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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1766 - (show annotations)
Tue Jun 15 16:15:50 2021 UTC (2 days, 12 hours ago) by ph10
File size: 299676 byte(s)
Error occurred while calculating annotation data.
Final source tidies for 8.45 release.
1 ChangeLog for PCRE
2 ------------------
3
4 Note that the PCRE 8.xx series (PCRE1) is now at end of life. All development
5 is happening in the PCRE2 10.xx series.
6
7
8 Version 8.45 15-June-2021
9 -------------------------
10
11 This is the final release of PCRE1. A few minor tidies are included.
12
13 1. CMakeLists.txt has two user-supplied patches applied, one to allow for the
14 setting of MODULE_PATH, and the other to support the generation of pcre-config
15 file and libpcre*.pc files.
16
17 2. There was a memory leak if a compile error occurred when there were more
18 than 20 named groups (Bugzilla #2613).
19
20 3. Fixed some typos in code and documentation.
21
22 4. Fixed a small (*MARK) bug in the interpreter (Bugzilla #2771).
23
24
25 Version 8.44 12 February-2020
26 -----------------------------
27
28 1. Setting --enable-jit=auto for an out-of-tree build failed because the
29 source directory wasn't in the search path for AC_TRY_COMPILE always. Patch
30 from Ross Burton.
31
32 2. Applied a patch from Michael Shigorin to fix 8.43 build on e2k arch
33 with lcc compiler (EDG frontend based); the problem it fixes is:
34
35 lcc: "pcrecpp.cc", line 74: error: declaration aliased to undefined entity
36 "_ZN7pcrecpp2RE6no_argE" [-Werror]
37
38 3. Change 2 for 8.43 omitted (*LF) from the list of start-of-pattern items. Now
39 added.
40
41 4. Fix ARMv5 JIT improper handling of labels right after a constant pool.
42
43 5. Small patch to pcreposix.c to set the erroroffset field to -1 immediately
44 after a successful compile, instead of at the start of matching to avoid a
45 sanitizer complaint (regexec is supposed to be thread safe).
46
47 6. Check the size of the number after (?C as it is read, in order to avoid
48 integer overflow.
49
50 7. Tidy up left shifts to avoid sanitize warnings; also fix one NULL deference
51 in pcretest.
52
53
54 Version 8.43 23-February-2019
55 -----------------------------
56
57 1. Some time ago the config macro SUPPORT_UTF8 was changed to SUPPORT_UTF
58 because it also applies to UTF-16 and UTF-32. However, this change was not made
59 in the pcre2cpp files; consequently the C++ wrapper has from then been compiled
60 with a bug in it, which would have been picked up by the unit test except that
61 it also had its UTF8 code cut out. The bug was in a global replace when moving
62 forward after matching an empty string.
63
64 2. The C++ wrapper got broken a long time ago (version 7.3, August 2007) when
65 (*CR) was invented (assuming it was the first such start-of-pattern option).
66 The wrapper could never handle such patterns because it wraps patterns in
67 (?:...)\z in order to support end anchoring. I have hacked in some code to fix
68 this, that is, move the wrapping till after any existing start-of-pattern
69 special settings.
70
71 3. "pcre2grep" (sic) was accidentally mentioned in an error message (fix was
72 ported from PCRE2).
73
74 4. Typo LCC_ALL for LC_ALL fixed in pcregrep.
75
76 5. In a pattern such as /[^\x{100}-\x{ffff}]*[\x80-\xff]/ which has a repeated
77 negative class with no characters less than 0x100 followed by a positive class
78 with only characters less than 0x100, the first class was incorrectly being
79 auto-possessified, causing incorrect match failures.
80
81 6. If the only branch in a conditional subpattern was anchored, the whole
82 subpattern was treated as anchored, when it should not have been, since the
83 assumed empty second branch cannot be anchored. Demonstrated by test patterns
84 such as /(?(1)^())b/ or /(?(?=^))b/.
85
86 7. Fix subject buffer overread in JIT when UTF is disabled and \X or \R has
87 a greater than 1 fixed quantifier. This issue was found by Yunho Kim.
88
89 8. If a pattern started with a subroutine call that had a quantifier with a
90 minimum of zero, an incorrect "match must start with this character" could be
91 recorded. Example: /(?&xxx)*ABC(?<xxx>XYZ)/ would (incorrectly) expect 'A' to
92 be the first character of a match.
93
94 9. Improve MAP_JIT flag usage on MacOS. Patch by Rich Siegel.
95
96
97 Version 8.42 20-March-2018
98 --------------------------
99
100 1. Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
101
102 2. Fixed outdated real_pcre definitions in pcre.h.in (patch by Evgeny Kotkov).
103
104 3. pcregrep was truncating components of file names to 128 characters when
105 processing files with the -r option, and also (some very odd code) truncating
106 path names to 512 characters. There is now a check on the absolute length of
107 full path file names, which may be up to 2047 characters long.
108
109 4. Using pcre_dfa_exec(), in UTF mode when UCP support was not defined, there
110 was the possibility of a false positive match when caselessly matching a "not
111 this character" item such as [^\x{1234}] (with a code point greater than 127)
112 because the "other case" variable was not being initialized.
113
114 5. Although pcre_jit_exec checks whether the pattern is compiled
115 in a given mode, it was also expected that at least one mode is available.
116 This is fixed and pcre_jit_exec returns with PCRE_ERROR_JIT_BADOPTION
117 when the pattern is not optimized by JIT at all.
118
119 6. The line number and related variables such as match counts in pcregrep
120 were all int variables, causing overflow when files with more than 2147483647
121 lines were processed (assuming 32-bit ints). They have all been changed to
122 unsigned long ints.
123
124 7. If a backreference with a minimum repeat count of zero was first in a
125 pattern, apart from assertions, an incorrect first matching character could be
126 recorded. For example, for the pattern /(?=(a))\1?b/, "b" was incorrectly set
127 as the first character of a match.
128
129 8. Fix out-of-bounds read for partial matching of /./ against an empty string
130 when the newline type is CRLF.
131
132 9. When matching using the the REG_STARTEND feature of the POSIX API with a
133 non-zero starting offset, unset capturing groups with lower numbers than a
134 group that did capture something were not being correctly returned as "unset"
135 (that is, with offset values of -1).
136
137 10. Matching the pattern /(*UTF)\C[^\v]+\x80/ against an 8-bit string
138 containing multi-code-unit characters caused bad behaviour and possibly a
139 crash. This issue was fixed for other kinds of repeat in release 8.37 by change
140 38, but repeating character classes were overlooked.
141
142 11. A small fix to pcregrep to avoid compiler warnings for -Wformat-overflow=2.
143
144 12. Added --enable-jit=auto support to configure.ac.
145
146 13. Fix misleading error message in configure.ac.
147
148
149 Version 8.41 05-July-2017
150 -------------------------
151
152 1. Fixed typo in CMakeLists.txt (wrong number of arguments for
153 PCRE_STATIC_RUNTIME (affects MSVC only).
154
155 2. Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
156 mode with --only-matching matched several lines, it restarted scanning at the
157 next line instead of moving on to the end of the matched string, which can be
158 several lines after the start.
159
160 3. Fix a missing else in the JIT compiler reported by 'idaifish'.
161
162 4. A (?# style comment is now ignored between a basic quantifier and a
163 following '+' or '?' (example: /X+(?#comment)?Y/.
164
165 5. Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
166 Pisar).
167
168 6. Fuzzers have reported issues in pcretest. These are NOT serious (it is,
169 after all, just a test program). However, to stop the reports, some easy ones
170 are fixed:
171
172 (a) Check for values < 256 when calling isprint() in pcretest.
173 (b) Give an error for too big a number after \O.
174
175 7. In the 32-bit library in non-UTF mode, an attempt to find a Unicode
176 property for a character with a code point greater than 0x10ffff (the Unicode
177 maximum) caused a crash.
178
179 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
180 encountered a character class with a possessive repeat, for example [a-f]{3}+.
181
182 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
183 length incorrectly, which could result in buffer overflow.
184
185 10. Remove redundant line of code (accidentally left in ages ago).
186
187 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
188 with namespace pcrecpp (Bugzilla #2084).
189
190 12. Remove a duplication typo in pcre_tables.c.
191
192 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
193 starting offset greater than zero.
194
195
196 Version 8.40 11-January-2017
197 ----------------------------
198
199 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
200 the match extended over a line boundary.
201
202 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
203 MSVC static compilation, putting the first patch under a new option.
204
205 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
206
207 4. Ignore "show all captures" (/=) for DFA matching.
208
209 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
210
211 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
212 without PCRE_UCP set, a negative character type such as \D in a positive
213 class should cause all characters greater than 255 to match, whatever else
214 is in the class. There was a bug that caused this not to happen if a
215 Unicode property item was added to such a class, for example [\D\P{Nd}] or
216 [\W\pL].
217
218 7. When pcretest was outputing information from a callout, the caret indicator
219 for the current position in the subject line was incorrect if it was after
220 an escape sequence for a character whose code point was greater than
221 \x{ff}.
222
223 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
224 the conditional was interpreted as a reference to capturing group 1 instead
225 of a test for recursion. Any group whose name began with R was
226 misinterpreted in this way. (The reference interpretation should only
227 happen if the group's name is precisely "R".)
228
229 9. A number of bugs have been mended relating to match start-up optimizations
230 when the first thing in a pattern is a positive lookahead. These all
231 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
232
233 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
234 both an initial 'X' and a following 'X'.
235 (b) Some patterns starting with an assertion that started with .* were
236 incorrectly optimized as having to match at the start of the subject or
237 after a newline. There are cases where this is not true, for example,
238 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
239 start with spaces. Starting .* in an assertion is no longer taken as an
240 indication of matching at the start (or after a newline).
241
242
243 Version 8.39 14-June-2016
244 -------------------------
245
246 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
247 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
248 misbehaved. This bug was found by the LLVM fuzzer.
249
250 2. Similar to the above, if an isolated \E was present between an item and its
251 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
252 bug was found by the LLVM fuzzer.
253
254 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
255 working correctly in UCP mode.
256
257 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
258 was set when the pmatch argument was NULL. It now returns REG_INVARG.
259
260 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
261
262 6. An empty \Q\E sequence between an item and its qualifier caused
263 pcre_compile() to misbehave when auto callouts were enabled. This bug was
264 found by the LLVM fuzzer.
265
266 7. If a pattern that was compiled with PCRE_EXTENDED started with white
267 space or a #-type comment that was followed by (?-x), which turns off
268 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
269 pcre_compile() assumed that (?-x) applied to the whole pattern and
270 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
271
272 8. A call of pcre_copy_named_substring() for a named substring whose number
273 was greater than the space in the ovector could cause a crash.
274
275 9. Yet another buffer overflow bug involved duplicate named groups with a
276 group that reset capture numbers (compare 8.38/7 below). Once again, I have
277 just allowed for more memory, even if not needed. (A proper fix is
278 implemented in PCRE2, but it involves a lot of refactoring.)
279
280 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
281 start of the match to be earlier than the end.
282
283 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
284
285 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
286 assertion, caused pcretest to generate incorrect output, and also to read
287 uninitialized memory (detected by ASAN or valgrind).
288
289 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
290 nested set of parentheses of sufficient size caused an overflow of the
291 compiling workspace (which was diagnosed, but of course is not desirable).
292
293 14. And yet another buffer overflow bug involving duplicate named groups, this
294 time nested, with a nested back reference. Yet again, I have just allowed
295 for more memory, because anything more needs all the refactoring that has
296 been done for PCRE2. An example pattern that provoked this bug is:
297 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
298 registered as CVE-2016-1283.
299
300 15. pcretest went into a loop if global matching was requested with an ovector
301 size less than 2. It now gives an error message. This bug was found by
302 afl-fuzz.
303
304 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
305 ("assertion expected") when (?(?C) was not followed by an opening
306 parenthesis.
307
308 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
309 actually affect anything, by sheer luck.
310
311 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
312 static compilation.
313
314 19. Modified the RunTest script to incorporate a valgrind suppressions file so
315 that certain errors, provoked by the SSE2 instruction set when JIT is used,
316 are ignored.
317
318 20. A racing condition is fixed in JIT reported by Mozilla.
319
320 21. Minor code refactor to avoid "array subscript is below array bounds"
321 compiler warning.
322
323 22. Minor code refactor to avoid "left shift of negative number" warning.
324
325 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
326 UCP support.
327
328 24. Refactor to avoid compiler warnings in pcrecpp.cc.
329
330 25. Refactor to fix a typo in pcre_jit_test.c
331
332 26. Patch to support compiling pcrecpp.cc with Intel compiler.
333
334
335 Version 8.38 23-November-2015
336 -----------------------------
337
338 1. If a group that contained a recursive back reference also contained a
339 forward reference subroutine call followed by a non-forward-reference
340 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
341 compile correct code, leading to undefined behaviour or an internally
342 detected error. This bug was discovered by the LLVM fuzzer.
343
344 2. Quantification of certain items (e.g. atomic back references) could cause
345 incorrect code to be compiled when recursive forward references were
346 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
347 This bug was discovered by the LLVM fuzzer.
348
349 3. A repeated conditional group whose condition was a reference by name caused
350 a buffer overflow if there was more than one group with the given name.
351 This bug was discovered by the LLVM fuzzer.
352
353 4. A recursive back reference by name within a group that had the same name as
354 another group caused a buffer overflow. For example:
355 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
356
357 5. A forward reference by name to a group whose number is the same as the
358 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
359 a buffer overflow at compile time. This bug was discovered by the LLVM
360 fuzzer.
361
362 6. A lookbehind assertion within a set of mutually recursive subpatterns could
363 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
364
365 7. Another buffer overflow bug involved duplicate named groups with a
366 reference between their definition, with a group that reset capture
367 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
368 fixed by always allowing for more memory, even if not needed. (A proper fix
369 is implemented in PCRE2, but it involves more refactoring.)
370
371 8. There was no check for integer overflow in subroutine calls such as (?123).
372
373 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
374 being treated as a literal 'l' instead of causing an error.
375
376 10. There was a buffer overflow if pcre_exec() was called with an ovector of
377 size 1. This bug was found by american fuzzy lop.
378
379 11. If a non-capturing group containing a conditional group that could match
380 an empty string was repeated, it was not identified as matching an empty
381 string itself. For example: /^(?:(?(1)x|)+)+$()/.
382
383 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
384 \a and \e in test subject lines.
385
386 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
387 instead of the EBCDIC value.
388
389 14. The handling of \c in an EBCDIC environment has been revised so that it is
390 now compatible with the specification in Perl's perlebcdic page.
391
392 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
393 ASCII/Unicode. This has now been added to the list of characters that are
394 recognized as white space in EBCDIC.
395
396 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
397 error (correctly) when used outside a class, but did not give an error
398 within a class.
399
400 17. \h within a class was incorrectly compiled in EBCDIC environments.
401
402 18. A pattern with an unmatched closing parenthesis that contained a backward
403 assertion which itself contained a forward reference caused buffer
404 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
405
406 19. JIT should return with error when the compiled pattern requires more stack
407 space than the maximum.
408
409 20. A possessively repeated conditional group that could match an empty string,
410 for example, /(?(R))*+/, was incorrectly compiled.
411
412 21. Fix infinite recursion in the JIT compiler when certain patterns such as
413 /(?:|a|){100}x/ are analysed.
414
415 22. Some patterns with character classes involving [: and \\ were incorrectly
416 compiled and could cause reading from uninitialized memory or an incorrect
417 error diagnosis.
418
419 23. Pathological patterns containing many nested occurrences of [: caused
420 pcre_compile() to run for a very long time.
421
422 24. A conditional group with only one branch has an implicit empty alternative
423 branch and must therefore be treated as potentially matching an empty
424 string.
425
426 25. If (?R was followed by - or + incorrect behaviour happened instead of a
427 diagnostic.
428
429 26. Arrange to give up on finding the minimum matching length for overly
430 complex patterns.
431
432 27. Similar to (4) above: in a pattern with duplicated named groups and an
433 occurrence of (?| it is possible for an apparently non-recursive back
434 reference to become recursive if a later named group with the relevant
435 number is encountered. This could lead to a buffer overflow. Wen Guanxing
436 from Venustech ADLAB discovered this bug.
437
438 28. If pcregrep was given the -q option with -c or -l, or when handling a
439 binary file, it incorrectly wrote output to stdout.
440
441 29. The JIT compiler did not restore the control verb head in case of *THEN
442 control verbs. This issue was found by Karl Skomski with a custom LLVM
443 fuzzer.
444
445 30. Error messages for syntax errors following \g and \k were giving inaccurate
446 offsets in the pattern.
447
448 31. Added a check for integer overflow in conditions (?(<digits>) and
449 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
450 fuzzer.
451
452 32. Handling recursive references such as (?2) when the reference is to a group
453 later in the pattern uses code that is very hacked about and error-prone.
454 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
455 give an internal error if it is obvious that compiling has gone wrong.
456
457 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
458 This issue was found by Karl Skomski with a custom LLVM fuzzer.
459
460 34. The JIT compiler should restore the control chain for empty possessive
461 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
462
463 35. Match limit check added to JIT recursion. This issue was found by Karl
464 Skomski with a custom LLVM fuzzer.
465
466 36. Yet another case similar to 27 above has been circumvented by an
467 unconditional allocation of extra memory. This issue is fixed "properly" in
468 PCRE2 by refactoring the way references are handled. Wen Guanxing
469 from Venustech ADLAB discovered this bug.
470
471 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
472 with a custom LLVM fuzzer.
473
474 38. Fixed a corner case of range optimization in JIT.
475
476 39. An incorrect error "overran compiling workspace" was given if there were
477 exactly enough group forward references such that the last one extended
478 into the workspace safety margin. The next one would have expanded the
479 workspace. The test for overflow was not including the safety margin.
480
481 40. A match limit issue is fixed in JIT which was found by Karl Skomski
482 with a custom LLVM fuzzer.
483
484 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
485 work under Windows. (Why has it taken so long for anyone to notice?)
486
487 42. In a character class such as [\W\p{Any}] where both a negative-type escape
488 ("not a word character") and a property escape were present, the property
489 escape was being ignored.
490
491 43. Fix crash caused by very long (*MARK) or (*THEN) names.
492
493 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
494 by a single ASCII character in a class item, was incorrectly compiled in
495 UCP mode. The POSIX class got lost, but only if the single character
496 followed it.
497
498 45. [:punct:] in UCP mode was matching some characters in the range 128-255
499 that should not have been matched.
500
501 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
502 class, all characters with code points greater than 255 are in the class.
503 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
504 such as \w are turned into Unicode properties), wide characters were not
505 correctly handled, and could fail to match.
506
507
508 Version 8.37 28-April-2015
509 --------------------------
510
511 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
512 for those parentheses to be closed with whatever has been captured so far.
513 However, it was failing to mark any other groups between the hightest
514 capture so far and the currrent group as "unset". Thus, the ovector for
515 those groups contained whatever was previously there. An example is the
516 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
517
518 2. If an assertion condition was quantified with a minimum of zero (an odd
519 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
520
521 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
522 unrecognized modifier, a crash could occur.
523
524 4. An attempt to do global matching in pcretest with a zero-length ovector
525 caused a crash.
526
527 5. Fixed a memory leak during matching that could occur for a subpattern
528 subroutine call (recursive or otherwise) if the number of captured groups
529 that had to be saved was greater than ten.
530
531 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
532 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
533 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
534
535 7. A UTF pattern containing a "not" match of a non-ASCII character and a
536 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
537
538 8. When a pattern is compiled, it remembers the highest back reference so that
539 when matching, if the ovector is too small, extra memory can be obtained to
540 use instead. A conditional subpattern whose condition is a check on a
541 capture having happened, such as, for example in the pattern
542 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
543 setting the highest backreference number. This mattered only if pcre_exec()
544 was called with an ovector that was too small to hold the capture, and there
545 was no other kind of back reference (a situation which is probably quite
546 rare). The effect of the bug was that the condition was always treated as
547 FALSE when the capture could not be consulted, leading to a incorrect
548 behaviour by pcre_exec(). This bug has been fixed.
549
550 9. A reference to a duplicated named group (either a back reference or a test
551 for being set in a conditional) that occurred in a part of the pattern where
552 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
553 to be incorrectly calculated, leading to overwriting.
554
555 10. A mutually recursive set of back references such as (\2)(\1) caused a
556 segfault at study time (while trying to find the minimum matching length).
557 The infinite loop is now broken (with the minimum length unset, that is,
558 zero).
559
560 11. If an assertion that was used as a condition was quantified with a minimum
561 of zero, matching went wrong. In particular, if the whole group had
562 unlimited repetition and could match an empty string, a segfault was
563 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
564 assertions to be quantified, but not if they are being used as conditions,
565 so the above pattern is faulted by Perl. PCRE has now been changed so that
566 it also rejects such patterns.
567
568 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
569 failed to allow the zero-repeat case if pcre2_exec() was called with an
570 ovector too small to capture the group.
571
572 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
573 Red Hat Product Security:
574
575 (a) A crash if /K and /F were both set with the option to save the compiled
576 pattern.
577
578 (b) Another crash if the option to print captured substrings in a callout
579 was combined with setting a null ovector, for example \O\C+ as a subject
580 string.
581
582 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
583 forward reference repeated a large (but limited) number of times within a
584 repeated outer group that has a zero minimum quantifier, caused incorrect
585 code to be compiled, leading to the error "internal error:
586 previously-checked referenced subpattern not found" when an incorrect
587 memory address was read. This bug was reported as "heap overflow",
588 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
589 CVE-2015-2325.
590
591 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
592 call within a group that also contained a recursive back reference caused
593 incorrect code to be compiled. This bug was reported as "heap overflow",
594 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
595 number CVE-2015-2326.
596
597 24. Computing the size of the JIT read-only data in advance has been a source
598 of various issues, and new ones are still appear unfortunately. To fix
599 existing and future issues, size computation is eliminated from the code,
600 and replaced by on-demand memory allocation.
601
602 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
603 adjacent to the end of the range, and the range contained characters with
604 more than one other case, caused incorrect behaviour when compiled in UTF
605 mode. In that example, the range a-j was left out of the class.
606
607 26. Fix JIT compilation of conditional blocks, which assertion
608 is converted to (*FAIL). E.g: /(?(?!))/.
609
610 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
611 discovered by the LLVM fuzzer.
612
613 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
614 when this assertion was used as a condition, for example (?(?!)a|b). In
615 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
616 error about an unsupported item.
617
618 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
619 possessification code could take exponential time to complete. A recursion
620 depth limit of 1000 has been imposed to limit the resources used by this
621 optimization.
622
623 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
624 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
625 because \S ensures they are all in the class. The code for doing this was
626 interacting badly with the code for computing the amount of space needed to
627 compile the pattern, leading to a buffer overflow. This bug was discovered
628 by the LLVM fuzzer.
629
630 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
631 other kinds of group caused stack overflow at compile time. This bug was
632 discovered by the LLVM fuzzer.
633
634 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
635 between a subroutine call and its quantifier was incorrectly compiled,
636 leading to buffer overflow or other errors. This bug was discovered by the
637 LLVM fuzzer.
638
639 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
640 assertion after (?(. The code was failing to check the character after
641 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
642 was discovered by the LLVM fuzzer.
643
644 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
645 a fixed maximum following a group that contains a subroutine reference was
646 incorrectly compiled and could trigger buffer overflow. This bug was
647 discovered by the LLVM fuzzer.
648
649 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
650 caused a stack overflow instead of the diagnosis of a non-fixed length
651 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
652
653 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
654 (e.g. /(?<=\Ka)/) could make pcregrep loop.
655
656 37. There was a similar problem to 36 in pcretest for global matches.
657
658 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
659 and a subsequent item in the pattern caused a non-match, backtracking over
660 the repeated \X did not stop, but carried on past the start of the subject,
661 causing reference to random memory and/or a segfault. There were also some
662 other cases where backtracking after \C could crash. This set of bugs was
663 discovered by the LLVM fuzzer.
664
665 39. The function for finding the minimum length of a matching string could take
666 a very long time if mutual recursion was present many times in a pattern,
667 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
668 method has been implemented. This infelicity was discovered by the LLVM
669 fuzzer.
670
671 40. Static linking against the PCRE library using the pkg-config module was
672 failing on missing pthread symbols.
673
674
675 Version 8.36 26-September-2014
676 ------------------------------
677
678 1. Got rid of some compiler warnings in the C++ modules that were shown up by
679 -Wmissing-field-initializers and -Wunused-parameter.
680
681 2. The tests for quantifiers being too big (greater than 65535) were being
682 applied after reading the number, and stupidly assuming that integer
683 overflow would give a negative number. The tests are now applied as the
684 numbers are read.
685
686 3. Tidy code in pcre_exec.c where two branches that used to be different are
687 now the same.
688
689 4. The JIT compiler did not generate match limit checks for certain
690 bracketed expressions with quantifiers. This may lead to exponential
691 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
692 issue should be resolved now.
693
694 5. Fixed an issue, which occures when nested alternatives are optimized
695 with table jumps.
696
697 6. Inserted two casts and changed some ints to size_t in the light of some
698 reported 64-bit compiler warnings (Bugzilla 1477).
699
700 7. Fixed a bug concerned with zero-minimum possessive groups that could match
701 an empty string, which sometimes were behaving incorrectly in the
702 interpreter (though correctly in the JIT matcher). This pcretest input is
703 an example:
704
705 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
706 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
707
708 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
709 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
710 for an empty string was breaking the inner loop and carrying on at a lower
711 level, when possessive repeated groups should always return to a higher
712 level as they have no backtrack points in them. The empty string test now
713 occurs at the outer level.
714
715 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
716 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
717
718 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
719 doesn't).
720
721 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
722 to hold a pattern to be incorrectly computed (too small) when there were
723 named back references to duplicated names. This could cause "internal
724 error: code overflow" or "double free or corruption" or other memory
725 handling errors.
726
727 11. When named subpatterns had the same prefixes, back references could be
728 confused. For example, in this pattern:
729
730 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
731
732 the reference to 'Name' was incorrectly treated as a reference to a
733 duplicate name.
734
735 12. A pattern such as /^s?c/mi8 where the optional character has more than
736 one "other case" was incorrectly compiled such that it would only try to
737 match starting at "c".
738
739 13. When a pattern starting with \s was studied, VT was not included in the
740 list of possible starting characters; this should have been part of the
741 8.34/18 patch.
742
743 14. If a character class started [\Qx]... where x is any character, the class
744 was incorrectly terminated at the ].
745
746 15. If a pattern that started with a caseless match for a character with more
747 than one "other case" was studied, PCRE did not set up the starting code
748 unit bit map for the list of possible characters. Now it does. This is an
749 optimization improvement, not a bug fix.
750
751 16. The Unicode data tables have been updated to Unicode 7.0.0.
752
753 17. Fixed a number of memory leaks in pcregrep.
754
755 18. Avoid a compiler warning (from some compilers) for a function call with
756 a cast that removes "const" from an lvalue by using an intermediate
757 variable (to which the compiler does not object).
758
759 19. Incorrect code was compiled if a group that contained an internal recursive
760 back reference was optional (had quantifier with a minimum of zero). This
761 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
762 caused segmentation faults because of stack overflows at compile time.
763
764 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
765 group that is quantified with an indefinite repeat, caused a compile-time
766 loop which used up all the system stack and provoked a segmentation fault.
767 This was not the same bug as 19 above.
768
769 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
770 Patch by Mike Frysinger.
771
772
773 Version 8.35 04-April-2014
774 --------------------------
775
776 1. A new flag is set, when property checks are present in an XCLASS.
777 When this flag is not set, PCRE can perform certain optimizations
778 such as studying these XCLASS-es.
779
780 2. The auto-possessification of character sets were improved: a normal
781 and an extended character set can be compared now. Furthermore
782 the JIT compiler optimizes more character set checks.
783
784 3. Got rid of some compiler warnings for potentially uninitialized variables
785 that show up only when compiled with -O2.
786
787 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
788 of a match later then the end of the match. The pcretest program was not
789 handling the case sensibly - it was outputting from the start to the next
790 binary zero. It now reports this situation in a message, and outputs the
791 text from the end to the start.
792
793 5. Fast forward search is improved in JIT. Instead of the first three
794 characters, any three characters with fixed position can be searched.
795 Search order: first, last, middle.
796
797 6. Improve character range checks in JIT. Characters are read by an inprecise
798 function now, which returns with an unknown value if the character code is
799 above a certain threshold (e.g: 256). The only limitation is that the value
800 must be bigger than the threshold as well. This function is useful when
801 the characters above the threshold are handled in the same way.
802
803 7. The macros whose names start with RAWUCHAR are placeholders for a future
804 mode in which only the bottom 21 bits of 32-bit data items are used. To
805 make this more memorable for those maintaining the code, the names have
806 been changed to start with UCHAR21, and an extensive comment has been added
807 to their definition.
808
809 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
810 to the export list in Makefile.am (they were accidentally omitted from the
811 8.34 tarball).
812
813 9. The informational output from pcretest used the phrase "starting byte set"
814 which is inappropriate for the 16-bit and 32-bit libraries. As the output
815 for "first char" and "need char" really means "non-UTF-char", I've changed
816 "byte" to "char", and slightly reworded the output. The documentation about
817 these values has also been (I hope) clarified.
818
819 10. Another JIT related optimization: use table jumps for selecting the correct
820 backtracking path, when more than four alternatives are present inside a
821 bracket.
822
823 11. Empty match is not possible, when the minimum length is greater than zero,
824 and there is no \K in the pattern. JIT should avoid empty match checks in
825 such cases.
826
827 12. In a caseless character class with UCP support, when a character with more
828 than one alternative case was not the first character of a range, not all
829 the alternative cases were added to the class. For example, s and \x{17f}
830 are both alternative cases for S: the class [RST] was handled correctly,
831 but [R-T] was not.
832
833 13. The configure.ac file always checked for pthread support when JIT was
834 enabled. This is not used in Windows, so I have put this test inside a
835 check for the presence of windows.h (which was already tested for).
836
837 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
838 The algorithm provides a way to skip certain starting offsets, and usually
839 faster than linear prefix searches.
840
841 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
842 as for 'fr_FR' and 'french'. For some reason, however, it then used the
843 Windows-specific input and output files, which have 'french' screwed in.
844 So this could never have worked. One of the problems with locales is that
845 they aren't always the same. I have now updated RunTest so that it checks
846 the output of the locale test (test 3) against three different output
847 files, and it allows the test to pass if any one of them matches. With luck
848 this should make the test pass on some versions of Solaris where it was
849 failing. Because of the uncertainty, the script did not used to stop if
850 test 3 failed; it now does. If further versions of a French locale ever
851 come to light, they can now easily be added.
852
853 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
854 there was a message during ./configure, but it did not stop. This now
855 provokes an error. The invalid example in README has been corrected.
856 If a value less than the minimum is given, the minimum value has always
857 been used, but now a warning is given.
858
859 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
860 was a bug in the test system, which is now fixed. Also, the list of various
861 configurations that are tested for each release did not have one with both
862 16/32 bits and --enable-bar-anycrlf. It now does.
863
864 18. pcretest was missing "-C bsr" for displaying the \R default setting.
865
866 19. Little endian PowerPC systems are supported now by the JIT compiler.
867
868 20. The fast forward newline mechanism could enter to an infinite loop on
869 certain invalid UTF-8 input. Although we don't support these cases
870 this issue can be fixed by a performance optimization.
871
872 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
873 not take account if existing stack usage. There is now a new global
874 variable called pcre_stack_guard that can be set to point to an external
875 function to check stack availability. It is called at the start of
876 processing every parenthesized group.
877
878 22. A typo in the code meant that in ungreedy mode the max/min qualifier
879 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
880 not match "ab".
881
882 23. When UTF was disabled, the JIT program reported some incorrect compile
883 errors. These messages are silenced now.
884
885 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
886 compiler.
887
888 25. Change all the temporary files used in RunGrepTest to be different to those
889 used by RunTest so that the tests can be run simultaneously, for example by
890 "make -j check".
891
892
893 Version 8.34 15-December-2013
894 -----------------------------
895
896 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
897 executable memory. Patch inspired by Carsten Klein.
898
899 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
900 this macro is never tested and has no effect, because the work to support
901 coverage involves only compiling and linking options and special targets in
902 the Makefile. The comment in config.h implied that defining the macro would
903 enable coverage support, which is totally false. There was also support for
904 setting this macro in the CMake files (my fault, I just copied it from
905 configure). SUPPORT_GCOV has now been removed.
906
907 3. Make a small performance improvement in strlen16() and strlen32() in
908 pcretest.
909
910 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
911 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
912 these cases). There was also one in pcretest.c.
913
914 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
915
916 6. In UTF mode, the code for checking whether a group could match an empty
917 string (which is used for indefinitely repeated groups to allow for
918 breaking an infinite loop) was broken when the group contained a repeated
919 negated single-character class with a character that occupied more than one
920 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
921 UTF-8 mode). The effect was undefined: the group might or might not be
922 deemed as matching an empty string, or the program might have crashed.
923
924 7. The code for checking whether a group could match an empty string was not
925 recognizing that \h, \H, \v, \V, and \R must match a character.
926
927 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
928 an empty string. If it can, pcretest shows this in its information output.
929
930 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
931 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
932 matched by pcre_exec() without using JIT:
933
934 (a) If the rest of the pattern did not match after a maximal run of
935 grapheme clusters, the code for backing up to try with fewer of them
936 did not always back up over a full grapheme when characters that do not
937 have the modifier quality were involved, e.g. Hangul syllables.
938
939 (b) If the match point in a subject started with modifier character, and
940 there was no match, the code could incorrectly back up beyond the match
941 point, and potentially beyond the first character in the subject,
942 leading to a segfault or an incorrect match result.
943
944 10. A conditional group with an assertion condition could lead to PCRE
945 recording an incorrect first data item for a match if no other first data
946 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
947 first data item, and therefore matched "ca" after "c" instead of at the
948 start.
949
950 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
951 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
952
953 12. The source of pcregrep now includes z/OS-specific code so that it can be
954 compiled for z/OS as part of the special z/OS distribution.
955
956 13. Added the -T and -TM options to pcretest.
957
958 14. The code in pcre_compile.c for creating the table of named capturing groups
959 has been refactored. Instead of creating the table dynamically during the
960 actual compiling pass, the information is remembered during the pre-compile
961 pass (on the stack unless there are more than 20 named groups, in which
962 case malloc() is used) and the whole table is created before the actual
963 compile happens. This has simplified the code (it is now nearly 150 lines
964 shorter) and prepared the way for better handling of references to groups
965 with duplicate names.
966
967 15. A back reference to a named subpattern when there is more than one of the
968 same name now checks them in the order in which they appear in the pattern.
969 The first one that is set is used for the reference. Previously only the
970 first one was inspected. This change makes PCRE more compatible with Perl.
971
972 16. Unicode character properties were updated from Unicode 6.3.0.
973
974 17. The compile-time code for auto-possessification has been refactored, based
975 on a patch by Zoltan Herczeg. It now happens after instead of during
976 compilation. The code is cleaner, and more cases are handled. The option
977 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
978 options in pcretest are provided to set it. It can also be set by
979 (*NO_AUTO_POSSESS) at the start of a pattern.
980
981 18. The character VT has been added to the default ("C" locale) set of
982 characters that match \s and are generally treated as white space,
983 following this same change in Perl 5.18. There is now no difference between
984 "Perl space" and "POSIX space". Whether VT is treated as white space in
985 other locales depends on the locale.
986
987 19. The code for checking named groups as conditions, either for being set or
988 for being recursed, has been refactored (this is related to 14 and 15
989 above). Processing unduplicated named groups should now be as fast at
990 numerical groups, and processing duplicated groups should be faster than
991 before.
992
993 20. Two patches to the CMake build system, by Alexander Barkov:
994
995 (1) Replace the "source" command by "." in CMakeLists.txt because
996 "source" is a bash-ism.
997
998 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
999 without these the CMake build does not work on Solaris.
1000
1001 21. Perl has changed its handling of \8 and \9. If there is no previously
1002 encountered capturing group of those numbers, they are treated as the
1003 literal characters 8 and 9 instead of a binary zero followed by the
1004 literals. PCRE now does the same.
1005
1006 22. Following Perl, added \o{} to specify codepoints in octal, making it
1007 possible to specify values greater than 0777 and also making them
1008 unambiguous.
1009
1010 23. Perl now gives an error for missing closing braces after \x{... instead of
1011 treating the string as literal. PCRE now does the same.
1012
1013 24. RunTest used to grumble if an inappropriate test was selected explicitly,
1014 but just skip it when running all tests. This make it awkward to run ranges
1015 of tests when one of them was inappropriate. Now it just skips any
1016 inappropriate tests, as it always did when running all tests.
1017
1018 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
1019 character types such as \d or \w, too many callouts were inserted, and the
1020 data that they returned was rubbish.
1021
1022 26. In UCP mode, \s was not matching two of the characters that Perl matches,
1023 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
1024 were matched by \h. The code has now been refactored so that the lists of
1025 the horizontal and vertical whitespace characters used for \h and \v (which
1026 are defined only in one place) are now also used for \s.
1027
1028 27. Add JIT support for the 64 bit TileGX architecture.
1029 Patch by Jiong Wang (Tilera Corporation).
1030
1031 28. Possessive quantifiers for classes (both explicit and automatically
1032 generated) now use special opcodes instead of wrapping in ONCE brackets.
1033
1034 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
1035 (because it's meaningless), this was not happening when PCRE_CASELESS was
1036 set. Not wrong, but inefficient.
1037
1038 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
1039 properties for \w, \d, etc) is present in a test regex. Otherwise if the
1040 test contains no characters greater than 255, Perl doesn't realise it
1041 should be using Unicode semantics.
1042
1043 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
1044 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
1045 does in Unicode mode.
1046
1047 32. Added the "forbid" facility to pcretest so that putting tests into the
1048 wrong test files can sometimes be quickly detected.
1049
1050 33. There is now a limit (default 250) on the depth of nesting of parentheses.
1051 This limit is imposed to control the amount of system stack used at compile
1052 time. It can be changed at build time by --with-parens-nest-limit=xxx or
1053 the equivalent in CMake.
1054
1055 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
1056 errors. Perl warns for these when in warning mode, but PCRE has no facility
1057 for giving warnings.
1058
1059 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
1060 However, this was not working for (?!) because it is optimized to (*FAIL),
1061 for which PCRE does not allow quantifiers. The optimization is now disabled
1062 when a quantifier follows (?!). I can't see any use for this, but it makes
1063 things uniform.
1064
1065 36. Perl no longer allows group names to start with digits, so I have made this
1066 change also in PCRE. It simplifies the code a bit.
1067
1068 37. In extended mode, Perl ignores spaces before a + that indicates a
1069 possessive quantifier. PCRE allowed a space before the quantifier, but not
1070 before the possessive +. It now does.
1071
1072 38. The use of \K (reset reported match start) within a repeated possessive
1073 group such as (a\Kb)*+ was not working.
1074
1075 40. Document that the same character tables must be used at compile time and
1076 run time, and that the facility to pass tables to pcre_exec() and
1077 pcre_dfa_exec() is for use only with saved/restored patterns.
1078
1079 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
1080 features for Builds with MSVC:
1081
1082 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
1083 builds). The .rc files can be used to set FileDescription and many other
1084 attributes.
1085
1086 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
1087 This allows higher-level build scripts which want .pdb files to avoid
1088 hard-coding the exact files needed."
1089
1090 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
1091 mean "start of word" and "end of word", respectively, as a transition aid.
1092
1093 43. A minimizing repeat of a class containing codepoints greater than 255 in
1094 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
1095 compiled to use the heap for recursion.
1096
1097 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
1098 is configured.
1099
1100
1101 Version 8.33 28-May-2013
1102 ------------------------
1103
1104 1. Added 'U' to some constants that are compared to unsigned integers, to
1105 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
1106 variables that are added to signed variables, to ensure the result is
1107 signed and can be negated.
1108
1109 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
1110 CMake config files.
1111
1112 3. Revise the creation of config.h.generic so that all boolean macros are
1113 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
1114 overriding via -D on the command line possible.
1115
1116 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
1117 to unsigned int is reported to make a quite noticeable speed difference in
1118 a specific Windows environment. Testing on Linux did also appear to show
1119 some benefit (and it is clearly not harmful). Also fixed the definition of
1120 Xop which should be unsigned.
1121
1122 5. Related to (4), changing the definition of the intermediate variable cc
1123 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
1124 improvements.
1125
1126 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
1127 unnecessary spaces.
1128
1129 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
1130 and later.
1131
1132 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
1133
1134 9. Optimizing fast_forward_start_bits in JIT.
1135
1136 10. Adding support for callouts in JIT, and fixing some issues revealed
1137 during this work. Namely:
1138
1139 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
1140
1141 (b) Minimum length was not checked before the matching is started.
1142
1143 11. The value of capture_last that is passed to callouts was incorrect in some
1144 cases when there was a capture on one path that was subsequently abandoned
1145 after a backtrack. Also, the capture_last value is now reset after a
1146 recursion, since all captures are also reset in this case.
1147
1148 12. The interpreter no longer returns the "too many substrings" error in the
1149 case when an overflowing capture is in a branch that is subsequently
1150 abandoned after a backtrack.
1151
1152 13. In the pathological case when an offset vector of size 2 is used, pcretest
1153 now prints out the matched string after a yield of 0 or 1.
1154
1155 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
1156 Only supported by the JIT compiler at the moment.
1157
1158 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
1159
1160 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
1161 offset of the starting point of the matching process, provided the offsets
1162 vector is large enough.
1163
1164 17. The \A escape now records a lookbehind value of 1, though its execution
1165 does not actually inspect the previous character. This is to ensure that,
1166 in partial multi-segment matching, at least one character from the old
1167 segment is retained when a new segment is processed. Otherwise, if there
1168 are no lookbehinds in the pattern, \A might match incorrectly at the start
1169 of a new segment.
1170
1171 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
1172
1173 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
1174 gives some modest performance improvement in 8-bit mode.
1175
1176 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
1177 be expressed in certain programming languages using Universal Character
1178 Names.
1179
1180 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
1181 which points out that "non characters" are not "characters that may not
1182 appear in Unicode strings" but rather "characters that are reserved for
1183 internal use and have only local meaning".
1184
1185 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
1186 there was a conditional group that depended on an assertion, if the
1187 assertion was false, the callout that immediately followed the alternation
1188 in the condition was skipped when pcre_exec() was used for matching.
1189
1190 23. Allow an explicit callout to be inserted before an assertion that is the
1191 condition for a conditional group, for compatibility with automatic
1192 callouts, which always insert a callout at this point.
1193
1194 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1195 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1196
1197 25. (*PRUNE) is now supported by the JIT compiler.
1198
1199 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1200
1201 27. Fix the case where there are two or more SKIPs with arguments that may be
1202 ignored.
1203
1204 28. (*SKIP) is now supported by the JIT compiler.
1205
1206 29. (*THEN) is now supported by the JIT compiler.
1207
1208 30. Update RunTest with additional test selector options.
1209
1210 31. The way PCRE handles backtracking verbs has been changed in two ways.
1211
1212 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1213 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1214 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1215 rather obscure rules do not always do the same thing.
1216
1217 (2) Previously, backtracking verbs were confined within assertions. This is
1218 no longer the case for positive assertions, except for (*ACCEPT). Again,
1219 this sometimes improves Perl compatibility, and sometimes does not.
1220
1221 32. A number of tests that were in test 2 because Perl did things differently
1222 have been moved to test 1, because either Perl or PCRE has changed, and
1223 these tests are now compatible.
1224
1225 32. Backtracking control verbs are now handled in the same way in JIT and
1226 interpreter.
1227
1228 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1229 contained a forward subroutine reference caused a compile error.
1230
1231 34. Auto-detect and optimize limited repetitions in JIT.
1232
1233 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1234 blocking (*UTF) etc.
1235
1236 36. In the interpreter, maximizing pattern repetitions for characters and
1237 character types now use tail recursion, which reduces stack usage.
1238
1239 37. The value of the max lookbehind was not correctly preserved if a compiled
1240 and saved regex was reloaded on a host of different endianness.
1241
1242 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1243 of the compiled pattern block, expand the flags field from 16 to 32 bits
1244 because it was almost full.
1245
1246 39. Try madvise first before posix_madvise.
1247
1248 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1249 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1250 original change locked out all matches of empty strings. This has been
1251 changed so that one match of an empty string per line is recognized.
1252 Subsequent searches on the same line (for colouring or for --only-matching,
1253 for example) do not recognize empty strings.
1254
1255 41. Applied a user patch to fix a number of spelling mistakes in comments.
1256
1257 42. Data lines longer than 65536 caused pcretest to crash.
1258
1259 43. Clarified the data type for length and startoffset arguments for pcre_exec
1260 and pcre_dfa_exec in the function-specific man pages, where they were
1261 explicitly stated to be in bytes, never having been updated. I also added
1262 some clarification to the pcreapi man page.
1263
1264 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1265 a segmentation fault.
1266
1267
1268 Version 8.32 30-November-2012
1269 -----------------------------
1270
1271 1. Improved JIT compiler optimizations for first character search and single
1272 character iterators.
1273
1274 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1275 Patch by Daniel Richard G.
1276
1277 3. Single character iterator optimizations in the JIT compiler.
1278
1279 4. Improved JIT compiler optimizations for character ranges.
1280
1281 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1282 Reported by Giuseppe D'Angelo.
1283
1284 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1285 of a line, was being set incorrectly in cases where .* appeared inside
1286 atomic brackets at the start of a pattern, or where there was a subsequent
1287 *PRUNE or *SKIP.
1288
1289 7. Improved instruction cache flush for POWER/PowerPC.
1290 Patch by Daniel Richard G.
1291
1292 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1293 grep:
1294
1295 (a) There is now no limit to the number of patterns to be matched.
1296
1297 (b) An error is given if a pattern is too long.
1298
1299 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1300 are now supported.
1301
1302 (d) --exclude-from and --include-from (multiple use) have been added.
1303
1304 (e) Exclusions and inclusions now apply to all files and directories, not
1305 just to those obtained from scanning a directory recursively.
1306
1307 (f) Multiple uses of -f and --file-list are now supported.
1308
1309 (g) In a Windows environment, the default for -d has been changed from
1310 "read" (the GNU grep default) to "skip", because otherwise the presence
1311 of a directory in the file list provokes an error.
1312
1313 (h) The documentation has been revised and clarified in places.
1314
1315 9. Improve the matching speed of capturing brackets.
1316
1317 10. Changed the meaning of \X so that it now matches a Unicode extended
1318 grapheme cluster.
1319
1320 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1321 out POSIX threads when JIT support is configured.
1322
1323 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1324
1325 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1326 structure could go wrong in environments where size_t is not the same size
1327 as int.
1328
1329 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1330 set.
1331
1332 15. The EBCDIC support had decayed; later updates to the code had included
1333 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1334 general tidy up of EBCDIC-related issues, and the documentation was also
1335 not quite right. There is now a test that can be run on ASCII systems to
1336 check some of the EBCDIC-related things (but is it not a full test).
1337
1338 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1339 in a small tidy to the code.
1340
1341 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1342
1343 18. If the --only-matching (-o) option in pcregrep is specified multiple
1344 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1345 substrings matched by the 1st and 2nd capturing parentheses. A separating
1346 string can be specified by --om-separator (default empty).
1347
1348 19. Improving the first n character searches.
1349
1350 20. Turn case lists for horizontal and vertical white space into macros so that
1351 they are defined only once.
1352
1353 21. This set of changes together give more compatible Unicode case-folding
1354 behaviour for characters that have more than one other case when UCP
1355 support is available.
1356
1357 (a) The Unicode property table now has offsets into a new table of sets of
1358 three or more characters that are case-equivalent. The MultiStage2.py
1359 script that generates these tables (the pcre_ucd.c file) now scans
1360 CaseFolding.txt instead of UnicodeData.txt for character case
1361 information.
1362
1363 (b) The code for adding characters or ranges of characters to a character
1364 class has been abstracted into a generalized function that also handles
1365 case-independence. In UTF-mode with UCP support, this uses the new data
1366 to handle characters with more than one other case.
1367
1368 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1369 whose other case is greater than 256 are now correctly matched
1370 caselessly. Previously, the high codepoint matched the low one, but not
1371 vice versa.
1372
1373 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1374 of the new class addition function, using character lists defined as
1375 macros alongside the case definitions of 20 above.
1376
1377 (e) Caseless back references now work with characters that have more than
1378 one other case.
1379
1380 (f) General caseless matching of characters with more than one other case
1381 is supported.
1382
1383 22. Unicode character properties were updated from Unicode 6.2.0
1384
1385 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1386
1387 24. Add support for 32-bit character strings, and UTF-32
1388
1389 25. Major JIT compiler update (code refactoring and bugfixing).
1390 Experimental Sparc 32 support is added.
1391
1392 26. Applied a modified version of Daniel Richard G's patch to create
1393 pcre.h.generic and config.h.generic by "make" instead of in the
1394 PrepareRelease script.
1395
1396 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1397 pcre_compile.c when checking for a zero character.
1398
1399 28. Introducing a native interface for JIT. Through this interface, the compiled
1400 machine code can be directly executed. The purpose of this interface is to
1401 provide fast pattern matching, so several sanity checks are not performed.
1402 However, feature tests are still performed. The new interface provides
1403 1.4x speedup compared to the old one.
1404
1405 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1406 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1407 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1408
1409 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1410 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1411 the "old" RFC 2279). Instead, it ended up passing a negative length to
1412 pcre_exec().
1413
1414 31. Add support for GCC's visibility feature to hide internal functions.
1415
1416 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1417 "unknown -C option" after outputting 0 or 1.
1418
1419 33. There is now support for generating a code coverage report for the test
1420 suite in environments where gcc is the compiler and lcov is installed. This
1421 is mainly for the benefit of the developers.
1422
1423 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1424 unaddressable using valgrind annotations, allowing valgrind to detect
1425 invalid memory accesses. This is mainly for the benefit of the developers.
1426
1427 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1428
1429 26. Give configure error if --enable-cpp but no C++ compiler found.
1430
1431
1432 Version 8.31 06-July-2012
1433 -------------------------
1434
1435 1. Fixing a wrong JIT test case and some compiler warnings.
1436
1437 2. Removed a bashism from the RunTest script.
1438
1439 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1440 to unsigned type, result still unsigned" that was given by an MS compiler
1441 on encountering the code "-sizeof(xxx)".
1442
1443 4. Partial matching support is added to the JIT compiler.
1444
1445 5. Fixed several bugs concerned with partial matching of items that consist
1446 of more than one character:
1447
1448 (a) /^(..)\1/ did not partially match "aba" because checking references was
1449 done on an "all or nothing" basis. This also applied to repeated
1450 references.
1451
1452 (b) \R did not give a hard partial match if \r was found at the end of the
1453 subject.
1454
1455 (c) \X did not give a hard partial match after matching one or more
1456 characters at the end of the subject.
1457
1458 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1459 a partial match for the string "\r".
1460
1461 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1462 a partial match for a CR character at the end of the subject string.
1463
1464 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1465 running pcretest, the text "(JIT)" added to the output whenever JIT is
1466 actually used to run the match.
1467
1468 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1469 or /S+[+] with a digit between 1 and 7.
1470
1471 8. OP_NOT now supports any UTF character not just single-byte ones.
1472
1473 9. (*MARK) control verb is now supported by the JIT compiler.
1474
1475 10. The command "./RunTest list" lists the available tests without actually
1476 running any of them. (Because I keep forgetting what they all are.)
1477
1478 11. Add PCRE_INFO_MAXLOOKBEHIND.
1479
1480 12. Applied a (slightly modified) user-supplied patch that improves performance
1481 when the heap is used for recursion (compiled with --disable-stack-for-
1482 recursion). Instead of malloc and free for each heap frame each time a
1483 logical recursion happens, frames are retained on a chain and re-used where
1484 possible. This sometimes gives as much as 30% improvement.
1485
1486 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1487 call.
1488
1489 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1490
1491 15. It is now possible to link pcretest with libedit as an alternative to
1492 libreadline.
1493
1494 16. (*COMMIT) control verb is now supported by the JIT compiler.
1495
1496 17. The Unicode data tables have been updated to Unicode 6.1.0.
1497
1498 18. Added --file-list option to pcregrep.
1499
1500 19. Added binary file support to pcregrep, including the -a, --binary-files,
1501 -I, and --text options.
1502
1503 20. The madvise function is renamed for posix_madvise for QNX compatibility
1504 reasons. Fixed by Giuseppe D'Angelo.
1505
1506 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1507
1508 22. Changed the option for creating symbolic links for 16-bit man pages from
1509 -s to -sf so that re-installing does not cause issues.
1510
1511 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1512
1513 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1514 matches in certain environments (the workspace was not being correctly
1515 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1516 some of the workspace data at the beginning of a restart.
1517
1518 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1519 was not doing so when it should - probably a typo introduced by SVN 528
1520 (change 8.10/14).
1521
1522 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1523 \w+ when the character tables indicated that \x{c4} was a word character.
1524 There were several related cases, all because the tests for doing a table
1525 lookup were testing for characters less than 127 instead of 255.
1526
1527 27. If a pattern contains capturing parentheses that are not used in a match,
1528 their slots in the ovector are set to -1. For those that are higher than
1529 any matched groups, this happens at the end of processing. In the case when
1530 there were back references that the ovector was too small to contain
1531 (causing temporary malloc'd memory to be used during matching), and the
1532 highest capturing number was not used, memory off the end of the ovector
1533 was incorrectly being set to -1. (It was using the size of the temporary
1534 memory instead of the true size.)
1535
1536 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1537 ovector size, it uses memory at the end of the block that it has got.
1538
1539 29. Check for an overlong MARK name and give an error at compile time. The
1540 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1541
1542 30. JIT compiler update.
1543
1544 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1545 Rill for the patch.
1546
1547 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1548
1549 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1550
1551 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1552 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1553 was enabled.
1554
1555 35. Improve JIT code generation for greedy plus quantifier.
1556
1557 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1558 1 to "aa" instead of to an empty string. The bug affected repeated groups
1559 that could potentially match an empty string.
1560
1561 37. Optimizing single character iterators in JIT.
1562
1563 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1564 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1565 codepoints that are too big for the mode are faulted, and in a UTF mode,
1566 disallowed codepoints are also faulted.
1567
1568 39. If PCRE was compiled with UTF support, in three places in the DFA
1569 matcher there was code that should only have been obeyed in UTF mode, but
1570 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1571 processing when bytes with values greater than 127 were present. In 16-bit
1572 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1573 both cases the values are those that cannot be the first data item in a UTF
1574 character. The three items that might have provoked this were recursions,
1575 possessively repeated groups, and atomic groups.
1576
1577 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1578 and pcregrep, because some OS require shared objects to be explicitly
1579 passed to ld, causing the link step to fail if they are not.
1580
1581 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1582 mode, patterns that started with \h* or \R* might be incorrectly matched.
1583
1584
1585 Version 8.30 04-February-2012
1586 -----------------------------
1587
1588 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1589 name is defined in ctype.h.
1590
1591 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1592 only in quite long subpatterns.
1593
1594 3. Removed the function pcre_info(), which has been obsolete and deprecated
1595 since it was replaced by pcre_fullinfo() in February 2000.
1596
1597 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1598 match a (*MARK), and the match failed at the start of the subject, a
1599 reference to memory before the start of the subject could occur. This bug
1600 was introduced by fix 17 of release 8.21.
1601
1602 5. A reference to an unset group with zero minimum repetition was giving
1603 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1604 /(another)?(\1?)test/ matched against "hello world test". This bug was
1605 introduced in release 8.13.
1606
1607 6. Add support for 16-bit character strings (a large amount of work involving
1608 many changes and refactorings).
1609
1610 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1611 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1612 from a file.
1613
1614 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1615 rounding is not applied in this particular case).
1616
1617 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1618 if they appear, or are escaped, in patterns.
1619
1620 10. Get rid of a number of -Wunused-but-set-variable warnings.
1621
1622 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1623 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1624 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1625 also returns the mark "x". This bug applied to capturing parentheses,
1626 non-capturing parentheses, and atomic parentheses. It also applied to some
1627 assertions.
1628
1629 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1630 information out of configure.ac instead of relying on pcre.h.generic, which
1631 is not stored in the repository.
1632
1633 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1634 -lreadline.
1635
1636 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1637
1638 15. Applied Graycode's patch to put the top-level frame on the stack rather
1639 than the heap when not using the stack for recursion. This gives a
1640 performance improvement in many cases when recursion is not deep.
1641
1642 16. Experimental code added to "pcretest -C" to output the stack frame size.
1643
1644
1645 Version 8.21 12-Dec-2011
1646 ------------------------
1647
1648 1. Updating the JIT compiler.
1649
1650 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1651 are added as well.
1652
1653 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1654 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1655 calling _pcre_jit_exec. Some extra comments are added.
1656
1657 4. (*MARK) settings inside atomic groups that do not contain any capturing
1658 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1659 was introduced by change 18 for 8.20.
1660
1661 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1662 ECMA-262 standard.
1663
1664 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1665 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1666 This bug was probably introduced by change 9 of 8.13.
1667
1668 7. While fixing 6 above, I noticed that a number of other items were being
1669 incorrectly rejected as "not fixed length". This arose partly because newer
1670 opcodes had not been added to the fixed-length checking code. I have (a)
1671 corrected the bug and added tests for these items, and (b) arranged for an
1672 error to occur if an unknown opcode is encountered while checking for fixed
1673 length instead of just assuming "not fixed length". The items that were
1674 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1675 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1676 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1677
1678 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1679 being incorrectly compiled and would have given unpredicatble results.
1680
1681 9. A possessively repeated subpattern with minimum repeat count greater than
1682 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1683 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1684 the first (A) could occur when it should not.
1685
1686 10. Add a cast and remove a redundant test from the code.
1687
1688 11. JIT should use pcre_malloc/pcre_free for allocation.
1689
1690 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1691 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1692 is anything other than /usr, -L is still shown).
1693
1694 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1695
1696 14. Perl does not support \N without a following name in a [] class; PCRE now
1697 also gives an error.
1698
1699 15. If a forward reference was repeated with an upper limit of around 2000,
1700 it caused the error "internal error: overran compiling workspace". The
1701 maximum number of forward references (including repeats) was limited by the
1702 internal workspace, and dependent on the LINK_SIZE. The code has been
1703 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1704 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1705 of around 200,000 forward references. While doing this, I also speeded up
1706 the filling in of repeated forward references.
1707
1708 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1709 incorrectly expecting the subject to contain another "a" after the start.
1710
1711 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1712 in the match, the SKIP should be ignored. This was not happening; instead
1713 the SKIP was being treated as NOMATCH. For patterns such as
1714 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1715 tested.
1716
1717 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1718 now much more compatible with Perl, in particular in cases where the result
1719 is a non-match for a non-anchored pattern. For example, if
1720 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1721 "m", where previously it did not return a name. A side effect of this
1722 change is that for partial matches, the last encountered mark name is
1723 returned, as for non matches. A number of tests that were previously not
1724 Perl-compatible have been moved into the Perl-compatible test files. The
1725 refactoring has had the pleasing side effect of removing one argument from
1726 the match() function, thus reducing its stack requirements.
1727
1728 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1729 subsequent uses of /S (without +) incorrectly behaved like /S+.
1730
1731 21. Retrieve executable code size support for the JIT compiler and fixing
1732 some warnings.
1733
1734 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1735 not work when the shorter character appeared right at the end of the
1736 subject string.
1737
1738 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1739 systems.
1740
1741 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1742 output it when the /M option is used in pcretest.
1743
1744 25. The CheckMan script was not being included in the distribution. Also, added
1745 an explicit "perl" to run Perl scripts from the PrepareRelease script
1746 because this is reportedly needed in Windows.
1747
1748 26. If study data was being save in a file and studying had not found a set of
1749 "starts with" bytes for the pattern, the data written to the file (though
1750 never used) was taken from uninitialized memory and so caused valgrind to
1751 complain.
1752
1753 27. Updated RunTest.bat as provided by Sheri Pierce.
1754
1755 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1756
1757 29. Computation of memory usage for the table of capturing group names was
1758 giving an unnecessarily large value.
1759
1760
1761 Version 8.20 21-Oct-2011
1762 ------------------------
1763
1764 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1765 a POSIX class. After further experiments with Perl, which convinced me that
1766 Perl has bugs and confusions, a closing square bracket is no longer allowed
1767 in a POSIX name. This bug also affected patterns with classes that started
1768 with full stops.
1769
1770 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1771 captured substring, but while checking the failing first alternative,
1772 substring 1 is temporarily captured. If the output vector supplied to
1773 pcre_exec() was not big enough for this capture, the yield of the function
1774 was still zero ("insufficient space for captured substrings"). This cannot
1775 be totally fixed without adding another stack variable, which seems a lot
1776 of expense for a edge case. However, I have improved the situation in cases
1777 such as /(a)(b)x|abc/ matched against "abc", where the return code
1778 indicates that fewer than the maximum number of slots in the ovector have
1779 been set.
1780
1781 3. Related to (2) above: when there are more back references in a pattern than
1782 slots in the output vector, pcre_exec() uses temporary memory during
1783 matching, and copies in the captures as far as possible afterwards. It was
1784 using the entire output vector, but this conflicts with the specification
1785 that only 2/3 is used for passing back captured substrings. Now it uses
1786 only the first 2/3, for compatibility. This is, of course, another edge
1787 case.
1788
1789 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1790 main code base, and can be used by building with --enable-jit. When this is
1791 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1792 runtime --no-jit option is given.
1793
1794 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1795 ovector, the return from the function was zero, implying that there were
1796 other matches that did not fit. The correct "exactly full" value is now
1797 returned.
1798
1799 6. If a subpattern that was called recursively or as a subroutine contained
1800 (*PRUNE) or any other control that caused it to give a non-standard return,
1801 invalid errors such as "Error -26 (nested recursion at the same subject
1802 position)" or even infinite loops could occur.
1803
1804 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1805 computing the minimum length on reaching *ACCEPT, and so ended up with the
1806 wrong value of 1 rather than 0. Further investigation indicates that
1807 computing a minimum subject length in the presence of *ACCEPT is difficult
1808 (think back references, subroutine calls), and so I have changed the code
1809 so that no minimum is registered for a pattern that contains *ACCEPT.
1810
1811 8. If (*THEN) was present in the first (true) branch of a conditional group,
1812 it was not handled as intended. [But see 16 below.]
1813
1814 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1815 Sheri Pierce.
1816
1817 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1818 the first byte in a match must be "a".
1819
1820 11. Change 17 for 8.13 increased the recursion depth for patterns like
1821 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1822 pattern contains any instances of (*THEN). If it does not, the old
1823 optimizations are restored. It would be nice to do this on a per-group
1824 basis, but at the moment that is not feasible.
1825
1826 12. In some environments, the output of pcretest -C is CRLF terminated. This
1827 broke RunTest's code that checks for the link size. A single white space
1828 character after the value is now allowed for.
1829
1830 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1831 For "fr", it uses the Windows-specific input and output files.
1832
1833 14. If (*THEN) appeared in a group that was called recursively or as a
1834 subroutine, it did not work as intended. [But see next item.]
1835
1836 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1837 pattern fragments (but not containing any | characters). If A and B are
1838 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1839 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1840 D. In other words, Perl considers parentheses that do not contain any |
1841 characters to be part of a surrounding alternative, whereas PCRE was
1842 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1843 differently. PCRE now behaves in the same way as Perl, except in the case
1844 of subroutine/recursion calls such as (?1) which have in any case always
1845 been different (but PCRE had them first :-).
1846
1847 16. Related to 15 above: Perl does not treat the | in a conditional group as
1848 creating alternatives. Such a group is treated in the same way as an
1849 ordinary group without any | characters when processing (*THEN). PCRE has
1850 been changed to match Perl's behaviour.
1851
1852 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1853 RunGrepTest script failed.
1854
1855 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1856 inevitable for groups that contain captures, but it can lead to a lot of
1857 stack use in large patterns. The old behaviour has been restored for atomic
1858 groups that do not contain any capturing parentheses.
1859
1860 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1861 suppress the check for a minimum subject length at run time. (If it was
1862 given to pcre_exec() or pcre_dfa_exec() it did work.)
1863
1864 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1865 fail to work when decoding hex characters in data strings in EBCDIC
1866 environments.
1867
1868 21. It appears that in at least one Mac OS environment, the isxdigit() function
1869 is implemented as a macro that evaluates to its argument more than once,
1870 contravening the C 90 Standard (I haven't checked a later standard). There
1871 was an instance in pcretest which caused it to go wrong when processing
1872 \x{...} escapes in subject strings. The has been rewritten to avoid using
1873 things like p++ in the argument of isxdigit().
1874
1875
1876 Version 8.13 16-Aug-2011
1877 ------------------------
1878
1879 1. The Unicode data tables have been updated to Unicode 6.0.0.
1880
1881 2. Two minor typos in pcre_internal.h have been fixed.
1882
1883 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1884 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1885 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1886
1887 4. There were a number of related bugs in the code for matching backrefences
1888 caselessly in UTF-8 mode when codes for the characters concerned were
1889 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1890 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1891 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1892 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1893 2-byte code at the end of the subject (it thought there wasn't enough data
1894 left).
1895
1896 5. Comprehensive information about what went wrong is now returned by
1897 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1898 as the output vector has at least 2 elements. The offset of the start of
1899 the failing character and a reason code are placed in the vector.
1900
1901 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1902 now returned is for the first byte of the failing character, instead of the
1903 last byte inspected. This is an incompatible change, but I hope it is small
1904 enough not to be a problem. It makes the returned offset consistent with
1905 pcre_exec() and pcre_dfa_exec().
1906
1907 7. pcretest now gives a text phrase as well as the error number when
1908 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1909 failure, the offset and reason code are output.
1910
1911 8. When \R was used with a maximizing quantifier it failed to skip backwards
1912 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1913 back over a single character (\n). This seems wrong (because it treated the
1914 two characters as a single entity when going forwards), conflicts with the
1915 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1916 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1917 has been changed.
1918
1919 9. Some internal refactoring has changed the processing so that the handling
1920 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1921 time (the PCRE_DOTALL option was changed this way some time ago: version
1922 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1923 which was always a bit of a fudge. It also means that there is one less
1924 argument for the match() function, which reduces its stack requirements
1925 slightly. This change also fixes an incompatibility with Perl: the pattern
1926 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1927
1928 10. More internal refactoring has drastically reduced the number of recursive
1929 calls to match() for possessively repeated groups such as (abc)++ when
1930 using pcre_exec().
1931
1932 11. While implementing 10, a number of bugs in the handling of groups were
1933 discovered and fixed:
1934
1935 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1936 (a|)*(?1) gave a compile-time internal error.
1937 ((a|)+)+ did not notice that the outer group could match an empty string.
1938 (^a|^)+ was not marked as anchored.
1939 (.*a|.*)+ was not marked as matching at start or after a newline.
1940
1941 12. Yet more internal refactoring has removed another argument from the match()
1942 function. Special calls to this function are now indicated by setting a
1943 value in a variable in the "match data" data block.
1944
1945 13. Be more explicit in pcre_study() instead of relying on "default" for
1946 opcodes that mean there is no starting character; this means that when new
1947 ones are added and accidentally left out of pcre_study(), testing should
1948 pick them up.
1949
1950 14. The -s option of pcretest has been documented for ages as being an old
1951 synonym of -m (show memory usage). I have changed it to mean "force study
1952 for every regex", that is, assume /S for every regex. This is similar to -i
1953 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1954 using it. It makes it easier to run collections of tests with and without
1955 study enabled, and thereby test pcre_study() more easily. All the standard
1956 tests are now run with and without -s (but some patterns can be marked as
1957 "never study" - see 20 below).
1958
1959 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1960 restoration of the capturing data to the outer values was not happening
1961 correctly.
1962
1963 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1964 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1965 pattern had matched an empty string, and so incorrectly returned a no
1966 match.
1967
1968 17. There was optimizing code for the last branch of non-capturing parentheses,
1969 and also for the obeyed branch of a conditional subexpression, which used
1970 tail recursion to cut down on stack usage. Unfortunately, now that there is
1971 the possibility of (*THEN) occurring in these branches, tail recursion is
1972 no longer possible because the return has to be checked for (*THEN). These
1973 two optimizations have therefore been removed. [But see 8.20/11 above.]
1974
1975 18. If a pattern containing \R was studied, it was assumed that \R always
1976 matched two bytes, thus causing the minimum subject length to be
1977 incorrectly computed because \R can also match just one byte.
1978
1979 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1980 was incorrectly computed.
1981
1982 20. If /S is present twice on a test pattern in pcretest input, it now
1983 *disables* studying, thereby overriding the use of -s on the command line
1984 (see 14 above). This is necessary for one or two tests to keep the output
1985 identical in both cases.
1986
1987 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1988 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1989
1990 22. When an atomic group that contained a capturing parenthesis was
1991 successfully matched, but the branch in which it appeared failed, the
1992 capturing was not being forgotten if a higher numbered group was later
1993 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1994 group 1 to "a", when in fact it should be unset. This applied to multi-
1995 branched capturing and non-capturing groups, repeated or not, and also to
1996 positive assertions (capturing in negative assertions does not happen
1997 in PCRE) and also to nested atomic groups.
1998
1999 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
2000 subject after a captured substring, to make it easier to tell which of a
2001 number of identical substrings has been captured.
2002
2003 24. The way atomic groups are processed by pcre_exec() has been changed so that
2004 if they are repeated, backtracking one repetition now resets captured
2005 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
2006 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
2007 "aaa". Previously, it would have been "a". As part of this code
2008 refactoring, the way recursive calls are handled has also been changed.
2009
2010 25. If an assertion condition captured any substrings, they were not passed
2011 back unless some other capturing happened later. For example, if
2012 (?(?=(a))a) was matched against "a", no capturing was returned.
2013
2014 26. When studying a pattern that contained subroutine calls or assertions,
2015 the code for finding the minimum length of a possible match was handling
2016 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
2017 group 1 called group 2 while simultaneously a separate group 2 called group
2018 1). A stack overflow occurred in this case. I have fixed this by limiting
2019 the recursion depth to 10.
2020
2021 27. Updated RunTest.bat in the distribution to the version supplied by Tom
2022 Fortmann. This supports explicit test numbers on the command line, and has
2023 argument validation and error reporting.
2024
2025 28. An instance of \X with an unlimited repeat could fail if at any point the
2026 first character it looked at was a mark character.
2027
2028 29. Some minor code refactoring concerning Unicode properties and scripts
2029 should reduce the stack requirement of match() slightly.
2030
2031 30. Added the '=' option to pcretest to check the setting of unused capturing
2032 slots at the end of the pattern, which are documented as being -1, but are
2033 not included in the return count.
2034
2035 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
2036 compiled something random. Now it gives a compile-time error (as does
2037 Perl).
2038
2039 32. A *MARK encountered during the processing of a positive assertion is now
2040 recorded and passed back (compatible with Perl).
2041
2042 33. If --only-matching or --colour was set on a pcregrep call whose pattern
2043 had alternative anchored branches, the search for a second match in a line
2044 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
2045 matched the line "0102" twice. The same bug affected patterns that started
2046 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
2047 twice.
2048
2049 34. Previously, PCRE did not allow quantification of assertions. However, Perl
2050 does, and because of capturing effects, quantifying parenthesized
2051 assertions may at times be useful. Quantifiers are now allowed for
2052 parenthesized assertions.
2053
2054 35. A minor code tidy in pcre_compile() when checking options for \R usage.
2055
2056 36. \g was being checked for fancy things in a character class, when it should
2057 just be a literal "g".
2058
2059 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
2060 appearance of a nested POSIX class supersedes an apparent external class.
2061 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
2062 unescaped square brackets may also appear as part of class names. For
2063 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
2064 more like Perl. (But see 8.20/1 above.)
2065
2066 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
2067 was because it thought it was \N{name}, which is not supported).
2068
2069 39. Add minix to OS list not supporting the -S option in pcretest.
2070
2071 40. PCRE tries to detect cases of infinite recursion at compile time, but it
2072 cannot analyze patterns in sufficient detail to catch mutual recursions
2073 such as ((?1))((?2)). There is now a runtime test that gives an error if a
2074 subgroup is called recursively as a subpattern for a second time at the
2075 same position in the subject string. In previous releases this might have
2076 been caught by the recursion limit, or it might have run out of stack.
2077
2078 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
2079 happen only once. PCRE was, however incorrectly giving a compile time error
2080 "recursive call could loop indefinitely" because it cannot analyze the
2081 pattern in sufficient detail. The compile time test no longer happens when
2082 PCRE is compiling a conditional subpattern, but actual runaway loops are
2083 now caught at runtime (see 40 above).
2084
2085 42. It seems that Perl allows any characters other than a closing parenthesis
2086 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
2087 has been changed to be the same.
2088
2089 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
2090 as not to get warnings when autogen.sh is called. Also changed
2091 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
2092
2093 44. To help people who use pcregrep to scan files containing exceedingly long
2094 lines, the following changes have been made:
2095
2096 (a) The default value of the buffer size parameter has been increased from
2097 8K to 20K. (The actual buffer used is three times this size.)
2098
2099 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
2100 PCRE is built.
2101
2102 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
2103 to be set at run time.
2104
2105 (d) Numerical values in pcregrep options can be followed by K or M, for
2106 example --buffer-size=50K.
2107
2108 (e) If a line being scanned overflows pcregrep's buffer, an error is now
2109 given and the return code is set to 2.
2110
2111 45. Add a pointer to the latest mark to the callout data block.
2112
2113 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
2114 partial match of an empty string instead of no match. This was specific to
2115 the use of ".".
2116
2117 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
2118 complete match instead of a partial match. This bug was dependent on both
2119 the PCRE_UTF8 and PCRE_DOTALL options being set.
2120
2121 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
2122 starting byte set, because \b was not being ignored.
2123
2124
2125 Version 8.12 15-Jan-2011
2126 ------------------------
2127
2128 1. Fixed some typos in the markup of the man pages, and wrote a script that
2129 checks for such things as part of the documentation building process.
2130
2131 2. On a big-endian 64-bit system, pcregrep did not correctly process the
2132 --match-limit and --recursion-limit options (added for 8.11). In
2133 particular, this made one of the standard tests fail. (The integer value
2134 went into the wrong half of a long int.)
2135
2136 3. If the --colour option was given to pcregrep with -v (invert match), it
2137 did strange things, either producing crazy output, or crashing. It should,
2138 of course, ignore a request for colour when reporting lines that do not
2139 match.
2140
2141 4. Another pcregrep bug caused similar problems if --colour was specified with
2142 -M (multiline) and the pattern match finished with a line ending.
2143
2144 5. In pcregrep, when a pattern that ended with a literal newline sequence was
2145 matched in multiline mode, the following line was shown as part of the
2146 match. This seems wrong, so I have changed it.
2147
2148 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
2149 the check for further matches in the same line (so they could be coloured)
2150 to overrun the end of the current line. If another match was found, it was
2151 incorrectly shown (and then shown again when found in the next line).
2152
2153 7. If pcregrep was compiled under Windows, there was a reference to the
2154 function pcregrep_exit() before it was defined. I am assuming this was
2155 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
2156 reported by a user. I've moved the definition above the reference.
2157
2158
2159 Version 8.11 10-Dec-2010
2160 ------------------------
2161
2162 1. (*THEN) was not working properly if there were untried alternatives prior
2163 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
2164 backtracked to try for "b" instead of moving to the next alternative branch
2165 at the same level (in this case, to look for "c"). The Perl documentation
2166 is clear that when (*THEN) is backtracked onto, it goes to the "next
2167 alternative in the innermost enclosing group".
2168
2169 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
2170 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
2171 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
2172 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
2173 (*THEN).
2174
2175 3. If \s appeared in a character class, it removed the VT character from
2176 the class, even if it had been included by some previous item, for example
2177 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
2178 of \s, but is part of the POSIX "space" class.)
2179
2180 4. A partial match never returns an empty string (because you can always
2181 match an empty string at the end of the subject); however the checking for
2182 an empty string was starting at the "start of match" point. This has been
2183 changed to the "earliest inspected character" point, because the returned
2184 data for a partial match starts at this character. This means that, for
2185 example, /(?<=abc)def/ gives a partial match for the subject "abc"
2186 (previously it gave "no match").
2187
2188 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2189 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2190 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2191 has an implication that the given string is incomplete (because a partial
2192 match is preferred over a full match). For this reason, these items now
2193 give a partial match in this situation. [Aside: previously, the one case
2194 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2195 match rather than a full match, which was wrong by the old rules, but is
2196 now correct.]
2197
2198 6. There was a bug in the handling of #-introduced comments, recognized when
2199 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2200 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2201 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2202 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2203 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2204 places in pcre_compile().
2205
2206 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2207 comments when looking ahead for named forward references to subpatterns,
2208 the only newline sequence it recognized was NL. It now handles newlines
2209 according to the set newline convention.
2210
2211 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2212 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2213 cater for a lack of strerror(). These oversights have been fixed.
2214
2215 9. Added --match-limit and --recursion-limit to pcregrep.
2216
2217 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2218
2219 11. When the -o option was used, pcregrep was setting a return code of 1, even
2220 when matches were found, and --line-buffered was not being honoured.
2221
2222 12. Added an optional parentheses number to the -o and --only-matching options
2223 of pcregrep.
2224
2225 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2226 can match an empty string. The code to do it in pcretest and pcredemo
2227 needed fixing:
2228
2229 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2230 only one byte after an empty string match just before CRLF (this case
2231 just got forgotten; "any" and "anycrlf" were OK).
2232
2233 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2234 mode when an empty string match preceded an ASCII character followed by
2235 a non-ASCII character. (The code for advancing by one character rather
2236 than one byte was nonsense.)
2237
2238 (c) The pcredemo.c sample program did not have any code at all to handle
2239 the cases when CRLF is a valid newline sequence.
2240
2241 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2242 as a starting offset was within the subject string. There is now a new
2243 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2244 negative or greater than the length of the string. In order to test this,
2245 pcretest is extended to allow the setting of negative starting offsets.
2246
2247 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2248 starting offset points to the beginning of a UTF-8 character was
2249 unnecessarily clumsy. I tidied it up.
2250
2251 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2252 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2253
2254 17. Nobody had reported that the --include_dir option, which was added in
2255 release 7.7 should have been called --include-dir (hyphen, not underscore)
2256 for compatibility with GNU grep. I have changed it to --include-dir, but
2257 left --include_dir as an undocumented synonym, and the same for
2258 --exclude-dir, though that is not available in GNU grep, at least as of
2259 release 2.5.4.
2260
2261 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2262 characters from a string of bytes) have been redefined so as not to use
2263 loops, in order to improve performance in some environments. At the same
2264 time, I abstracted some of the common code into auxiliary macros to save
2265 repetition (this should not affect the compiled code).
2266
2267 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2268 compile-time error is now given if \c is not followed by an ASCII
2269 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2270 different, and any byte value is allowed.)
2271
2272 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2273 START_OPTIMIZE option, which is now allowed at compile time - but just
2274 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2275 to pcregrep and other applications that have no direct access to PCRE
2276 options. The new /Y option in pcretest sets this option when calling
2277 pcre_compile().
2278
2279 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2280 back references. Groups containing recursive back references were forced to
2281 be atomic by that change, but in the case of named groups, the amount of
2282 memory required was incorrectly computed, leading to "Failed: internal
2283 error: code overflow". This has been fixed.
2284
2285 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2286 pcretest.c, to avoid build problems in some Borland environments.
2287
2288
2289 Version 8.10 25-Jun-2010
2290 ------------------------
2291
2292 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2293 THEN.
2294
2295 2. (*ACCEPT) was not working when inside an atomic group.
2296
2297 3. Inside a character class, \B is treated as a literal by default, but
2298 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2299 causes the error). The code is unchanged, but I tidied the documentation.
2300
2301 4. Inside a character class, PCRE always treated \R and \X as literals,
2302 whereas Perl faults them if its -w option is set. I have changed PCRE so
2303 that it faults them when PCRE_EXTRA is set.
2304
2305 5. Added support for \N, which always matches any character other than
2306 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2307
2308 6. When compiling pcregrep with newer versions of gcc which may have
2309 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2310 declared with attribute warn_unused_result" were given. Just casting the
2311 result to (void) does not stop the warnings; a more elaborate fudge is
2312 needed. I've used a macro to implement this.
2313
2314 7. Minor change to pcretest.c to avoid a compiler warning.
2315
2316 8. Added four artifical Unicode properties to help with an option to make
2317 \s etc use properties (see next item). The new properties are: Xan
2318 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2319
2320 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2321 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2322 this option. Modified pcretest to add /W to test this facility. Added
2323 REG_UCP to make it available via the POSIX interface.
2324
2325 10. Added --line-buffered to pcregrep.
2326
2327 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2328 studied, and the match started with a letter with a code point greater than
2329 127 whose first byte was different to the first byte of the other case of
2330 the letter, the other case of this starting letter was not recognized
2331 (#976).
2332
2333 12. If a pattern that was studied started with a repeated Unicode property
2334 test, for example, \p{Nd}+, there was the theoretical possibility of
2335 setting up an incorrect bitmap of starting bytes, but fortunately it could
2336 not have actually happened in practice until change 8 above was made (it
2337 added property types that matched character-matching opcodes).
2338
2339 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2340 possible starting bytes for non-anchored patterns.
2341
2342 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2343 \R, and also a number of cases that involve Unicode properties, both
2344 explicit and implicit when PCRE_UCP is set.
2345
2346 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2347 input, it could crash or give wrong results if characters with values
2348 greater than 0xc0 were present in the subject string. (Detail: it assumed
2349 UTF-8 input when processing these items.)
2350
2351 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2352 size_t is 64-bit (#991).
2353
2354 17. Added a check for running out of memory when PCRE is compiled with
2355 --disable-stack-for-recursion (#990).
2356
2357 18. If the last data line in a file for pcretest does not have a newline on
2358 the end, a newline was missing in the output.
2359
2360 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2361 less than 128) in its various bitmaps. However, there is a facility for
2362 generating tables according to the current locale when PCRE is compiled. It
2363 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2364 characters, are recognized by isspace() and therefore were getting set in
2365 these tables, and indeed these tables seem to approximate to ISO 8859. This
2366 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2367 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2368 which of course cannot start UTF-8 characters. I have changed the code so
2369 that only real ASCII characters (less than 128) and the correct starting
2370 bytes for UTF-8 encodings are set for characters greater than 127 when in
2371 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2372 altogether.)
2373
2374 20. Added the /T option to pcretest so as to be able to run tests with non-
2375 standard character tables, thus making it possible to include the tests
2376 used for 19 above in the standard set of tests.
2377
2378 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2379 reference to a subpattern the other side of a comment that contains an
2380 opening parenthesis caused either an internal compiling error, or a
2381 reference to the wrong subpattern.
2382
2383
2384 Version 8.02 19-Mar-2010
2385 ------------------------
2386
2387 1. The Unicode data tables have been updated to Unicode 5.2.0.
2388
2389 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2390 configured.
2391
2392 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2393 original author of that file, following a query about its status.
2394
2395 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2396 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2397
2398 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2399 quantifier applied to a forward-referencing subroutine call, could compile
2400 incorrect code or give the error "internal error: previously-checked
2401 referenced subpattern not found".
2402
2403 6. Both MS Visual Studio and Symbian OS have problems with initializing
2404 variables to point to external functions. For these systems, therefore,
2405 pcre_malloc etc. are now initialized to local functions that call the
2406 relevant global functions.
2407
2408 7. There were two entries missing in the vectors called coptable and poptable
2409 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2410 I've fixed the data, and added a kludgy way of testing at compile time that
2411 the lengths are correct (equal to the number of opcodes).
2412
2413 8. Following on from 7, I added a similar kludge to check the length of the
2414 eint vector in pcreposix.c.
2415
2416 9. Error texts for pcre_compile() are held as one long string to avoid too
2417 much relocation at load time. To find a text, the string is searched,
2418 counting zeros. There was no check for running off the end of the string,
2419 which could happen if a new error number was added without updating the
2420 string.
2421
2422 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2423
2424 11. \K was not working if it appeared in an atomic group or in a group that
2425 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2426 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2427 the assertion is positive, but not if it is negative.
2428
2429 12. Change 11 fortuitously reduced the size of the stack frame used in the
2430 "match()" function of pcre_exec.c by one pointer. Forthcoming
2431 implementation of support for (*MARK) will need an extra pointer on the
2432 stack; I have reserved it now, so that the stack frame size does not
2433 decrease.
2434
2435 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2436 item in branch that calls a recursion is a subroutine call - as in the
2437 second branch in the above example - was incorrectly given the compile-
2438 time error "recursive call could loop indefinitely" because pcre_compile()
2439 was not correctly checking the subroutine for matching a non-empty string.
2440
2441 14. The checks for overrunning compiling workspace could trigger after an
2442 overrun had occurred. This is a "should never occur" error, but it can be
2443 triggered by pathological patterns such as hundreds of nested parentheses.
2444 The checks now trigger 100 bytes before the end of the workspace.
2445
2446 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2447
2448
2449 Version 8.01 19-Jan-2010
2450 ------------------------
2451
2452 1. If a pattern contained a conditional subpattern with only one branch (in
2453 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2454 computed the wrong minimum data length (which is of course zero for such
2455 subpatterns). This could cause incorrect "no match" results.
2456
2457 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2458 the pattern is reset in the first branch, pcre_compile() failed with
2459 "internal error: code overflow at offset...". This happened only when
2460 the reset was to the original external option setting. (An optimization
2461 abstracts leading options settings into an external setting, which was the
2462 cause of this.)
2463
2464 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2465 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2466 assertion pattern did not match (meaning that the assertion was true), it
2467 was incorrectly treated as false if the SKIP had been reached during the
2468 matching. This also applied to assertions used as conditions.
2469
2470 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2471 assertion subpattern, including such a pattern used as a condition,
2472 unpredictable results occurred, instead of the error return
2473 PCRE_ERROR_DFA_UITEM.
2474
2475 5. The C++ GlobalReplace function was not working like Perl for the special
2476 situation when an empty string is matched. It now does the fancy magic
2477 stuff that is necessary.
2478
2479 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2480 removed. (These were left over from very, very early versions of PCRE.)
2481
2482 7. Some cosmetic changes to the code to make life easier when compiling it
2483 as part of something else:
2484
2485 (a) Change DEBUG to PCRE_DEBUG.
2486
2487 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2488 called "current" as "current_branch", to prevent a collision with the
2489 Linux macro when compiled as a kernel module.
2490
2491 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2492 prevent a collision with the Linux macro when compiled as a kernel
2493 module.
2494
2495 8. In pcre_compile() there are some checks for integer overflows that used to
2496 cast potentially large values to (double). This has been changed to that
2497 when building, a check for int64_t is made, and if it is found, it is used
2498 instead, thus avoiding the use of floating point arithmetic. (There is no
2499 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2500 double.
2501
2502 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2503 2005 (difference between two addresses compared to an unsigned value).
2504
2505 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2506 custom one, because of the following reported problem in Windows:
2507
2508 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2509 under Win32.
2510 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2511 therefore missing the function definition.
2512 - The compiler thus generates a "C" signature for the test function.
2513 - The linker fails to find the "C" function.
2514 - PCRE fails to configure if asked to do so against libbz2.
2515
2516 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2517 messages were output:
2518
2519 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2520 rerunning libtoolize, to keep the correct libtool macros in-tree.
2521 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2522
2523 I have done both of these things.
2524
2525 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2526 most of the time, it *can* run out if it is given a pattern that contains a
2527 runaway infinite recursion. I updated the discussion in the pcrestack man
2528 page.
2529
2530 13. Now that we have gone to the x.xx style of version numbers, the minor
2531 version may start with zero. Using 08 or 09 is a bad idea because users
2532 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2533 interpreted as invalid octal numbers. I've updated the previous comment in
2534 configure.ac, and also added a check that gives an error if 08 or 09 are
2535 used.
2536
2537 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2538 causing partial matching to fail when the end of the subject matched \W
2539 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2540
2541 15. There were some discrepancies between the declarations in pcre_internal.h
2542 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2543 their definitions. The declarations used "const uschar *" and the
2544 definitions used USPTR. Even though USPTR is normally defined as "const
2545 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2546 reported that: "This difference in casting confuses some C++ compilers, for
2547 example, SunCC recognizes above declarations as different functions and
2548 generates broken code for hbpcre." I have changed the declarations to use
2549 USPTR.
2550
2551 16. GNU libtool is named differently on some systems. The autogen.sh script now
2552 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2553 (FreeBSD).
2554
2555 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2556 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2557 comment: "Figure out how to create a longlong from a string: strtoll and
2558 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2559 instance, but it only takes 2 args instead of 3!"
2560
2561 18. A subtle bug concerned with back references has been fixed by a change of
2562 specification, with a corresponding code fix. A pattern such as
2563 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2564 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2565 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2566 same bug. Such groups have to be quantified to be useful, or contained
2567 inside another quantified group. (If there's no repetition, the reference
2568 can never match.) The problem arises because, having left the group and
2569 moved on to the rest of the pattern, a later failure that backtracks into
2570 the group uses the captured value from the final iteration of the group
2571 rather than the correct earlier one. I have fixed this in PCRE by forcing
2572 any group that contains a reference to itself to be an atomic group; that
2573 is, there cannot be any backtracking into it once it has completed. This is
2574 similar to recursive and subroutine calls.
2575
2576
2577 Version 8.00 19-Oct-09
2578 ----------------------
2579
2580 1. The table for translating pcre_compile() error codes into POSIX error codes
2581 was out-of-date, and there was no check on the pcre_compile() error code
2582 being within the table. This could lead to an OK return being given in
2583 error.
2584
2585 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2586 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2587 in a Windows environment.
2588
2589 3. The pcregrep --count option prints the count for each file even when it is
2590 zero, as does GNU grep. However, pcregrep was also printing all files when
2591 --files-with-matches was added. Now, when both options are given, it prints
2592 counts only for those files that have at least one match. (GNU grep just
2593 prints the file name in this circumstance, but including the count seems
2594 more useful - otherwise, why use --count?) Also ensured that the
2595 combination -clh just lists non-zero counts, with no names.
2596
2597 4. The long form of the pcregrep -F option was incorrectly implemented as
2598 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2599 but it seems right to fix it, and I didn't think it was worth preserving
2600 the old behaviour.
2601
2602 5. The command line items --regex=pattern and --regexp=pattern were not
2603 recognized by pcregrep, which required --regex pattern or --regexp pattern
2604 (with a space rather than an '='). The man page documented the '=' forms,
2605 which are compatible with GNU grep; these now work.
2606
2607 6. No libpcreposix.pc file was created for pkg-config; there was just
2608 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2609
2610 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2611 when UCP support is not needed, by modifying the Python script that
2612 generates it from Unicode data files. This should not matter if the module
2613 is correctly used as a library, but I received one complaint about 50K of
2614 unwanted data. My guess is that the person linked everything into his
2615 program rather than using a library. Anyway, it does no harm.
2616
2617 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2618 was a minimum greater than 1 for a wide character in a possessive
2619 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2620 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2621 character. Chaos in the form of incorrect output or a compiling loop could
2622 result.
2623
2624 9. The restrictions on what a pattern can contain when partial matching is
2625 requested for pcre_exec() have been removed. All patterns can now be
2626 partially matched by this function. In addition, if there are at least two
2627 slots in the offset vector, the offset of the earliest inspected character
2628 for the match and the offset of the end of the subject are set in them when
2629 PCRE_ERROR_PARTIAL is returned.
2630
2631 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2632 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2633 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2634 and may be more useful for multi-segment matching.
2635
2636 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2637 used to be given if ever the end of the subject was reached; now it is
2638 given only if matching could not proceed because another character was
2639 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2640 string "Z", which now yields "no match" instead of "partial match". In the
2641 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2642 final character ended with (*FAIL).
2643
2644 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2645 if the pattern had a "must contain" character that was already found in the
2646 earlier partial match, unless partial matching was again requested. For
2647 example, with the pattern /dog.(body)?/, the "must contain" character is
2648 "g". If the first part-match was for the string "dog", restarting with
2649 "sbody" failed. This bug has been fixed.
2650
2651 13. The string returned by pcre_dfa_exec() after a partial match has been
2652 changed so that it starts at the first inspected character rather than the
2653 first character of the match. This makes a difference only if the pattern
2654 starts with a lookbehind assertion or \b or \B (\K is not supported by
2655 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2656 matching functions compatible, and I think it's the right thing to do.
2657
2658 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2659 so that the demonstration program is easily available in environments where
2660 PCRE has not been installed from source.
2661
2662 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2663 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2664 library.
2665
2666 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2667 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2668 is not the first non-POSIX option to be added. Clearly some people find
2669 these options useful.
2670
2671 17. If a caller to the POSIX matching function regexec() passes a non-zero
2672 value for nmatch with a NULL value for pmatch, the value of
2673 nmatch is forced to zero.
2674
2675 18. RunGrepTest did not have a test for the availability of the -u option of
2676 the diff command, as RunTest does. It now checks in the same way as
2677 RunTest, and also checks for the -b option.
2678
2679 19. If an odd number of negated classes containing just a single character
2680 interposed, within parentheses, between a forward reference to a named
2681 subpattern and the definition of the subpattern, compilation crashed with
2682 an internal error, complaining that it could not find the referenced
2683 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2684 [The bug was that it was starting one character too far in when skipping
2685 over the character class, thus treating the ] as data rather than
2686 terminating the class. This meant it could skip too much.]
2687
2688 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2689 /g option in pcretest when the pattern contains \K, which makes it possible
2690 to have an empty string match not at the start, even when the pattern is
2691 anchored. Updated pcretest and pcredemo to use this option.
2692
2693 21. If the maximum number of capturing subpatterns in a recursion was greater
2694 than the maximum at the outer level, the higher number was returned, but
2695 with unset values at the outer level. The correct (outer level) value is
2696 now given.
2697
2698 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2699 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2700 make it do so. The string so far is captured, making this feature
2701 compatible with Perl.
2702
2703 23. The tests have been re-organized, adding tests 11 and 12, to make it
2704 possible to check the Perl 5.10 features against Perl 5.10.
2705
2706 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2707 pattern matches a fixed length string. PCRE did not allow this; now it
2708 does. Neither allows recursion.
2709
2710 25. I finally figured out how to implement a request to provide the minimum
2711 length of subject string that was needed in order to match a given pattern.
2712 (It was back references and recursion that I had previously got hung up
2713 on.) This code has now been added to pcre_study(); it finds a lower bound
2714 to the length of subject needed. It is not necessarily the greatest lower
2715 bound, but using it to avoid searching strings that are too short does give
2716 some useful speed-ups. The value is available to calling programs via
2717 pcre_fullinfo().
2718
2719 26. While implementing 25, I discovered to my embarrassment that pcretest had
2720 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2721 study optimizations had never been tested with that matching function.
2722 Oops. What is worse, even when it was passed study data, there was a bug in
2723 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2724 were also very few tests of studied patterns with pcre_dfa_exec().
2725
2726 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2727 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2728 on the other side of the coin, they are no longer allowed to have different
2729 names, because these cannot be distinguished in PCRE, and this has caused
2730 confusion. (This is a difference from Perl.)
2731
2732 28. When duplicate subpattern names are present (necessarily with different
2733 numbers, as required by 27 above), and a test is made by name in a
2734 conditional pattern, either for a subpattern having been matched, or for
2735 recursion in such a pattern, all the associated numbered subpatterns are
2736 tested, and the overall condition is true if the condition is true for any
2737 one of them. This is the way Perl works, and is also more like the way
2738 testing by number works.
2739
2740
2741 Version 7.9 11-Apr-09
2742 ---------------------
2743
2744 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2745 (pcretest), all targets were linked against these libraries. This included
2746 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2747 libraries. This caused unwanted dependencies to be created. This problem
2748 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2749 pcretest is linked with readline.
2750
2751 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2752 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2753 moved outside it again, because FALSE and TRUE are already defined in AIX,
2754 but BOOL is not.
2755
2756 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2757 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2758
2759 4. The pcregrep documentation said spaces were inserted as well as colons (or
2760 hyphens) following file names and line numbers when outputting matching
2761 lines. This is not true; no spaces are inserted. I have also clarified the
2762 wording for the --colour (or --color) option.
2763
2764 5. In pcregrep, when --colour was used with -o, the list of matching strings
2765 was not coloured; this is different to GNU grep, so I have changed it to be
2766 the same.
2767
2768 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2769 each matching line was coloured. Now it goes on to look for further matches
2770 of any of the test patterns, which is the same behaviour as GNU grep.
2771
2772 7. A pattern that could match an empty string could cause pcregrep to loop; it
2773 doesn't make sense to accept an empty string match in pcregrep, so I have
2774 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2775 seems to be how GNU grep behaves. [But see later change 40 for release
2776 8.33.]
2777
2778 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2779 start or after a newline", because the conditional assertion was not being
2780 correctly handled. The rule now is that both the assertion and what follows
2781 in the first alternative must satisfy the test.
2782
2783 9. If auto-callout was enabled in a pattern with a conditional group whose
2784 condition was an assertion, PCRE could crash during matching, both with
2785 pcre_exec() and pcre_dfa_exec().
2786
2787 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2788 used for matching.
2789
2790 11. Unicode property support in character classes was not working for
2791 characters (bytes) greater than 127 when not in UTF-8 mode.
2792
2793 12. Added the -M command line option to pcretest.
2794
2795 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2796
2797 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2798
2799 16. Added comments and documentation about mis-use of no_arg in the C++
2800 wrapper.
2801
2802 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2803 from Martin Jerabek that uses macro names for all relevant character and
2804 string constants.
2805
2806 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2807 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2808 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2809 these, but not everybody uses configure.
2810
2811 19. A conditional group that had only one branch was not being correctly
2812 recognized as an item that could match an empty string. This meant that an
2813 enclosing group might also not be so recognized, causing infinite looping
2814 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2815 with the subject "ab", where knowledge that the repeated group can match
2816 nothing is needed in order to break the loop.
2817
2818 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2819 exec(), but without supplying a callout function, matching went wrong.
2820
2821 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2822 leak if the size of the offset vector was greater than 30. When the vector
2823 is smaller, the saved offsets during recursion go onto a local stack
2824 vector, but for larger vectors malloc() is used. It was failing to free
2825 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2826 error, in fact).
2827
2828 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2829 heapframe that is used only when UTF-8 support is enabled. This caused no
2830 problem, but was untidy.
2831
2832 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2833 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2834 included within another project.
2835
2836 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2837 slightly modified by me:
2838
2839 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2840 not building pcregrep.
2841
2842 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2843 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2844
2845 25. Forward references, both numeric and by name, in patterns that made use of
2846 duplicate group numbers, could behave incorrectly or give incorrect errors,
2847 because when scanning forward to find the reference group, PCRE was not
2848 taking into account the duplicate group numbers. A pattern such as
2849 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2850
2851 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2852 the feature of a custom pointer more persuasive (as requested by a user).
2853
2854 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2855 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2856 pre-defined.
2857
2858 28. Added support for (*UTF8) at the start of a pattern.
2859
2860 29. Arrange for flags added by the "release type" setting in CMake to be shown
2861 in the configuration summary.
2862
2863
2864 Version 7.8 05-Sep-08
2865 ---------------------
2866
2867 1. Replaced UCP searching code with optimized version as implemented for Ad
2868 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2869 stage table and inline lookup instead of a function, giving speed ups of 2
2870 to 5 times on some simple patterns that I tested. Permission was given to
2871 distribute the MultiStage2.py script that generates the tables (it's not in
2872 the tarball, but is in the Subversion repository).
2873
2874 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2875 scripts.
2876
2877 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2878 a group with a zero qualifier. The result of the study could be incorrect,
2879 or the function might crash, depending on the pattern.
2880
2881 4. Caseless matching was not working for non-ASCII characters in back
2882 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2883 It now works when Unicode Property Support is available.
2884
2885 5. In pcretest, an escape such as \x{de} in the data was always generating
2886 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2887 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2888 truncation.
2889
2890 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2891
2892 7. Added two (int) casts to pcregrep when printing the difference of two
2893 pointers, in case they are 64-bit values.
2894
2895 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2896 test 2 if it fails.
2897
2898 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2899 and a #define of that name to empty if it is not externally set. This is to
2900 allow users of MSVC to set it if necessary.
2901
2902 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2903 the convenience functions in the pcre_get.c source file.
2904
2905 11. An option change at the start of a pattern that had top-level alternatives
2906 could cause overwriting and/or a crash. This command provoked a crash in
2907 some environments:
2908
2909 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2910
2911 This potential security problem was recorded as CVE-2008-2371.
2912
2913 12. For a pattern where the match had to start at the beginning or immediately
2914 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2915 pcre_dfa_exec() could read past the end of the passed subject if there was
2916 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2917 pcretest so that it places the subject at the end of its malloc-ed buffer.
2918
2919 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2920 exec() might read past the end of the data buffer in UTF-8 mode.
2921
2922 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2923 the data contained the byte 0x85 as part of a UTF-8 character within its
2924 first line. This applied both to normal and DFA matching.
2925
2926 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2927 /^[^d]*?$/8 failed to match "abc".
2928
2929 16. Added a missing copyright notice to pcrecpp_internal.h.
2930
2931 17. Make it more clear in the documentation that values returned from
2932 pcre_exec() in ovector are byte offsets, not character counts.
2933
2934 18. Tidied a few places to stop certain compilers from issuing warnings.
2935
2936 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2937 supplied by Stefan Weber. I made a further small update for 7.8 because
2938 there is a change of source arrangements: the pcre_searchfuncs.c module is
2939 replaced by pcre_ucd.c.
2940
2941
2942 Version 7.7 07-May-08
2943 ---------------------
2944
2945 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2946 a string to a long long, pretend we don't even have a long long." This is
2947 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2948
2949 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2950 pre-7.6 versions, which defined a global no_arg variable instead of putting
2951 it in the RE class. (See also #8 below.)
2952
2953 3. Remove a line of dead code, identified by coverity and reported by Nuno
2954 Lopes.
2955
2956 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2957
2958 (1) The include/exclude patterns were being applied to the whole pathnames
2959 of files, instead of just to the final components.
2960
2961 (2) If there was more than one level of directory, the subdirectories were
2962 skipped unless they satisfied the include/exclude conditions. This is
2963 inconsistent with GNU grep (and could even be seen as contrary to the
2964 pcregrep specification - which I improved to make it absolutely clear).
2965 The action now is always to scan all levels of directory, and just
2966 apply the include/exclude patterns to regular files.
2967
2968 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2969 --exclude_dir in the tests to avoid scanning .svn directories.
2970
2971 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2972 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2973 doesn't support NULs in patterns.
2974
2975 7. Added some missing "const"s to declarations of static tables in
2976 pcre_compile.c and pcre_dfa_exec.c.
2977
2978 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2979 caused by fix #2 above. (Subsequently also a second patch to fix the
2980 first patch. And a third patch - this was a messy problem.)
2981
2982 9. Applied Craig's patch to remove the use of push_back().
2983
2984 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2985 matching function regexec().
2986
2987 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2988 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2989 references. PCRE supports relative numbers with this syntax (I don't think
2990 Oniguruma does).
2991
2992 12. Previously, a group with a zero repeat such as (...){0} was completely
2993 omitted from the compiled regex. However, this means that if the group
2994 was called as a subroutine from elsewhere in the pattern, things went wrong
2995 (an internal error was given). Such groups are now left in the compiled
2996 pattern, with a new opcode that causes them to be skipped at execution
2997 time.
2998
2999 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
3000 to the way PCRE behaves:
3001
3002 (a) A lone ] character is dis-allowed (Perl treats it as data).
3003
3004 (b) A back reference to an unmatched subpattern matches an empty string
3005 (Perl fails the current match path).
3006
3007 (c) A data ] in a character class must be notated as \] because if the
3008 first data character in a class is ], it defines an empty class. (In
3009 Perl it is not possible to have an empty class.) The empty class []
3010 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
3011 The negative empty class [^] matches any one character, independently
3012 of the DOTALL setting.
3013
3014 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
3015 non-existent subpattern following a character class starting with ']' and
3016 containing () gave an internal compiling error instead of "reference to
3017 non-existent subpattern". Fortunately, when the pattern did exist, the
3018 compiled code was correct. (When scanning forwards to check for the
3019 existence of the subpattern, it was treating the data ']' as terminating
3020 the class, so got the count wrong. When actually compiling, the reference
3021 was subsequently set up correctly.)
3022
3023 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
3024 it was being rejected as not supported by pcre_dfa_exec(), even though
3025 other assertions are supported. I have made pcre_dfa_exec() support
3026 (*FAIL).
3027
3028 16. The implementation of 13c above involved the invention of a new opcode,
3029 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
3030 cannot be changed at match time, I realized I could make a small
3031 improvement to matching performance by compiling OP_ALLANY instead of
3032 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
3033 on the OP_ANY path.
3034
3035 17. Compiling pcretest on Windows with readline support failed without the
3036 following two fixes: (1) Make the unistd.h include conditional on
3037 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
3038
3039 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
3040 ncurses library to be included for pcretest when ReadLine support is
3041 requested, but also to allow for it to be overridden. This patch came from
3042 Daniel Bergström.
3043
3044 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
3045 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
3046 any errors with the current Unicode tables. Thanks to Peter Kankowski for
3047 spotting this.
3048
3049
3050 Version 7.6 28-Jan-08
3051 ---------------------
3052
3053 1. A character class containing a very large number of characters with
3054 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
3055 overflow.
3056
3057 2. Patch to cut out the "long long" test in pcrecpp_unittest when
3058 HAVE_LONG_LONG is not defined.
3059
3060 3. Applied Christian Ehrlicher's patch to update the CMake build files to
3061 bring them up to date and include new features. This patch includes:
3062
3063 - Fixed PH's badly added libz and libbz2 support.
3064 - Fixed a problem with static linking.
3065 - Added pcredemo. [But later removed - see 7 below.]
3066 - Fixed dftables problem and added an option.
3067 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
3068 HAVE_LONG_LONG.
3069 - Added readline support for pcretest.
3070 - Added an listing of the option settings after cmake has run.
3071
3072 4. A user submitted a patch to Makefile that makes it easy to create
3073 "pcre.dll" under mingw when using Configure/Make. I added stuff to
3074 Makefile.am that cause it to include this special target, without
3075 affecting anything else. Note that the same mingw target plus all
3076 the other distribution libraries and programs are now supported
3077 when configuring with CMake (see 6 below) instead of with
3078 Configure/Make.
3079
3080 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
3081 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
3082 exported in the Windows port". It has not yet been confirmed that the patch
3083 solves the problem, but it does no harm.
3084
3085 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
3086 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
3087 with CMake, and also correct the comment about stack recursion.
3088
3089 7. Remove the automatic building of pcredemo from the ./configure system and
3090 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
3091 of a program that users should build themselves after PCRE is installed, so
3092 building it automatically is not really right. What is more, it gave
3093 trouble in some build environments.
3094
3095 8. Further tidies to CMakeLists.txt from Sheri and Christian.
3096
3097
3098 Version 7.5 10-Jan-08
3099 ---------------------
3100
3101 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
3102 values in parens when parsing an RE using the C++ wrapper."
3103
3104 2. Negative specials like \S did not work in character classes in UTF-8 mode.
3105 Characters greater than 255 were excluded from the class instead of being
3106 included.
3107
3108 3. The same bug as (2) above applied to negated POSIX classes such as
3109 [:^space:].
3110
3111 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
3112 defined or documented. It seems to have been a typo for PCRE_STATIC, so
3113 I have changed it.
3114
3115 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
3116 first named subpattern) and a construct such as (?&a) would reference the
3117 first named subpattern whose name started with "a" (in other words, the
3118 length check was missing). Both these problems are fixed. "Subpattern name
3119 expected" is now given for (?&) (a zero-length name), and this patch also
3120 makes it give the same error for \k'' (previously it complained that that
3121 was a reference to a non-existent subpattern).
3122
3123 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
3124 this is right because (?- can be followed by option settings as well as by
3125 digits. I have, however, made the messages clearer.
3126
3127 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
3128 than the number used in the conditional) now cause a compile-time error.
3129 This is actually not compatible with Perl, which accepts such patterns, but
3130 treats the conditional as always being FALSE (as PCRE used to), but it
3131 seems to me that giving a diagnostic is better.
3132
3133 8. Change "alphameric" to the more common word "alphanumeric" in comments
3134 and messages.
3135
3136 9. Fix two occurrences of "backslash" in comments that should have been
3137 "backspace".
3138
3139 10. Remove two redundant lines of code that can never be obeyed (their function
3140 was moved elsewhere).
3141
3142 11. The program that makes PCRE's Unicode character property table had a bug
3143 which caused it to generate incorrect table entries for sequences of
3144 characters that have the same character type, but are in different scripts.
3145 It amalgamated them into a single range, with the script of the first of
3146 them. In other words, some characters were in the wrong script. There were
3147 thirteen such cases, affecting characters in the following ranges:
3148
3149 U+002b0 - U+002c1
3150 U+0060c - U+0060d
3151 U+0061e - U+00612
3152 U+0064b - U+0065e
3153 U+0074d - U+0076d
3154 U+01800 - U+01805
3155 U+01d00 - U+01d77
3156 U+01d9b - U+01dbf
3157 U+0200b - U+0200f
3158 U+030fc - U+030fe
3159 U+03260 - U+0327f
3160 U+0fb46 - U+0fbb1
3161 U+10450 - U+1049d
3162
3163 12. The -o option (show only the matching part of a line) for pcregrep was not
3164 compatible with GNU grep in that, if there was more than one match in a
3165 line, it showed only the first of them. It now behaves in the same way as
3166 GNU grep.
3167
3168 13. If the -o and -v options were combined for pcregrep, it printed a blank
3169 line for every non-matching line. GNU grep prints nothing, and pcregrep now
3170 does the same. The return code can be used to tell if there were any
3171 non-matching lines.
3172
3173 14. Added --file-offsets and --line-offsets to pcregrep.
3174
3175 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
3176 infinitely looping recursion. The bug was that positive lookaheads were not
3177 being skipped when checking for a possible empty match (negative lookaheads
3178 and both kinds of lookbehind were skipped).
3179
3180 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
3181 inclusion of <windows.h> to before rather than after the definition of
3182 INVALID_FILE_ATTRIBUTES (patch from David Byron).
3183
3184 17. Specifying a possessive quantifier with a specific limit for a Unicode
3185 character property caused pcre_compile() to compile bad code, which led at
3186 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
3187 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
3188 caused the error; without that there was no problem.
3189
3190 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3191
3192 19. Added --enable-pcretest-libreadline.
3193
3194 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3195 RE::GlobalReplace(). As a result, the number of replacements returned was
3196 double what it should be. I removed one of the increments, but Craig sent a
3197 later patch that removed the other one (the right fix) and added unit tests
3198 that check the return values (which was not done before).
3199
3200 21. Several CMake things:
3201
3202 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3203 the names libpcre and libpcreposix, not just pcre and pcreposix.
3204
3205 (2) The above change means that pcretest and pcregrep are now correctly
3206 linked with the newly-built libraries, not previously installed ones.
3207
3208 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
3209
3210 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3211 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3212 UTF-8 newline character). The key issue is that the pattern starts .*;
3213 this means that the match must be either at the beginning, or after a
3214 newline. The bug was in the code for advancing after a failed match and
3215 checking that the new position followed a newline. It was not taking
3216 account of UTF-8 characters correctly.
3217
3218 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3219 character classes. PCRE was not treating the sequence [:...:] as a
3220 character class unless the ... were all letters. Perl, however, seems to
3221 allow any characters between [: and :], though of course it rejects as
3222 unknown any "names" that contain non-letters, because all the known class
3223 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3224 for example, whereas PCRE did not - it did not recognize a POSIX character
3225 class. This seemed a bit dangerous, so the code has been changed to be
3226 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3227 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3228 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3229 Perl does, and where it didn't before.
3230
3231 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3232 Windows environments %n is disabled by default.
3233
3234
3235 Version 7.4 21-Sep-07
3236 ---------------------
3237
3238 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3239 means that a class such as [\s] counted as "explicit reference to CR or
3240 LF". That isn't really right - the whole point of the change was to try to
3241 help when there was an actual mention of one of the two characters. So now
3242 the change happens only if \r or \n (or a literal CR or LF) character is
3243 encountered.
3244
3245 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3246 of both had grown to the point where there were only 3 bits left.
3247 Fortunately, there was spare space in the data structure, and so I have
3248 moved the internal flags into a new 16-bit field to free up more option
3249 bits.
3250
3251 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3252 but did not set the internal JCHANGED flag - either of these is enough to
3253 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3254 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3255 start sets both bits.
3256
3257 4. Added options (at build time, compile time, exec time) to change \R from
3258 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3259
3260 5. doc/pcresyntax.html was missing from the distribution.
3261
3262 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3263 compatibility, even though it is no longer used.
3264
3265 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3266 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3267 windows.h file is present (where different names are used). [This was
3268 reversed later after testing - see 16 below.]
3269
3270 8. Changed all #include <config.h> to #include "config.h". There were also
3271 some further <pcre.h> cases that I changed to "pcre.h".
3272
3273 9. When pcregrep was used with the --colour option, it missed the line ending
3274 sequence off the lines that it output.
3275
3276 10. It was pointed out to me that arrays of string pointers cause lots of
3277 relocations when a shared library is dynamically loaded. A technique of
3278 using a single long string with a table of offsets can drastically reduce
3279 these. I have refactored PCRE in four places to do this. The result is
3280 dramatic:
3281
3282 Originally: 290
3283 After changing UCP table: 187
3284 After changing error message table: 43
3285 After changing table of "verbs" 36
3286 After changing table of Posix names 22
3287
3288 Thanks to the folks working on Gregex for glib for this insight.
3289
3290 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3291 unicode-properties was also set.
3292
3293 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3294
3295 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3296 checked only for CRLF.
3297
3298 14. Added casts to pcretest.c to avoid compiler warnings.
3299
3300 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3301
3302 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3303 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3304 entirely. This removes changes made in 7 above.
3305
3306 17. The CMake files have been updated, and there is now more information about
3307 building with CMake in the NON-UNIX-USE document.
3308
3309
3310 Version 7.3 28-Aug-07
3311 ---------------------
3312
3313 1. In the rejigging of the build system that eventually resulted in 7.1, the
3314 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3315 brackets there is not right, since it causes compilers to look for an
3316 installed pcre.h, not the version that is in the source that is being
3317 compiled (which of course may be different). I have changed it back to:
3318
3319 #include "pcre.h"
3320
3321 I have a vague recollection that the change was concerned with compiling in
3322 different directories, but in the new build system, that is taken care of
3323 by the VPATH setting the Makefile.
3324
3325 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3326 when the subject happened to end in the byte 0x85 (e.g. if the last
3327 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3328 characters but of course it shouldn't be taken as a newline when it is part
3329 of another character. The bug was that, for an unlimited repeat of . in
3330 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3331 characters when looking for a newline.
3332
3333 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3334
3335 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3336 in debug output.
3337
3338 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3339 long printing in the pcrecpp unittest when running under MinGW.
3340
3341 6. ESC_K was left out of the EBCDIC table.
3342
3343 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3344 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3345 limit also applies to "virtual nesting" when a pattern is recursive, and in
3346 this case 1000 isn't so big. I have been able to remove this limit at the
3347 expense of backing off one optimization in certain circumstances. Normally,
3348 when pcre_exec() would call its internal match() function recursively and
3349 immediately return the result unconditionally, it uses a "tail recursion"
3350 feature to save stack. However, when a subpattern that can match an empty
3351 string has an unlimited repetition quantifier, it no longer makes this
3352 optimization. That gives it a stack frame in which to save the data for
3353 checking that an empty string has been matched. Previously this was taken
3354 from the 1000-entry workspace that had been reserved. So now there is no
3355 explicit limit, but more stack is used.
3356
3357 8. Applied Daniel's patches to solve problems with the import/export magic
3358 syntax that is required for Windows, and which was going wrong for the
3359 pcreposix and pcrecpp parts of the library. These were overlooked when this
3360 problem was solved for the main library.
3361
3362 9. There were some crude static tests to avoid integer overflow when computing
3363 the size of patterns that contain repeated groups with explicit upper
3364 limits. As the maximum quantifier is 65535, the maximum group length was
3365 set at 30,000 so that the product of these two numbers did not overflow a
3366 32-bit integer. However, it turns out that people want to use groups that
3367 are longer than 30,000 bytes (though not repeat them that many times).
3368 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3369 made it possible to implement the integer overflow checks in a much more
3370 dynamic way, which I have now done. The artificial limitation on group
3371 length has been removed - we now have only the limit on the total length of
3372 the compiled pattern, which depends on the LINK_SIZE setting.
3373
3374 10. Fixed a bug in the documentation for get/copy named substring when
3375 duplicate names are permitted. If none of the named substrings are set, the
3376 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3377 empty string.
3378
3379 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3380 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3381 because the ] is interpreted as the first data character and the
3382 terminating ] is not found. PCRE has been made compatible with Perl in this
3383 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3384 cause memory overwriting.
3385
3386 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3387 string has been matched (to stop an infinite loop). It was not recognizing
3388 a conditional subpattern that could match an empty string if that
3389 subpattern was within another subpattern. For example, it looped when
3390 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3391 condition was not nested. This bug has been fixed.
3392
3393 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3394 past the start of the subject in the presence of bytes with the top bit
3395 set, for example "\x8aBCD".
3396
3397 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3398 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3399
3400 14. Optimized (?!) to (*FAIL).
3401
3402 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3403 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3404 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3405 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3406 does: it's just the validity check that is more restrictive.
3407
3408 16. Inserted checks for integer overflows during escape sequence (backslash)
3409 processing, and also fixed erroneous offset values for syntax errors during
3410 backslash processing.
3411
3412 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3413 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3414
3415 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3416 caused an overrun.
3417
3418 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3419 something other than just ASCII characters) inside a group that had an
3420 unlimited repeat caused a loop at compile time (while checking to see
3421 whether the group could match an empty string).
3422
3423 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3424 [\P{Any}] did so. (Error in the code for printing property names.)
3425
3426 21. An orphan \E inside a character class could cause a crash.
3427
3428 22. A repeated capturing bracket such as (A)? could cause a wild memory
3429 reference during compilation.
3430
3431 23. There are several functions in pcre_compile() that scan along a compiled
3432 expression for various reasons (e.g. to see if it's fixed length for look
3433 behind). There were bugs in these functions when a repeated \p or \P was
3434 present in the pattern. These operators have additional parameters compared
3435 with \d, etc, and these were not being taken into account when moving along
3436 the compiled data. Specifically:
3437
3438 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3439 length.
3440
3441 (b) An item such as \pL+ within a repeated group could cause crashes or
3442 loops.
3443
3444 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3445 "reference to non-existent subpattern" error.
3446
3447 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3448
3449 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3450 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3451
3452 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3453
3454 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3455 character were causing crashes (broken optimization).
3456
3457 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3458 \p or \P) caused a compile-time loop.
3459
3460 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3461 break. For example, the unstudied pattern [\r\n]A does not match the string
3462 "\r\nA" because change 7.0/46 below moves the current point on by two
3463 characters after failing to match at the start. However, the pattern \nA
3464 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3465 the same is true. There doesn't seem any very clean way out of this, but
3466 what I have chosen to do makes the common cases work: PCRE now takes note
3467 of whether there can be an explicit match for \r or \n anywhere in the
3468 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3469 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3470 pattern has explicit CR or LF references.
3471
3472 29. Added (*CR) etc for changing newline setting at start of pattern.
3473
3474
3475 Version 7.2 19-Jun-07
3476 ---------------------
3477
3478 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3479 which is apparently normally available under Windows.
3480
3481 2. Re-jig the pcregrep tests with different newline settings in an attempt
3482 to make them independent of the local environment's newline setting.
3483
3484 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3485
3486 4. Some of the "internals" tests were previously cut out when the link size
3487 was not 2, because the output contained actual offsets. The recent new
3488 "Z" feature of pcretest means that these can be cut out, making the tests
3489 usable with all link sizes.
3490
3491 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3492 stack recursion. This gives a massive performance boost under BSD, but just
3493 a small improvement under Linux. However, it saves one field in the frame
3494 in all cases.
3495
3496 6. Added more features from the forthcoming Perl 5.10:
3497
3498 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3499 recursion call. It refers to the nth most recently opened parentheses.
3500
3501 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3502 to be opened parentheses.
3503
3504 (c) Conditions that refer to capturing parentheses can be specified
3505 relatively, for example, (?(-2)... or (?(+3)...
3506
3507 (d) \K resets the start of the current match so that everything before
3508 is not part of it.
3509
3510 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3511
3512 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3513 reference syntax.
3514
3515 (g) (?| introduces a group in which the numbering of parentheses in each
3516 alternative starts with the same number.
3517
3518 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3519
3520 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3521 PCRE_INFO_JCHANGED.
3522
3523 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3524 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3525 for detecting groups that can match an empty string.
3526
3527 9. A pattern with a very large number of alternatives (more than several
3528 hundred) was running out of internal workspace during the pre-compile
3529 phase, where pcre_compile() figures out how much memory will be needed. A
3530 bit of new cunning has reduced the workspace needed for groups with
3531 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3532 workspace instead of running out of the 4096 that are available.
3533
3534 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3535
3536 11. Applied patch from Google to remove an optimization that didn't quite work.
3537 The report of the bug said:
3538
3539 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3540 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3541 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3542
3543 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3544 it matched the wrong number of bytes.
3545
3546
3547 Version 7.1 24-Apr-07
3548 ---------------------
3549
3550 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3551 that is more "standard", making use of automake and other Autotools. There
3552 is some re-arrangement of the files and adjustment of comments consequent
3553 on this.
3554
3555 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3556 for recursive directory scanning broke on some systems because the files
3557 are not scanned in any specific order and on different systems the order
3558 was different. A call to "sort" has been inserted into RunGrepTest for the
3559 approprate test as a short-term fix. In the longer term there may be an
3560 alternative.
3561
3562 3. I had an email from Eric Raymond about problems translating some of PCRE's
3563 man pages to HTML (despite the fact that I distribute HTML pages, some
3564 people do their own conversions for various reasons). The problems
3565 concerned the use of low-level troff macros .br and .in. I have therefore
3566 removed all such uses from the man pages (some were redundant, some could
3567 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3568 HTML has been updated to handle .nf/.fi and to complain if it encounters
3569 .br or .in.
3570
3571 4. Updated comments in configure.ac that get placed in config.h.in and also
3572 arranged for config.h to be included in the distribution, with the name
3573 config.h.generic, for the benefit of those who have to compile without
3574 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3575
3576 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3577 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3578 makevp.bat for the current PCRE, using the additional files
3579 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3580
3581 6. A Windows user reported a minor discrepancy with test 2, which turned out
3582 to be caused by a trailing space on an input line that had got lost in his
3583 copy. The trailing space was an accident, so I've just removed it.
3584
3585 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3586 that is needed.
3587
3588 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3589 as "const" (a) because they are and (b) because it helps the PHP
3590 maintainers who have recently made a script to detect big data structures
3591 in the php code that should be moved to the .rodata section. I remembered
3592 to update Builducptable as well, so it won't revert if ucptable.h is ever
3593 re-created.
3594
3595 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3596 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3597 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3598 support is not required. This saves 1.5-2K of code, which is important in
3599 some applications.
3600
3601 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3602 so as not to refer to the tables, even though these functions will never be
3603 called when UTF-8 support is disabled. Otherwise there are problems with a
3604 shared library.
3605
3606 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3607
3608 (a) It was defining its arguments as char * instead of void *.
3609
3610 (b) It was assuming that all moves were upwards in memory; this was true
3611 a long time ago when I wrote it, but is no longer the case.
3612
3613 The emulated memove() is provided for those environments that have neither
3614 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3615 is clearly not the case, as these two bugs were recently reported.
3616
3617 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3618 and Detrail to create the HTML documentation, the .txt form of the man
3619 pages, and it removes trailing spaces from listed files. It also creates
3620 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3621 case, it wraps all the #defines with #ifndefs. This script should be run
3622 before "make dist".
3623
3624 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3625 with Unicode property support.
3626
3627 (a) For a maximizing quantifier, if the two different cases of the
3628 character were of different lengths in their UTF-8 codings (there are
3629 some cases like this - I found 11), and the matching function had to
3630 back up over a mixture of the two cases, it incorrectly assumed they
3631 were both the same length.
3632
3633 (b) When PCRE was configured to use the heap rather than the stack for
3634 recursion during matching, it was not correctly preserving the data for
3635 the other case of a UTF-8 character when checking ahead for a match
3636 while processing a minimizing repeat. If the check also involved
3637 matching a wide character, but failed, corruption could cause an
3638 erroneous result when trying to check for a repeat of the original
3639 character.
3640
3641 13. Some tidying changes to the testing mechanism:
3642
3643 (a) The RunTest script now detects the internal link size and whether there
3644 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3645 values substituted by "configure". (The RunGrepTest script already did
3646 this for UTF-8.) The configure.ac script no longer substitutes the
3647 relevant variables.
3648
3649 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3650 with length and offset values. This means that the output is different
3651 for different internal link sizes. Test 2 is skipped for link sizes
3652 other than 2 because of this, bypassing the problem. Unfortunately,
3653 there was also a test in test 3 (the locale tests) that used /B and
3654 failed for link sizes other than 2. Rather than cut the whole test out,
3655 I have added a new /Z option to pcretest that replaces the length and
3656 offset values with spaces. This is now used to make test 3 independent
3657 of link size. (Test 2 will be tidied up later.)
3658
3659 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3660 segmentation fault instead of returning the appropriate error message.
3661
3662 15. In multiline mode when the newline sequence was set to "any", the pattern
3663 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3664 This doesn't seem right; it now treats the CRLF combination as the line
3665 ending, and so does not match in that case. It's only a pattern such as ^$
3666 that would hit this one: something like ^ABC$ would have failed after \r
3667 and then tried again after \r\n.
3668
3669 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3670 in an attempt to make files that differ only in their line terminators
3671 compare equal. This works on Linux.
3672
3673 17. Under certain error circumstances pcregrep might try to free random memory
3674 as it exited. This is now fixed, thanks to valgrind.
3675
3676 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3677 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3678 was because its rules for how to advance for /g after matching an empty
3679 string at the end of a line did not allow for this case. They now check for
3680 it specially.
3681
3682 20. pcretest is supposed to handle patterns and data of any length, by
3683 extending its buffers when necessary. It was getting this wrong when the
3684 buffer for a data line had to be extended.
3685
3686 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3687 CRLF as a newline sequence.
3688
3689 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3690 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3691 I have nevertheless tidied it up.
3692
3693 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3694
3695 24. Added a man page for pcre-config.
3696
3697
3698 Version 7.0 19-Dec-06
3699 ---------------------
3700
3701 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3702 moving to gcc 4.1.1.
3703
3704 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3705 sys/time.h, which is documented as needed for this function. It doesn't
3706 seem to matter on Linux, but it showed up on some releases of OS X.
3707
3708 3. It seems that there are systems where bytes whose values are greater than
3709 127 match isprint() in the "C" locale. The "C" locale should be the
3710 default when a C program starts up. In most systems, only ASCII printing
3711 characters match isprint(). This difference caused the output from pcretest
3712 to vary, making some of the tests fail. I have changed pcretest so that:
3713
3714 (a) When it is outputting text in the compiled version of a pattern, bytes
3715 other than 32-126 are always shown as hex escapes.
3716
3717 (b) When it is outputting text that is a matched part of a subject string,
3718 it does the same, unless a different locale has been set for the match
3719 (using the /L modifier). In this case, it uses isprint() to decide.
3720
3721 4. Fixed a major bug that caused incorrect computation of the amount of memory
3722 required for a compiled pattern when options that changed within the
3723 pattern affected the logic of the preliminary scan that determines the
3724 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3725 that the computed length was too small. The symptoms of this bug were
3726 either the PCRE error "internal error: code overflow" from pcre_compile(),
3727 or a glibc crash with a message such as "pcretest: free(): invalid next
3728 size (fast)". Examples of patterns that provoked this bug (shown in
3729 pcretest format) are:
3730
3731 /(?-x: )/x
3732 /(?x)(?-x: \s*#\s*)/
3733 /((?i)[\x{c0}])/8
3734 /(?i:[\x{c0}])/8
3735
3736 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3737 is now done differently.
3738
3739 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3740 wrapper classes; (b) implement a new function in the C++ scanner that is
3741 more efficient than the old way of doing things because it avoids levels of
3742 recursion in the regex matching; (c) add a paragraph to the documentation
3743 for the FullMatch() function.
3744
3745 6. The escape sequence \n was being treated as whatever was defined as
3746 "newline". Not only was this contrary to the documentation, which states
3747 that \n is character 10 (hex 0A), but it also went horribly wrong when
3748 "newline" was defined as CRLF. This has been fixed.
3749
3750 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3751 was being set to -1 for the "end of line" case (supposedly a value that no
3752 character can have). Though this value is never used (the check for end of
3753 line is "zero bytes in current character"), it caused compiler complaints.
3754 I've changed it to 0xffffffff.
3755
3756 8. In pcre_version.c, the version string was being built by a sequence of
3757 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3758 string (as it is for production releases) called a macro with an empty
3759 argument. The C standard says the result of this is undefined. The gcc
3760 compiler treats it as an empty string (which was what was wanted) but it is
3761 reported that Visual C gives an error. The source has been hacked around to
3762 avoid this problem.
3763
3764 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3765 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3766 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3767 of them did).
3768
3769 10. Originally, pcretest opened its input and output without "b"; then I was
3770 told that "b" was needed in some environments, so it was added for release
3771 5.0 to both the input and output. (It makes no difference on Unix-like
3772 systems.) Later I was told that it is wrong for the input on Windows. I've
3773 now abstracted the modes into two macros, to make it easier to fiddle with
3774 them, and removed "b" from the input mode under Windows.
3775
3776 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3777
3778 12. Added -help and --help to pcretest as an official way of being reminded
3779 of the options.
3780
3781 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3782 and pcrecpp.cc because they annoy compilers at high warning levels.
3783
3784 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3785
3786 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3787 scripts are not C programs :-) and which was not noticed because it works
3788 on Linux.
3789
3790 16. pcretest is supposed to handle any length of pattern and data line (as one
3791 line or as a continued sequence of lines) by extending its input buffer if
3792 necessary. This feature was broken for very long pattern lines, leading to
3793 a string of junk being passed to pcre_compile() if the pattern was longer
3794 than about 50K.
3795
3796 17. I have done a major re-factoring of the way pcre_compile() computes the
3797 amount of memory needed for a compiled pattern. Previously, there was code
3798 that made a preliminary scan of the pattern in order to do this. That was
3799 OK when PCRE was new, but as the facilities have expanded, it has become
3800 harder and harder to keep it in step with the real compile phase, and there
3801 have been a number of bugs (see for example, 4 above). I have now found a
3802 cunning way of running the real compile function in a "fake" mode that
3803 enables it to compute how much memory it would need, while actually only
3804 ever using a few hundred bytes of working memory and without too many
3805 tests of the mode. This should make future maintenance and development
3806 easier. A side effect of this work is that the limit of 200 on the nesting
3807 depth of parentheses has been removed (though this was never a serious
3808 limitation, I suspect). However, there is a downside: pcre_compile() now
3809 runs more slowly than before (30% or more, depending on the pattern). I
3810 hope this isn't a big issue. There is no effect on runtime performance.
3811
3812 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3813 newline (only possible for the last line of a file) and it was a
3814 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3815
3816 19. Added additional timing features to pcretest. (1) The -tm option now times
3817 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3818 separate command line item, by a number that specifies the number of
3819 repeats to use when timing. The default is 50000; this gives better
3820 precision, but takes uncomfortably long for very large patterns.
3821
3822 20. Extended pcre_study() to be more clever in cases where a branch of a
3823 subpattern has no definite first character. For example, (a*|b*)[cd] would
3824 previously give no result from pcre_study(). Now it recognizes that the
3825 first character must be a, b, c, or d.
3826
3827 21. There was an incorrect error "recursive call could loop indefinitely" if
3828 a subpattern (or the entire pattern) that was being tested for matching an
3829 empty string contained only one non-empty item after a nested subpattern.
3830 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3831 incorrectly, because the \d was being skipped in the check.
3832
3833 22. The pcretest program now has a new pattern option /B and a command line
3834 option -b, which is equivalent to adding /B to every pattern. This causes
3835 it to show the compiled bytecode, without the additional information that
3836 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3837 is the same as /B/I).
3838
3839 23. A new optimization is now able automatically to treat some sequences such
3840 as a*b as a*+b. More specifically, if something simple (such as a character
3841 or a simple class like \d) has an unlimited quantifier, and is followed by
3842 something that cannot possibly match the quantified thing, the quantifier
3843 is automatically "possessified".
3844
3845 24. A recursive reference to a subpattern whose number was greater than 39
3846 went wrong under certain circumstances in UTF-8 mode. This bug could also
3847 have affected the operation of pcre_study().
3848
3849 25. Realized that a little bit of performance could be had by replacing
3850 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3851
3852 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3853
3854 27. Possessive quantifiers such as a++ were previously implemented by turning
3855 them into atomic groups such as ($>a+). Now they have their own opcodes,
3856 which improves performance. This includes the automatically created ones
3857 from 23 above.
3858
3859 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3860 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3861 the first matched character to be a colon. This applied both to named and
3862 numbered groups.
3863
3864 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3865
3866 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3867 building PCRE on Windows easier, so I have included it in the distribution.
3868
3869 31. There is now a check in pcretest against a ridiculously large number being
3870 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3871 loop, the loop is abandoned.
3872
3873 32. Forward references to subpatterns in conditions such as (?(2)...) where
3874 subpattern 2 is defined later cause pcre_compile() to search forwards in
3875 the pattern for the relevant set of parentheses. This search went wrong
3876 when there were unescaped parentheses in a character class, parentheses
3877 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3878
3879 33. "Subroutine" calls and backreferences were previously restricted to
3880 referencing subpatterns earlier in the regex. This restriction has now
3881 been removed.
3882
3883 34. Added a number of extra features that are going to be in Perl 5.10. On the
3884 whole, these are just syntactic alternatives for features that PCRE had
3885 previously implemented using the Python syntax or my own invention. The
3886 other formats are all retained for compatibility.
3887
3888 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3889 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3890 also .NET compatible.
3891
3892 (b) A recursion or subroutine call to a named group can now be defined as
3893 (?&name) as well as (?P>name).
3894
3895 (c) A backreference to a named group can now be defined as \k<name> or
3896 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3897 5.10, are also .NET compatible.
3898
3899 (d) A conditional reference to a named group can now use the syntax
3900 (?(<name>) or (?('name') as well as (?(name).
3901
3902 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3903 groups (named and numbered) that are never evaluated inline, but can be
3904 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3905 is always false. There may be only one alternative in such a group.
3906
3907 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3908 as the simple (?(R). The condition is true only if the most recent
3909 recursion is that of the given number or name. It does not search out
3910 through the entire recursion stack.
3911
3912 (g) The escape \gN or \g{N} has been added, where N is a positive or
3913 negative number, specifying an absolute or relative reference.
3914
3915 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3916 some "unreachable code" warnings.
3917
3918 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3919 things, this adds five new scripts.
3920
3921 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3922 There were also incompatibilities regarding the handling of \Q..\E inside
3923 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3924 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3925
3926 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3927 matches an empty string, and forcibly breaks the loop. There were bugs in
3928 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3929 against aaaa the result was just "a" rather than "aaaa", for example. Two
3930 separate and independent bugs (that affected different cases) have been
3931 fixed.
3932
3933 39. Refactored the code to abolish the use of different opcodes for small
3934 capturing bracket numbers. This is a tidy that I avoided doing when I
3935 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3936 The new approach is not only tidier, it makes it possible to reduce the
3937 memory needed to fix the previous bug (38).
3938
3939 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3940 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3941 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3942 mode.
3943
3944 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3945 report.
3946
3947 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3948 copy construction and assignment in the C++ wrapper.
3949
3950 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3951 couple of bugs that could have given wrong results in the "--newline=crlf"
3952 case.
3953
3954 44. Added a number of casts and did some reorganization of signed/unsigned int
3955 variables following suggestions from Dair Grant. Also renamed the variable
3956 "this" as "item" because it is a C++ keyword.
3957
3958 45. Arranged for dftables to add
3959
3960 #include "pcre_internal.h"
3961
3962 to pcre_chartables.c because without it, gcc 4.x may remove the array
3963 definition from the final binary if PCRE is built into a static library and
3964 dead code stripping is activated.
3965
3966 46. For an unanchored pattern, if a match attempt fails at the start of a
3967 newline sequence, and the newline setting is CRLF or ANY, and the next two
3968 characters are CRLF, advance by two characters instead of one.
3969
3970
3971 Version 6.7 04-Jul-06
3972 ---------------------
3973
3974 1. In order to handle tests when input lines are enormously long, pcretest has
3975 been re-factored so that it automatically extends its buffers when
3976 necessary. The code is crude, but this _is_ just a test program. The
3977 default size has been increased from 32K to 50K.
3978
3979 2. The code in pcre_study() was using the value of the re argument before
3980 testing it for NULL. (Of course, in any sensible call of the function, it
3981 won't be NULL.)
3982
3983 3. The memmove() emulation function in pcre_internal.h, which is used on
3984 systems that lack both memmove() and bcopy() - that is, hardly ever -
3985 was missing a "static" storage class specifier.
3986
3987 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3988 containing an extended class (one that cannot be represented by a bitmap
3989 because it contains high-valued characters or Unicode property items, e.g.
3990 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3991 pattern, but PCRE should not loop if you do not (it no longer does).
3992 [Detail: two cases were found: (a) a repeated subpattern containing an
3993 extended class; (b) a recursive reference to a subpattern that followed a
3994 previous extended class. It wasn't skipping over the extended class
3995 correctly when UTF-8 mode was not set.]
3996
3997 5. A negated single-character class was not being recognized as fixed-length
3998 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3999 compile error "lookbehind assertion is not fixed length".
4000
4001 6. The RunPerlTest auxiliary script was showing an unexpected difference
4002 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
4003 write a Perl script that can interpret lines of an input file either as
4004 byte characters or as UTF-8, which is what "perltest" was being required to
4005 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
4006 can't do is switch easily at run time between having the "use utf8;" pragma
4007 or not. In the end, I fudged it by using the RunPerlTest script to insert
4008 "use utf8;" explicitly for the UTF-8 tests.
4009
4010 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
4011 the end of the subject string, contrary to the documentation and to what
4012 Perl does. This was true of both matching functions. Now it matches only at
4013 the start of the subject and immediately after *internal* newlines.
4014
4015 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
4016 a pointer to an int instead of a pointer to an unsigned long int. This
4017 caused problems on 64-bit systems.
4018
4019 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
4020 instance of the 'standard' template library not being so standard".
4021
4022 10. There was no check on the number of named subpatterns nor the maximum
4023 length of a subpattern name. The product of these values is used to compute
4024 the size of the memory block for a compiled pattern. By supplying a very
4025 long subpattern name and a large number of named subpatterns, the size
4026 computation could be caused to overflow. This is now prevented by limiting
4027 the length of names to 32 characters, and the number of named subpatterns
4028 to 10,000.
4029
4030 11. Subpatterns that are repeated with specific counts have to be replicated in
4031 the compiled pattern. The size of memory for this was computed from the
4032 length of the subpattern and the repeat count. The latter is limited to
4033 65535, but there was no limit on the former, meaning that integer overflow
4034 could in principle occur. The compiled length of a repeated subpattern is
4035 now limited to 30,000 bytes in order to prevent this.
4036
4037 12. Added the optional facility to have named substrings with the same name.
4038
4039 13. Added the ability to use a named substring as a condition, using the
4040 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
4041 are numbers (not recommended). Forward references are permitted.
4042
4043 14. Added forward references in named backreferences (if you see what I mean).
4044
4045 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
4046 pattern could run off the end of the subject. For example, the pattern
4047 "(?s)(.{1,5})"8 did this with the subject "ab".
4048
4049 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
4050 PCRE_CASELESS was set when matching characters that were quantified with ?
4051 or *.
4052
4053 17. A character class other than a single negated character that had a minimum
4054 but no maximum quantifier - for example [ab]{6,} - was not handled
4055 correctly by pce_dfa_exec(). It would match only one character.
4056
4057 18. A valid (though odd) pattern that looked like a POSIX character
4058 class but used an invalid character after [ (for example [[,abc,]]) caused
4059 pcre_compile() to give the error "Failed: internal error: code overflow" or
4060 in some cases to crash with a glibc free() error. This could even happen if
4061 the pattern terminated after [[ but there just happened to be a sequence of
4062 letters, a binary zero, and a closing ] in the memory that followed.
4063
4064 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
4065 over the years. Originally (before any Unicode support), just the bottom 8
4066 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
4067 output from "man perlunicode" includes this:
4068
4069 The regular expression compiler produces polymorphic opcodes. That
4070 is, the pattern adapts to the data and automatically switches to
4071 the Unicode character scheme when presented with Unicode data--or
4072 instead uses a traditional byte scheme when presented with byte
4073 data.
4074
4075 Sadly, a wide octal escape does not cause a switch, and in a string with
4076 no other multibyte characters, these octal escapes are treated as before.
4077 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
4078 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
4079 Unicode string.
4080
4081 I have not perpetrated such confusion in PCRE. Up till now, it took just
4082 the bottom 8 bits, as in old Perl. I have now made octal escapes with
4083 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
4084 translate to the appropriate multibyte character.
4085
4086 29. Applied some refactoring to reduce the number of warnings from Microsoft
4087 and Borland compilers. This has included removing the fudge introduced
4088 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
4089 a warning about an unused variable.
4090
4091 21. PCRE has not included VT (character 0x0b) in the set of whitespace
4092 characters since release 4.0, because Perl (from release 5.004) does not.
4093 [Or at least, is documented not to: some releases seem to be in conflict
4094 with the documentation.] However, when a pattern was studied with
4095 pcre_study() and all its branches started with \s, PCRE still included VT
4096 as a possible starting character. Of course, this did no harm; it just
4097 caused an unnecessary match attempt.
4098
4099 22. Removed a now-redundant internal flag bit that recorded the fact that case
4100 dependency changed within the pattern. This was once needed for "required
4101 byte" processing, but is no longer used. This recovers a now-scarce options
4102 bit. Also moved the least significant internal flag bit to the most-
4103 significant bit of the word, which was not previously used (hangover from
4104 the days when it was an int rather than a uint) to free up another bit for
4105 the future.
4106
4107 23. Added support for CRLF line endings as well as CR and LF. As well as the
4108 default being selectable at build time, it can now be changed at runtime
4109 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
4110 specify that it is scanning data with non-default line endings.
4111
4112 24. Changed the definition of CXXLINK to make it agree with the definition of
4113 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
4114
4115 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
4116 recursions. This makes a big different to stack usage for some patterns.
4117
4118 26. If a subpattern containing a named recursion or subroutine reference such
4119 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
4120 the space required for the compiled pattern went wrong and gave too small a
4121 value. Depending on the environment, this could lead to "Failed: internal
4122 error: code overflow at offset 49" or "glibc detected double free or
4123 corruption" errors.
4124
4125 27. Applied patches from Google (a) to support the new newline modes and (b) to
4126 advance over multibyte UTF-8 characters in GlobalReplace.
4127
4128 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
4129 difference for some implementation of PCRE in some Windows version.
4130
4131 29. Added some extra testing facilities to pcretest:
4132
4133 \q<number> in a data line sets the "match limit" value
4134 \Q<number> in a data line sets the "match recursion limt" value
4135 -S <number> sets the stack size, where <number> is in megabytes
4136
4137 The -S option isn't available for Windows.
4138
4139
4140 Version 6.6 06-Feb-06
4141 ---------------------
4142
4143 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
4144 in pcreposix.h. I have copied the definition from pcre.h.
4145
4146 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
4147 because pcre.h is no longer a built file.
4148
4149 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
4150 not normally included in the compiled code.
4151
4152
4153 Version 6.5 01-Feb-06
4154 ---------------------
4155
4156 1. When using the partial match feature with pcre_dfa_exec(), it was not
4157 anchoring the second and subsequent partial matches at the new starting
4158 point. This could lead to incorrect results. For example, with the pattern
4159 /1234/, partially matching against "123" and then "a4" gave a match.
4160
4161 2. Changes to pcregrep:
4162
4163 (a) All non-match returns from pcre_exec() were being treated as failures
4164 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
4165 error message is output. Some extra information is given for the
4166 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
4167 probably the only errors that are likely to be caused by users (by
4168 specifying a regex that has nested indefinite repeats, for instance).
4169 If there are more than 20 of these errors, pcregrep is abandoned.
4170
4171 (b) A binary zero was treated as data while matching, but terminated the
4172 output line if it was written out. This has been fixed: binary zeroes
4173 are now no different to any other data bytes.
4174
4175 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
4176 used to set a locale for matching. The --locale=xxxx long option has
4177 been added (no short equivalent) to specify a locale explicitly on the
4178 pcregrep command, overriding the environment variables.
4179
4180 (d) When -B was used with -n, some line numbers in the output were one less
4181 than they should have been.
4182
4183 (e) Added the -o (--only-matching) option.
4184
4185 (f) If -A or -C was used with -c (count only), some lines of context were
4186 accidentally printed for the final match.
4187
4188 (g) Added the -H (--with-filename) option.
4189
4190 (h) The combination of options -rh failed to suppress file names for files
4191 that were found from directory arguments.
4192
4193 (i) Added the -D (--devices) and -d (--directories) options.
4194
4195 (j) Added the -F (--fixed-strings) option.
4196
4197 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4198
4199 (l) Added the --colo(u)r option.
4200
4201 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4202 is not present by default.
4203
4204 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4205 items such as (?R) or (?1), when the recursion could match a number of
4206 alternatives. If it matched one of the alternatives, but subsequently,
4207 outside the recursion, there was a failure, the code tried to back up into
4208 the recursion. However, because of the way PCRE is implemented, this is not
4209 possible, and the result was an incorrect result from the match.
4210
4211 In order to prevent this happening, the specification of recursion has
4212 been changed so that all such subpatterns are automatically treated as
4213 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4214
4215 4. I had overlooked the fact that, in some locales, there are characters for
4216 which isalpha() is true but neither isupper() nor islower() are true. In
4217 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4218 and ordfeminine) are like this. This affected the treatment of \w and \W
4219 when they appeared in character classes, but not when they appeared outside
4220 a character class. The bit map for "word" characters is now created
4221 separately from the results of isalnum() instead of just taking it from the
4222 upper, lower, and digit maps. (Plus the underscore character, of course.)
4223
4224 5. The above bug also affected the handling of POSIX character classes such as
4225 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4226 permanent tables. Instead, the bit maps for such a class were previously
4227 created as the appropriate unions of the upper, lower, and digit bitmaps.
4228 Now they are created by subtraction from the [[:word:]] class, which has
4229 its own bitmap.
4230
4231 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4232 It is created by subtracting the vertical space characters (\x09, \x0a,
4233 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4234 subtraction was done in the overall bitmap for a character class, meaning
4235 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4236 be recognized. This bug has been fixed.
4237
4238 7. Patches from the folks at Google:
4239
4240 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4241 real life, but is still worth protecting against".
4242
4243 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4244 regular expressions".
4245
4246 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4247 have it.
4248
4249 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4250 "configure" and the latter not, in order to fix a problem somebody had
4251 with compiling the Arg class on HP-UX.
4252
4253 (e) Improve the error-handling of the C++ wrapper a little bit.
4254
4255 (f) New tests for checking recursion limiting.
4256
4257 8. The pcre_memmove() function, which is used only if the environment does not
4258 have a standard memmove() function (and is therefore rarely compiled),
4259 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4260 returning a result (though PCRE never actually uses the result).
4261
4262 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4263 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4264 returned instead of calling malloc() with an overflowing number that would
4265 most likely cause subsequent chaos.
4266
4267 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4268
4269 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4270 with this option is matched, the nmatch and pmatch options of regexec() are
4271 ignored.
4272
4273 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4274 provided in case anyone wants to the the POSIX interface with UTF-8
4275 strings.
4276
4277 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4278 C++ linking (needed for some HP-UX environments).
4279
4280 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4281 (unused parameter) and in the pcre_printint() function (omitted "default"
4282 switch label when the default is to do nothing).
4283
4284 15. Added some code to make it possible, when PCRE is compiled as a C++
4285 library, to replace subject pointers for pcre_exec() with a smart pointer
4286 class, thus making it possible to process discontinuous strings.
4287
4288 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4289 much the same function. They were added by different people who were trying
4290 to make PCRE easy to compile on non-Unix systems. It has been suggested
4291 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4292 for compiling on Windows systems. I have therefore replaced it with
4293 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4294 defaults to "extern" for C or "extern C" for C++, which works fine on
4295 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4296 SCOPE with something explicit in config.h. In addition:
4297
4298 (a) pcreposix.h still had just "extern" instead of either of these macros;
4299 I have replaced it with PCRE_DATA_SCOPE.
4300
4301 (b) Functions such as _pcre_xclass(), which are internal to the library,
4302 but external in the C sense, all had PCRE_EXPORT in their definitions.
4303 This is apparently wrong for the Windows case, so I have removed it.
4304 (It makes no difference on Unix-like systems.)
4305
4306 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4307 of recursive calls to match(). This is different to MATCH_LIMIT because
4308 that limits the total number of calls to match(), not all of which increase
4309 the depth of recursion. Limiting the recursion depth limits the amount of
4310 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4311 when PCRE is compiled, and changed at run time. A patch from Google adds
4312 this functionality to the C++ interface.
4313
4314 18. Changes to the handling of Unicode character properties:
4315
4316 (a) Updated the table to Unicode 4.1.0.
4317
4318 (b) Recognize characters that are not in the table as "Cn" (undefined).
4319
4320 (c) I revised the way the table is implemented to a much improved format
4321 which includes recognition of ranges. It now supports the ranges that
4322 are defined in UnicodeData.txt, and it also amalgamates other
4323 characters into ranges. This has reduced the number of entries in the
4324 table from around 16,000 to around 3,000, thus reducing its size
4325 considerably. I realized I did not need to use a tree structure after
4326 all - a binary chop search is just as efficient. Having reduced the
4327 number of entries, I extended their size from 6 bytes to 8 bytes to
4328 allow for more data.
4329
4330 (d) Added support for Unicode script names via properties such as \p{Han}.
4331
4332 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4333 matching that character.
4334
4335 20. When matching a repeated Unicode property with a minimum greater than zero,
4336 (for example \pL{2,}), PCRE could look past the end of the subject if it
4337 reached it while seeking the minimum number of characters. This could
4338 happen only if some of the characters were more than one byte long, because
4339 there is a check for at least the minimum number of bytes.
4340
4341 21. Refactored the implementation of \p and \P so as to be more general, to
4342 allow for more different types of property in future. This has changed the
4343 compiled form incompatibly. Anybody with saved compiled patterns that use
4344 \p or \P will have to recompile them.
4345
4346 22. Added "Any" and "L&" to the supported property types.
4347
4348 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4349 but give a compile time error if the value is greater than 0xff.
4350
4351 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4352 accidentally not being installed or uninstalled.
4353
4354 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4355 made were to insert the current release number. This seemed silly, because
4356 it made things harder for people building PCRE on systems that don't run
4357 "configure". I have turned pcre.h into a distributed file, no longer built
4358 by "configure", with the version identification directly included. There is
4359 no longer a pcre.h.in file.
4360
4361 However, this change necessitated a change to the pcre-config script as
4362 well. It is built from pcre-config.in, and one of the substitutions was the
4363 release number. I have updated configure.ac so that ./configure now finds
4364 the release number by grepping pcre.h.
4365
4366 26. Added the ability to run the tests under valgrind.
4367
4368
4369 Version 6.4 05-Sep-05
4370 ---------------------
4371
4372 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4373 "--" to be printed when multiple files were scanned, even when none of the
4374 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4375 consider it to be a bug, and have restored the previous behaviour.
4376
4377 2. A couple of code tidies to get rid of compiler warnings.
4378
4379 3. The pcretest program used to cheat by referring to symbols in the library
4380 whose names begin with _pcre_. These are internal symbols that are not
4381 really supposed to be visible externally, and in some environments it is
4382 possible to suppress them. The cheating is now confined to including
4383 certain files from the library's source, which is a bit cleaner.
4384
4385 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4386 file's purpose clearer.
4387
4388 5. Reorganized pcre_ucp_findchar().
4389
4390
4391 Version 6.3 15-Aug-05
4392 ---------------------
4393
4394 1. The file libpcre.pc.in did not have general read permission in the tarball.
4395
4396 2. There were some problems when building without C++ support:
4397
4398 (a) If C++ support was not built, "make install" and "make test" still
4399 tried to test it.
4400
4401 (b) There were problems when the value of CXX was explicitly set. Some
4402 changes have been made to try to fix these, and ...
4403
4404 (c) --disable-cpp can now be used to explicitly disable C++ support.
4405
4406 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4407 backslash in a target when C++ was disabled. This confuses some
4408 versions of "make", apparently. Using an intermediate variable solves
4409 this. (Same for CPP_LOBJ.)
4410
4411 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4412 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4413 necessary on certain architectures.
4414
4415 4. Added a setting of -export-symbols-regex to the link command to remove
4416 those symbols that are exported in the C sense, but actually are local
4417 within the library, and not documented. Their names all begin with
4418 "_pcre_". This is not a perfect job, because (a) we have to except some
4419 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4420 available (and never for static libraries). I have made a note to try to
4421 find a way round (a) in the future.
4422
4423
4424 Version 6.2 01-Aug-05
4425 ---------------------
4426
4427 1. There was no test for integer overflow of quantifier values. A construction
4428 such as {1111111111111111} would give undefined results. What is worse, if
4429 a minimum quantifier for a parenthesized subpattern overflowed and became
4430 negative, the calculation of the memory size went wrong. This could have
4431 led to memory overwriting.
4432
4433 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4434
4435 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4436 operating environments where this matters.
4437
4438 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4439 PCRE options from within the C++ wrapper.
4440
4441 5. Named capturing subpatterns were not being correctly counted when a pattern
4442 was compiled. This caused two problems: (a) If there were more than 100
4443 such subpatterns, the calculation of the memory needed for the whole
4444 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4445 back references of the form \12, where the number was greater than 9, were
4446 not recognized as back references, even though there were sufficient
4447 previous subpatterns.
4448
4449 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4450 versions of gcc, e.g. 2.95.4.
4451
4452
4453 Version 6.1 21-Jun-05
4454 ---------------------
4455
4456 1. There was one reference to the variable "posix" in pcretest.c that was not
4457 surrounded by "#if !defined NOPOSIX".
4458
4459 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4460 the cross-check on the old pcre_info() function, for the benefit of the
4461 cut-down version of PCRE that is currently imported into Exim.
4462
4463 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4464 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4465 patterns that start (?i) but I don't think that matters. The use of (?i) is
4466 just an example; this all applies to the other options as well.
4467
4468 4. Since libtool seems to echo the compile commands it is issuing, the output
4469 from "make" can be reduced a bit by putting "@" in front of each libtool
4470 compile command.
4471
4472 5. Patch from the folks at Google for configure.in to be a bit more thorough
4473 in checking for a suitable C++ installation before trying to compile the
4474 C++ stuff. This should fix a reported problem when a compiler was present,
4475 but no suitable headers.
4476
4477 6. The man pages all had just "PCRE" as their title. I have changed them to
4478 be the relevant file name. I have also arranged that these names are
4479 retained in the file doc/pcre.txt, which is a concatenation in text format
4480 of all the man pages except the little individual ones for each function.
4481
4482 7. The NON-UNIX-USE file had not been updated for the different set of source
4483 files that come with release 6. I also added a few comments about the C++
4484 wrapper.
4485
4486
4487 Version 6.0 07-Jun-05
4488 ---------------------
4489
4490 1. Some minor internal re-organization to help with my DFA experiments.
4491
4492 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4493 didn't matter for the library itself when fully configured, but did matter
4494 when compiling without UCP support, or within Exim, where the ucp files are
4495 not imported.
4496
4497 3. Refactoring of the library code to split up the various functions into
4498 different source modules. The addition of the new DFA matching code (see
4499 below) to a single monolithic source would have made it really too
4500 unwieldy, quite apart from causing all the code to be include in a
4501 statically linked application, when only some functions are used. This is
4502 relevant even without the DFA addition now that patterns can be compiled in
4503 one application and matched in another.
4504
4505 The downside of splitting up is that there have to be some external
4506 functions and data tables that are used internally in different modules of
4507 the library but which are not part of the API. These have all had their
4508 names changed to start with "_pcre_" so that they are unlikely to clash
4509 with other external names.
4510
4511 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4512 a different (DFA) algorithm. Although it is slower than the original
4513 function, it does have some advantages for certain types of matching
4514 problem.
4515
4516 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4517 including restarting after a partial match.
4518
4519 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4520 defined when compiling for Windows was sent to me. I have put it into the
4521 code, though I have no means of testing or verifying it.
4522
4523 7. Added the pcre_refcount() auxiliary function.
4524
4525 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4526 match before or at the first newline in the subject string. In pcretest,
4527 the /f option on a pattern can be used to set this.
4528
4529 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4530 would behave wrongly. This has been present in PCRE since release 4.0.
4531
4532 10. A number of changes to the pcregrep command:
4533
4534 (a) Refactored how -x works; insert ^(...)$ instead of setting
4535 PCRE_ANCHORED and checking the length, in preparation for adding
4536 something similar for -w.
4537
4538 (b) Added the -w (match as a word) option.
4539
4540 (c) Refactored the way lines are read and buffered so as to have more
4541 than one at a time available.
4542
4543 (d) Implemented a pcregrep test script.
4544
4545 (e) Added the -M (multiline match) option. This allows patterns to match
4546 over several lines of the subject. The buffering ensures that at least
4547 8K, or the rest of the document (whichever is the shorter) is available
4548 for matching (and similarly the previous 8K for lookbehind assertions).
4549
4550 (f) Changed the --help output so that it now says
4551
4552 -w, --word-regex(p)
4553
4554 instead of two lines, one with "regex" and the other with "regexp"
4555 because that confused at least one person since the short forms are the
4556 same. (This required a bit of code, as the output is generated
4557 automatically from a table. It wasn't just a text change.)
4558
4559 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4560 option but starts with a hyphen. Could be a pattern or a path name
4561 starting with a hyphen, for instance.
4562
4563 (h) "-" can be given as a file name to represent stdin.
4564
4565 (i) When file names are being printed, "(standard input)" is used for
4566 the standard input, for compatibility with GNU grep. Previously
4567 "<stdin>" was used.
4568
4569 (j) The option --label=xxx can be used to supply a name to be used for
4570 stdin when file names are being printed. There is no short form.
4571
4572 (k) Re-factored the options decoding logic because we are going to add
4573 two more options that take data. Such options can now be given in four
4574 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4575
4576 (l) Added the -A, -B, and -C options for requesting that lines of context
4577 around matches be printed.
4578
4579 (m) Added the -L option to print the names of files that do not contain
4580 any matching lines, that is, the complement of -l.
4581
4582 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4583 continue to scan other files.
4584
4585 (o) The -s option was incorrectly implemented. For compatibility with other
4586 greps, it now suppresses the error message for a non-existent or non-
4587 accessible file (but not the return code). There is a new option called
4588 -q that suppresses the output of matching lines, which was what -s was
4589 previously doing.
4590
4591 (p) Added --include and --exclude options to specify files for inclusion
4592 and exclusion when recursing.
4593
4594 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4595 Hopefully, it now does.
4596
4597 12. Missing cast in pcre_study().
4598
4599 13. Added an "uninstall" target to the makefile.
4600
4601 14. Replaced "extern" in the function prototypes in Makefile.in with
4602 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4603 world, but is set differently for Windows.
4604
4605 15. Added a second compiling function called pcre_compile2(). The only
4606 difference is that it has an extra argument, which is a pointer to an
4607 integer error code. When there is a compile-time failure, this is set
4608 non-zero, in addition to the error test pointer being set to point to an
4609 error message. The new argument may be NULL if no error number is required
4610 (but then you may as well call pcre_compile(), which is now just a
4611 wrapper). This facility is provided because some applications need a
4612 numeric error indication, but it has also enabled me to tidy up the way
4613 compile-time errors are handled in the POSIX wrapper.
4614
4615 16. Added VPATH=.libs to the makefile; this should help when building with one
4616 prefix path and installing with another. (Or so I'm told by someone who
4617 knows more about this stuff than I do.)
4618
4619 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4620 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4621 match everything, including newlines. This is not POSIX-compatible, but
4622 somebody wanted the feature. From pcretest it can be activated by using
4623 both the P and the s flags.
4624
4625 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4626
4627 19. libpcre.pc was being incorrectly installed as executable.
4628
4629 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4630 it now also looks for '\r' so that it will work unmodified on Windows.
4631
4632 21. Added Google's contributed C++ wrapper to the distribution.
4633
4634 22. Added some untidy missing memory free() calls in pcretest, to keep
4635 Electric Fence happy when testing.
4636
4637
4638
4639 Version 5.0 13-Sep-04
4640 ---------------------
4641
4642 1. Internal change: literal characters are no longer packed up into items
4643 containing multiple characters in a single byte-string. Each character
4644 is now matched using a separate opcode. However, there may be more than one
4645 byte in the character in UTF-8 mode.
4646
4647 2. The pcre_callout_block structure has two new fields: pattern_position and
4648 next_item_length. These contain the offset in the pattern to the next match
4649 item, and its length, respectively.
4650
4651 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4652 insertion of callouts before each pattern item. Added the /C option to
4653 pcretest to make use of this.
4654
4655 4. On the advice of a Windows user, the lines
4656
4657 #if defined(_WIN32) || defined(WIN32)
4658 _setmode( _fileno( stdout ), 0x8000 );
4659 #endif /* defined(_WIN32) || defined(WIN32) */
4660
4661 have been added to the source of pcretest. This apparently does useful
4662 magic in relation to line terminators.
4663
4664 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4665 for the benefit of those environments where the "b" makes a difference.
4666
4667 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4668 to know about it. I have put a hack into configure.in that adds in code
4669 to set GCC=yes if CC=icc. This seems to end up at a point in the
4670 generated configure script that is early enough to affect the setting of
4671 compiler options, which is what is needed, but I have no means of testing
4672 whether it really works. (The user who reported this had patched the
4673 generated configure script, which of course I cannot do.)
4674
4675 LATER: After change 22 below (new libtool files), the configure script
4676 seems to know about icc (and also ecc). Therefore, I have commented out
4677 this hack in configure.in.
4678
4679 7. Added support for pkg-config (2 patches were sent in).
4680
4681 8. Negated POSIX character classes that used a combination of internal tables
4682 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4683 [[:^ascii]]. Typically, they would match almost any characters. The other
4684 POSIX classes were not broken in this way.
4685
4686 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4687 to find the match, as PCRE was deluded into thinking that the match had to
4688 start at the start point or following a newline. The same bug applied to
4689 patterns with negative forward assertions or any backward assertions
4690 preceding ".*" at the start, unless the pattern required a fixed first
4691 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4692
4693 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4694 starting at the last subject character, bytes beyond the end of the subject
4695 string were read.
4696
4697 11. Renamed the variable "class" as "classbits" to make life easier for C++
4698 users. (Previously there was a macro definition, but it apparently wasn't
4699 enough.)
4700
4701 12. Added the new field "tables" to the extra data so that tables can be passed
4702 in at exec time, or the internal tables can be re-selected. This allows
4703 a compiled regex to be saved and re-used at a later time by a different
4704 program that might have everything at different addresses.
4705
4706 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4707 -R library as well as a -L library.
4708
4709 14. The debugging options of pcretest (-d on the command line or D on a
4710 pattern) showed incorrect output for anything following an extended class
4711 that contained multibyte characters and which was followed by a quantifier.
4712
4713 15. Added optional support for general category Unicode character properties
4714 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4715 support. It adds about 90K to the size of the library. The meanings of the
4716 inbuilt class escapes such as \d and \s have NOT been changed.
4717
4718 16. Updated pcredemo.c to include calls to free() to release the memory for the
4719 compiled pattern.
4720
4721 17. The generated file chartables.c was being created in the source directory
4722 instead of in the building directory. This caused the build to fail if the
4723 source directory was different from the building directory, and was
4724 read-only.
4725
4726 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4727 file. No doubt somebody will tell me if they don't make sense... Also added
4728 Dan Mooney's comments about building on OpenVMS.
4729
4730 19. Added support for partial matching via the PCRE_PARTIAL option for
4731 pcre_exec() and the \P data escape in pcretest.
4732
4733 20. Extended pcretest with 3 new pattern features:
4734
4735 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4736 write the compiled pattern to the file whose name is "rest-of-line".
4737 This is a straight binary dump of the data, with the saved pointer to
4738 the character tables forced to be NULL. The study data, if any, is
4739 written too. After writing, pcretest reads a new pattern.
4740
4741 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4742 compiled pattern from the given file. There must not be any
4743 occurrences of "<" in the file name (pretty unlikely); if there are,
4744 pcretest will instead treat the initial "<" as a pattern delimiter.
4745 After reading in the pattern, pcretest goes on to read data lines as
4746 usual.
4747
4748 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4749 and 16-bit fields in a compiled pattern, to simulate a pattern that
4750 was compiled on a host of opposite endianness.
4751
4752 21. The pcre-exec() function can now cope with patterns that were compiled on
4753 hosts of opposite endianness, with this restriction:
4754
4755 As for any compiled expression that is saved and used later, the tables
4756 pointer field cannot be preserved; the extra_data field in the arguments
4757 to pcre_exec() should be used to pass in a tables address if a value
4758 other than the default internal tables were used at compile time.
4759
4760 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4761 now diagnosed as an error. Previously, most of the time, a negative number
4762 would have been treated as zero, but if in addition "ovector" was passed as
4763 NULL, a crash could occur.
4764
4765 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4766 new versions from the libtool 1.5 distribution (the last one is a copy of
4767 a file called libtool.m4). This seems to have fixed the need to patch
4768 "configure" to support Darwin 1.3 (which I used to do). However, I still
4769 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4770 workstation).
4771
4772 24. Changed the PCRE licence to be the more standard "BSD" licence.
4773
4774
4775 Version 4.5 01-Dec-03
4776 ---------------------
4777
4778 1. There has been some re-arrangement of the code for the match() function so
4779 that it can be compiled in a version that does not call itself recursively.
4780 Instead, it keeps those local variables that need separate instances for
4781 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4782 needs to "recurse". Keeping track of where control must go is done by means
4783 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4784 hide most of the details from the main code, and operates only if
4785 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4786 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4787 operating.
4788
4789 To make it easier for callers to provide specially tailored get/free
4790 functions for this usage, two new functions, pcre_stack_malloc, and
4791 pcre_stack_free, are used. They are always called in strict stacking order,
4792 and the size of block requested is always the same.
4793
4794 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4795 PCRE has been compiled to use the stack or the heap for recursion. The
4796 -C option of pcretest uses this to show which version is compiled.
4797
4798 A new data escape \S, is added to pcretest; it causes the amounts of store
4799 obtained and freed by both kinds of malloc/free at match time to be added
4800 to the output.
4801
4802 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4803 what's available on my current Linux desktop machine.
4804
4805 3. When matching a UTF-8 string, the test for a valid string at the start has
4806 been extended. If start_offset is not zero, PCRE now checks that it points
4807 to a byte that is the start of a UTF-8 character. If not, it returns
4808 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4809 this is necessary because there may be backward assertions in the pattern.
4810 When matching the same subject several times, it may save resources to use
4811 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4812
4813 4. The code for checking the validity of UTF-8 strings has been tightened so
4814 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4815 containing "overlong sequences".
4816
4817 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4818 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4819 should have been "&", but it just so happened that all the cases this let
4820 through by mistake were picked up later in the function.
4821
4822 6. I had used a variable called "isblank" - this is a C99 function, causing
4823 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4824
4825 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4826 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4827 so the version is shown; (c) stop "make test" echoing "./RunTest".
4828
4829 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4830
4831 9. The prototype for memmove() for systems that don't have it was using
4832 size_t, but the inclusion of the header that defines size_t was later. I've
4833 moved the #includes for the C headers earlier to avoid this.
4834
4835 10. Added some adjustments to the code to make it easier to compiler on certain
4836 special systems:
4837
4838 (a) Some "const" qualifiers were missing.
4839 (b) Added the macro EXPORT before all exported functions; by default this
4840 is defined to be empty.
4841 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4842 that it reads its output file name as an argument instead of writing
4843 to the standard output and assuming this can be redirected.
4844
4845 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4846 class containing characters with values greater than 255, PCRE compilation
4847 went into a loop.
4848
4849 12. A recursive reference to a subpattern that was within another subpattern
4850 that had a minimum quantifier of zero caused PCRE to crash. For example,
4851 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4852 recursion. If the recursively-called subpattern itself had a zero repeat,
4853 that was OK.
4854
4855 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4856 buffer into which it was copied (for escape processing) was still set at
4857 1024, so long lines caused crashes.
4858
4859 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4860 "internal error: code overflow...". This applied to any character class
4861 that was followed by a possessive quantifier.
4862
4863 15. Modified the Makefile to add libpcre.la as a prerequisite for
4864 libpcreposix.la because I was told this is needed for a parallel build to
4865 work.
4866
4867 16. If a pattern that contained .* following optional items at the start was
4868 studied, the wrong optimizing data was generated, leading to matching
4869 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4870 matching string must start with a or b or c. The correct conclusion for
4871 this pattern is that a match can start with any character.
4872
4873
4874 Version 4.4 13-Aug-03
4875 ---------------------
4876
4877 1. In UTF-8 mode, a character class containing characters with values between
4878 127 and 255 was not handled correctly if the compiled pattern was studied.
4879 In fixing this, I have also improved the studying algorithm for such
4880 classes (slightly).
4881
4882 2. Three internal functions had redundant arguments passed to them. Removal
4883 might give a very teeny performance improvement.
4884
4885 3. Documentation bug: the value of the capture_top field in a callout is *one
4886 more than* the number of the hightest numbered captured substring.
4887
4888 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4889 in incorrectly linking with a previously installed version. They now link
4890 explicitly with libpcre.la.
4891
4892 5. configure.in no longer needs to recognize Cygwin specially.
4893
4894 6. A problem in pcre.in for Windows platforms is fixed.
4895
4896 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4897 pcretest, it used to include the size of the study block as part of its
4898 output. Unfortunately, the structure contains a field that has a different
4899 size on different hardware architectures. This meant that the tests that
4900 showed this size failed. As the block is currently always of a fixed size,
4901 this information isn't actually particularly useful in pcretest output, so
4902 I have just removed it.
4903
4904 8. Three pre-processor statements accidentally did not start in column 1.
4905 Sadly, there are *still* compilers around that complain, even though
4906 standard C has not required this for well over a decade. Sigh.
4907
4908 9. In pcretest, the code for checking callouts passed small integers in the
4909 callout_data field, which is a void * field. However, some picky compilers
4910 complained about the casts involved for this on 64-bit systems. Now
4911 pcretest passes the address of the small integer instead, which should get
4912 rid of the warnings.
4913
4914 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4915 both compile and run time, and gives an error if an invalid UTF-8 sequence
4916 is found. There is a option for disabling this check in cases where the
4917 string is known to be correct and/or the maximum performance is wanted.
4918
4919 11. In response to a bug report, I changed one line in Makefile.in from
4920
4921 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4922 to
4923 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4924
4925 to look similar to other lines, but I have no way of telling whether this
4926 is the right thing to do, as I do not use Windows. No doubt I'll get told
4927 if it's wrong...
4928
4929
4930 Version 4.3 21-May-03
4931 ---------------------
4932
4933 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4934 Makefile.
4935
4936 2. Some refactoring to improve the quality of the code:
4937
4938 (i) The utf8_table... variables are now declared "const".
4939
4940 (ii) The code for \cx, which used the "case flipping" table to upper case
4941 lower case letters, now just substracts 32. This is ASCII-specific,
4942 but the whole concept of \cx is ASCII-specific, so it seems
4943 reasonable.
4944
4945 (iii) PCRE was using its character types table to recognize decimal and
4946 hexadecimal digits in the pattern. This is silly, because it handles
4947 only 0-9, a-f, and A-F, but the character types table is locale-
4948 specific, which means strange things might happen. A private
4949 table is now used for this - though it costs 256 bytes, a table is
4950 much faster than multiple explicit tests. Of course, the standard
4951 character types table is still used for matching digits in subject
4952 strings against \d.
4953
4954 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4955 ending in _t are). So I've renamed it as ESC_tee.
4956
4957 3. The first argument for regexec() in the POSIX wrapper should have been
4958 defined as "const".
4959
4960 4. Changed pcretest to use malloc() for its buffers so that they can be
4961 Electric Fenced for debugging.
4962
4963 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4964 to read one or more bytes before the start of the subject string. Often this
4965 had no effect on PCRE's behaviour, but in some circumstances it could
4966 provoke a segmentation fault.
4967
4968 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4969 to try to read one or more bytes before the start of the subject string.
4970
4971 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4972 UTF-8 support could misbehave in various ways if the subject string
4973 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4974 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4975 back over UTF-8 characters.)
4976
4977
4978 Version 4.2 14-Apr-03
4979 ---------------------
4980
4981 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4982
4983 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4984 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4985 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4986 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4987 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4988 and BUILD_EXEEXT
4989 Note: automatic setting of the BUILD variables is not yet working
4990 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4991 compile-time but not at link-time
4992 [LINK]: use for linking executables only
4993 make different versions for Windows and non-Windows
4994 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4995 libraries
4996 [LINK_FOR_BUILD]: new variable
4997 [OBJEXT]: use throughout
4998 [EXEEXT]: use throughout
4999 <winshared>: new target
5000 <wininstall>: new target
5001 <dftables.o>: use native compiler
5002 <dftables>: use native linker
5003 <install>: handle Windows platform correctly
5004 <clean>: ditto
5005 <check>: ditto
5006 copy DLL to top builddir before testing
5007
5008 As part of these changes, -no-undefined was removed again. This was reported
5009 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
5010 in any case.
5011
5012 3. Some tidies to get rid of compiler warnings:
5013
5014 . In the match_data structure, match_limit was an unsigned long int, whereas
5015 match_call_count was an int. I've made them both unsigned long ints.
5016
5017 . In pcretest the fact that a const uschar * doesn't automatically cast to
5018 a void * provoked a warning.
5019
5020 . Turning on some more compiler warnings threw up some "shadow" variables
5021 and a few more missing casts.
5022
5023 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
5024 option, a class that contained a single character with a value between 128
5025 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
5026
5027 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
5028 option, a class that contained several characters, but with at least one
5029 whose value was between 128 and 255 caused PCRE to crash.
5030
5031
5032 Version 4.1 12-Mar-03
5033 ---------------------
5034
5035 1. Compiling with gcc -pedantic found a couple of places where casts were
5036 needed, and a string in dftables.c that was longer than standard compilers are
5037 required to support.
5038
5039 2. Compiling with Sun's compiler found a few more places where the code could
5040 be tidied up in order to avoid warnings.
5041
5042 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
5043 first of these names is deprecated in the latest Autoconf in favour of the name
5044 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
5045 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
5046 analogy I have changed it to CFLAGS_FOR_BUILD.
5047
5048 4. Added -no-undefined to the linking command in the Makefile, because this is
5049 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
5050 linking step for the pcreposix library.
5051
5052 5. PCRE was failing to diagnose the case of two named groups with the same
5053 name.
5054
5055 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
5056 literal character that is needed in the subject for a match, and scans along to
5057 ensure that it is present before embarking on the full matching process. This
5058 saves time in cases of nested unlimited repeats that are never going to match.
5059 Problem: the scan can take a lot of time if the subject is very long (e.g.
5060 megabytes), thus penalizing straightforward matches. It is now done only if the
5061 amount of subject to be scanned is less than 1000 bytes.
5062
5063 7. A lesser problem with the same optimization is that it was recording the
5064 first character of an anchored pattern as "needed", thus provoking a search
5065 right along the subject, even when the first match of the pattern was going to
5066 fail. The "needed" character is now not set for anchored patterns, unless it
5067 follows something in the pattern that is of non-fixed length. Thus, it still
5068 fulfils its original purpose of finding quick non-matches in cases of nested
5069 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
5070
5071
5072 Version 4.0 17-Feb-03
5073 ---------------------
5074
5075 1. If a comment in an extended regex that started immediately after a meta-item
5076 extended to the end of string, PCRE compiled incorrect data. This could lead to
5077 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
5078
5079 2. Moved to autoconf 2.53 and libtool 1.4.2.
5080
5081 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
5082 the special perltest8 script is no longer needed - all the tests can be run
5083 from a single perltest script.
5084
5085 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
5086 by \s. It has now been removed in PCRE. This means it isn't recognized as
5087 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
5088 class [:space:] *does* include VT, thereby creating a mess.
5089
5090 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
5091 space and tab.
5092
5093 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
5094 its new features into the main test script, reducing the number of scripts.
5095
5096 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
5097 were backward compatible, and made the (?i) apply to the whole pattern, as if
5098 /i were given. Now it behaves more logically, and applies the option setting
5099 only to what follows. PCRE has been changed to follow suit. However, if it
5100 finds options settings right at the start of the pattern, it extracts them into
5101 the global options, as before. Thus, they show up in the info data.
5102
5103 8. Added support for the \Q...\E escape sequence. Characters in between are
5104 treated as literals. This is slightly different from Perl in that $ and @ are
5105 also handled as literals inside the quotes. In Perl, they will cause variable
5106 interpolation. Note the following examples:
5107
5108 Pattern PCRE matches Perl matches
5109
5110 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
5111 \Qabc\$xyz\E abc\$xyz abc\$xyz
5112 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
5113
5114 For compatibility with Perl, \Q...\E sequences are recognized inside character
5115 classes as well as outside them.
5116
5117 9. Re-organized 3 code statements in pcretest to avoid "overflow in
5118 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
5119 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
5120 signed/unsigned warnings.
5121
5122 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
5123 option for pcretest, so I've replaced it by a simple function that does just
5124 that job.
5125
5126 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
5127 "pcregrep -".
5128
5129 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
5130 Java package. This provides some syntactic sugar for simple cases of what my
5131 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
5132 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
5133 item, you can use this simplified notation. Note that only makes sense with
5134 greedy quantifiers. Consequently, the use of the possessive quantifier forces
5135 greediness, whatever the setting of the PCRE_UNGREEDY option.
5136
5137 13. A change of greediness default within a pattern was not taking effect at
5138 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
5139 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
5140 was abstracted outside.
5141
5142 14. PCRE now supports the \G assertion. It is true when the current matching
5143 position is at the start point of the match. This differs from \A when the
5144 starting offset is non-zero. Used with the /g option of pcretest (or similar
5145 code), it works in the same way as it does for Perl's /g option. If all
5146 alternatives of a regex begin with \G, the expression is anchored to the start
5147 match position, and the "anchored" flag is set in the compiled expression.
5148
5149 15. Some bugs concerning the handling of certain option changes within patterns
5150 have been fixed. These applied to options other than (?ims). For example,
5151 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
5152 been the other way round. Some of this was related to change 7 above.
5153
5154 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
5155 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
5156 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
5157 POSIX classes only within a class (e.g. /[[:alpha:]]/).
5158
5159 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
5160 mode. Unlike ".", it always matches newline, whatever the setting of
5161 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
5162 assertions. Perl allows it, but it doesn't (in general) work because it can't
5163 calculate the length of the lookbehind. At least, that's the case for Perl
5164 5.8.0 - I've been told they are going to document that it doesn't work in
5165 future.
5166
5167 18. Added an error diagnosis for escapes that PCRE does not support: these are
5168 \L, \l, \N, \P, \p, \U, \u, and \X.
5169
5170 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
5171 reading past the end of the pattern in cases such as /[abcd/.
5172
5173 20. PCRE was getting more memory than necessary for patterns with classes that
5174 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
5175
5176 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
5177 compiling PCRE for use with Virtual Pascal.
5178
5179 22. Small fix to the Makefile to make it work properly if the build is done
5180 outside the source tree.
5181
5182 23. Added a new extension: a condition to go with recursion. If a conditional
5183 subpattern starts with (?(R) the "true" branch is used if recursion has
5184 happened, whereas the "false" branch is used only at the top level.
5185
5186 24. When there was a very long string of literal characters (over 255 bytes
5187 without UTF support, over 250 bytes with UTF support), the computation of how
5188 much memory was required could be incorrect, leading to segfaults or other
5189 strange effects.
5190
5191 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5192 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5193 there was a subsequent back reference to those brackets. This meant that, for
5194 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5195 possible to check for precisely this case. All we can do is abandon the
5196 optimization if .* occurs inside capturing brackets when there are any back
5197 references whatsoever. (See below for a better fix that came later.)
5198
5199 26. The handling of the optimization for finding the first character of a
5200 non-anchored pattern, and for finding a character that is required later in the
5201 match were failing in some cases. This didn't break the matching; it just
5202 failed to optimize when it could. The way this is done has been re-implemented.
5203
5204 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5205
5206 28. Added a new feature that provides some of the functionality that Perl
5207 provides with (?{...}). The facility is termed a "callout". The way it is done
5208 in PCRE is for the caller to provide an optional function, by setting
5209 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5210 global variable. By default it is unset, which disables all calling out. To get
5211 the function called, the regex must include (?C) at appropriate points. This
5212 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5213 This provides a means of identifying different callout points. When PCRE
5214 reaches such a point in the regex, if pcre_callout has been set, the external
5215 function is called. It is provided with data in a structure called
5216 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5217 matching continues; if it returns a non-zero value, the match at the current
5218 point fails. However, backtracking will occur if possible. [This was changed
5219 later and other features added - see item 49 below.]
5220
5221 29. pcretest is upgraded to test the callout functionality. It provides a
5222 callout function that displays information. By default, it shows the start of
5223 the match and the current position in the text. There are some new data escapes
5224 to vary what happens:
5225
5226 \C+ in addition, show current contents of captured substrings
5227 \C- do not supply a callout function
5228 \C!n return 1 when callout number n is reached
5229 \C!n!m return 1 when callout number n is reached for the mth time
5230
5231 30. If pcregrep was called with the -l option and just a single file name, it
5232 output "<stdin>" if a match was found, instead of the file name.
5233
5234 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5235 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5236 pcre_exec(). This saves a malloc/free per call. The default value of
5237 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5238 when configuring.
5239
5240 32. The default maximum size of a compiled pattern is 64K. There have been a
5241 few cases of people hitting this limit. The code now uses macros to handle the
5242 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5243 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5244 configuring. Tests 2 and 5 work only with 2-byte links because they output
5245 debugging information about compiled patterns.
5246
5247 33. Internal code re-arrangements:
5248
5249 (a) Moved the debugging function for printing out a compiled regex into
5250 its own source file (printint.c) and used #include to pull it into
5251 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5252 separate copies.
5253
5254 (b) Defined the list of op-code names for debugging as a macro in
5255 internal.h so that it is next to the definition of the opcodes.
5256
5257 (c) Defined a table of op-code lengths for simpler skipping along compiled
5258 code. This is again a macro in internal.h so that it is next to the
5259 definition of the opcodes.
5260
5261 34. Added support for recursive calls to individual subpatterns, along the
5262 lines of Robin Houston's patch (but implemented somewhat differently).
5263
5264 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5265 allow it to read and process whole directories in Win32. This code was
5266 contributed by Lionel Fourquaux; it has not been tested by me.
5267
5268 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5269 used to name a group. Names consist of alphanumerics and underscores, and must
5270 be unique. Back references use the syntax (?P=name) and recursive calls use
5271 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5272 numbers. The function pcre_fullinfo() can be used after compilation to extract
5273 a name/number map. There are three relevant calls:
5274
5275 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5276 PCRE_INFO_NAMECOUNT yields the number of entries
5277 PCRE_INFO_NAMETABLE yields a pointer to the map.
5278
5279 The map is a vector of fixed-size entries. The size of each entry depends on
5280 the length of the longest name used. The first two bytes of each entry are the
5281 group number, most significant byte first. There follows the corresponding
5282 name, zero terminated. The names are in alphabetical order.
5283
5284 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5285 case instead of 255. Making it the same both with and without UTF-8 support
5286 means that the same test output works with both.
5287
5288 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5289 calling malloc() with a zero argument.
5290
5291 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5292 optimization. I've improved things by keeping a bitmap of backreferences with
5293 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5294 fact referenced, the optimization can be applied. It is unlikely that a
5295 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5296 the match to follow \n) will appear inside brackets with a number greater than
5297 31, but if it does, any back reference > 31 suppresses the optimization.
5298
5299 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5300 of disabling numbered capturing parentheses. Any opening parenthesis that is
5301 not followed by ? behaves as if it were followed by ?: but named parentheses
5302 can still be used for capturing (and they will acquire numbers in the usual
5303 way).
5304
5305 41. Redesigned the return codes from the match() function into yes/no/error so
5306 that errors can be passed back from deep inside the nested calls. A malloc
5307 failure while inside a recursive subpattern call now causes the
5308 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5309
5310 42. It is now possible to set a limit on the number of times the match()
5311 function is called in a call to pcre_exec(). This facility makes it possible to
5312 limit the amount of recursion and backtracking, though not in a directly
5313 obvious way, because the match() function is used in a number of different
5314 circumstances. The count starts from zero for each position in the subject
5315 string (for non-anchored patterns). The default limit is, for compatibility, a
5316 large number, namely 10 000 000. You can change this in two ways:
5317
5318 (a) When configuring PCRE before making, you can use --with-match-limit=n
5319 to set a default value for the compiled library.
5320
5321 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5322 a different value is set. See 45 below.
5323
5324 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5325
5326 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5327 of things that can be changed at compile time. The first argument specifies
5328 what is wanted and the second points to where the information is to be placed.
5329 The current list of available information is:
5330
5331 PCRE_CONFIG_UTF8
5332
5333 The output is an integer that is set to one if UTF-8 support is available;
5334 otherwise it is set to zero.
5335
5336 PCRE_CONFIG_NEWLINE
5337
5338 The output is an integer that it set to the value of the code that is used for
5339 newline. It is either LF (10) or CR (13).
5340
5341 PCRE_CONFIG_LINK_SIZE
5342
5343 The output is an integer that contains the number of bytes used for internal
5344 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5345
5346 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5347
5348 The output is an integer that contains the threshold above which the POSIX
5349 interface uses malloc() for output vectors. See item 31 above.
5350
5351 PCRE_CONFIG_MATCH_LIMIT
5352
5353 The output is an unsigned integer that contains the default limit of the number
5354 of match() calls in a pcre_exec() execution. See 42 above.
5355
5356 44. pcretest has been upgraded by the addition of the -C option. This causes it
5357 to extract all the available output from the new pcre_config() function, and to
5358 output it. The program then exits immediately.
5359
5360 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5361 order to support additional features. One way would have been to define
5362 pcre_exec2() (for example) with extra arguments, but this would not have been
5363 extensible, and would also have required all calls to the original function to
5364 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5365 is used for passing in "extra" data from pcre_study().
5366
5367 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5368 contains the following fields:
5369
5370 flags a bitmap indicating which of the following fields are set
5371 study_data opaque data from pcre_study()
5372 match_limit a way of specifying a limit on match() calls for a specific
5373 call to pcre_exec()
5374 callout_data data for callouts (see 49 below)
5375
5376 The flag bits are also defined in pcre.h, and are
5377
5378 PCRE_EXTRA_STUDY_DATA
5379 PCRE_EXTRA_MATCH_LIMIT
5380 PCRE_EXTRA_CALLOUT_DATA
5381
5382 The pcre_study() function now returns one of these new pcre_extra blocks, with
5383 the actual study data pointed to by the study_data field, and the
5384 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5385 before. That is, this change is entirely upwards-compatible and requires no
5386 change to existing code.
5387
5388 If you want to pass in additional data to pcre_exec(), you can either place it
5389 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5390 block.
5391
5392 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5393 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5394 times with different match limits, until it finds the minimum value needed for
5395 pcre_exec() to complete. The value is then output. This can be instructive; for
5396 most simple matches the number is quite small, but for pathological cases it
5397 gets very large very quickly.
5398
5399 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5400 returns the size of the data block pointed to by the study_data field in a
5401 pcre_extra block, that is, the value that was passed as the argument to
5402 pcre_malloc() when PCRE was getting memory in which to place the information
5403 created by pcre_study(). The fourth argument should point to a size_t variable.
5404 pcretest has been extended so that this information is shown after a successful
5405 pcre_study() call when information about the compiled regex is being displayed.
5406
5407 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5408 because what follows is always an absolute path. (Later: it turns out that this
5409 is more than cosmetic for MinGW, because it doesn't like empty path
5410 components.)
5411
5412 49. Some changes have been made to the callout feature (see 28 above):
5413
5414 (i) A callout function now has three choices for what it returns:
5415
5416 0 => success, carry on matching
5417 > 0 => failure at this point, but backtrack if possible
5418 < 0 => serious error, return this value from pcre_exec()
5419
5420 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5421 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5422 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5423 use by callout functions. It will never be used by PCRE itself.
5424
5425 (ii) The pcre_extra structure (see 45 above) has a void * field called
5426 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5427 pcre_callout_block structure has a field of the same name. The contents of
5428 the field passed in the pcre_extra structure are passed to the callout
5429 function in the corresponding field in the callout block. This makes it
5430 easier to use the same callout-containing regex from multiple threads. For
5431 testing, the pcretest program has a new data escape
5432
5433 \C*n pass the number n (may be negative) as callout_data
5434
5435 If the callout function in pcretest receives a non-zero value as
5436 callout_data, it returns that value.
5437
5438 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5439 there were some redundant $(CFLAGS) in commands that are now specified as
5440 $(LINK), which already includes $(CFLAGS).
5441
5442 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5443 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5444 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5445 one-byte characters throughout. Note that case-insensitive matching applies
5446 only to characters whose values are less than 256. PCRE doesn't support the
5447 notion of cases for higher-valued characters.
5448
5449 (i) A character class whose characters are all within 0-255 is handled as
5450 a bit map, and the map is inverted for negative classes. Previously, a
5451 character > 255 always failed to match such a class; however it should
5452 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5453
5454 (ii) A negated character class with a single character < 255 is coded as
5455 "not this character" (OP_NOT). This wasn't working properly when the test
5456 character was multibyte, either singly or repeated.
5457
5458 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5459 mode, for example: \x{100}{2,3}.
5460
5461 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5462 singly or repeated) now correctly test multibyte characters. However,
5463 PCRE doesn't recognize any characters with values greater than 255 as
5464 digits, spaces, or word characters. Such characters always match \D, \S,
5465 and \W, and never match \d, \s, or \w.
5466
5467 (v) Classes may now contain characters and character ranges with values
5468 greater than 255. For example: [ab\x{100}-\x{400}].
5469
5470 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5471 PCRE in UTF-8 mode.
5472
5473 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5474 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5475 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5476 value.)
5477
5478 53. The single man page has become too large. I have therefore split it up into
5479 a number of separate man pages. These also give rise to individual HTML pages;
5480 these are now put in a separate directory, and there is an index.html page that
5481 lists them all. Some hyperlinking between the pages has been installed.
5482
5483 54. Added convenience functions for handling named capturing parentheses.
5484
5485 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5486 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5487 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5488 are faulted.
5489
5490 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5491 calling configure. These values are used when compiling the dftables.c program
5492 which is run to generate the source of the default character tables. They
5493 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5494 you will need to set these values.
5495
5496 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5497
5498
5499 Version 3.9 02-Jan-02
5500 ---------------------
5501
5502 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5503
5504 2. If --disable-static was given, the building process failed when trying to
5505 build pcretest and pcregrep. (For some reason it was using libtool to compile
5506 them, which is not right, as they aren't part of the library.)
5507
5508
5509 Version 3.8 18-Dec-01
5510 ---------------------
5511
5512 1. The experimental UTF-8 code was completely screwed up. It was packing the
5513 bytes in the wrong order. How dumb can you get?
5514
5515
5516 Version 3.7 29-Oct-01
5517 ---------------------
5518
5519 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5520 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5521 this didn't happen under Solaris 8, where I normally test things.
5522
5523 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5524 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5525 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5526 this fix an hour or so after the initial 3.7 release.)
5527
5528
5529 Version 3.6 23-Oct-01
5530 ---------------------
5531
5532 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5533 offsets passed as NULL with zero offset count.
5534
5535 2. The config.guess and config.sub files had not been updated when I moved to
5536 the latest autoconf.
5537
5538
5539 Version 3.5 15-Aug-01
5540 ---------------------
5541
5542 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5543 had been forgotten.
5544
5545 2. By using declared but undefined structures, we can avoid using "void"
5546 definitions in pcre.h while keeping the internal definitions of the structures
5547 private.
5548
5549 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5550 user point of view, this means that both static and shared libraries are built
5551 by default, but this can be individually controlled. More of the work of
5552 handling this static/shared cases is now inside libtool instead of PCRE's make
5553 file.
5554
5555 4. The pcretest utility is now installed along with pcregrep because it is
5556 useful for users (to test regexs) and by doing this, it automatically gets
5557 relinked by libtool. The documentation has been turned into a man page, so
5558 there are now .1, .txt, and .html versions in /doc.
5559
5560 5. Upgrades to pcregrep:
5561 (i) Added long-form option names like gnu grep.
5562 (ii) Added --help to list all options with an explanatory phrase.
5563 (iii) Added -r, --recursive to recurse into sub-directories.
5564 (iv) Added -f, --file to read patterns from a file.
5565
5566 6. pcre_exec() was referring to its "code" argument before testing that
5567 argument for NULL (and giving an error if it was NULL).
5568
5569 7. Upgraded Makefile.in to allow for compiling in a different directory from
5570 the source directory.
5571
5572 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5573 options bits, the pointer it was passed was to an int instead of to an unsigned
5574 long int. This mattered only on 64-bit systems.
5575
5576 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5577 generated) instead of pcre.in, which it its source. Also made the same change
5578 in several of the .c files.
5579
5580 10. A new release of gcc defines printf() as a macro, which broke pcretest
5581 because it had an ifdef in the middle of a string argument for printf(). Fixed
5582 by using separate calls to printf().
5583
5584 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5585 script, to force use of CR or LF instead of \n in the source. On non-Unix
5586 systems, the value can be set in config.h.
5587
5588 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5589 absolute limit. Changed the text of the error message to make this clear, and
5590 likewise updated the man page.
5591
5592 13. The limit of 99 on the number of capturing subpatterns has been removed.
5593 The new limit is 65535, which I hope will not be a "real" limit.
5594
5595
5596 Version 3.4 22-Aug-00
5597 ---------------------
5598
5599 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5600
5601 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5602
5603
5604 Version 3.3 01-Aug-00
5605 ---------------------
5606
5607 1. If an octal character was given, but the value was greater than \377, it
5608 was not getting masked to the least significant bits, as documented. This could
5609 lead to crashes in some systems.
5610
5611 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5612 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5613
5614 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5615 These just pass their arguments on to (pcre_free)(), but they are provided
5616 because some uses of PCRE bind it to non-C systems that can call its functions,
5617 but cannot call free() or pcre_free() directly.
5618
5619 4. Add "make test" as a synonym for "make check". Corrected some comments in
5620 the Makefile.
5621
5622 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5623 Makefile.
5624
5625 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5626 command called pgrep for grepping around the active processes.
5627
5628 7. Added the beginnings of support for UTF-8 character strings.
5629
5630 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5631 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5632 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5633 out for the ar command.)
5634
5635
5636 Version 3.2 12-May-00
5637 ---------------------
5638
5639 This is purely a bug fixing release.
5640
5641 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5642 of ZA. This was just one example of several cases that could provoke this bug,
5643 which was introduced by change 9 of version 2.00. The code for breaking
5644 infinite loops after an iteration that matches an empty string was't working
5645 correctly.
5646
5647 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5648 when matched against abbab (for example). After matching an empty string, it
5649 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5650 caused it to match further down the string than it should.
5651
5652 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5653 was there because it doesn't seem to be needed, and it causes trouble on some
5654 systems, as it is not a Standard C header. It has been removed.
5655
5656 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5657 were reported on the Macintosh. The changes were from
5658
5659 while ((c = *(++ptr)) != 0 && c != '\n');
5660 to
5661 while ((c = *(++ptr)) != 0 && c != '\n') ;
5662
5663 Totally extraordinary, but if that's what it takes...
5664
5665 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5666 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5667 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5668 assumes the way PCRE uses memmove() (always moving upwards).
5669
5670 6. PCRE is being used in one environment where strchr() is not available. There
5671 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5672 faster code anyway.
5673
5674
5675 Version 3.1 09-Feb-00
5676 ---------------------
5677
5678 The only change in this release is the fixing of some bugs in Makefile.in for
5679 the "install" target:
5680
5681 (1) It was failing to install pcreposix.h.
5682
5683 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5684
5685
5686 Version 3.0 01-Feb-00
5687 ---------------------
5688
5689 1. Add support for the /+ modifier to perltest (to output $` like it does in
5690 pcretest).
5691
5692 2. Add support for the /g modifier to perltest.
5693
5694 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5695 matches null strings.
5696
5697 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5698 pattern. Perl treats empty patterns specially - it reuses the most recent
5699 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5700 effect.
5701
5702 5. The POSIX interface was broken in that it was just handing over the POSIX
5703 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5704 required a bigger vector, with some working space on the end. This means that
5705 the POSIX wrapper now has to get and free some memory, and copy the results.
5706
5707 6. Added some simple autoconf support, placing the test data and the
5708 documentation in separate directories, re-organizing some of the
5709 information files, and making it build pcre-config (a GNU standard). Also added
5710 libtool support for building PCRE as a shared library, which is now the
5711 default.
5712
5713 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5714 09 are not valid octal constants. Single digits will be used for minor values
5715 less than 10.
5716
5717 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5718 existing programs that set these in the POSIX interface can use PCRE without
5719 modification.
5720
5721 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5722 return all that pcre_info() returns, plus additional data. The pcre_info()
5723 function is retained for compatibility, but is considered to be obsolete.
5724
5725 10. Added experimental recursion feature (?R) to handle one common case that
5726 Perl 5.6 will be able to do with (?p{...}).
5727
5728 11. Added support for POSIX character classes like [:alpha:], which Perl is
5729 adopting.
5730
5731
5732 Version 2.08 31-Aug-99
5733 ----------------------
5734
5735 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5736 trying to match at the startoffset position, but instead was moving forward to
5737 the next newline as if a previous match had failed.
5738
5739 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5740 and could get into a loop if a null string was matched other than at the start
5741 of the subject.
5742
5743 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5744 be distinguished at compile time, and for completeness also added PCRE_DATE.
5745
5746 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5747 in GnuWin32 environments.
5748
5749
5750 Version 2.07 29-Jul-99
5751 ----------------------
5752
5753 1. The documentation is now supplied in plain text form and HTML as well as in
5754 the form of man page sources.
5755
5756 2. C++ compilers don't like assigning (void *) values to other pointer types.
5757 In particular this affects malloc(). Although there is no problem in Standard
5758 C, I've put in casts to keep C++ compilers happy.
5759
5760 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5761 should be (const char *).
5762
5763 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5764 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5765 However, I haven't made this a standard facility. The documentation doesn't
5766 mention it, and the Makefile doesn't support it.
5767
5768 5. The Makefile now contains an "install" target, with editable destinations at
5769 the top of the file. The pcretest program is not installed.
5770
5771 6. pgrep -V now gives the PCRE version number and date.
5772
5773 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5774 causing the entire string to be ignored, instead of just the last character.
5775
5776 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5777 non-matching string, it can take a very, very long time, even for strings of
5778 quite modest length, because of the nested recursion. PCRE now does better in
5779 some of these cases. It does this by remembering the last required literal
5780 character in the pattern, and pre-searching the subject to ensure it is present
5781 before running the real match. In other words, it applies a heuristic to detect
5782 some types of certain failure quickly, and in the above example, if presented
5783 with a string that has no trailing " it gives "no match" very quickly.
5784
5785 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5786 other alternatives are tried instead.
5787
5788
5789 Version 2.06 09-Jun-99
5790 ----------------------
5791
5792 1. Change pcretest's output for amount of store used to show just the code
5793 space, because the remainder (the data block) varies in size between 32-bit and
5794 64-bit systems.
5795
5796 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5797 start matching at. This allows lookbehinds to work when searching for multiple
5798 occurrences in a string.
5799
5800 3. Added additional options to pcretest for testing multiple occurrences:
5801
5802 /+ outputs the rest of the string that follows a match
5803 /g loops for multiple occurrences, using the new startoffset argument
5804 /G loops for multiple occurrences by passing an incremented pointer
5805
5806 4. PCRE wasn't doing the "first character" optimization for patterns starting
5807 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5808 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5809 the letter 'x'. On long subject strings, this gives a significant speed-up.
5810
5811
5812 Version 2.05 21-Apr-99
5813 ----------------------
5814
5815 1. Changed the type of magic_number from int to long int so that it works
5816 properly on 16-bit systems.
5817
5818 2. Fixed a bug which caused patterns starting with .* not to work correctly
5819 when the subject string contained newline characters. PCRE was assuming
5820 anchoring for such patterns in all cases, which is not correct because .* will
5821 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5822 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5823 must be retried after every newline in the subject.
5824
5825
5826 Version 2.04 18-Feb-99
5827 ----------------------
5828
5829 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5830 computation of the store needed to hold the pattern was incorrect (too large).
5831 If such patterns were nested a few deep, this could multiply and become a real
5832 problem.
5833
5834 2. Added /M option to pcretest to show the memory requirement of a specific
5835 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5836
5837 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5838 compiled in such a way that the backtracking after subsequent failure was
5839 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5840 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5841
5842
5843 Version 2.03 02-Feb-99
5844 ----------------------
5845
5846 1. Fixed typo and small mistake in man page.
5847
5848 2. Added 4th condition (GPL supersedes if conflict) and created separate
5849 LICENCE file containing the conditions.
5850
5851 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5852 Perl, that is the internal \ allows the delimiter to be included in the
5853 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5854 the final delimiter, add \ to the end of the pattern (to test the error).
5855
5856 4. Added the convenience functions for extracting substrings after a successful
5857 match. Updated pcretest to make it able to test these functions.
5858
5859
5860 Version 2.02 14-Jan-99
5861 ----------------------
5862
5863 1. Initialized the working variables associated with each extraction so that
5864 their saving and restoring doesn't refer to uninitialized store.
5865
5866 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5867 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5868 fix the problem.
5869
5870 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5871 calls, and wasn't printing the correct value for compiling calls. Increased the
5872 default value of LOOPREPEAT, and the number of significant figures in the
5873 times.
5874
5875 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5876
5877 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5878 a building problem on Windows NT with a FAT file system.
5879
5880
5881 Version 2.01 21-Oct-98
5882 ----------------------
5883
5884 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5885 to character tables built by pcre_maketables() in the current locale. If NULL
5886 is passed, the default tables are used.
5887
5888
5889 Version 2.00 24-Sep-98
5890 ----------------------
5891
5892 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5893 it any more.
5894
5895 2. Allow quantification of (?>) groups, and make it work correctly.
5896
5897 3. The first character computation wasn't working for (?>) groups.
5898
5899 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5900 end of the subject) and add 5.005's \z, which really does match only at the
5901 very end of the subject.
5902
5903 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5904
5905 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5906 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5907 localized options. All options to pcre_study() were also removed.
5908
5909 7. Add other new features from 5.005:
5910
5911 $(?<= positive lookbehind
5912 $(?<! negative lookbehind
5913 (?imsx-imsx) added the unsetting capability
5914 such a setting is global if at outer level; local otherwise
5915 (?imsx-imsx:) non-capturing groups with option setting
5916 (?(cond)re|re) conditional pattern matching
5917
5918 A backreference to itself in a repeated group matches the previous
5919 captured string.
5920
5921 8. General tidying up of studying (both automatic and via "study")
5922 consequential on the addition of new assertions.
5923
5924 9. As in 5.005, unlimited repeated groups that could match an empty substring
5925 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5926 runtime if any iteration does actually match an empty substring.
5927
5928 10. Include the RunTest script in the distribution.
5929
5930 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5931 discrepancies, some of which were old and were also with respect to 5.004. They
5932 have now been fixed.
5933
5934
5935 Version 1.09 28-Apr-98
5936 ----------------------
5937
5938 1. A negated single character class followed by a quantifier with a minimum
5939 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5940 program crashes, or just wrong answers. This did not apply to negated classes
5941 containing more than one character, or to minima other than one.
5942
5943
5944 Version 1.08 27-Mar-98
5945 ----------------------
5946
5947 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5948
5949 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5950 latter must appear before anything that relies on it in the pattern.
5951
5952
5953 Version 1.07 16-Feb-98
5954 ----------------------
5955
5956 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5957 repeat of a potentially empty string).
5958
5959
5960 Version 1.06 23-Jan-98
5961 ----------------------
5962
5963 1. Added Markus Oberhumer's little patches for C++.
5964
5965 2. Literal strings longer than 255 characters were broken.
5966
5967
5968 Version 1.05 23-Dec-97
5969 ----------------------
5970
5971 1. Negated character classes containing more than one character were failing if
5972 PCRE_CASELESS was set at run time.
5973
5974
5975 Version 1.04 19-Dec-97
5976 ----------------------
5977
5978 1. Corrected the man page, where some "const" qualifiers had been omitted.
5979
5980 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5981 input syntax.
5982
5983 3. Fixed memory leak which occurred when a regex with back references was
5984 matched with an offsets vector that wasn't big enough. The temporary memory
5985 that is used in this case wasn't being freed if the match failed.
5986
5987 4. Tidied pcretest to ensure it frees memory that it gets.
5988
5989 5. Temporary memory was being obtained in the case where the passed offsets
5990 vector was exactly big enough.
5991
5992 6. Corrected definition of offsetof() from change 5 below.
5993
5994 7. I had screwed up change 6 below and broken the rules for the use of
5995 setjmp(). Now fixed.
5996
5997
5998 Version 1.03 18-Dec-97
5999 ----------------------
6000
6001 1. A erroneous regex with a missing opening parenthesis was correctly
6002 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
6003 on some systems.
6004
6005 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
6006 it was reported that one broken compiler failed on the former because "code" is
6007 also an independent variable.
6008
6009 3. The erroneous regex a[]b caused an array overrun reference.
6010
6011 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
6012 fail on data ending with that character. (It was going on too far, and checking
6013 the next character, typically a binary zero.) This was specific to the
6014 optimized code for single-character negative classes.
6015
6016 5. Added a contributed patch from the TIN world which does the following:
6017
6018 + Add an undef for memmove, in case the the system defines a macro for it.
6019
6020 + Add a definition of offsetof(), in case there isn't one. (I don't know
6021 the reason behind this - offsetof() is part of the ANSI standard - but
6022 it does no harm).
6023
6024 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
6025 most of the places where whitespace preceded '#'. I have given up and
6026 allowed the remaining 2 cases to be at the margin.
6027
6028 + Rename some variables in pcre to eliminate shadowing. This seems very
6029 pedantic, but does no harm, of course.
6030
6031 6. Moved the call to setjmp() into its own function, to get rid of warnings
6032 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
6033
6034 7. Constructs such as \d{8,} were compiling into the equivalent of
6035 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
6036 outcome, but in this particular case used more store than had been allocated,
6037 which caused the bug to be discovered because it threw up an internal error.
6038
6039 8. The debugging code in both pcre and pcretest for outputting the compiled
6040 form of a regex was going wrong in the case of back references followed by
6041 curly-bracketed repeats.
6042
6043
6044 Version 1.02 12-Dec-97
6045 ----------------------
6046
6047 1. Typos in pcre.3 and comments in the source fixed.
6048
6049 2. Applied a contributed patch to get rid of places where it used to remove
6050 'const' from variables, and fixed some signed/unsigned and uninitialized
6051 variable warnings.
6052
6053 3. Added the "runtest" target to Makefile.
6054