. That gives it a stack frame in which to save the data for
3111 checking that an empty string has been matched. Previously this was taken
3112 from the 1000-entry workspace that had been reserved. So now there is no
3113 explicit limit, but more stack is used.
3114
3115 8. Applied Daniel's patches to solve problems with the import/export magic
3116 syntax that is required for Windows, and which was going wrong for the
3117 pcreposix and pcrecpp parts of the library. These were overlooked when this
3118 problem was solved for the main library.
3119
3120 9. There were some crude static tests to avoid integer overflow when computing
3121 the size of patterns that contain repeated groups with explicit upper
3122 limits. As the maximum quantifier is 65535, the maximum group length was
3123 set at 30,000 so that the product of these two numbers did not overflow a
3124 32-bit integer. However, it turns out that people want to use groups that
3125 are longer than 30,000 bytes (though not repeat them that many times).
3126 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3127 made it possible to implement the integer overflow checks in a much more
3128 dynamic way, which I have now done. The artificial limitation on group
3129 length has been removed - we now have only the limit on the total length of
3130 the compiled pattern, which depends on the LINK_SIZE setting.
3131
3132 10. Fixed a bug in the documentation for get/copy named substring when
3133 duplicate names are permitted. If none of the named substrings are set, the
3134 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3135 empty string.
3136
3137 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3138 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3139 because the ] is interpreted as the first data character and the
3140 terminating ] is not found. PCRE has been made compatible with Perl in this
3141 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3142 cause memory overwriting.
3143
3144 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3145 string has been matched (to stop an infinite loop). It was not recognizing
3146 a conditional subpattern that could match an empty string if that
3147 subpattern was within another subpattern. For example, it looped when
3148 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3149 condition was not nested. This bug has been fixed.
3150
3151 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3152 past the start of the subject in the presence of bytes with the top bit
3153 set, for example "\x8aBCD".
3154
3155 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3156 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3157
3158 14. Optimized (?!) to (*FAIL).
3159
3160 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3161 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3162 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3163 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3164 does: it's just the validity check that is more restrictive.
3165
3166 16. Inserted checks for integer overflows during escape sequence (backslash)
3167 processing, and also fixed erroneous offset values for syntax errors during
3168 backslash processing.
3169
3170 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3171 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3172
3173 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3174 caused an overrun.
3175
3176 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3177 something other than just ASCII characters) inside a group that had an
3178 unlimited repeat caused a loop at compile time (while checking to see
3179 whether the group could match an empty string).
3180
3181 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3182 [\P{Any}] did so. (Error in the code for printing property names.)
3183
3184 21. An orphan \E inside a character class could cause a crash.
3185
3186 22. A repeated capturing bracket such as (A)? could cause a wild memory
3187 reference during compilation.
3188
3189 23. There are several functions in pcre_compile() that scan along a compiled
3190 expression for various reasons (e.g. to see if it's fixed length for look
3191 behind). There were bugs in these functions when a repeated \p or \P was
3192 present in the pattern. These operators have additional parameters compared
3193 with \d, etc, and these were not being taken into account when moving along
3194 the compiled data. Specifically:
3195
3196 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3197 length.
3198
3199 (b) An item such as \pL+ within a repeated group could cause crashes or
3200 loops.
3201
3202 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3203 "reference to non-existent subpattern" error.
3204
3205 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3206
3207 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3208 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3209
3210 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3211
3212 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3213 character were causing crashes (broken optimization).
3214
3215 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3216 \p or \P) caused a compile-time loop.
3217
3218 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3219 break. For example, the unstudied pattern [\r\n]A does not match the string
3220 "\r\nA" because change 7.0/46 below moves the current point on by two
3221 characters after failing to match at the start. However, the pattern \nA
3222 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3223 the same is true. There doesn't seem any very clean way out of this, but
3224 what I have chosen to do makes the common cases work: PCRE now takes note
3225 of whether there can be an explicit match for \r or \n anywhere in the
3226 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3227 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3228 pattern has explicit CR or LF references.
3229
3230 29. Added (*CR) etc for changing newline setting at start of pattern.
3231
3232
3233 Version 7.2 19-Jun-07
3234 ---------------------
3235
3236 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3237 which is apparently normally available under Windows.
3238
3239 2. Re-jig the pcregrep tests with different newline settings in an attempt
3240 to make them independent of the local environment's newline setting.
3241
3242 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3243
3244 4. Some of the "internals" tests were previously cut out when the link size
3245 was not 2, because the output contained actual offsets. The recent new
3246 "Z" feature of pcretest means that these can be cut out, making the tests
3247 usable with all link sizes.
3248
3249 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3250 stack recursion. This gives a massive performance boost under BSD, but just
3251 a small improvement under Linux. However, it saves one field in the frame
3252 in all cases.
3253
3254 6. Added more features from the forthcoming Perl 5.10:
3255
3256 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3257 recursion call. It refers to the nth most recently opened parentheses.
3258
3259 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3260 to be opened parentheses.
3261
3262 (c) Conditions that refer to capturing parentheses can be specified
3263 relatively, for example, (?(-2)... or (?(+3)...
3264
3265 (d) \K resets the start of the current match so that everything before
3266 is not part of it.
3267
3268 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3269
3270 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3271 reference syntax.
3272
3273 (g) (?| introduces a group in which the numbering of parentheses in each
3274 alternative starts with the same number.
3275
3276 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3277
3278 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3279 PCRE_INFO_JCHANGED.
3280
3281 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3282 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3283 for detecting groups that can match an empty string.
3284
3285 9. A pattern with a very large number of alternatives (more than several
3286 hundred) was running out of internal workspace during the pre-compile
3287 phase, where pcre_compile() figures out how much memory will be needed. A
3288 bit of new cunning has reduced the workspace needed for groups with
3289 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3290 workspace instead of running out of the 4096 that are available.
3291
3292 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3293
3294 11. Applied patch from Google to remove an optimization that didn't quite work.
3295 The report of the bug said:
3296
3297 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3298 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3299 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3300
3301 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3302 it matched the wrong number of bytes.
3303
3304
3305 Version 7.1 24-Apr-07
3306 ---------------------
3307
3308 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3309 that is more "standard", making use of automake and other Autotools. There
3310 is some re-arrangement of the files and adjustment of comments consequent
3311 on this.
3312
3313 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3314 for recursive directory scanning broke on some systems because the files
3315 are not scanned in any specific order and on different systems the order
3316 was different. A call to "sort" has been inserted into RunGrepTest for the
3317 approprate test as a short-term fix. In the longer term there may be an
3318 alternative.
3319
3320 3. I had an email from Eric Raymond about problems translating some of PCRE's
3321 man pages to HTML (despite the fact that I distribute HTML pages, some
3322 people do their own conversions for various reasons). The problems
3323 concerned the use of low-level troff macros .br and .in. I have therefore
3324 removed all such uses from the man pages (some were redundant, some could
3325 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3326 HTML has been updated to handle .nf/.fi and to complain if it encounters
3327 .br or .in.
3328
3329 4. Updated comments in configure.ac that get placed in config.h.in and also
3330 arranged for config.h to be included in the distribution, with the name
3331 config.h.generic, for the benefit of those who have to compile without
3332 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3333
3334 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3335 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3336 makevp.bat for the current PCRE, using the additional files
3337 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3338
3339 6. A Windows user reported a minor discrepancy with test 2, which turned out
3340 to be caused by a trailing space on an input line that had got lost in his
3341 copy. The trailing space was an accident, so I've just removed it.
3342
3343 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3344 that is needed.
3345
3346 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3347 as "const" (a) because they are and (b) because it helps the PHP
3348 maintainers who have recently made a script to detect big data structures
3349 in the php code that should be moved to the .rodata section. I remembered
3350 to update Builducptable as well, so it won't revert if ucptable.h is ever
3351 re-created.
3352
3353 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3354 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3355 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3356 support is not required. This saves 1.5-2K of code, which is important in
3357 some applications.
3358
3359 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3360 so as not to refer to the tables, even though these functions will never be
3361 called when UTF-8 support is disabled. Otherwise there are problems with a
3362 shared library.
3363
3364 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3365
3366 (a) It was defining its arguments as char * instead of void *.
3367
3368 (b) It was assuming that all moves were upwards in memory; this was true
3369 a long time ago when I wrote it, but is no longer the case.
3370
3371 The emulated memove() is provided for those environments that have neither
3372 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3373 is clearly not the case, as these two bugs were recently reported.
3374
3375 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3376 and Detrail to create the HTML documentation, the .txt form of the man
3377 pages, and it removes trailing spaces from listed files. It also creates
3378 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3379 case, it wraps all the #defines with #ifndefs. This script should be run
3380 before "make dist".
3381
3382 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3383 with Unicode property support.
3384
3385 (a) For a maximizing quantifier, if the two different cases of the
3386 character were of different lengths in their UTF-8 codings (there are
3387 some cases like this - I found 11), and the matching function had to
3388 back up over a mixture of the two cases, it incorrectly assumed they
3389 were both the same length.
3390
3391 (b) When PCRE was configured to use the heap rather than the stack for
3392 recursion during matching, it was not correctly preserving the data for
3393 the other case of a UTF-8 character when checking ahead for a match
3394 while processing a minimizing repeat. If the check also involved
3395 matching a wide character, but failed, corruption could cause an
3396 erroneous result when trying to check for a repeat of the original
3397 character.
3398
3399 13. Some tidying changes to the testing mechanism:
3400
3401 (a) The RunTest script now detects the internal link size and whether there
3402 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3403 values substituted by "configure". (The RunGrepTest script already did
3404 this for UTF-8.) The configure.ac script no longer substitutes the
3405 relevant variables.
3406
3407 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3408 with length and offset values. This means that the output is different
3409 for different internal link sizes. Test 2 is skipped for link sizes
3410 other than 2 because of this, bypassing the problem. Unfortunately,
3411 there was also a test in test 3 (the locale tests) that used /B and
3412 failed for link sizes other than 2. Rather than cut the whole test out,
3413 I have added a new /Z option to pcretest that replaces the length and
3414 offset values with spaces. This is now used to make test 3 independent
3415 of link size. (Test 2 will be tidied up later.)
3416
3417 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3418 segmentation fault instead of returning the appropriate error message.
3419
3420 15. In multiline mode when the newline sequence was set to "any", the pattern
3421 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3422 This doesn't seem right; it now treats the CRLF combination as the line
3423 ending, and so does not match in that case. It's only a pattern such as ^$
3424 that would hit this one: something like ^ABC$ would have failed after \r
3425 and then tried again after \r\n.
3426
3427 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3428 in an attempt to make files that differ only in their line terminators
3429 compare equal. This works on Linux.
3430
3431 17. Under certain error circumstances pcregrep might try to free random memory
3432 as it exited. This is now fixed, thanks to valgrind.
3433
3434 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3435 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3436 was because its rules for how to advance for /g after matching an empty
3437 string at the end of a line did not allow for this case. They now check for
3438 it specially.
3439
3440 20. pcretest is supposed to handle patterns and data of any length, by
3441 extending its buffers when necessary. It was getting this wrong when the
3442 buffer for a data line had to be extended.
3443
3444 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3445 CRLF as a newline sequence.
3446
3447 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3448 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3449 I have nevertheless tidied it up.
3450
3451 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3452
3453 24. Added a man page for pcre-config.
3454
3455
3456 Version 7.0 19-Dec-06
3457 ---------------------
3458
3459 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3460 moving to gcc 4.1.1.
3461
3462 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3463 sys/time.h, which is documented as needed for this function. It doesn't
3464 seem to matter on Linux, but it showed up on some releases of OS X.
3465
3466 3. It seems that there are systems where bytes whose values are greater than
3467 127 match isprint() in the "C" locale. The "C" locale should be the
3468 default when a C program starts up. In most systems, only ASCII printing
3469 characters match isprint(). This difference caused the output from pcretest
3470 to vary, making some of the tests fail. I have changed pcretest so that:
3471
3472 (a) When it is outputting text in the compiled version of a pattern, bytes
3473 other than 32-126 are always shown as hex escapes.
3474
3475 (b) When it is outputting text that is a matched part of a subject string,
3476 it does the same, unless a different locale has been set for the match
3477 (using the /L modifier). In this case, it uses isprint() to decide.
3478
3479 4. Fixed a major bug that caused incorrect computation of the amount of memory
3480 required for a compiled pattern when options that changed within the
3481 pattern affected the logic of the preliminary scan that determines the
3482 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3483 that the computed length was too small. The symptoms of this bug were
3484 either the PCRE error "internal error: code overflow" from pcre_compile(),
3485 or a glibc crash with a message such as "pcretest: free(): invalid next
3486 size (fast)". Examples of patterns that provoked this bug (shown in
3487 pcretest format) are:
3488
3489 /(?-x: )/x
3490 /(?x)(?-x: \s*#\s*)/
3491 /((?i)[\x{c0}])/8
3492 /(?i:[\x{c0}])/8
3493
3494 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3495 is now done differently.
3496
3497 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3498 wrapper classes; (b) implement a new function in the C++ scanner that is
3499 more efficient than the old way of doing things because it avoids levels of
3500 recursion in the regex matching; (c) add a paragraph to the documentation
3501 for the FullMatch() function.
3502
3503 6. The escape sequence \n was being treated as whatever was defined as
3504 "newline". Not only was this contrary to the documentation, which states
3505 that \n is character 10 (hex 0A), but it also went horribly wrong when
3506 "newline" was defined as CRLF. This has been fixed.
3507
3508 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3509 was being set to -1 for the "end of line" case (supposedly a value that no
3510 character can have). Though this value is never used (the check for end of
3511 line is "zero bytes in current character"), it caused compiler complaints.
3512 I've changed it to 0xffffffff.
3513
3514 8. In pcre_version.c, the version string was being built by a sequence of
3515 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3516 string (as it is for production releases) called a macro with an empty
3517 argument. The C standard says the result of this is undefined. The gcc
3518 compiler treats it as an empty string (which was what was wanted) but it is
3519 reported that Visual C gives an error. The source has been hacked around to
3520 avoid this problem.
3521
3522 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3523 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3524 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3525 of them did).
3526
3527 10. Originally, pcretest opened its input and output without "b"; then I was
3528 told that "b" was needed in some environments, so it was added for release
3529 5.0 to both the input and output. (It makes no difference on Unix-like
3530 systems.) Later I was told that it is wrong for the input on Windows. I've
3531 now abstracted the modes into two macros, to make it easier to fiddle with
3532 them, and removed "b" from the input mode under Windows.
3533
3534 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3535
3536 12. Added -help and --help to pcretest as an official way of being reminded
3537 of the options.
3538
3539 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3540 and pcrecpp.cc because they annoy compilers at high warning levels.
3541
3542 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3543
3544 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3545 scripts are not C programs :-) and which was not noticed because it works
3546 on Linux.
3547
3548 16. pcretest is supposed to handle any length of pattern and data line (as one
3549 line or as a continued sequence of lines) by extending its input buffer if
3550 necessary. This feature was broken for very long pattern lines, leading to
3551 a string of junk being passed to pcre_compile() if the pattern was longer
3552 than about 50K.
3553
3554 17. I have done a major re-factoring of the way pcre_compile() computes the
3555 amount of memory needed for a compiled pattern. Previously, there was code
3556 that made a preliminary scan of the pattern in order to do this. That was
3557 OK when PCRE was new, but as the facilities have expanded, it has become
3558 harder and harder to keep it in step with the real compile phase, and there
3559 have been a number of bugs (see for example, 4 above). I have now found a
3560 cunning way of running the real compile function in a "fake" mode that
3561 enables it to compute how much memory it would need, while actually only
3562 ever using a few hundred bytes of working memory and without too many
3563 tests of the mode. This should make future maintenance and development
3564 easier. A side effect of this work is that the limit of 200 on the nesting
3565 depth of parentheses has been removed (though this was never a serious
3566 limitation, I suspect). However, there is a downside: pcre_compile() now
3567 runs more slowly than before (30% or more, depending on the pattern). I
3568 hope this isn't a big issue. There is no effect on runtime performance.
3569
3570 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3571 newline (only possible for the last line of a file) and it was a
3572 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3573
3574 19. Added additional timing features to pcretest. (1) The -tm option now times
3575 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3576 separate command line item, by a number that specifies the number of
3577 repeats to use when timing. The default is 50000; this gives better
3578 precision, but takes uncomfortably long for very large patterns.
3579
3580 20. Extended pcre_study() to be more clever in cases where a branch of a
3581 subpattern has no definite first character. For example, (a*|b*)[cd] would
3582 previously give no result from pcre_study(). Now it recognizes that the
3583 first character must be a, b, c, or d.
3584
3585 21. There was an incorrect error "recursive call could loop indefinitely" if
3586 a subpattern (or the entire pattern) that was being tested for matching an
3587 empty string contained only one non-empty item after a nested subpattern.
3588 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3589 incorrectly, because the \d was being skipped in the check.
3590
3591 22. The pcretest program now has a new pattern option /B and a command line
3592 option -b, which is equivalent to adding /B to every pattern. This causes
3593 it to show the compiled bytecode, without the additional information that
3594 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3595 is the same as /B/I).
3596
3597 23. A new optimization is now able automatically to treat some sequences such
3598 as a*b as a*+b. More specifically, if something simple (such as a character
3599 or a simple class like \d) has an unlimited quantifier, and is followed by
3600 something that cannot possibly match the quantified thing, the quantifier
3601 is automatically "possessified".
3602
3603 24. A recursive reference to a subpattern whose number was greater than 39
3604 went wrong under certain circumstances in UTF-8 mode. This bug could also
3605 have affected the operation of pcre_study().
3606
3607 25. Realized that a little bit of performance could be had by replacing
3608 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3609
3610 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3611
3612 27. Possessive quantifiers such as a++ were previously implemented by turning
3613 them into atomic groups such as ($>a+). Now they have their own opcodes,
3614 which improves performance. This includes the automatically created ones
3615 from 23 above.
3616
3617 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3618 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3619 the first matched character to be a colon. This applied both to named and
3620 numbered groups.
3621
3622 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3623
3624 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3625 building PCRE on Windows easier, so I have included it in the distribution.
3626
3627 31. There is now a check in pcretest against a ridiculously large number being
3628 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3629 loop, the loop is abandoned.
3630
3631 32. Forward references to subpatterns in conditions such as (?(2)...) where
3632 subpattern 2 is defined later cause pcre_compile() to search forwards in
3633 the pattern for the relevant set of parentheses. This search went wrong
3634 when there were unescaped parentheses in a character class, parentheses
3635 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3636
3637 33. "Subroutine" calls and backreferences were previously restricted to
3638 referencing subpatterns earlier in the regex. This restriction has now
3639 been removed.
3640
3641 34. Added a number of extra features that are going to be in Perl 5.10. On the
3642 whole, these are just syntactic alternatives for features that PCRE had
3643 previously implemented using the Python syntax or my own invention. The
3644 other formats are all retained for compatibility.
3645
3646 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3647 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3648 also .NET compatible.
3649
3650 (b) A recursion or subroutine call to a named group can now be defined as
3651 (?&name) as well as (?P>name).
3652
3653 (c) A backreference to a named group can now be defined as \k<name> or
3654 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3655 5.10, are also .NET compatible.
3656
3657 (d) A conditional reference to a named group can now use the syntax
3658 (?(<name>) or (?('name') as well as (?(name).
3659
3660 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3661 groups (named and numbered) that are never evaluated inline, but can be
3662 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3663 is always false. There may be only one alternative in such a group.
3664
3665 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3666 as the simple (?(R). The condition is true only if the most recent
3667 recursion is that of the given number or name. It does not search out
3668 through the entire recursion stack.
3669
3670 (g) The escape \gN or \g{N} has been added, where N is a positive or
3671 negative number, specifying an absolute or relative reference.
3672
3673 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3674 some "unreachable code" warnings.
3675
3676 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3677 things, this adds five new scripts.
3678
3679 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3680 There were also incompatibilities regarding the handling of \Q..\E inside
3681 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3682 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3683
3684 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3685 matches an empty string, and forcibly breaks the loop. There were bugs in
3686 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3687 against aaaa the result was just "a" rather than "aaaa", for example. Two
3688 separate and independent bugs (that affected different cases) have been
3689 fixed.
3690
3691 39. Refactored the code to abolish the use of different opcodes for small
3692 capturing bracket numbers. This is a tidy that I avoided doing when I
3693 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3694 The new approach is not only tidier, it makes it possible to reduce the
3695 memory needed to fix the previous bug (38).
3696
3697 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3698 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3699 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3700 mode.
3701
3702 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3703 report.
3704
3705 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3706 copy construction and assignment in the C++ wrapper.
3707
3708 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3709 couple of bugs that could have given wrong results in the "--newline=crlf"
3710 case.
3711
3712 44. Added a number of casts and did some reorganization of signed/unsigned int
3713 variables following suggestions from Dair Grant. Also renamed the variable
3714 "this" as "item" because it is a C++ keyword.
3715
3716 45. Arranged for dftables to add
3717
3718 #include "pcre_internal.h"
3719
3720 to pcre_chartables.c because without it, gcc 4.x may remove the array
3721 definition from the final binary if PCRE is built into a static library and
3722 dead code stripping is activated.
3723
3724 46. For an unanchored pattern, if a match attempt fails at the start of a
3725 newline sequence, and the newline setting is CRLF or ANY, and the next two
3726 characters are CRLF, advance by two characters instead of one.
3727
3728
3729 Version 6.7 04-Jul-06
3730 ---------------------
3731
3732 1. In order to handle tests when input lines are enormously long, pcretest has
3733 been re-factored so that it automatically extends its buffers when
3734 necessary. The code is crude, but this _is_ just a test program. The
3735 default size has been increased from 32K to 50K.
3736
3737 2. The code in pcre_study() was using the value of the re argument before
3738 testing it for NULL. (Of course, in any sensible call of the function, it
3739 won't be NULL.)
3740
3741 3. The memmove() emulation function in pcre_internal.h, which is used on
3742 systems that lack both memmove() and bcopy() - that is, hardly ever -
3743 was missing a "static" storage class specifier.
3744
3745 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3746 containing an extended class (one that cannot be represented by a bitmap
3747 because it contains high-valued characters or Unicode property items, e.g.
3748 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3749 pattern, but PCRE should not loop if you do not (it no longer does).
3750 [Detail: two cases were found: (a) a repeated subpattern containing an
3751 extended class; (b) a recursive reference to a subpattern that followed a
3752 previous extended class. It wasn't skipping over the extended class
3753 correctly when UTF-8 mode was not set.]
3754
3755 5. A negated single-character class was not being recognized as fixed-length
3756 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3757 compile error "lookbehind assertion is not fixed length".
3758
3759 6. The RunPerlTest auxiliary script was showing an unexpected difference
3760 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3761 write a Perl script that can interpret lines of an input file either as
3762 byte characters or as UTF-8, which is what "perltest" was being required to
3763 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3764 can't do is switch easily at run time between having the "use utf8;" pragma
3765 or not. In the end, I fudged it by using the RunPerlTest script to insert
3766 "use utf8;" explicitly for the UTF-8 tests.
3767
3768 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3769 the end of the subject string, contrary to the documentation and to what
3770 Perl does. This was true of both matching functions. Now it matches only at
3771 the start of the subject and immediately after *internal* newlines.
3772
3773 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3774 a pointer to an int instead of a pointer to an unsigned long int. This
3775 caused problems on 64-bit systems.
3776
3777 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3778 instance of the 'standard' template library not being so standard".
3779
3780 10. There was no check on the number of named subpatterns nor the maximum
3781 length of a subpattern name. The product of these values is used to compute
3782 the size of the memory block for a compiled pattern. By supplying a very
3783 long subpattern name and a large number of named subpatterns, the size
3784 computation could be caused to overflow. This is now prevented by limiting
3785 the length of names to 32 characters, and the number of named subpatterns
3786 to 10,000.
3787
3788 11. Subpatterns that are repeated with specific counts have to be replicated in
3789 the compiled pattern. The size of memory for this was computed from the
3790 length of the subpattern and the repeat count. The latter is limited to
3791 65535, but there was no limit on the former, meaning that integer overflow
3792 could in principle occur. The compiled length of a repeated subpattern is
3793 now limited to 30,000 bytes in order to prevent this.
3794
3795 12. Added the optional facility to have named substrings with the same name.
3796
3797 13. Added the ability to use a named substring as a condition, using the
3798 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3799 are numbers (not recommended). Forward references are permitted.
3800
3801 14. Added forward references in named backreferences (if you see what I mean).
3802
3803 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3804 pattern could run off the end of the subject. For example, the pattern
3805 "(?s)(.{1,5})"8 did this with the subject "ab".
3806
3807 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3808 PCRE_CASELESS was set when matching characters that were quantified with ?
3809 or *.
3810
3811 17. A character class other than a single negated character that had a minimum
3812 but no maximum quantifier - for example [ab]{6,} - was not handled
3813 correctly by pce_dfa_exec(). It would match only one character.
3814
3815 18. A valid (though odd) pattern that looked like a POSIX character
3816 class but used an invalid character after [ (for example [[,abc,]]) caused
3817 pcre_compile() to give the error "Failed: internal error: code overflow" or
3818 in some cases to crash with a glibc free() error. This could even happen if
3819 the pattern terminated after [[ but there just happened to be a sequence of
3820 letters, a binary zero, and a closing ] in the memory that followed.
3821
3822 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3823 over the years. Originally (before any Unicode support), just the bottom 8
3824 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3825 output from "man perlunicode" includes this:
3826
3827 The regular expression compiler produces polymorphic opcodes. That
3828 is, the pattern adapts to the data and automatically switches to
3829 the Unicode character scheme when presented with Unicode data--or
3830 instead uses a traditional byte scheme when presented with byte
3831 data.
3832
3833 Sadly, a wide octal escape does not cause a switch, and in a string with
3834 no other multibyte characters, these octal escapes are treated as before.
3835 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3836 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3837 Unicode string.
3838
3839 I have not perpetrated such confusion in PCRE. Up till now, it took just
3840 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3841 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3842 translate to the appropriate multibyte character.
3843
3844 29. Applied some refactoring to reduce the number of warnings from Microsoft
3845 and Borland compilers. This has included removing the fudge introduced
3846 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3847 a warning about an unused variable.
3848
3849 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3850 characters since release 4.0, because Perl (from release 5.004) does not.
3851 [Or at least, is documented not to: some releases seem to be in conflict
3852 with the documentation.] However, when a pattern was studied with
3853 pcre_study() and all its branches started with \s, PCRE still included VT
3854 as a possible starting character. Of course, this did no harm; it just
3855 caused an unnecessary match attempt.
3856
3857 22. Removed a now-redundant internal flag bit that recorded the fact that case
3858 dependency changed within the pattern. This was once needed for "required
3859 byte" processing, but is no longer used. This recovers a now-scarce options
3860 bit. Also moved the least significant internal flag bit to the most-
3861 significant bit of the word, which was not previously used (hangover from
3862 the days when it was an int rather than a uint) to free up another bit for
3863 the future.
3864
3865 23. Added support for CRLF line endings as well as CR and LF. As well as the
3866 default being selectable at build time, it can now be changed at runtime
3867 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3868 specify that it is scanning data with non-default line endings.
3869
3870 24. Changed the definition of CXXLINK to make it agree with the definition of
3871 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3872
3873 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3874 recursions. This makes a big different to stack usage for some patterns.
3875
3876 26. If a subpattern containing a named recursion or subroutine reference such
3877 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3878 the space required for the compiled pattern went wrong and gave too small a
3879 value. Depending on the environment, this could lead to "Failed: internal
3880 error: code overflow at offset 49" or "glibc detected double free or
3881 corruption" errors.
3882
3883 27. Applied patches from Google (a) to support the new newline modes and (b) to
3884 advance over multibyte UTF-8 characters in GlobalReplace.
3885
3886 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3887 difference for some implementation of PCRE in some Windows version.
3888
3889 29. Added some extra testing facilities to pcretest:
3890
3891 \q<number> in a data line sets the "match limit" value
3892 \Q<number> in a data line sets the "match recursion limt" value
3893 -S <number> sets the stack size, where <number> is in megabytes
3894
3895 The -S option isn't available for Windows.
3896
3897
3898 Version 6.6 06-Feb-06
3899 ---------------------
3900
3901 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3902 in pcreposix.h. I have copied the definition from pcre.h.
3903
3904 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3905 because pcre.h is no longer a built file.
3906
3907 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3908 not normally included in the compiled code.
3909
3910
3911 Version 6.5 01-Feb-06
3912 ---------------------
3913
3914 1. When using the partial match feature with pcre_dfa_exec(), it was not
3915 anchoring the second and subsequent partial matches at the new starting
3916 point. This could lead to incorrect results. For example, with the pattern
3917 /1234/, partially matching against "123" and then "a4" gave a match.
3918
3919 2. Changes to pcregrep:
3920
3921 (a) All non-match returns from pcre_exec() were being treated as failures
3922 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3923 error message is output. Some extra information is given for the
3924 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3925 probably the only errors that are likely to be caused by users (by
3926 specifying a regex that has nested indefinite repeats, for instance).
3927 If there are more than 20 of these errors, pcregrep is abandoned.
3928
3929 (b) A binary zero was treated as data while matching, but terminated the
3930 output line if it was written out. This has been fixed: binary zeroes
3931 are now no different to any other data bytes.
3932
3933 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3934 used to set a locale for matching. The --locale=xxxx long option has
3935 been added (no short equivalent) to specify a locale explicitly on the
3936 pcregrep command, overriding the environment variables.
3937
3938 (d) When -B was used with -n, some line numbers in the output were one less
3939 than they should have been.
3940
3941 (e) Added the -o (--only-matching) option.
3942
3943 (f) If -A or -C was used with -c (count only), some lines of context were
3944 accidentally printed for the final match.
3945
3946 (g) Added the -H (--with-filename) option.
3947
3948 (h) The combination of options -rh failed to suppress file names for files
3949 that were found from directory arguments.
3950
3951 (i) Added the -D (--devices) and -d (--directories) options.
3952
3953 (j) Added the -F (--fixed-strings) option.
3954
3955 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3956
3957 (l) Added the --colo(u)r option.
3958
3959 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3960 is not present by default.
3961
3962 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3963 items such as (?R) or (?1), when the recursion could match a number of
3964 alternatives. If it matched one of the alternatives, but subsequently,
3965 outside the recursion, there was a failure, the code tried to back up into
3966 the recursion. However, because of the way PCRE is implemented, this is not
3967 possible, and the result was an incorrect result from the match.
3968
3969 In order to prevent this happening, the specification of recursion has
3970 been changed so that all such subpatterns are automatically treated as
3971 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3972
3973 4. I had overlooked the fact that, in some locales, there are characters for
3974 which isalpha() is true but neither isupper() nor islower() are true. In
3975 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3976 and ordfeminine) are like this. This affected the treatment of \w and \W
3977 when they appeared in character classes, but not when they appeared outside
3978 a character class. The bit map for "word" characters is now created
3979 separately from the results of isalnum() instead of just taking it from the
3980 upper, lower, and digit maps. (Plus the underscore character, of course.)
3981
3982 5. The above bug also affected the handling of POSIX character classes such as
3983 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3984 permanent tables. Instead, the bit maps for such a class were previously
3985 created as the appropriate unions of the upper, lower, and digit bitmaps.
3986 Now they are created by subtraction from the [[:word:]] class, which has
3987 its own bitmap.
3988
3989 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3990 It is created by subtracting the vertical space characters (\x09, \x0a,
3991 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3992 subtraction was done in the overall bitmap for a character class, meaning
3993 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3994 be recognized. This bug has been fixed.
3995
3996 7. Patches from the folks at Google:
3997
3998 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3999 real life, but is still worth protecting against".
4000
4001 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4002 regular expressions".
4003
4004 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4005 have it.
4006
4007 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4008 "configure" and the latter not, in order to fix a problem somebody had
4009 with compiling the Arg class on HP-UX.
4010
4011 (e) Improve the error-handling of the C++ wrapper a little bit.
4012
4013 (f) New tests for checking recursion limiting.
4014
4015 8. The pcre_memmove() function, which is used only if the environment does not
4016 have a standard memmove() function (and is therefore rarely compiled),
4017 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4018 returning a result (though PCRE never actually uses the result).
4019
4020 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4021 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4022 returned instead of calling malloc() with an overflowing number that would
4023 most likely cause subsequent chaos.
4024
4025 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4026
4027 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4028 with this option is matched, the nmatch and pmatch options of regexec() are
4029 ignored.
4030
4031 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4032 provided in case anyone wants to the the POSIX interface with UTF-8
4033 strings.
4034
4035 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4036 C++ linking (needed for some HP-UX environments).
4037
4038 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4039 (unused parameter) and in the pcre_printint() function (omitted "default"
4040 switch label when the default is to do nothing).
4041
4042 15. Added some code to make it possible, when PCRE is compiled as a C++
4043 library, to replace subject pointers for pcre_exec() with a smart pointer
4044 class, thus making it possible to process discontinuous strings.
4045
4046 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4047 much the same function. They were added by different people who were trying
4048 to make PCRE easy to compile on non-Unix systems. It has been suggested
4049 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4050 for compiling on Windows systems. I have therefore replaced it with
4051 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4052 defaults to "extern" for C or "extern C" for C++, which works fine on
4053 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4054 SCOPE with something explicit in config.h. In addition:
4055
4056 (a) pcreposix.h still had just "extern" instead of either of these macros;
4057 I have replaced it with PCRE_DATA_SCOPE.
4058
4059 (b) Functions such as _pcre_xclass(), which are internal to the library,
4060 but external in the C sense, all had PCRE_EXPORT in their definitions.
4061 This is apparently wrong for the Windows case, so I have removed it.
4062 (It makes no difference on Unix-like systems.)
4063
4064 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4065 of recursive calls to match(). This is different to MATCH_LIMIT because
4066 that limits the total number of calls to match(), not all of which increase
4067 the depth of recursion. Limiting the recursion depth limits the amount of
4068 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4069 when PCRE is compiled, and changed at run time. A patch from Google adds
4070 this functionality to the C++ interface.
4071
4072 18. Changes to the handling of Unicode character properties:
4073
4074 (a) Updated the table to Unicode 4.1.0.
4075
4076 (b) Recognize characters that are not in the table as "Cn" (undefined).
4077
4078 (c) I revised the way the table is implemented to a much improved format
4079 which includes recognition of ranges. It now supports the ranges that
4080 are defined in UnicodeData.txt, and it also amalgamates other
4081 characters into ranges. This has reduced the number of entries in the
4082 table from around 16,000 to around 3,000, thus reducing its size
4083 considerably. I realized I did not need to use a tree structure after
4084 all - a binary chop search is just as efficient. Having reduced the
4085 number of entries, I extended their size from 6 bytes to 8 bytes to
4086 allow for more data.
4087
4088 (d) Added support for Unicode script names via properties such as \p{Han}.
4089
4090 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4091 matching that character.
4092
4093 20. When matching a repeated Unicode property with a minimum greater than zero,
4094 (for example \pL{2,}), PCRE could look past the end of the subject if it
4095 reached it while seeking the minimum number of characters. This could
4096 happen only if some of the characters were more than one byte long, because
4097 there is a check for at least the minimum number of bytes.
4098
4099 21. Refactored the implementation of \p and \P so as to be more general, to
4100 allow for more different types of property in future. This has changed the
4101 compiled form incompatibly. Anybody with saved compiled patterns that use
4102 \p or \P will have to recompile them.
4103
4104 22. Added "Any" and "L&" to the supported property types.
4105
4106 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4107 but give a compile time error if the value is greater than 0xff.
4108
4109 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4110 accidentally not being installed or uninstalled.
4111
4112 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4113 made were to insert the current release number. This seemed silly, because
4114 it made things harder for people building PCRE on systems that don't run
4115 "configure". I have turned pcre.h into a distributed file, no longer built
4116 by "configure", with the version identification directly included. There is
4117 no longer a pcre.h.in file.
4118
4119 However, this change necessitated a change to the pcre-config script as
4120 well. It is built from pcre-config.in, and one of the substitutions was the
4121 release number. I have updated configure.ac so that ./configure now finds
4122 the release number by grepping pcre.h.
4123
4124 26. Added the ability to run the tests under valgrind.
4125
4126
4127 Version 6.4 05-Sep-05
4128 ---------------------
4129
4130 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4131 "--" to be printed when multiple files were scanned, even when none of the
4132 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4133 consider it to be a bug, and have restored the previous behaviour.
4134
4135 2. A couple of code tidies to get rid of compiler warnings.
4136
4137 3. The pcretest program used to cheat by referring to symbols in the library
4138 whose names begin with _pcre_. These are internal symbols that are not
4139 really supposed to be visible externally, and in some environments it is
4140 possible to suppress them. The cheating is now confined to including
4141 certain files from the library's source, which is a bit cleaner.
4142
4143 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4144 file's purpose clearer.
4145
4146 5. Reorganized pcre_ucp_findchar().
4147
4148
4149 Version 6.3 15-Aug-05
4150 ---------------------
4151
4152 1. The file libpcre.pc.in did not have general read permission in the tarball.
4153
4154 2. There were some problems when building without C++ support:
4155
4156 (a) If C++ support was not built, "make install" and "make test" still
4157 tried to test it.
4158
4159 (b) There were problems when the value of CXX was explicitly set. Some
4160 changes have been made to try to fix these, and ...
4161
4162 (c) --disable-cpp can now be used to explicitly disable C++ support.
4163
4164 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4165 backslash in a target when C++ was disabled. This confuses some
4166 versions of "make", apparently. Using an intermediate variable solves
4167 this. (Same for CPP_LOBJ.)
4168
4169 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4170 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4171 necessary on certain architectures.
4172
4173 4. Added a setting of -export-symbols-regex to the link command to remove
4174 those symbols that are exported in the C sense, but actually are local
4175 within the library, and not documented. Their names all begin with
4176 "_pcre_". This is not a perfect job, because (a) we have to except some
4177 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4178 available (and never for static libraries). I have made a note to try to
4179 find a way round (a) in the future.
4180
4181
4182 Version 6.2 01-Aug-05
4183 ---------------------
4184
4185 1. There was no test for integer overflow of quantifier values. A construction
4186 such as {1111111111111111} would give undefined results. What is worse, if
4187 a minimum quantifier for a parenthesized subpattern overflowed and became
4188 negative, the calculation of the memory size went wrong. This could have
4189 led to memory overwriting.
4190
4191 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4192
4193 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4194 operating environments where this matters.
4195
4196 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4197 PCRE options from within the C++ wrapper.
4198
4199 5. Named capturing subpatterns were not being correctly counted when a pattern
4200 was compiled. This caused two problems: (a) If there were more than 100
4201 such subpatterns, the calculation of the memory needed for the whole
4202 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4203 back references of the form \12, where the number was greater than 9, were
4204 not recognized as back references, even though there were sufficient
4205 previous subpatterns.
4206
4207 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4208 versions of gcc, e.g. 2.95.4.
4209
4210
4211 Version 6.1 21-Jun-05
4212 ---------------------
4213
4214 1. There was one reference to the variable "posix" in pcretest.c that was not
4215 surrounded by "#if !defined NOPOSIX".
4216
4217 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4218 the cross-check on the old pcre_info() function, for the benefit of the
4219 cut-down version of PCRE that is currently imported into Exim.
4220
4221 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4222 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4223 patterns that start (?i) but I don't think that matters. The use of (?i) is
4224 just an example; this all applies to the other options as well.
4225
4226 4. Since libtool seems to echo the compile commands it is issuing, the output
4227 from "make" can be reduced a bit by putting "@" in front of each libtool
4228 compile command.
4229
4230 5. Patch from the folks at Google for configure.in to be a bit more thorough
4231 in checking for a suitable C++ installation before trying to compile the
4232 C++ stuff. This should fix a reported problem when a compiler was present,
4233 but no suitable headers.
4234
4235 6. The man pages all had just "PCRE" as their title. I have changed them to
4236 be the relevant file name. I have also arranged that these names are
4237 retained in the file doc/pcre.txt, which is a concatenation in text format
4238 of all the man pages except the little individual ones for each function.
4239
4240 7. The NON-UNIX-USE file had not been updated for the different set of source
4241 files that come with release 6. I also added a few comments about the C++
4242 wrapper.
4243
4244
4245 Version 6.0 07-Jun-05
4246 ---------------------
4247
4248 1. Some minor internal re-organization to help with my DFA experiments.
4249
4250 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4251 didn't matter for the library itself when fully configured, but did matter
4252 when compiling without UCP support, or within Exim, where the ucp files are
4253 not imported.
4254
4255 3. Refactoring of the library code to split up the various functions into
4256 different source modules. The addition of the new DFA matching code (see
4257 below) to a single monolithic source would have made it really too
4258 unwieldy, quite apart from causing all the code to be include in a
4259 statically linked application, when only some functions are used. This is
4260 relevant even without the DFA addition now that patterns can be compiled in
4261 one application and matched in another.
4262
4263 The downside of splitting up is that there have to be some external
4264 functions and data tables that are used internally in different modules of
4265 the library but which are not part of the API. These have all had their
4266 names changed to start with "_pcre_" so that they are unlikely to clash
4267 with other external names.
4268
4269 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4270 a different (DFA) algorithm. Although it is slower than the original
4271 function, it does have some advantages for certain types of matching
4272 problem.
4273
4274 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4275 including restarting after a partial match.
4276
4277 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4278 defined when compiling for Windows was sent to me. I have put it into the
4279 code, though I have no means of testing or verifying it.
4280
4281 7. Added the pcre_refcount() auxiliary function.
4282
4283 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4284 match before or at the first newline in the subject string. In pcretest,
4285 the /f option on a pattern can be used to set this.
4286
4287 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4288 would behave wrongly. This has been present in PCRE since release 4.0.
4289
4290 10. A number of changes to the pcregrep command:
4291
4292 (a) Refactored how -x works; insert ^(...)$ instead of setting
4293 PCRE_ANCHORED and checking the length, in preparation for adding
4294 something similar for -w.
4295
4296 (b) Added the -w (match as a word) option.
4297
4298 (c) Refactored the way lines are read and buffered so as to have more
4299 than one at a time available.
4300
4301 (d) Implemented a pcregrep test script.
4302
4303 (e) Added the -M (multiline match) option. This allows patterns to match
4304 over several lines of the subject. The buffering ensures that at least
4305 8K, or the rest of the document (whichever is the shorter) is available
4306 for matching (and similarly the previous 8K for lookbehind assertions).
4307
4308 (f) Changed the --help output so that it now says
4309
4310 -w, --word-regex(p)
4311
4312 instead of two lines, one with "regex" and the other with "regexp"
4313 because that confused at least one person since the short forms are the
4314 same. (This required a bit of code, as the output is generated
4315 automatically from a table. It wasn't just a text change.)
4316
4317 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4318 option but starts with a hyphen. Could be a pattern or a path name
4319 starting with a hyphen, for instance.
4320
4321 (h) "-" can be given as a file name to represent stdin.
4322
4323 (i) When file names are being printed, "(standard input)" is used for
4324 the standard input, for compatibility with GNU grep. Previously
4325 "<stdin>" was used.
4326
4327 (j) The option --label=xxx can be used to supply a name to be used for
4328 stdin when file names are being printed. There is no short form.
4329
4330 (k) Re-factored the options decoding logic because we are going to add
4331 two more options that take data. Such options can now be given in four
4332 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4333
4334 (l) Added the -A, -B, and -C options for requesting that lines of context
4335 around matches be printed.
4336
4337 (m) Added the -L option to print the names of files that do not contain
4338 any matching lines, that is, the complement of -l.
4339
4340 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4341 continue to scan other files.
4342
4343 (o) The -s option was incorrectly implemented. For compatibility with other
4344 greps, it now suppresses the error message for a non-existent or non-
4345 accessible file (but not the return code). There is a new option called
4346 -q that suppresses the output of matching lines, which was what -s was
4347 previously doing.
4348
4349 (p) Added --include and --exclude options to specify files for inclusion
4350 and exclusion when recursing.
4351
4352 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4353 Hopefully, it now does.
4354
4355 12. Missing cast in pcre_study().
4356
4357 13. Added an "uninstall" target to the makefile.
4358
4359 14. Replaced "extern" in the function prototypes in Makefile.in with
4360 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4361 world, but is set differently for Windows.
4362
4363 15. Added a second compiling function called pcre_compile2(). The only
4364 difference is that it has an extra argument, which is a pointer to an
4365 integer error code. When there is a compile-time failure, this is set
4366 non-zero, in addition to the error test pointer being set to point to an
4367 error message. The new argument may be NULL if no error number is required
4368 (but then you may as well call pcre_compile(), which is now just a
4369 wrapper). This facility is provided because some applications need a
4370 numeric error indication, but it has also enabled me to tidy up the way
4371 compile-time errors are handled in the POSIX wrapper.
4372
4373 16. Added VPATH=.libs to the makefile; this should help when building with one
4374 prefix path and installing with another. (Or so I'm told by someone who
4375 knows more about this stuff than I do.)
4376
4377 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4378 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4379 match everything, including newlines. This is not POSIX-compatible, but
4380 somebody wanted the feature. From pcretest it can be activated by using
4381 both the P and the s flags.
4382
4383 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4384
4385 19. libpcre.pc was being incorrectly installed as executable.
4386
4387 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4388 it now also looks for '\r' so that it will work unmodified on Windows.
4389
4390 21. Added Google's contributed C++ wrapper to the distribution.
4391
4392 22. Added some untidy missing memory free() calls in pcretest, to keep
4393 Electric Fence happy when testing.
4394
4395
4396
4397 Version 5.0 13-Sep-04
4398 ---------------------
4399
4400 1. Internal change: literal characters are no longer packed up into items
4401 containing multiple characters in a single byte-string. Each character
4402 is now matched using a separate opcode. However, there may be more than one
4403 byte in the character in UTF-8 mode.
4404
4405 2. The pcre_callout_block structure has two new fields: pattern_position and
4406 next_item_length. These contain the offset in the pattern to the next match
4407 item, and its length, respectively.
4408
4409 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4410 insertion of callouts before each pattern item. Added the /C option to
4411 pcretest to make use of this.
4412
4413 4. On the advice of a Windows user, the lines
4414
4415 #if defined(_WIN32) || defined(WIN32)
4416 _setmode( _fileno( stdout ), 0x8000 );
4417 #endif /* defined(_WIN32) || defined(WIN32) */
4418
4419 have been added to the source of pcretest. This apparently does useful
4420 magic in relation to line terminators.
4421
4422 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4423 for the benefit of those environments where the "b" makes a difference.
4424
4425 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4426 to know about it. I have put a hack into configure.in that adds in code
4427 to set GCC=yes if CC=icc. This seems to end up at a point in the
4428 generated configure script that is early enough to affect the setting of
4429 compiler options, which is what is needed, but I have no means of testing
4430 whether it really works. (The user who reported this had patched the
4431 generated configure script, which of course I cannot do.)
4432
4433 LATER: After change 22 below (new libtool files), the configure script
4434 seems to know about icc (and also ecc). Therefore, I have commented out
4435 this hack in configure.in.
4436
4437 7. Added support for pkg-config (2 patches were sent in).
4438
4439 8. Negated POSIX character classes that used a combination of internal tables
4440 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4441 [[:^ascii]]. Typically, they would match almost any characters. The other
4442 POSIX classes were not broken in this way.
4443
4444 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4445 to find the match, as PCRE was deluded into thinking that the match had to
4446 start at the start point or following a newline. The same bug applied to
4447 patterns with negative forward assertions or any backward assertions
4448 preceding ".*" at the start, unless the pattern required a fixed first
4449 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4450
4451 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4452 starting at the last subject character, bytes beyond the end of the subject
4453 string were read.
4454
4455 11. Renamed the variable "class" as "classbits" to make life easier for C++
4456 users. (Previously there was a macro definition, but it apparently wasn't
4457 enough.)
4458
4459 12. Added the new field "tables" to the extra data so that tables can be passed
4460 in at exec time, or the internal tables can be re-selected. This allows
4461 a compiled regex to be saved and re-used at a later time by a different
4462 program that might have everything at different addresses.
4463
4464 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4465 -R library as well as a -L library.
4466
4467 14. The debugging options of pcretest (-d on the command line or D on a
4468 pattern) showed incorrect output for anything following an extended class
4469 that contained multibyte characters and which was followed by a quantifier.
4470
4471 15. Added optional support for general category Unicode character properties
4472 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4473 support. It adds about 90K to the size of the library. The meanings of the
4474 inbuilt class escapes such as \d and \s have NOT been changed.
4475
4476 16. Updated pcredemo.c to include calls to free() to release the memory for the
4477 compiled pattern.
4478
4479 17. The generated file chartables.c was being created in the source directory
4480 instead of in the building directory. This caused the build to fail if the
4481 source directory was different from the building directory, and was
4482 read-only.
4483
4484 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4485 file. No doubt somebody will tell me if they don't make sense... Also added
4486 Dan Mooney's comments about building on OpenVMS.
4487
4488 19. Added support for partial matching via the PCRE_PARTIAL option for
4489 pcre_exec() and the \P data escape in pcretest.
4490
4491 20. Extended pcretest with 3 new pattern features:
4492
4493 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4494 write the compiled pattern to the file whose name is "rest-of-line".
4495 This is a straight binary dump of the data, with the saved pointer to
4496 the character tables forced to be NULL. The study data, if any, is
4497 written too. After writing, pcretest reads a new pattern.
4498
4499 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4500 compiled pattern from the given file. There must not be any
4501 occurrences of "<" in the file name (pretty unlikely); if there are,
4502 pcretest will instead treat the initial "<" as a pattern delimiter.
4503 After reading in the pattern, pcretest goes on to read data lines as
4504 usual.
4505
4506 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4507 and 16-bit fields in a compiled pattern, to simulate a pattern that
4508 was compiled on a host of opposite endianness.
4509
4510 21. The pcre-exec() function can now cope with patterns that were compiled on
4511 hosts of opposite endianness, with this restriction:
4512
4513 As for any compiled expression that is saved and used later, the tables
4514 pointer field cannot be preserved; the extra_data field in the arguments
4515 to pcre_exec() should be used to pass in a tables address if a value
4516 other than the default internal tables were used at compile time.
4517
4518 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4519 now diagnosed as an error. Previously, most of the time, a negative number
4520 would have been treated as zero, but if in addition "ovector" was passed as
4521 NULL, a crash could occur.
4522
4523 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4524 new versions from the libtool 1.5 distribution (the last one is a copy of
4525 a file called libtool.m4). This seems to have fixed the need to patch
4526 "configure" to support Darwin 1.3 (which I used to do). However, I still
4527 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4528 workstation).
4529
4530 24. Changed the PCRE licence to be the more standard "BSD" licence.
4531
4532
4533 Version 4.5 01-Dec-03
4534 ---------------------
4535
4536 1. There has been some re-arrangement of the code for the match() function so
4537 that it can be compiled in a version that does not call itself recursively.
4538 Instead, it keeps those local variables that need separate instances for
4539 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4540 needs to "recurse". Keeping track of where control must go is done by means
4541 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4542 hide most of the details from the main code, and operates only if
4543 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4544 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4545 operating.
4546
4547 To make it easier for callers to provide specially tailored get/free
4548 functions for this usage, two new functions, pcre_stack_malloc, and
4549 pcre_stack_free, are used. They are always called in strict stacking order,
4550 and the size of block requested is always the same.
4551
4552 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4553 PCRE has been compiled to use the stack or the heap for recursion. The
4554 -C option of pcretest uses this to show which version is compiled.
4555
4556 A new data escape \S, is added to pcretest; it causes the amounts of store
4557 obtained and freed by both kinds of malloc/free at match time to be added
4558 to the output.
4559
4560 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4561 what's available on my current Linux desktop machine.
4562
4563 3. When matching a UTF-8 string, the test for a valid string at the start has
4564 been extended. If start_offset is not zero, PCRE now checks that it points
4565 to a byte that is the start of a UTF-8 character. If not, it returns
4566 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4567 this is necessary because there may be backward assertions in the pattern.
4568 When matching the same subject several times, it may save resources to use
4569 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4570
4571 4. The code for checking the validity of UTF-8 strings has been tightened so
4572 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4573 containing "overlong sequences".
4574
4575 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4576 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4577 should have been "&", but it just so happened that all the cases this let
4578 through by mistake were picked up later in the function.
4579
4580 6. I had used a variable called "isblank" - this is a C99 function, causing
4581 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4582
4583 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4584 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4585 so the version is shown; (c) stop "make test" echoing "./RunTest".
4586
4587 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4588
4589 9. The prototype for memmove() for systems that don't have it was using
4590 size_t, but the inclusion of the header that defines size_t was later. I've
4591 moved the #includes for the C headers earlier to avoid this.
4592
4593 10. Added some adjustments to the code to make it easier to compiler on certain
4594 special systems:
4595
4596 (a) Some "const" qualifiers were missing.
4597 (b) Added the macro EXPORT before all exported functions; by default this
4598 is defined to be empty.
4599 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4600 that it reads its output file name as an argument instead of writing
4601 to the standard output and assuming this can be redirected.
4602
4603 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4604 class containing characters with values greater than 255, PCRE compilation
4605 went into a loop.
4606
4607 12. A recursive reference to a subpattern that was within another subpattern
4608 that had a minimum quantifier of zero caused PCRE to crash. For example,
4609 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4610 recursion. If the recursively-called subpattern itself had a zero repeat,
4611 that was OK.
4612
4613 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4614 buffer into which it was copied (for escape processing) was still set at
4615 1024, so long lines caused crashes.
4616
4617 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4618 "internal error: code overflow...". This applied to any character class
4619 that was followed by a possessive quantifier.
4620
4621 15. Modified the Makefile to add libpcre.la as a prerequisite for
4622 libpcreposix.la because I was told this is needed for a parallel build to
4623 work.
4624
4625 16. If a pattern that contained .* following optional items at the start was
4626 studied, the wrong optimizing data was generated, leading to matching
4627 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4628 matching string must start with a or b or c. The correct conclusion for
4629 this pattern is that a match can start with any character.
4630
4631
4632 Version 4.4 13-Aug-03
4633 ---------------------
4634
4635 1. In UTF-8 mode, a character class containing characters with values between
4636 127 and 255 was not handled correctly if the compiled pattern was studied.
4637 In fixing this, I have also improved the studying algorithm for such
4638 classes (slightly).
4639
4640 2. Three internal functions had redundant arguments passed to them. Removal
4641 might give a very teeny performance improvement.
4642
4643 3. Documentation bug: the value of the capture_top field in a callout is *one
4644 more than* the number of the hightest numbered captured substring.
4645
4646 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4647 in incorrectly linking with a previously installed version. They now link
4648 explicitly with libpcre.la.
4649
4650 5. configure.in no longer needs to recognize Cygwin specially.
4651
4652 6. A problem in pcre.in for Windows platforms is fixed.
4653
4654 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4655 pcretest, it used to include the size of the study block as part of its
4656 output. Unfortunately, the structure contains a field that has a different
4657 size on different hardware architectures. This meant that the tests that
4658 showed this size failed. As the block is currently always of a fixed size,
4659 this information isn't actually particularly useful in pcretest output, so
4660 I have just removed it.
4661
4662 8. Three pre-processor statements accidentally did not start in column 1.
4663 Sadly, there are *still* compilers around that complain, even though
4664 standard C has not required this for well over a decade. Sigh.
4665
4666 9. In pcretest, the code for checking callouts passed small integers in the
4667 callout_data field, which is a void * field. However, some picky compilers
4668 complained about the casts involved for this on 64-bit systems. Now
4669 pcretest passes the address of the small integer instead, which should get
4670 rid of the warnings.
4671
4672 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4673 both compile and run time, and gives an error if an invalid UTF-8 sequence
4674 is found. There is a option for disabling this check in cases where the
4675 string is known to be correct and/or the maximum performance is wanted.
4676
4677 11. In response to a bug report, I changed one line in Makefile.in from
4678
4679 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4680 to
4681 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4682
4683 to look similar to other lines, but I have no way of telling whether this
4684 is the right thing to do, as I do not use Windows. No doubt I'll get told
4685 if it's wrong...
4686
4687
4688 Version 4.3 21-May-03
4689 ---------------------
4690
4691 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4692 Makefile.
4693
4694 2. Some refactoring to improve the quality of the code:
4695
4696 (i) The utf8_table... variables are now declared "const".
4697
4698 (ii) The code for \cx, which used the "case flipping" table to upper case
4699 lower case letters, now just substracts 32. This is ASCII-specific,
4700 but the whole concept of \cx is ASCII-specific, so it seems
4701 reasonable.
4702
4703 (iii) PCRE was using its character types table to recognize decimal and
4704 hexadecimal digits in the pattern. This is silly, because it handles
4705 only 0-9, a-f, and A-F, but the character types table is locale-
4706 specific, which means strange things might happen. A private
4707 table is now used for this - though it costs 256 bytes, a table is
4708 much faster than multiple explicit tests. Of course, the standard
4709 character types table is still used for matching digits in subject
4710 strings against \d.
4711
4712 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4713 ending in _t are). So I've renamed it as ESC_tee.
4714
4715 3. The first argument for regexec() in the POSIX wrapper should have been
4716 defined as "const".
4717
4718 4. Changed pcretest to use malloc() for its buffers so that they can be
4719 Electric Fenced for debugging.
4720
4721 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4722 to read one or more bytes before the start of the subject string. Often this
4723 had no effect on PCRE's behaviour, but in some circumstances it could
4724 provoke a segmentation fault.
4725
4726 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4727 to try to read one or more bytes before the start of the subject string.
4728
4729 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4730 UTF-8 support could misbehave in various ways if the subject string
4731 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4732 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4733 back over UTF-8 characters.)
4734
4735
4736 Version 4.2 14-Apr-03
4737 ---------------------
4738
4739 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4740
4741 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4742 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4743 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4744 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4745 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4746 and BUILD_EXEEXT
4747 Note: automatic setting of the BUILD variables is not yet working
4748 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4749 compile-time but not at link-time
4750 [LINK]: use for linking executables only
4751 make different versions for Windows and non-Windows
4752 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4753 libraries
4754 [LINK_FOR_BUILD]: new variable
4755 [OBJEXT]: use throughout
4756 [EXEEXT]: use throughout
4757 <winshared>: new target
4758 <wininstall>: new target
4759 <dftables.o>: use native compiler
4760 <dftables>: use native linker
4761 <install>: handle Windows platform correctly
4762 <clean>: ditto
4763 <check>: ditto
4764 copy DLL to top builddir before testing
4765
4766 As part of these changes, -no-undefined was removed again. This was reported
4767 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4768 in any case.
4769
4770 3. Some tidies to get rid of compiler warnings:
4771
4772 . In the match_data structure, match_limit was an unsigned long int, whereas
4773 match_call_count was an int. I've made them both unsigned long ints.
4774
4775 . In pcretest the fact that a const uschar * doesn't automatically cast to
4776 a void * provoked a warning.
4777
4778 . Turning on some more compiler warnings threw up some "shadow" variables
4779 and a few more missing casts.
4780
4781 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4782 option, a class that contained a single character with a value between 128
4783 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4784
4785 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4786 option, a class that contained several characters, but with at least one
4787 whose value was between 128 and 255 caused PCRE to crash.
4788
4789
4790 Version 4.1 12-Mar-03
4791 ---------------------
4792
4793 1. Compiling with gcc -pedantic found a couple of places where casts were
4794 needed, and a string in dftables.c that was longer than standard compilers are
4795 required to support.
4796
4797 2. Compiling with Sun's compiler found a few more places where the code could
4798 be tidied up in order to avoid warnings.
4799
4800 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4801 first of these names is deprecated in the latest Autoconf in favour of the name
4802 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4803 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4804 analogy I have changed it to CFLAGS_FOR_BUILD.
4805
4806 4. Added -no-undefined to the linking command in the Makefile, because this is
4807 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4808 linking step for the pcreposix library.
4809
4810 5. PCRE was failing to diagnose the case of two named groups with the same
4811 name.
4812
4813 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4814 literal character that is needed in the subject for a match, and scans along to
4815 ensure that it is present before embarking on the full matching process. This
4816 saves time in cases of nested unlimited repeats that are never going to match.
4817 Problem: the scan can take a lot of time if the subject is very long (e.g.
4818 megabytes), thus penalizing straightforward matches. It is now done only if the
4819 amount of subject to be scanned is less than 1000 bytes.
4820
4821 7. A lesser problem with the same optimization is that it was recording the
4822 first character of an anchored pattern as "needed", thus provoking a search
4823 right along the subject, even when the first match of the pattern was going to
4824 fail. The "needed" character is now not set for anchored patterns, unless it
4825 follows something in the pattern that is of non-fixed length. Thus, it still
4826 fulfils its original purpose of finding quick non-matches in cases of nested
4827 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4828
4829
4830 Version 4.0 17-Feb-03
4831 ---------------------
4832
4833 1. If a comment in an extended regex that started immediately after a meta-item
4834 extended to the end of string, PCRE compiled incorrect data. This could lead to
4835 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4836
4837 2. Moved to autoconf 2.53 and libtool 1.4.2.
4838
4839 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4840 the special perltest8 script is no longer needed - all the tests can be run
4841 from a single perltest script.
4842
4843 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4844 by \s. It has now been removed in PCRE. This means it isn't recognized as
4845 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4846 class [:space:] *does* include VT, thereby creating a mess.
4847
4848 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4849 space and tab.
4850
4851 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4852 its new features into the main test script, reducing the number of scripts.
4853
4854 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4855 were backward compatible, and made the (?i) apply to the whole pattern, as if
4856 /i were given. Now it behaves more logically, and applies the option setting
4857 only to what follows. PCRE has been changed to follow suit. However, if it
4858 finds options settings right at the start of the pattern, it extracts them into
4859 the global options, as before. Thus, they show up in the info data.
4860
4861 8. Added support for the \Q...\E escape sequence. Characters in between are
4862 treated as literals. This is slightly different from Perl in that $ and @ are
4863 also handled as literals inside the quotes. In Perl, they will cause variable
4864 interpolation. Note the following examples:
4865
4866 Pattern PCRE matches Perl matches
4867
4868 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4869 \Qabc\$xyz\E abc\$xyz abc\$xyz
4870 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4871
4872 For compatibility with Perl, \Q...\E sequences are recognized inside character
4873 classes as well as outside them.
4874
4875 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4876 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4877 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4878 signed/unsigned warnings.
4879
4880 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4881 option for pcretest, so I've replaced it by a simple function that does just
4882 that job.
4883
4884 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4885 "pcregrep -".
4886
4887 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4888 Java package. This provides some syntactic sugar for simple cases of what my
4889 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4890 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4891 item, you can use this simplified notation. Note that only makes sense with
4892 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4893 greediness, whatever the setting of the PCRE_UNGREEDY option.
4894
4895 13. A change of greediness default within a pattern was not taking effect at
4896 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4897 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4898 was abstracted outside.
4899
4900 14. PCRE now supports the \G assertion. It is true when the current matching
4901 position is at the start point of the match. This differs from \A when the
4902 starting offset is non-zero. Used with the /g option of pcretest (or similar
4903 code), it works in the same way as it does for Perl's /g option. If all
4904 alternatives of a regex begin with \G, the expression is anchored to the start
4905 match position, and the "anchored" flag is set in the compiled expression.
4906
4907 15. Some bugs concerning the handling of certain option changes within patterns
4908 have been fixed. These applied to options other than (?ims). For example,
4909 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4910 been the other way round. Some of this was related to change 7 above.
4911
4912 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4913 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4914 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4915 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4916
4917 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4918 mode. Unlike ".", it always matches newline, whatever the setting of
4919 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4920 assertions. Perl allows it, but it doesn't (in general) work because it can't
4921 calculate the length of the lookbehind. At least, that's the case for Perl
4922 5.8.0 - I've been told they are going to document that it doesn't work in
4923 future.
4924
4925 18. Added an error diagnosis for escapes that PCRE does not support: these are
4926 \L, \l, \N, \P, \p, \U, \u, and \X.
4927
4928 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4929 reading past the end of the pattern in cases such as /[abcd/.
4930
4931 20. PCRE was getting more memory than necessary for patterns with classes that
4932 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4933
4934 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4935 compiling PCRE for use with Virtual Pascal.
4936
4937 22. Small fix to the Makefile to make it work properly if the build is done
4938 outside the source tree.
4939
4940 23. Added a new extension: a condition to go with recursion. If a conditional
4941 subpattern starts with (?(R) the "true" branch is used if recursion has
4942 happened, whereas the "false" branch is used only at the top level.
4943
4944 24. When there was a very long string of literal characters (over 255 bytes
4945 without UTF support, over 250 bytes with UTF support), the computation of how
4946 much memory was required could be incorrect, leading to segfaults or other
4947 strange effects.
4948
4949 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4950 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4951 there was a subsequent back reference to those brackets. This meant that, for
4952 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4953 possible to check for precisely this case. All we can do is abandon the
4954 optimization if .* occurs inside capturing brackets when there are any back
4955 references whatsoever. (See below for a better fix that came later.)
4956
4957 26. The handling of the optimization for finding the first character of a
4958 non-anchored pattern, and for finding a character that is required later in the
4959 match were failing in some cases. This didn't break the matching; it just
4960 failed to optimize when it could. The way this is done has been re-implemented.
4961
4962 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4963
4964 28. Added a new feature that provides some of the functionality that Perl
4965 provides with (?{...}). The facility is termed a "callout". The way it is done
4966 in PCRE is for the caller to provide an optional function, by setting
4967 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4968 global variable. By default it is unset, which disables all calling out. To get
4969 the function called, the regex must include (?C) at appropriate points. This
4970 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4971 This provides a means of identifying different callout points. When PCRE
4972 reaches such a point in the regex, if pcre_callout has been set, the external
4973 function is called. It is provided with data in a structure called
4974 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4975 matching continues; if it returns a non-zero value, the match at the current
4976 point fails. However, backtracking will occur if possible. [This was changed
4977 later and other features added - see item 49 below.]
4978
4979 29. pcretest is upgraded to test the callout functionality. It provides a
4980 callout function that displays information. By default, it shows the start of
4981 the match and the current position in the text. There are some new data escapes
4982 to vary what happens:
4983
4984 \C+ in addition, show current contents of captured substrings
4985 \C- do not supply a callout function
4986 \C!n return 1 when callout number n is reached
4987 \C!n!m return 1 when callout number n is reached for the mth time
4988
4989 30. If pcregrep was called with the -l option and just a single file name, it
4990 output "<stdin>" if a match was found, instead of the file name.
4991
4992 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4993 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4994 pcre_exec(). This saves a malloc/free per call. The default value of
4995 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4996 when configuring.
4997
4998 32. The default maximum size of a compiled pattern is 64K. There have been a
4999 few cases of people hitting this limit. The code now uses macros to handle the
5000 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5001 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5002 configuring. Tests 2 and 5 work only with 2-byte links because they output
5003 debugging information about compiled patterns.
5004
5005 33. Internal code re-arrangements:
5006
5007 (a) Moved the debugging function for printing out a compiled regex into
5008 its own source file (printint.c) and used #include to pull it into
5009 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5010 separate copies.
5011
5012 (b) Defined the list of op-code names for debugging as a macro in
5013 internal.h so that it is next to the definition of the opcodes.
5014
5015 (c) Defined a table of op-code lengths for simpler skipping along compiled
5016 code. This is again a macro in internal.h so that it is next to the
5017 definition of the opcodes.
5018
5019 34. Added support for recursive calls to individual subpatterns, along the
5020 lines of Robin Houston's patch (but implemented somewhat differently).
5021
5022 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5023 allow it to read and process whole directories in Win32. This code was
5024 contributed by Lionel Fourquaux; it has not been tested by me.
5025
5026 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5027 used to name a group. Names consist of alphanumerics and underscores, and must
5028 be unique. Back references use the syntax (?P=name) and recursive calls use
5029 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5030 numbers. The function pcre_fullinfo() can be used after compilation to extract
5031 a name/number map. There are three relevant calls:
5032
5033 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5034 PCRE_INFO_NAMECOUNT yields the number of entries
5035 PCRE_INFO_NAMETABLE yields a pointer to the map.
5036
5037 The map is a vector of fixed-size entries. The size of each entry depends on
5038 the length of the longest name used. The first two bytes of each entry are the
5039 group number, most significant byte first. There follows the corresponding
5040 name, zero terminated. The names are in alphabetical order.
5041
5042 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5043 case instead of 255. Making it the same both with and without UTF-8 support
5044 means that the same test output works with both.
5045
5046 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5047 calling malloc() with a zero argument.
5048
5049 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5050 optimization. I've improved things by keeping a bitmap of backreferences with
5051 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5052 fact referenced, the optimization can be applied. It is unlikely that a
5053 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5054 the match to follow \n) will appear inside brackets with a number greater than
5055 31, but if it does, any back reference > 31 suppresses the optimization.
5056
5057 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5058 of disabling numbered capturing parentheses. Any opening parenthesis that is
5059 not followed by ? behaves as if it were followed by ?: but named parentheses
5060 can still be used for capturing (and they will acquire numbers in the usual
5061 way).
5062
5063 41. Redesigned the return codes from the match() function into yes/no/error so
5064 that errors can be passed back from deep inside the nested calls. A malloc
5065 failure while inside a recursive subpattern call now causes the
5066 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5067
5068 42. It is now possible to set a limit on the number of times the match()
5069 function is called in a call to pcre_exec(). This facility makes it possible to
5070 limit the amount of recursion and backtracking, though not in a directly
5071 obvious way, because the match() function is used in a number of different
5072 circumstances. The count starts from zero for each position in the subject
5073 string (for non-anchored patterns). The default limit is, for compatibility, a
5074 large number, namely 10 000 000. You can change this in two ways:
5075
5076 (a) When configuring PCRE before making, you can use --with-match-limit=n
5077 to set a default value for the compiled library.
5078
5079 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5080 a different value is set. See 45 below.
5081
5082 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5083
5084 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5085 of things that can be changed at compile time. The first argument specifies
5086 what is wanted and the second points to where the information is to be placed.
5087 The current list of available information is:
5088
5089 PCRE_CONFIG_UTF8
5090
5091 The output is an integer that is set to one if UTF-8 support is available;
5092 otherwise it is set to zero.
5093
5094 PCRE_CONFIG_NEWLINE
5095
5096 The output is an integer that it set to the value of the code that is used for
5097 newline. It is either LF (10) or CR (13).
5098
5099 PCRE_CONFIG_LINK_SIZE
5100
5101 The output is an integer that contains the number of bytes used for internal
5102 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5103
5104 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5105
5106 The output is an integer that contains the threshold above which the POSIX
5107 interface uses malloc() for output vectors. See item 31 above.
5108
5109 PCRE_CONFIG_MATCH_LIMIT
5110
5111 The output is an unsigned integer that contains the default limit of the number
5112 of match() calls in a pcre_exec() execution. See 42 above.
5113
5114 44. pcretest has been upgraded by the addition of the -C option. This causes it
5115 to extract all the available output from the new pcre_config() function, and to
5116 output it. The program then exits immediately.
5117
5118 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5119 order to support additional features. One way would have been to define
5120 pcre_exec2() (for example) with extra arguments, but this would not have been
5121 extensible, and would also have required all calls to the original function to
5122 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5123 is used for passing in "extra" data from pcre_study().
5124
5125 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5126 contains the following fields:
5127
5128 flags a bitmap indicating which of the following fields are set
5129 study_data opaque data from pcre_study()
5130 match_limit a way of specifying a limit on match() calls for a specific
5131 call to pcre_exec()
5132 callout_data data for callouts (see 49 below)
5133
5134 The flag bits are also defined in pcre.h, and are
5135
5136 PCRE_EXTRA_STUDY_DATA
5137 PCRE_EXTRA_MATCH_LIMIT
5138 PCRE_EXTRA_CALLOUT_DATA
5139
5140 The pcre_study() function now returns one of these new pcre_extra blocks, with
5141 the actual study data pointed to by the study_data field, and the
5142 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5143 before. That is, this change is entirely upwards-compatible and requires no
5144 change to existing code.
5145
5146 If you want to pass in additional data to pcre_exec(), you can either place it
5147 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5148 block.
5149
5150 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5151 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5152 times with different match limits, until it finds the minimum value needed for
5153 pcre_exec() to complete. The value is then output. This can be instructive; for
5154 most simple matches the number is quite small, but for pathological cases it
5155 gets very large very quickly.
5156
5157 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5158 returns the size of the data block pointed to by the study_data field in a
5159 pcre_extra block, that is, the value that was passed as the argument to
5160 pcre_malloc() when PCRE was getting memory in which to place the information
5161 created by pcre_study(). The fourth argument should point to a size_t variable.
5162 pcretest has been extended so that this information is shown after a successful
5163 pcre_study() call when information about the compiled regex is being displayed.
5164
5165 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5166 because what follows is always an absolute path. (Later: it turns out that this
5167 is more than cosmetic for MinGW, because it doesn't like empty path
5168 components.)
5169
5170 49. Some changes have been made to the callout feature (see 28 above):
5171
5172 (i) A callout function now has three choices for what it returns:
5173
5174 0 => success, carry on matching
5175 > 0 => failure at this point, but backtrack if possible
5176 < 0 => serious error, return this value from pcre_exec()
5177
5178 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5179 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5180 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5181 use by callout functions. It will never be used by PCRE itself.
5182
5183 (ii) The pcre_extra structure (see 45 above) has a void * field called
5184 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5185 pcre_callout_block structure has a field of the same name. The contents of
5186 the field passed in the pcre_extra structure are passed to the callout
5187 function in the corresponding field in the callout block. This makes it
5188 easier to use the same callout-containing regex from multiple threads. For
5189 testing, the pcretest program has a new data escape
5190
5191 \C*n pass the number n (may be negative) as callout_data
5192
5193 If the callout function in pcretest receives a non-zero value as
5194 callout_data, it returns that value.
5195
5196 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5197 there were some redundant $(CFLAGS) in commands that are now specified as
5198 $(LINK), which already includes $(CFLAGS).
5199
5200 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5201 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5202 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5203 one-byte characters throughout. Note that case-insensitive matching applies
5204 only to characters whose values are less than 256. PCRE doesn't support the
5205 notion of cases for higher-valued characters.
5206
5207 (i) A character class whose characters are all within 0-255 is handled as
5208 a bit map, and the map is inverted for negative classes. Previously, a
5209 character > 255 always failed to match such a class; however it should
5210 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5211
5212 (ii) A negated character class with a single character < 255 is coded as
5213 "not this character" (OP_NOT). This wasn't working properly when the test
5214 character was multibyte, either singly or repeated.
5215
5216 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5217 mode, for example: \x{100}{2,3}.
5218
5219 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5220 singly or repeated) now correctly test multibyte characters. However,
5221 PCRE doesn't recognize any characters with values greater than 255 as
5222 digits, spaces, or word characters. Such characters always match \D, \S,
5223 and \W, and never match \d, \s, or \w.
5224
5225 (v) Classes may now contain characters and character ranges with values
5226 greater than 255. For example: [ab\x{100}-\x{400}].
5227
5228 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5229 PCRE in UTF-8 mode.
5230
5231 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5232 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5233 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5234 value.)
5235
5236 53. The single man page has become too large. I have therefore split it up into
5237 a number of separate man pages. These also give rise to individual HTML pages;
5238 these are now put in a separate directory, and there is an index.html page that
5239 lists them all. Some hyperlinking between the pages has been installed.
5240
5241 54. Added convenience functions for handling named capturing parentheses.
5242
5243 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5244 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5245 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5246 are faulted.
5247
5248 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5249 calling configure. These values are used when compiling the dftables.c program
5250 which is run to generate the source of the default character tables. They
5251 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5252 you will need to set these values.
5253
5254 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5255
5256
5257 Version 3.9 02-Jan-02
5258 ---------------------
5259
5260 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5261
5262 2. If --disable-static was given, the building process failed when trying to
5263 build pcretest and pcregrep. (For some reason it was using libtool to compile
5264 them, which is not right, as they aren't part of the library.)
5265
5266
5267 Version 3.8 18-Dec-01
5268 ---------------------
5269
5270 1. The experimental UTF-8 code was completely screwed up. It was packing the
5271 bytes in the wrong order. How dumb can you get?
5272
5273
5274 Version 3.7 29-Oct-01
5275 ---------------------
5276
5277 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5278 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5279 this didn't happen under Solaris 8, where I normally test things.
5280
5281 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5282 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5283 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5284 this fix an hour or so after the initial 3.7 release.)
5285
5286
5287 Version 3.6 23-Oct-01
5288 ---------------------
5289
5290 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5291 offsets passed as NULL with zero offset count.
5292
5293 2. The config.guess and config.sub files had not been updated when I moved to
5294 the latest autoconf.
5295
5296
5297 Version 3.5 15-Aug-01
5298 ---------------------
5299
5300 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5301 had been forgotten.
5302
5303 2. By using declared but undefined structures, we can avoid using "void"
5304 definitions in pcre.h while keeping the internal definitions of the structures
5305 private.
5306
5307 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5308 user point of view, this means that both static and shared libraries are built
5309 by default, but this can be individually controlled. More of the work of
5310 handling this static/shared cases is now inside libtool instead of PCRE's make
5311 file.
5312
5313 4. The pcretest utility is now installed along with pcregrep because it is
5314 useful for users (to test regexs) and by doing this, it automatically gets
5315 relinked by libtool. The documentation has been turned into a man page, so
5316 there are now .1, .txt, and .html versions in /doc.
5317
5318 5. Upgrades to pcregrep:
5319 (i) Added long-form option names like gnu grep.
5320 (ii) Added --help to list all options with an explanatory phrase.
5321 (iii) Added -r, --recursive to recurse into sub-directories.
5322 (iv) Added -f, --file to read patterns from a file.
5323
5324 6. pcre_exec() was referring to its "code" argument before testing that
5325 argument for NULL (and giving an error if it was NULL).
5326
5327 7. Upgraded Makefile.in to allow for compiling in a different directory from
5328 the source directory.
5329
5330 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5331 options bits, the pointer it was passed was to an int instead of to an unsigned
5332 long int. This mattered only on 64-bit systems.
5333
5334 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5335 generated) instead of pcre.in, which it its source. Also made the same change
5336 in several of the .c files.
5337
5338 10. A new release of gcc defines printf() as a macro, which broke pcretest
5339 because it had an ifdef in the middle of a string argument for printf(). Fixed
5340 by using separate calls to printf().
5341
5342 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5343 script, to force use of CR or LF instead of \n in the source. On non-Unix
5344 systems, the value can be set in config.h.
5345
5346 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5347 absolute limit. Changed the text of the error message to make this clear, and
5348 likewise updated the man page.
5349
5350 13. The limit of 99 on the number of capturing subpatterns has been removed.
5351 The new limit is 65535, which I hope will not be a "real" limit.
5352
5353
5354 Version 3.4 22-Aug-00
5355 ---------------------
5356
5357 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5358
5359 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5360
5361
5362 Version 3.3 01-Aug-00
5363 ---------------------
5364
5365 1. If an octal character was given, but the value was greater than \377, it
5366 was not getting masked to the least significant bits, as documented. This could
5367 lead to crashes in some systems.
5368
5369 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5370 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5371
5372 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5373 These just pass their arguments on to (pcre_free)(), but they are provided
5374 because some uses of PCRE bind it to non-C systems that can call its functions,
5375 but cannot call free() or pcre_free() directly.
5376
5377 4. Add "make test" as a synonym for "make check". Corrected some comments in
5378 the Makefile.
5379
5380 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5381 Makefile.
5382
5383 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5384 command called pgrep for grepping around the active processes.
5385
5386 7. Added the beginnings of support for UTF-8 character strings.
5387
5388 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5389 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5390 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5391 out for the ar command.)
5392
5393
5394 Version 3.2 12-May-00
5395 ---------------------
5396
5397 This is purely a bug fixing release.
5398
5399 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5400 of ZA. This was just one example of several cases that could provoke this bug,
5401 which was introduced by change 9 of version 2.00. The code for breaking
5402 infinite loops after an iteration that matches an empty string was't working
5403 correctly.
5404
5405 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5406 when matched against abbab (for example). After matching an empty string, it
5407 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5408 caused it to match further down the string than it should.
5409
5410 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5411 was there because it doesn't seem to be needed, and it causes trouble on some
5412 systems, as it is not a Standard C header. It has been removed.
5413
5414 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5415 were reported on the Macintosh. The changes were from
5416
5417 while ((c = *(++ptr)) != 0 && c != '\n');
5418 to
5419 while ((c = *(++ptr)) != 0 && c != '\n') ;
5420
5421 Totally extraordinary, but if that's what it takes...
5422
5423 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5424 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5425 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5426 assumes the way PCRE uses memmove() (always moving upwards).
5427
5428 6. PCRE is being used in one environment where strchr() is not available. There
5429 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5430 faster code anyway.
5431
5432
5433 Version 3.1 09-Feb-00
5434 ---------------------
5435
5436 The only change in this release is the fixing of some bugs in Makefile.in for
5437 the "install" target:
5438
5439 (1) It was failing to install pcreposix.h.
5440
5441 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5442
5443
5444 Version 3.0 01-Feb-00
5445 ---------------------
5446
5447 1. Add support for the /+ modifier to perltest (to output $` like it does in
5448 pcretest).
5449
5450 2. Add support for the /g modifier to perltest.
5451
5452 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5453 matches null strings.
5454
5455 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5456 pattern. Perl treats empty patterns specially - it reuses the most recent
5457 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5458 effect.
5459
5460 5. The POSIX interface was broken in that it was just handing over the POSIX
5461 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5462 required a bigger vector, with some working space on the end. This means that
5463 the POSIX wrapper now has to get and free some memory, and copy the results.
5464
5465 6. Added some simple autoconf support, placing the test data and the
5466 documentation in separate directories, re-organizing some of the
5467 information files, and making it build pcre-config (a GNU standard). Also added
5468 libtool support for building PCRE as a shared library, which is now the
5469 default.
5470
5471 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5472 09 are not valid octal constants. Single digits will be used for minor values
5473 less than 10.
5474
5475 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5476 existing programs that set these in the POSIX interface can use PCRE without
5477 modification.
5478
5479 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5480 return all that pcre_info() returns, plus additional data. The pcre_info()
5481 function is retained for compatibility, but is considered to be obsolete.
5482
5483 10. Added experimental recursion feature (?R) to handle one common case that
5484 Perl 5.6 will be able to do with (?p{...}).
5485
5486 11. Added support for POSIX character classes like [:alpha:], which Perl is
5487 adopting.
5488
5489
5490 Version 2.08 31-Aug-99
5491 ----------------------
5492
5493 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5494 trying to match at the startoffset position, but instead was moving forward to
5495 the next newline as if a previous match had failed.
5496
5497 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5498 and could get into a loop if a null string was matched other than at the start
5499 of the subject.
5500
5501 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5502 be distinguished at compile time, and for completeness also added PCRE_DATE.
5503
5504 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5505 in GnuWin32 environments.
5506
5507
5508 Version 2.07 29-Jul-99
5509 ----------------------
5510
5511 1. The documentation is now supplied in plain text form and HTML as well as in
5512 the form of man page sources.
5513
5514 2. C++ compilers don't like assigning (void *) values to other pointer types.
5515 In particular this affects malloc(). Although there is no problem in Standard
5516 C, I've put in casts to keep C++ compilers happy.
5517
5518 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5519 should be (const char *).
5520
5521 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5522 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5523 However, I haven't made this a standard facility. The documentation doesn't
5524 mention it, and the Makefile doesn't support it.
5525
5526 5. The Makefile now contains an "install" target, with editable destinations at
5527 the top of the file. The pcretest program is not installed.
5528
5529 6. pgrep -V now gives the PCRE version number and date.
5530
5531 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5532 causing the entire string to be ignored, instead of just the last character.
5533
5534 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5535 non-matching string, it can take a very, very long time, even for strings of
5536 quite modest length, because of the nested recursion. PCRE now does better in
5537 some of these cases. It does this by remembering the last required literal
5538 character in the pattern, and pre-searching the subject to ensure it is present
5539 before running the real match. In other words, it applies a heuristic to detect
5540 some types of certain failure quickly, and in the above example, if presented
5541 with a string that has no trailing " it gives "no match" very quickly.
5542
5543 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5544 other alternatives are tried instead.
5545
5546
5547 Version 2.06 09-Jun-99
5548 ----------------------
5549
5550 1. Change pcretest's output for amount of store used to show just the code
5551 space, because the remainder (the data block) varies in size between 32-bit and
5552 64-bit systems.
5553
5554 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5555 start matching at. This allows lookbehinds to work when searching for multiple
5556 occurrences in a string.
5557
5558 3. Added additional options to pcretest for testing multiple occurrences:
5559
5560 /+ outputs the rest of the string that follows a match
5561 /g loops for multiple occurrences, using the new startoffset argument
5562 /G loops for multiple occurrences by passing an incremented pointer
5563
5564 4. PCRE wasn't doing the "first character" optimization for patterns starting
5565 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5566 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5567 the letter 'x'. On long subject strings, this gives a significant speed-up.
5568
5569
5570 Version 2.05 21-Apr-99
5571 ----------------------
5572
5573 1. Changed the type of magic_number from int to long int so that it works
5574 properly on 16-bit systems.
5575
5576 2. Fixed a bug which caused patterns starting with .* not to work correctly
5577 when the subject string contained newline characters. PCRE was assuming
5578 anchoring for such patterns in all cases, which is not correct because .* will
5579 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5580 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5581 must be retried after every newline in the subject.
5582
5583
5584 Version 2.04 18-Feb-99
5585 ----------------------
5586
5587 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5588 computation of the store needed to hold the pattern was incorrect (too large).
5589 If such patterns were nested a few deep, this could multiply and become a real
5590 problem.
5591
5592 2. Added /M option to pcretest to show the memory requirement of a specific
5593 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5594
5595 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5596 compiled in such a way that the backtracking after subsequent failure was
5597 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5598 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5599
5600
5601 Version 2.03 02-Feb-99
5602 ----------------------
5603
5604 1. Fixed typo and small mistake in man page.
5605
5606 2. Added 4th condition (GPL supersedes if conflict) and created separate
5607 LICENCE file containing the conditions.
5608
5609 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5610 Perl, that is the internal \ allows the delimiter to be included in the
5611 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5612 the final delimiter, add \ to the end of the pattern (to test the error).
5613
5614 4. Added the convenience functions for extracting substrings after a successful
5615 match. Updated pcretest to make it able to test these functions.
5616
5617
5618 Version 2.02 14-Jan-99
5619 ----------------------
5620
5621 1. Initialized the working variables associated with each extraction so that
5622 their saving and restoring doesn't refer to uninitialized store.
5623
5624 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5625 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5626 fix the problem.
5627
5628 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5629 calls, and wasn't printing the correct value for compiling calls. Increased the
5630 default value of LOOPREPEAT, and the number of significant figures in the
5631 times.
5632
5633 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5634
5635 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5636 a building problem on Windows NT with a FAT file system.
5637
5638
5639 Version 2.01 21-Oct-98
5640 ----------------------
5641
5642 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5643 to character tables built by pcre_maketables() in the current locale. If NULL
5644 is passed, the default tables are used.
5645
5646
5647 Version 2.00 24-Sep-98
5648 ----------------------
5649
5650 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5651 it any more.
5652
5653 2. Allow quantification of (?>) groups, and make it work correctly.
5654
5655 3. The first character computation wasn't working for (?>) groups.
5656
5657 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5658 end of the subject) and add 5.005's \z, which really does match only at the
5659 very end of the subject.
5660
5661 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5662
5663 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5664 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5665 localized options. All options to pcre_study() were also removed.
5666
5667 7. Add other new features from 5.005:
5668
5669 $(?<= positive lookbehind
5670 $(?<! negative lookbehind
5671 (?imsx-imsx) added the unsetting capability
5672 such a setting is global if at outer level; local otherwise
5673 (?imsx-imsx:) non-capturing groups with option setting
5674 (?(cond)re|re) conditional pattern matching
5675
5676 A backreference to itself in a repeated group matches the previous
5677 captured string.
5678
5679 8. General tidying up of studying (both automatic and via "study")
5680 consequential on the addition of new assertions.
5681
5682 9. As in 5.005, unlimited repeated groups that could match an empty substring
5683 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5684 runtime if any iteration does actually match an empty substring.
5685
5686 10. Include the RunTest script in the distribution.
5687
5688 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5689 discrepancies, some of which were old and were also with respect to 5.004. They
5690 have now been fixed.
5691
5692
5693 Version 1.09 28-Apr-98
5694 ----------------------
5695
5696 1. A negated single character class followed by a quantifier with a minimum
5697 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5698 program crashes, or just wrong answers. This did not apply to negated classes
5699 containing more than one character, or to minima other than one.
5700
5701
5702 Version 1.08 27-Mar-98
5703 ----------------------
5704
5705 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5706
5707 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5708 latter must appear before anything that relies on it in the pattern.
5709
5710
5711 Version 1.07 16-Feb-98
5712 ----------------------
5713
5714 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5715 repeat of a potentially empty string).
5716
5717
5718 Version 1.06 23-Jan-98
5719 ----------------------
5720
5721 1. Added Markus Oberhumer's little patches for C++.
5722
5723 2. Literal strings longer than 255 characters were broken.
5724
5725
5726 Version 1.05 23-Dec-97
5727 ----------------------
5728
5729 1. Negated character classes containing more than one character were failing if
5730 PCRE_CASELESS was set at run time.
5731
5732
5733 Version 1.04 19-Dec-97
5734 ----------------------
5735
5736 1. Corrected the man page, where some "const" qualifiers had been omitted.
5737
5738 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5739 input syntax.
5740
5741 3. Fixed memory leak which occurred when a regex with back references was
5742 matched with an offsets vector that wasn't big enough. The temporary memory
5743 that is used in this case wasn't being freed if the match failed.
5744
5745 4. Tidied pcretest to ensure it frees memory that it gets.
5746
5747 5. Temporary memory was being obtained in the case where the passed offsets
5748 vector was exactly big enough.
5749
5750 6. Corrected definition of offsetof() from change 5 below.
5751
5752 7. I had screwed up change 6 below and broken the rules for the use of
5753 setjmp(). Now fixed.
5754
5755
5756 Version 1.03 18-Dec-97
5757 ----------------------
5758
5759 1. A erroneous regex with a missing opening parenthesis was correctly
5760 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5761 on some systems.
5762
5763 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5764 it was reported that one broken compiler failed on the former because "code" is
5765 also an independent variable.
5766
5767 3. The erroneous regex a[]b caused an array overrun reference.
5768
5769 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5770 fail on data ending with that character. (It was going on too far, and checking
5771 the next character, typically a binary zero.) This was specific to the
5772 optimized code for single-character negative classes.
5773
5774 5. Added a contributed patch from the TIN world which does the following:
5775
5776 + Add an undef for memmove, in case the the system defines a macro for it.
5777
5778 + Add a definition of offsetof(), in case there isn't one. (I don't know
5779 the reason behind this - offsetof() is part of the ANSI standard - but
5780 it does no harm).
5781
5782 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5783 most of the places where whitespace preceded '#'. I have given up and
5784 allowed the remaining 2 cases to be at the margin.
5785
5786 + Rename some variables in pcre to eliminate shadowing. This seems very
5787 pedantic, but does no harm, of course.
5788
5789 6. Moved the call to setjmp() into its own function, to get rid of warnings
5790 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5791
5792 7. Constructs such as \d{8,} were compiling into the equivalent of
5793 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5794 outcome, but in this particular case used more store than had been allocated,
5795 which caused the bug to be discovered because it threw up an internal error.
5796
5797 8. The debugging code in both pcre and pcretest for outputting the compiled
5798 form of a regex was going wrong in the case of back references followed by
5799 curly-bracketed repeats.
5800
5801
5802 Version 1.02 12-Dec-97
5803 ----------------------
5804
5805 1. Typos in pcre.3 and comments in the source fixed.
5806
5807 2. Applied a contributed patch to get rid of places where it used to remove
5808 'const' from variables, and fixed some signed/unsigned and uninitialized
5809 variable warnings.
5810
5811 3. Added the "runtest" target to Makefile.
5812
5813 4. Set default compiler flag to -O2 rather than just -O.
5814
5815
5816 Version 1.01 19-Nov-97
5817 ----------------------
5818
5819 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5820 like /([ab]*)*/, that is, for classes with more than one character in them.
5821
5822 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5823 as /((?>a*))*/ (a PCRE_EXTRA facility).
5824
5825
5826 Version 1.00 18-Nov-97
5827 ----------------------
5828
5829 1. Added compile-time macros to support systems such as SunOS4 which don't have
5830 memmove() or strerror() but have other things that can be used instead.
5831
5832 2. Arranged that "make clean" removes the executables.
5833
5834
5835 Version 0.99 27-Oct-97
5836 ----------------------
5837
5838 1. Fixed bug in code for optimizing classes with only one character. It was
5839 initializing a 32-byte map regardless, which could cause it to run off the end
5840 of the memory it had got.
5841
5842 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5843
5844
5845 Version 0.98 22-Oct-97
5846 ----------------------
5847
5848 1. Fixed bug in code for handling temporary memory usage when there are more
5849 back references than supplied space in the ovector. This could cause segfaults.
5850
5851
5852 Version 0.97 21-Oct-97
5853 ----------------------
5854
5855 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5856
5857 2. Optimized negated single characters not to use a bit map.
5858
5859 3. Brought error texts together as macro definitions; clarified some of them;
5860 fixed one that was wrong - it said "range out of order" when it meant "invalid
5861 escape sequence".
5862
5863 4. Changed some char * arguments to const char *.
5864
5865 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5866
5867 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5868 pcretest.
5869
5870
5871 Version 0.96 16-Oct-97
5872 ----------------------
5873
5874 1. Added a simple "pgrep" utility to the distribution.
5875
5876 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5877 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5878 where "ddd" means "one or more decimal digits".
5879
5880 3. Fixed serious bug. If a pattern had a back reference, but the call to
5881 pcre_exec() didn't supply a large enough ovector to record the related
5882 identifying subpattern, the match always failed. PCRE now remembers the number
5883 of the largest back reference, and gets some temporary memory in which to save
5884 the offsets during matching if necessary, in order to ensure that
5885 backreferences always work.
5886
5887 4. Increased the compatibility with Perl in a number of ways:
5888
5889 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5890 to request this handling. The option can be set at compile or exec time.
5891
5892 (b) $ matches before a terminating newline by default; an option
5893 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5894 mode). The option can be set at compile or exec time.
5895
5896 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5897 the same as Perl's. If the decimal number it represents is less than 10
5898 or there aren't that many previous left capturing parentheses, an octal
5899 escape is read. Inside a character class, it's always an octal escape,
5900 even if it is a single digit.
5901
5902 (d) An escaped but undefined alphabetic character is taken as a literal,
5903 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5904 escapes.
5905
5906 (e) {0} is now permitted. (The previous item is removed from the compiled
5907 pattern).
5908
5909 5. Changed all the names of code files so that the basic parts are no longer
5910 than 10 characters, and abolished the teeny "globals.c" file.
5911
5912 6. Changed the handling of character classes; they are now done with a 32-byte
5913 bit map always.
5914
5915 7. Added the -d and /D options to pcretest to make it possible to look at the
5916 internals of compilation without having to recompile pcre.
5917
5918
5919 Version 0.95 23-Sep-97
5920 ----------------------
5921
5922 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5923 \x20 at the start of a run of normal characters. These were being treated as
5924 real characters, instead of the source characters being re-checked.
5925
5926
5927 Version 0.94 18-Sep-97
5928 ----------------------
5929
5930 1. The functions are now thread-safe, with the caveat that the global variables
5931 containing pointers to malloc() and free() or alternative functions are the
5932 same for all threads.
5933
5934 2. Get pcre_study() to generate a bitmap of initial characters for non-
5935 anchored patterns when this is possible, and use it if passed to pcre_exec().
5936
5937
5938 Version 0.93 15-Sep-97
5939 ----------------------
5940
5941 1. /(b)|(:+)/ was computing an incorrect first character.
5942
5943 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5944 but not actually doing anything yet.
5945
5946 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5947 as Perl does (e.g. [-az] or [az-]).
5948
5949 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5950 all possible positions.
5951
5952 5. Split up into different modules to avoid including unneeded functions in a
5953 compiled binary. However, compile and exec are still in one module. The "study"
5954 function is split off.
5955
5956 6. The character tables are now in a separate module whose source is generated
5957 by an auxiliary program - but can then be edited by hand if required. There are
5958 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5959 toupper() in the code.
5960
5961 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5962 make them global. Abolish the function for setting them, as the caller can now
5963 set them directly.
5964
5965
5966 Version 0.92 11-Sep-97
5967 ----------------------
5968
5969 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5970 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5971
5972 2. Caseless matching was not working in character classes if the characters in
5973 the pattern were in upper case.
5974
5975 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5976
5977 4. Make PCRE_ANCHORED public and accept as a compile option.
5978
5979 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5980 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5981 pass them.
5982
5983 6. Give an error if bad option bits passed at compile or run time.
5984
5985 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5986 pcretest to cause it to pass that flag.
5987
5988 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5989 options, and the first character, if set.
5990
5991 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5992
5993
5994 Version 0.91 10-Sep-97
5995 ----------------------
5996
5997 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5998 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5999
6000 2. PCRE was looping on encountering an indefinitely repeated back reference to
6001 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
6002 Perl does - treats the match as successful.
6003
6004 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5