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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1758 - (show annotations)
Fri Nov 29 11:03:43 2019 UTC (10 days, 7 hours ago) by zherczeg
File size: 298744 byte(s)
Error occurred while calculating annotation data.
Fix ARMv5 JIT improper handling of labels right after a constant pool.
1 ChangeLog for PCRE
2 ------------------
3
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
6
7
8 Version 8.44-RC1 01 March-2019
9 ------------------------------
10
11 1. Setting --enable-jit=auto for an out-of-tree build failed because the
12 source directory wasn't in the search path for AC_TRY_COMPILE always. Patch
13 from Ross Burton.
14
15 2. Applied a patch from Michael Shigorin to fix 8.43 build on e2k arch
16 with lcc compiler (EDG frontend based); the problem it fixes is:
17
18 lcc: "pcrecpp.cc", line 74: error: declaration aliased to undefined entity
19 "_ZN7pcrecpp2RE6no_argE" [-Werror]
20
21 3. Change 2 for 8.43 omitted (*LF) from the list of start-of-pattern items. Now
22 added.
23
24 4. Fix ARMv5 JIT improper handling of labels right after a constant pool.
25
26
27 Version 8.43 23-February-2019
28 -----------------------------
29
30 1. Some time ago the config macro SUPPORT_UTF8 was changed to SUPPORT_UTF
31 because it also applies to UTF-16 and UTF-32. However, this change was not made
32 in the pcre2cpp files; consequently the C++ wrapper has from then been compiled
33 with a bug in it, which would have been picked up by the unit test except that
34 it also had its UTF8 code cut out. The bug was in a global replace when moving
35 forward after matching an empty string.
36
37 2. The C++ wrapper got broken a long time ago (version 7.3, August 2007) when
38 (*CR) was invented (assuming it was the first such start-of-pattern option).
39 The wrapper could never handle such patterns because it wraps patterns in
40 (?:...)\z in order to support end anchoring. I have hacked in some code to fix
41 this, that is, move the wrapping till after any existing start-of-pattern
42 special settings.
43
44 3. "pcre2grep" (sic) was accidentally mentioned in an error message (fix was
45 ported from PCRE2).
46
47 4. Typo LCC_ALL for LC_ALL fixed in pcregrep.
48
49 5. In a pattern such as /[^\x{100}-\x{ffff}]*[\x80-\xff]/ which has a repeated
50 negative class with no characters less than 0x100 followed by a positive class
51 with only characters less than 0x100, the first class was incorrectly being
52 auto-possessified, causing incorrect match failures.
53
54 6. If the only branch in a conditional subpattern was anchored, the whole
55 subpattern was treated as anchored, when it should not have been, since the
56 assumed empty second branch cannot be anchored. Demonstrated by test patterns
57 such as /(?(1)^())b/ or /(?(?=^))b/.
58
59 7. Fix subject buffer overread in JIT when UTF is disabled and \X or \R has
60 a greater than 1 fixed quantifier. This issue was found by Yunho Kim.
61
62 8. If a pattern started with a subroutine call that had a quantifier with a
63 minimum of zero, an incorrect "match must start with this character" could be
64 recorded. Example: /(?&xxx)*ABC(?<xxx>XYZ)/ would (incorrectly) expect 'A' to
65 be the first character of a match.
66
67 9. Improve MAP_JIT flag usage on MacOS. Patch by Rich Siegel.
68
69
70 Version 8.42 20-March-2018
71 --------------------------
72
73 1. Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
74
75 2. Fixed outdated real_pcre definitions in pcre.h.in (patch by Evgeny Kotkov).
76
77 3. pcregrep was truncating components of file names to 128 characters when
78 processing files with the -r option, and also (some very odd code) truncating
79 path names to 512 characters. There is now a check on the absolute length of
80 full path file names, which may be up to 2047 characters long.
81
82 4. Using pcre_dfa_exec(), in UTF mode when UCP support was not defined, there
83 was the possibility of a false positive match when caselessly matching a "not
84 this character" item such as [^\x{1234}] (with a code point greater than 127)
85 because the "other case" variable was not being initialized.
86
87 5. Although pcre_jit_exec checks whether the pattern is compiled
88 in a given mode, it was also expected that at least one mode is available.
89 This is fixed and pcre_jit_exec returns with PCRE_ERROR_JIT_BADOPTION
90 when the pattern is not optimized by JIT at all.
91
92 6. The line number and related variables such as match counts in pcregrep
93 were all int variables, causing overflow when files with more than 2147483647
94 lines were processed (assuming 32-bit ints). They have all been changed to
95 unsigned long ints.
96
97 7. If a backreference with a minimum repeat count of zero was first in a
98 pattern, apart from assertions, an incorrect first matching character could be
99 recorded. For example, for the pattern /(?=(a))\1?b/, "b" was incorrectly set
100 as the first character of a match.
101
102 8. Fix out-of-bounds read for partial matching of /./ against an empty string
103 when the newline type is CRLF.
104
105 9. When matching using the the REG_STARTEND feature of the POSIX API with a
106 non-zero starting offset, unset capturing groups with lower numbers than a
107 group that did capture something were not being correctly returned as "unset"
108 (that is, with offset values of -1).
109
110 10. Matching the pattern /(*UTF)\C[^\v]+\x80/ against an 8-bit string
111 containing multi-code-unit characters caused bad behaviour and possibly a
112 crash. This issue was fixed for other kinds of repeat in release 8.37 by change
113 38, but repeating character classes were overlooked.
114
115 11. A small fix to pcregrep to avoid compiler warnings for -Wformat-overflow=2.
116
117 12. Added --enable-jit=auto support to configure.ac.
118
119 13. Fix misleading error message in configure.ac.
120
121
122 Version 8.41 05-July-2017
123 -------------------------
124
125 1. Fixed typo in CMakeLists.txt (wrong number of arguments for
126 PCRE_STATIC_RUNTIME (affects MSVC only).
127
128 2. Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
129 mode with --only-matching matched several lines, it restarted scanning at the
130 next line instead of moving on to the end of the matched string, which can be
131 several lines after the start.
132
133 3. Fix a missing else in the JIT compiler reported by 'idaifish'.
134
135 4. A (?# style comment is now ignored between a basic quantifier and a
136 following '+' or '?' (example: /X+(?#comment)?Y/.
137
138 5. Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
139 Pisar).
140
141 6. Fuzzers have reported issues in pcretest. These are NOT serious (it is,
142 after all, just a test program). However, to stop the reports, some easy ones
143 are fixed:
144
145 (a) Check for values < 256 when calling isprint() in pcretest.
146 (b) Give an error for too big a number after \O.
147
148 7. In the 32-bit library in non-UTF mode, an attempt to find a Unicode
149 property for a character with a code point greater than 0x10ffff (the Unicode
150 maximum) caused a crash.
151
152 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
153 encountered a character class with a possessive repeat, for example [a-f]{3}+.
154
155 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
156 length incorrectly, which could result in buffer overflow.
157
158 10. Remove redundant line of code (accidentally left in ages ago).
159
160 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
161 with namespace pcrecpp (Bugzilla #2084).
162
163 12. Remove a duplication typo in pcre_tables.c.
164
165 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
166 starting offset greater than zero.
167
168
169 Version 8.40 11-January-2017
170 ----------------------------
171
172 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
173 the match extended over a line boundary.
174
175 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
176 MSVC static compilation, putting the first patch under a new option.
177
178 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
179
180 4. Ignore "show all captures" (/=) for DFA matching.
181
182 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
183
184 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
185 without PCRE_UCP set, a negative character type such as \D in a positive
186 class should cause all characters greater than 255 to match, whatever else
187 is in the class. There was a bug that caused this not to happen if a
188 Unicode property item was added to such a class, for example [\D\P{Nd}] or
189 [\W\pL].
190
191 7. When pcretest was outputing information from a callout, the caret indicator
192 for the current position in the subject line was incorrect if it was after
193 an escape sequence for a character whose code point was greater than
194 \x{ff}.
195
196 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
197 the conditional was interpreted as a reference to capturing group 1 instead
198 of a test for recursion. Any group whose name began with R was
199 misinterpreted in this way. (The reference interpretation should only
200 happen if the group's name is precisely "R".)
201
202 9. A number of bugs have been mended relating to match start-up optimizations
203 when the first thing in a pattern is a positive lookahead. These all
204 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
205
206 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
207 both an initial 'X' and a following 'X'.
208 (b) Some patterns starting with an assertion that started with .* were
209 incorrectly optimized as having to match at the start of the subject or
210 after a newline. There are cases where this is not true, for example,
211 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
212 start with spaces. Starting .* in an assertion is no longer taken as an
213 indication of matching at the start (or after a newline).
214
215
216 Version 8.39 14-June-2016
217 -------------------------
218
219 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
220 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
221 misbehaved. This bug was found by the LLVM fuzzer.
222
223 2. Similar to the above, if an isolated \E was present between an item and its
224 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
225 bug was found by the LLVM fuzzer.
226
227 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
228 working correctly in UCP mode.
229
230 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
231 was set when the pmatch argument was NULL. It now returns REG_INVARG.
232
233 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
234
235 6. An empty \Q\E sequence between an item and its qualifier caused
236 pcre_compile() to misbehave when auto callouts were enabled. This bug was
237 found by the LLVM fuzzer.
238
239 7. If a pattern that was compiled with PCRE_EXTENDED started with white
240 space or a #-type comment that was followed by (?-x), which turns off
241 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
242 pcre_compile() assumed that (?-x) applied to the whole pattern and
243 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
244
245 8. A call of pcre_copy_named_substring() for a named substring whose number
246 was greater than the space in the ovector could cause a crash.
247
248 9. Yet another buffer overflow bug involved duplicate named groups with a
249 group that reset capture numbers (compare 8.38/7 below). Once again, I have
250 just allowed for more memory, even if not needed. (A proper fix is
251 implemented in PCRE2, but it involves a lot of refactoring.)
252
253 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
254 start of the match to be earlier than the end.
255
256 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
257
258 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
259 assertion, caused pcretest to generate incorrect output, and also to read
260 uninitialized memory (detected by ASAN or valgrind).
261
262 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
263 nested set of parentheses of sufficient size caused an overflow of the
264 compiling workspace (which was diagnosed, but of course is not desirable).
265
266 14. And yet another buffer overflow bug involving duplicate named groups, this
267 time nested, with a nested back reference. Yet again, I have just allowed
268 for more memory, because anything more needs all the refactoring that has
269 been done for PCRE2. An example pattern that provoked this bug is:
270 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
271 registered as CVE-2016-1283.
272
273 15. pcretest went into a loop if global matching was requested with an ovector
274 size less than 2. It now gives an error message. This bug was found by
275 afl-fuzz.
276
277 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
278 ("assertion expected") when (?(?C) was not followed by an opening
279 parenthesis.
280
281 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
282 actually affect anything, by sheer luck.
283
284 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
285 static compilation.
286
287 19. Modified the RunTest script to incorporate a valgrind suppressions file so
288 that certain errors, provoked by the SSE2 instruction set when JIT is used,
289 are ignored.
290
291 20. A racing condition is fixed in JIT reported by Mozilla.
292
293 21. Minor code refactor to avoid "array subscript is below array bounds"
294 compiler warning.
295
296 22. Minor code refactor to avoid "left shift of negative number" warning.
297
298 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
299 UCP support.
300
301 24. Refactor to avoid compiler warnings in pcrecpp.cc.
302
303 25. Refactor to fix a typo in pcre_jit_test.c
304
305 26. Patch to support compiling pcrecpp.cc with Intel compiler.
306
307
308 Version 8.38 23-November-2015
309 -----------------------------
310
311 1. If a group that contained a recursive back reference also contained a
312 forward reference subroutine call followed by a non-forward-reference
313 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
314 compile correct code, leading to undefined behaviour or an internally
315 detected error. This bug was discovered by the LLVM fuzzer.
316
317 2. Quantification of certain items (e.g. atomic back references) could cause
318 incorrect code to be compiled when recursive forward references were
319 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
320 This bug was discovered by the LLVM fuzzer.
321
322 3. A repeated conditional group whose condition was a reference by name caused
323 a buffer overflow if there was more than one group with the given name.
324 This bug was discovered by the LLVM fuzzer.
325
326 4. A recursive back reference by name within a group that had the same name as
327 another group caused a buffer overflow. For example:
328 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
329
330 5. A forward reference by name to a group whose number is the same as the
331 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
332 a buffer overflow at compile time. This bug was discovered by the LLVM
333 fuzzer.
334
335 6. A lookbehind assertion within a set of mutually recursive subpatterns could
336 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
337
338 7. Another buffer overflow bug involved duplicate named groups with a
339 reference between their definition, with a group that reset capture
340 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
341 fixed by always allowing for more memory, even if not needed. (A proper fix
342 is implemented in PCRE2, but it involves more refactoring.)
343
344 8. There was no check for integer overflow in subroutine calls such as (?123).
345
346 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
347 being treated as a literal 'l' instead of causing an error.
348
349 10. There was a buffer overflow if pcre_exec() was called with an ovector of
350 size 1. This bug was found by american fuzzy lop.
351
352 11. If a non-capturing group containing a conditional group that could match
353 an empty string was repeated, it was not identified as matching an empty
354 string itself. For example: /^(?:(?(1)x|)+)+$()/.
355
356 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
357 \a and \e in test subject lines.
358
359 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
360 instead of the EBCDIC value.
361
362 14. The handling of \c in an EBCDIC environment has been revised so that it is
363 now compatible with the specification in Perl's perlebcdic page.
364
365 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
366 ASCII/Unicode. This has now been added to the list of characters that are
367 recognized as white space in EBCDIC.
368
369 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
370 error (correctly) when used outside a class, but did not give an error
371 within a class.
372
373 17. \h within a class was incorrectly compiled in EBCDIC environments.
374
375 18. A pattern with an unmatched closing parenthesis that contained a backward
376 assertion which itself contained a forward reference caused buffer
377 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
378
379 19. JIT should return with error when the compiled pattern requires more stack
380 space than the maximum.
381
382 20. A possessively repeated conditional group that could match an empty string,
383 for example, /(?(R))*+/, was incorrectly compiled.
384
385 21. Fix infinite recursion in the JIT compiler when certain patterns such as
386 /(?:|a|){100}x/ are analysed.
387
388 22. Some patterns with character classes involving [: and \\ were incorrectly
389 compiled and could cause reading from uninitialized memory or an incorrect
390 error diagnosis.
391
392 23. Pathological patterns containing many nested occurrences of [: caused
393 pcre_compile() to run for a very long time.
394
395 24. A conditional group with only one branch has an implicit empty alternative
396 branch and must therefore be treated as potentially matching an empty
397 string.
398
399 25. If (?R was followed by - or + incorrect behaviour happened instead of a
400 diagnostic.
401
402 26. Arrange to give up on finding the minimum matching length for overly
403 complex patterns.
404
405 27. Similar to (4) above: in a pattern with duplicated named groups and an
406 occurrence of (?| it is possible for an apparently non-recursive back
407 reference to become recursive if a later named group with the relevant
408 number is encountered. This could lead to a buffer overflow. Wen Guanxing
409 from Venustech ADLAB discovered this bug.
410
411 28. If pcregrep was given the -q option with -c or -l, or when handling a
412 binary file, it incorrectly wrote output to stdout.
413
414 29. The JIT compiler did not restore the control verb head in case of *THEN
415 control verbs. This issue was found by Karl Skomski with a custom LLVM
416 fuzzer.
417
418 30. Error messages for syntax errors following \g and \k were giving inaccurate
419 offsets in the pattern.
420
421 31. Added a check for integer overflow in conditions (?(<digits>) and
422 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
423 fuzzer.
424
425 32. Handling recursive references such as (?2) when the reference is to a group
426 later in the pattern uses code that is very hacked about and error-prone.
427 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
428 give an internal error if it is obvious that compiling has gone wrong.
429
430 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
431 This issue was found by Karl Skomski with a custom LLVM fuzzer.
432
433 34. The JIT compiler should restore the control chain for empty possessive
434 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
435
436 35. Match limit check added to JIT recursion. This issue was found by Karl
437 Skomski with a custom LLVM fuzzer.
438
439 36. Yet another case similar to 27 above has been circumvented by an
440 unconditional allocation of extra memory. This issue is fixed "properly" in
441 PCRE2 by refactoring the way references are handled. Wen Guanxing
442 from Venustech ADLAB discovered this bug.
443
444 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
445 with a custom LLVM fuzzer.
446
447 38. Fixed a corner case of range optimization in JIT.
448
449 39. An incorrect error "overran compiling workspace" was given if there were
450 exactly enough group forward references such that the last one extended
451 into the workspace safety margin. The next one would have expanded the
452 workspace. The test for overflow was not including the safety margin.
453
454 40. A match limit issue is fixed in JIT which was found by Karl Skomski
455 with a custom LLVM fuzzer.
456
457 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
458 work under Windows. (Why has it taken so long for anyone to notice?)
459
460 42. In a character class such as [\W\p{Any}] where both a negative-type escape
461 ("not a word character") and a property escape were present, the property
462 escape was being ignored.
463
464 43. Fix crash caused by very long (*MARK) or (*THEN) names.
465
466 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
467 by a single ASCII character in a class item, was incorrectly compiled in
468 UCP mode. The POSIX class got lost, but only if the single character
469 followed it.
470
471 45. [:punct:] in UCP mode was matching some characters in the range 128-255
472 that should not have been matched.
473
474 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
475 class, all characters with code points greater than 255 are in the class.
476 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
477 such as \w are turned into Unicode properties), wide characters were not
478 correctly handled, and could fail to match.
479
480
481 Version 8.37 28-April-2015
482 --------------------------
483
484 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
485 for those parentheses to be closed with whatever has been captured so far.
486 However, it was failing to mark any other groups between the hightest
487 capture so far and the currrent group as "unset". Thus, the ovector for
488 those groups contained whatever was previously there. An example is the
489 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
490
491 2. If an assertion condition was quantified with a minimum of zero (an odd
492 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
493
494 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
495 unrecognized modifier, a crash could occur.
496
497 4. An attempt to do global matching in pcretest with a zero-length ovector
498 caused a crash.
499
500 5. Fixed a memory leak during matching that could occur for a subpattern
501 subroutine call (recursive or otherwise) if the number of captured groups
502 that had to be saved was greater than ten.
503
504 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
505 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
506 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
507
508 7. A UTF pattern containing a "not" match of a non-ASCII character and a
509 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
510
511 8. When a pattern is compiled, it remembers the highest back reference so that
512 when matching, if the ovector is too small, extra memory can be obtained to
513 use instead. A conditional subpattern whose condition is a check on a
514 capture having happened, such as, for example in the pattern
515 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
516 setting the highest backreference number. This mattered only if pcre_exec()
517 was called with an ovector that was too small to hold the capture, and there
518 was no other kind of back reference (a situation which is probably quite
519 rare). The effect of the bug was that the condition was always treated as
520 FALSE when the capture could not be consulted, leading to a incorrect
521 behaviour by pcre_exec(). This bug has been fixed.
522
523 9. A reference to a duplicated named group (either a back reference or a test
524 for being set in a conditional) that occurred in a part of the pattern where
525 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
526 to be incorrectly calculated, leading to overwriting.
527
528 10. A mutually recursive set of back references such as (\2)(\1) caused a
529 segfault at study time (while trying to find the minimum matching length).
530 The infinite loop is now broken (with the minimum length unset, that is,
531 zero).
532
533 11. If an assertion that was used as a condition was quantified with a minimum
534 of zero, matching went wrong. In particular, if the whole group had
535 unlimited repetition and could match an empty string, a segfault was
536 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
537 assertions to be quantified, but not if they are being used as conditions,
538 so the above pattern is faulted by Perl. PCRE has now been changed so that
539 it also rejects such patterns.
540
541 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
542 failed to allow the zero-repeat case if pcre2_exec() was called with an
543 ovector too small to capture the group.
544
545 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
546 Red Hat Product Security:
547
548 (a) A crash if /K and /F were both set with the option to save the compiled
549 pattern.
550
551 (b) Another crash if the option to print captured substrings in a callout
552 was combined with setting a null ovector, for example \O\C+ as a subject
553 string.
554
555 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
556 forward reference repeated a large (but limited) number of times within a
557 repeated outer group that has a zero minimum quantifier, caused incorrect
558 code to be compiled, leading to the error "internal error:
559 previously-checked referenced subpattern not found" when an incorrect
560 memory address was read. This bug was reported as "heap overflow",
561 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
562 CVE-2015-2325.
563
564 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
565 call within a group that also contained a recursive back reference caused
566 incorrect code to be compiled. This bug was reported as "heap overflow",
567 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
568 number CVE-2015-2326.
569
570 24. Computing the size of the JIT read-only data in advance has been a source
571 of various issues, and new ones are still appear unfortunately. To fix
572 existing and future issues, size computation is eliminated from the code,
573 and replaced by on-demand memory allocation.
574
575 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
576 adjacent to the end of the range, and the range contained characters with
577 more than one other case, caused incorrect behaviour when compiled in UTF
578 mode. In that example, the range a-j was left out of the class.
579
580 26. Fix JIT compilation of conditional blocks, which assertion
581 is converted to (*FAIL). E.g: /(?(?!))/.
582
583 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
584 discovered by the LLVM fuzzer.
585
586 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
587 when this assertion was used as a condition, for example (?(?!)a|b). In
588 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
589 error about an unsupported item.
590
591 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
592 possessification code could take exponential time to complete. A recursion
593 depth limit of 1000 has been imposed to limit the resources used by this
594 optimization.
595
596 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
597 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
598 because \S ensures they are all in the class. The code for doing this was
599 interacting badly with the code for computing the amount of space needed to
600 compile the pattern, leading to a buffer overflow. This bug was discovered
601 by the LLVM fuzzer.
602
603 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
604 other kinds of group caused stack overflow at compile time. This bug was
605 discovered by the LLVM fuzzer.
606
607 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
608 between a subroutine call and its quantifier was incorrectly compiled,
609 leading to buffer overflow or other errors. This bug was discovered by the
610 LLVM fuzzer.
611
612 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
613 assertion after (?(. The code was failing to check the character after
614 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
615 was discovered by the LLVM fuzzer.
616
617 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
618 a fixed maximum following a group that contains a subroutine reference was
619 incorrectly compiled and could trigger buffer overflow. This bug was
620 discovered by the LLVM fuzzer.
621
622 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
623 caused a stack overflow instead of the diagnosis of a non-fixed length
624 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
625
626 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
627 (e.g. /(?<=\Ka)/) could make pcregrep loop.
628
629 37. There was a similar problem to 36 in pcretest for global matches.
630
631 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
632 and a subsequent item in the pattern caused a non-match, backtracking over
633 the repeated \X did not stop, but carried on past the start of the subject,
634 causing reference to random memory and/or a segfault. There were also some
635 other cases where backtracking after \C could crash. This set of bugs was
636 discovered by the LLVM fuzzer.
637
638 39. The function for finding the minimum length of a matching string could take
639 a very long time if mutual recursion was present many times in a pattern,
640 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
641 method has been implemented. This infelicity was discovered by the LLVM
642 fuzzer.
643
644 40. Static linking against the PCRE library using the pkg-config module was
645 failing on missing pthread symbols.
646
647
648 Version 8.36 26-September-2014
649 ------------------------------
650
651 1. Got rid of some compiler warnings in the C++ modules that were shown up by
652 -Wmissing-field-initializers and -Wunused-parameter.
653
654 2. The tests for quantifiers being too big (greater than 65535) were being
655 applied after reading the number, and stupidly assuming that integer
656 overflow would give a negative number. The tests are now applied as the
657 numbers are read.
658
659 3. Tidy code in pcre_exec.c where two branches that used to be different are
660 now the same.
661
662 4. The JIT compiler did not generate match limit checks for certain
663 bracketed expressions with quantifiers. This may lead to exponential
664 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
665 issue should be resolved now.
666
667 5. Fixed an issue, which occures when nested alternatives are optimized
668 with table jumps.
669
670 6. Inserted two casts and changed some ints to size_t in the light of some
671 reported 64-bit compiler warnings (Bugzilla 1477).
672
673 7. Fixed a bug concerned with zero-minimum possessive groups that could match
674 an empty string, which sometimes were behaving incorrectly in the
675 interpreter (though correctly in the JIT matcher). This pcretest input is
676 an example:
677
678 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
679 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
680
681 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
682 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
683 for an empty string was breaking the inner loop and carrying on at a lower
684 level, when possessive repeated groups should always return to a higher
685 level as they have no backtrack points in them. The empty string test now
686 occurs at the outer level.
687
688 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
689 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
690
691 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
692 doesn't).
693
694 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
695 to hold a pattern to be incorrectly computed (too small) when there were
696 named back references to duplicated names. This could cause "internal
697 error: code overflow" or "double free or corruption" or other memory
698 handling errors.
699
700 11. When named subpatterns had the same prefixes, back references could be
701 confused. For example, in this pattern:
702
703 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
704
705 the reference to 'Name' was incorrectly treated as a reference to a
706 duplicate name.
707
708 12. A pattern such as /^s?c/mi8 where the optional character has more than
709 one "other case" was incorrectly compiled such that it would only try to
710 match starting at "c".
711
712 13. When a pattern starting with \s was studied, VT was not included in the
713 list of possible starting characters; this should have been part of the
714 8.34/18 patch.
715
716 14. If a character class started [\Qx]... where x is any character, the class
717 was incorrectly terminated at the ].
718
719 15. If a pattern that started with a caseless match for a character with more
720 than one "other case" was studied, PCRE did not set up the starting code
721 unit bit map for the list of possible characters. Now it does. This is an
722 optimization improvement, not a bug fix.
723
724 16. The Unicode data tables have been updated to Unicode 7.0.0.
725
726 17. Fixed a number of memory leaks in pcregrep.
727
728 18. Avoid a compiler warning (from some compilers) for a function call with
729 a cast that removes "const" from an lvalue by using an intermediate
730 variable (to which the compiler does not object).
731
732 19. Incorrect code was compiled if a group that contained an internal recursive
733 back reference was optional (had quantifier with a minimum of zero). This
734 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
735 caused segmentation faults because of stack overflows at compile time.
736
737 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
738 group that is quantified with an indefinite repeat, caused a compile-time
739 loop which used up all the system stack and provoked a segmentation fault.
740 This was not the same bug as 19 above.
741
742 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
743 Patch by Mike Frysinger.
744
745
746 Version 8.35 04-April-2014
747 --------------------------
748
749 1. A new flag is set, when property checks are present in an XCLASS.
750 When this flag is not set, PCRE can perform certain optimizations
751 such as studying these XCLASS-es.
752
753 2. The auto-possessification of character sets were improved: a normal
754 and an extended character set can be compared now. Furthermore
755 the JIT compiler optimizes more character set checks.
756
757 3. Got rid of some compiler warnings for potentially uninitialized variables
758 that show up only when compiled with -O2.
759
760 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
761 of a match later then the end of the match. The pcretest program was not
762 handling the case sensibly - it was outputting from the start to the next
763 binary zero. It now reports this situation in a message, and outputs the
764 text from the end to the start.
765
766 5. Fast forward search is improved in JIT. Instead of the first three
767 characters, any three characters with fixed position can be searched.
768 Search order: first, last, middle.
769
770 6. Improve character range checks in JIT. Characters are read by an inprecise
771 function now, which returns with an unknown value if the character code is
772 above a certain threshold (e.g: 256). The only limitation is that the value
773 must be bigger than the threshold as well. This function is useful when
774 the characters above the threshold are handled in the same way.
775
776 7. The macros whose names start with RAWUCHAR are placeholders for a future
777 mode in which only the bottom 21 bits of 32-bit data items are used. To
778 make this more memorable for those maintaining the code, the names have
779 been changed to start with UCHAR21, and an extensive comment has been added
780 to their definition.
781
782 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
783 to the export list in Makefile.am (they were accidentally omitted from the
784 8.34 tarball).
785
786 9. The informational output from pcretest used the phrase "starting byte set"
787 which is inappropriate for the 16-bit and 32-bit libraries. As the output
788 for "first char" and "need char" really means "non-UTF-char", I've changed
789 "byte" to "char", and slightly reworded the output. The documentation about
790 these values has also been (I hope) clarified.
791
792 10. Another JIT related optimization: use table jumps for selecting the correct
793 backtracking path, when more than four alternatives are present inside a
794 bracket.
795
796 11. Empty match is not possible, when the minimum length is greater than zero,
797 and there is no \K in the pattern. JIT should avoid empty match checks in
798 such cases.
799
800 12. In a caseless character class with UCP support, when a character with more
801 than one alternative case was not the first character of a range, not all
802 the alternative cases were added to the class. For example, s and \x{17f}
803 are both alternative cases for S: the class [RST] was handled correctly,
804 but [R-T] was not.
805
806 13. The configure.ac file always checked for pthread support when JIT was
807 enabled. This is not used in Windows, so I have put this test inside a
808 check for the presence of windows.h (which was already tested for).
809
810 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
811 The algorithm provides a way to skip certain starting offsets, and usually
812 faster than linear prefix searches.
813
814 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
815 as for 'fr_FR' and 'french'. For some reason, however, it then used the
816 Windows-specific input and output files, which have 'french' screwed in.
817 So this could never have worked. One of the problems with locales is that
818 they aren't always the same. I have now updated RunTest so that it checks
819 the output of the locale test (test 3) against three different output
820 files, and it allows the test to pass if any one of them matches. With luck
821 this should make the test pass on some versions of Solaris where it was
822 failing. Because of the uncertainty, the script did not used to stop if
823 test 3 failed; it now does. If further versions of a French locale ever
824 come to light, they can now easily be added.
825
826 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
827 there was a message during ./configure, but it did not stop. This now
828 provokes an error. The invalid example in README has been corrected.
829 If a value less than the minimum is given, the minimum value has always
830 been used, but now a warning is given.
831
832 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
833 was a bug in the test system, which is now fixed. Also, the list of various
834 configurations that are tested for each release did not have one with both
835 16/32 bits and --enable-bar-anycrlf. It now does.
836
837 18. pcretest was missing "-C bsr" for displaying the \R default setting.
838
839 19. Little endian PowerPC systems are supported now by the JIT compiler.
840
841 20. The fast forward newline mechanism could enter to an infinite loop on
842 certain invalid UTF-8 input. Although we don't support these cases
843 this issue can be fixed by a performance optimization.
844
845 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
846 not take account if existing stack usage. There is now a new global
847 variable called pcre_stack_guard that can be set to point to an external
848 function to check stack availability. It is called at the start of
849 processing every parenthesized group.
850
851 22. A typo in the code meant that in ungreedy mode the max/min qualifier
852 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
853 not match "ab".
854
855 23. When UTF was disabled, the JIT program reported some incorrect compile
856 errors. These messages are silenced now.
857
858 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
859 compiler.
860
861 25. Change all the temporary files used in RunGrepTest to be different to those
862 used by RunTest so that the tests can be run simultaneously, for example by
863 "make -j check".
864
865
866 Version 8.34 15-December-2013
867 -----------------------------
868
869 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
870 executable memory. Patch inspired by Carsten Klein.
871
872 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
873 this macro is never tested and has no effect, because the work to support
874 coverage involves only compiling and linking options and special targets in
875 the Makefile. The comment in config.h implied that defining the macro would
876 enable coverage support, which is totally false. There was also support for
877 setting this macro in the CMake files (my fault, I just copied it from
878 configure). SUPPORT_GCOV has now been removed.
879
880 3. Make a small performance improvement in strlen16() and strlen32() in
881 pcretest.
882
883 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
884 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
885 these cases). There was also one in pcretest.c.
886
887 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
888
889 6. In UTF mode, the code for checking whether a group could match an empty
890 string (which is used for indefinitely repeated groups to allow for
891 breaking an infinite loop) was broken when the group contained a repeated
892 negated single-character class with a character that occupied more than one
893 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
894 UTF-8 mode). The effect was undefined: the group might or might not be
895 deemed as matching an empty string, or the program might have crashed.
896
897 7. The code for checking whether a group could match an empty string was not
898 recognizing that \h, \H, \v, \V, and \R must match a character.
899
900 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
901 an empty string. If it can, pcretest shows this in its information output.
902
903 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
904 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
905 matched by pcre_exec() without using JIT:
906
907 (a) If the rest of the pattern did not match after a maximal run of
908 grapheme clusters, the code for backing up to try with fewer of them
909 did not always back up over a full grapheme when characters that do not
910 have the modifier quality were involved, e.g. Hangul syllables.
911
912 (b) If the match point in a subject started with modifier character, and
913 there was no match, the code could incorrectly back up beyond the match
914 point, and potentially beyond the first character in the subject,
915 leading to a segfault or an incorrect match result.
916
917 10. A conditional group with an assertion condition could lead to PCRE
918 recording an incorrect first data item for a match if no other first data
919 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
920 first data item, and therefore matched "ca" after "c" instead of at the
921 start.
922
923 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
924 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
925
926 12. The source of pcregrep now includes z/OS-specific code so that it can be
927 compiled for z/OS as part of the special z/OS distribution.
928
929 13. Added the -T and -TM options to pcretest.
930
931 14. The code in pcre_compile.c for creating the table of named capturing groups
932 has been refactored. Instead of creating the table dynamically during the
933 actual compiling pass, the information is remembered during the pre-compile
934 pass (on the stack unless there are more than 20 named groups, in which
935 case malloc() is used) and the whole table is created before the actual
936 compile happens. This has simplified the code (it is now nearly 150 lines
937 shorter) and prepared the way for better handling of references to groups
938 with duplicate names.
939
940 15. A back reference to a named subpattern when there is more than one of the
941 same name now checks them in the order in which they appear in the pattern.
942 The first one that is set is used for the reference. Previously only the
943 first one was inspected. This change makes PCRE more compatible with Perl.
944
945 16. Unicode character properties were updated from Unicode 6.3.0.
946
947 17. The compile-time code for auto-possessification has been refactored, based
948 on a patch by Zoltan Herczeg. It now happens after instead of during
949 compilation. The code is cleaner, and more cases are handled. The option
950 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
951 options in pcretest are provided to set it. It can also be set by
952 (*NO_AUTO_POSSESS) at the start of a pattern.
953
954 18. The character VT has been added to the default ("C" locale) set of
955 characters that match \s and are generally treated as white space,
956 following this same change in Perl 5.18. There is now no difference between
957 "Perl space" and "POSIX space". Whether VT is treated as white space in
958 other locales depends on the locale.
959
960 19. The code for checking named groups as conditions, either for being set or
961 for being recursed, has been refactored (this is related to 14 and 15
962 above). Processing unduplicated named groups should now be as fast at
963 numerical groups, and processing duplicated groups should be faster than
964 before.
965
966 20. Two patches to the CMake build system, by Alexander Barkov:
967
968 (1) Replace the "source" command by "." in CMakeLists.txt because
969 "source" is a bash-ism.
970
971 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
972 without these the CMake build does not work on Solaris.
973
974 21. Perl has changed its handling of \8 and \9. If there is no previously
975 encountered capturing group of those numbers, they are treated as the
976 literal characters 8 and 9 instead of a binary zero followed by the
977 literals. PCRE now does the same.
978
979 22. Following Perl, added \o{} to specify codepoints in octal, making it
980 possible to specify values greater than 0777 and also making them
981 unambiguous.
982
983 23. Perl now gives an error for missing closing braces after \x{... instead of
984 treating the string as literal. PCRE now does the same.
985
986 24. RunTest used to grumble if an inappropriate test was selected explicitly,
987 but just skip it when running all tests. This make it awkward to run ranges
988 of tests when one of them was inappropriate. Now it just skips any
989 inappropriate tests, as it always did when running all tests.
990
991 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
992 character types such as \d or \w, too many callouts were inserted, and the
993 data that they returned was rubbish.
994
995 26. In UCP mode, \s was not matching two of the characters that Perl matches,
996 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
997 were matched by \h. The code has now been refactored so that the lists of
998 the horizontal and vertical whitespace characters used for \h and \v (which
999 are defined only in one place) are now also used for \s.
1000
1001 27. Add JIT support for the 64 bit TileGX architecture.
1002 Patch by Jiong Wang (Tilera Corporation).
1003
1004 28. Possessive quantifiers for classes (both explicit and automatically
1005 generated) now use special opcodes instead of wrapping in ONCE brackets.
1006
1007 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
1008 (because it's meaningless), this was not happening when PCRE_CASELESS was
1009 set. Not wrong, but inefficient.
1010
1011 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
1012 properties for \w, \d, etc) is present in a test regex. Otherwise if the
1013 test contains no characters greater than 255, Perl doesn't realise it
1014 should be using Unicode semantics.
1015
1016 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
1017 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
1018 does in Unicode mode.
1019
1020 32. Added the "forbid" facility to pcretest so that putting tests into the
1021 wrong test files can sometimes be quickly detected.
1022
1023 33. There is now a limit (default 250) on the depth of nesting of parentheses.
1024 This limit is imposed to control the amount of system stack used at compile
1025 time. It can be changed at build time by --with-parens-nest-limit=xxx or
1026 the equivalent in CMake.
1027
1028 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
1029 errors. Perl warns for these when in warning mode, but PCRE has no facility
1030 for giving warnings.
1031
1032 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
1033 However, this was not working for (?!) because it is optimized to (*FAIL),
1034 for which PCRE does not allow quantifiers. The optimization is now disabled
1035 when a quantifier follows (?!). I can't see any use for this, but it makes
1036 things uniform.
1037
1038 36. Perl no longer allows group names to start with digits, so I have made this
1039 change also in PCRE. It simplifies the code a bit.
1040
1041 37. In extended mode, Perl ignores spaces before a + that indicates a
1042 possessive quantifier. PCRE allowed a space before the quantifier, but not
1043 before the possessive +. It now does.
1044
1045 38. The use of \K (reset reported match start) within a repeated possessive
1046 group such as (a\Kb)*+ was not working.
1047
1048 40. Document that the same character tables must be used at compile time and
1049 run time, and that the facility to pass tables to pcre_exec() and
1050 pcre_dfa_exec() is for use only with saved/restored patterns.
1051
1052 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
1053 features for Builds with MSVC:
1054
1055 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
1056 builds). The .rc files can be used to set FileDescription and many other
1057 attributes.
1058
1059 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
1060 This allows higher-level build scripts which want .pdb files to avoid
1061 hard-coding the exact files needed."
1062
1063 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
1064 mean "start of word" and "end of word", respectively, as a transition aid.
1065
1066 43. A minimizing repeat of a class containing codepoints greater than 255 in
1067 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
1068 compiled to use the heap for recursion.
1069
1070 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
1071 is configured.
1072
1073
1074 Version 8.33 28-May-2013
1075 ------------------------
1076
1077 1. Added 'U' to some constants that are compared to unsigned integers, to
1078 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
1079 variables that are added to signed variables, to ensure the result is
1080 signed and can be negated.
1081
1082 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
1083 CMake config files.
1084
1085 3. Revise the creation of config.h.generic so that all boolean macros are
1086 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
1087 overriding via -D on the command line possible.
1088
1089 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
1090 to unsigned int is reported to make a quite noticeable speed difference in
1091 a specific Windows environment. Testing on Linux did also appear to show
1092 some benefit (and it is clearly not harmful). Also fixed the definition of
1093 Xop which should be unsigned.
1094
1095 5. Related to (4), changing the definition of the intermediate variable cc
1096 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
1097 improvements.
1098
1099 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
1100 unnecessary spaces.
1101
1102 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
1103 and later.
1104
1105 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
1106
1107 9. Optimizing fast_forward_start_bits in JIT.
1108
1109 10. Adding support for callouts in JIT, and fixing some issues revealed
1110 during this work. Namely:
1111
1112 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
1113
1114 (b) Minimum length was not checked before the matching is started.
1115
1116 11. The value of capture_last that is passed to callouts was incorrect in some
1117 cases when there was a capture on one path that was subsequently abandoned
1118 after a backtrack. Also, the capture_last value is now reset after a
1119 recursion, since all captures are also reset in this case.
1120
1121 12. The interpreter no longer returns the "too many substrings" error in the
1122 case when an overflowing capture is in a branch that is subsequently
1123 abandoned after a backtrack.
1124
1125 13. In the pathological case when an offset vector of size 2 is used, pcretest
1126 now prints out the matched string after a yield of 0 or 1.
1127
1128 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
1129 Only supported by the JIT compiler at the moment.
1130
1131 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
1132
1133 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
1134 offset of the starting point of the matching process, provided the offsets
1135 vector is large enough.
1136
1137 17. The \A escape now records a lookbehind value of 1, though its execution
1138 does not actually inspect the previous character. This is to ensure that,
1139 in partial multi-segment matching, at least one character from the old
1140 segment is retained when a new segment is processed. Otherwise, if there
1141 are no lookbehinds in the pattern, \A might match incorrectly at the start
1142 of a new segment.
1143
1144 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
1145
1146 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
1147 gives some modest performance improvement in 8-bit mode.
1148
1149 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
1150 be expressed in certain programming languages using Universal Character
1151 Names.
1152
1153 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
1154 which points out that "non characters" are not "characters that may not
1155 appear in Unicode strings" but rather "characters that are reserved for
1156 internal use and have only local meaning".
1157
1158 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
1159 there was a conditional group that depended on an assertion, if the
1160 assertion was false, the callout that immediately followed the alternation
1161 in the condition was skipped when pcre_exec() was used for matching.
1162
1163 23. Allow an explicit callout to be inserted before an assertion that is the
1164 condition for a conditional group, for compatibility with automatic
1165 callouts, which always insert a callout at this point.
1166
1167 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1168 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1169
1170 25. (*PRUNE) is now supported by the JIT compiler.
1171
1172 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1173
1174 27. Fix the case where there are two or more SKIPs with arguments that may be
1175 ignored.
1176
1177 28. (*SKIP) is now supported by the JIT compiler.
1178
1179 29. (*THEN) is now supported by the JIT compiler.
1180
1181 30. Update RunTest with additional test selector options.
1182
1183 31. The way PCRE handles backtracking verbs has been changed in two ways.
1184
1185 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1186 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1187 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1188 rather obscure rules do not always do the same thing.
1189
1190 (2) Previously, backtracking verbs were confined within assertions. This is
1191 no longer the case for positive assertions, except for (*ACCEPT). Again,
1192 this sometimes improves Perl compatibility, and sometimes does not.
1193
1194 32. A number of tests that were in test 2 because Perl did things differently
1195 have been moved to test 1, because either Perl or PCRE has changed, and
1196 these tests are now compatible.
1197
1198 32. Backtracking control verbs are now handled in the same way in JIT and
1199 interpreter.
1200
1201 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1202 contained a forward subroutine reference caused a compile error.
1203
1204 34. Auto-detect and optimize limited repetitions in JIT.
1205
1206 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1207 blocking (*UTF) etc.
1208
1209 36. In the interpreter, maximizing pattern repetitions for characters and
1210 character types now use tail recursion, which reduces stack usage.
1211
1212 37. The value of the max lookbehind was not correctly preserved if a compiled
1213 and saved regex was reloaded on a host of different endianness.
1214
1215 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1216 of the compiled pattern block, expand the flags field from 16 to 32 bits
1217 because it was almost full.
1218
1219 39. Try madvise first before posix_madvise.
1220
1221 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1222 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1223 original change locked out all matches of empty strings. This has been
1224 changed so that one match of an empty string per line is recognized.
1225 Subsequent searches on the same line (for colouring or for --only-matching,
1226 for example) do not recognize empty strings.
1227
1228 41. Applied a user patch to fix a number of spelling mistakes in comments.
1229
1230 42. Data lines longer than 65536 caused pcretest to crash.
1231
1232 43. Clarified the data type for length and startoffset arguments for pcre_exec
1233 and pcre_dfa_exec in the function-specific man pages, where they were
1234 explicitly stated to be in bytes, never having been updated. I also added
1235 some clarification to the pcreapi man page.
1236
1237 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1238 a segmentation fault.
1239
1240
1241 Version 8.32 30-November-2012
1242 -----------------------------
1243
1244 1. Improved JIT compiler optimizations for first character search and single
1245 character iterators.
1246
1247 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1248 Patch by Daniel Richard G.
1249
1250 3. Single character iterator optimizations in the JIT compiler.
1251
1252 4. Improved JIT compiler optimizations for character ranges.
1253
1254 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1255 Reported by Giuseppe D'Angelo.
1256
1257 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1258 of a line, was being set incorrectly in cases where .* appeared inside
1259 atomic brackets at the start of a pattern, or where there was a subsequent
1260 *PRUNE or *SKIP.
1261
1262 7. Improved instruction cache flush for POWER/PowerPC.
1263 Patch by Daniel Richard G.
1264
1265 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1266 grep:
1267
1268 (a) There is now no limit to the number of patterns to be matched.
1269
1270 (b) An error is given if a pattern is too long.
1271
1272 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1273 are now supported.
1274
1275 (d) --exclude-from and --include-from (multiple use) have been added.
1276
1277 (e) Exclusions and inclusions now apply to all files and directories, not
1278 just to those obtained from scanning a directory recursively.
1279
1280 (f) Multiple uses of -f and --file-list are now supported.
1281
1282 (g) In a Windows environment, the default for -d has been changed from
1283 "read" (the GNU grep default) to "skip", because otherwise the presence
1284 of a directory in the file list provokes an error.
1285
1286 (h) The documentation has been revised and clarified in places.
1287
1288 9. Improve the matching speed of capturing brackets.
1289
1290 10. Changed the meaning of \X so that it now matches a Unicode extended
1291 grapheme cluster.
1292
1293 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1294 out POSIX threads when JIT support is configured.
1295
1296 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1297
1298 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1299 structure could go wrong in environments where size_t is not the same size
1300 as int.
1301
1302 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1303 set.
1304
1305 15. The EBCDIC support had decayed; later updates to the code had included
1306 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1307 general tidy up of EBCDIC-related issues, and the documentation was also
1308 not quite right. There is now a test that can be run on ASCII systems to
1309 check some of the EBCDIC-related things (but is it not a full test).
1310
1311 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1312 in a small tidy to the code.
1313
1314 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1315
1316 18. If the --only-matching (-o) option in pcregrep is specified multiple
1317 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1318 substrings matched by the 1st and 2nd capturing parentheses. A separating
1319 string can be specified by --om-separator (default empty).
1320
1321 19. Improving the first n character searches.
1322
1323 20. Turn case lists for horizontal and vertical white space into macros so that
1324 they are defined only once.
1325
1326 21. This set of changes together give more compatible Unicode case-folding
1327 behaviour for characters that have more than one other case when UCP
1328 support is available.
1329
1330 (a) The Unicode property table now has offsets into a new table of sets of
1331 three or more characters that are case-equivalent. The MultiStage2.py
1332 script that generates these tables (the pcre_ucd.c file) now scans
1333 CaseFolding.txt instead of UnicodeData.txt for character case
1334 information.
1335
1336 (b) The code for adding characters or ranges of characters to a character
1337 class has been abstracted into a generalized function that also handles
1338 case-independence. In UTF-mode with UCP support, this uses the new data
1339 to handle characters with more than one other case.
1340
1341 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1342 whose other case is greater than 256 are now correctly matched
1343 caselessly. Previously, the high codepoint matched the low one, but not
1344 vice versa.
1345
1346 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1347 of the new class addition function, using character lists defined as
1348 macros alongside the case definitions of 20 above.
1349
1350 (e) Caseless back references now work with characters that have more than
1351 one other case.
1352
1353 (f) General caseless matching of characters with more than one other case
1354 is supported.
1355
1356 22. Unicode character properties were updated from Unicode 6.2.0
1357
1358 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1359
1360 24. Add support for 32-bit character strings, and UTF-32
1361
1362 25. Major JIT compiler update (code refactoring and bugfixing).
1363 Experimental Sparc 32 support is added.
1364
1365 26. Applied a modified version of Daniel Richard G's patch to create
1366 pcre.h.generic and config.h.generic by "make" instead of in the
1367 PrepareRelease script.
1368
1369 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1370 pcre_compile.c when checking for a zero character.
1371
1372 28. Introducing a native interface for JIT. Through this interface, the compiled
1373 machine code can be directly executed. The purpose of this interface is to
1374 provide fast pattern matching, so several sanity checks are not performed.
1375 However, feature tests are still performed. The new interface provides
1376 1.4x speedup compared to the old one.
1377
1378 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1379 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1380 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1381
1382 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1383 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1384 the "old" RFC 2279). Instead, it ended up passing a negative length to
1385 pcre_exec().
1386
1387 31. Add support for GCC's visibility feature to hide internal functions.
1388
1389 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1390 "unknown -C option" after outputting 0 or 1.
1391
1392 33. There is now support for generating a code coverage report for the test
1393 suite in environments where gcc is the compiler and lcov is installed. This
1394 is mainly for the benefit of the developers.
1395
1396 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1397 unaddressable using valgrind annotations, allowing valgrind to detect
1398 invalid memory accesses. This is mainly for the benefit of the developers.
1399
1400 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1401
1402 26. Give configure error if --enable-cpp but no C++ compiler found.
1403
1404
1405 Version 8.31 06-July-2012
1406 -------------------------
1407
1408 1. Fixing a wrong JIT test case and some compiler warnings.
1409
1410 2. Removed a bashism from the RunTest script.
1411
1412 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1413 to unsigned type, result still unsigned" that was given by an MS compiler
1414 on encountering the code "-sizeof(xxx)".
1415
1416 4. Partial matching support is added to the JIT compiler.
1417
1418 5. Fixed several bugs concerned with partial matching of items that consist
1419 of more than one character:
1420
1421 (a) /^(..)\1/ did not partially match "aba" because checking references was
1422 done on an "all or nothing" basis. This also applied to repeated
1423 references.
1424
1425 (b) \R did not give a hard partial match if \r was found at the end of the
1426 subject.
1427
1428 (c) \X did not give a hard partial match after matching one or more
1429 characters at the end of the subject.
1430
1431 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1432 a partial match for the string "\r".
1433
1434 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1435 a partial match for a CR character at the end of the subject string.
1436
1437 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1438 running pcretest, the text "(JIT)" added to the output whenever JIT is
1439 actually used to run the match.
1440
1441 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1442 or /S+[+] with a digit between 1 and 7.
1443
1444 8. OP_NOT now supports any UTF character not just single-byte ones.
1445
1446 9. (*MARK) control verb is now supported by the JIT compiler.
1447
1448 10. The command "./RunTest list" lists the available tests without actually
1449 running any of them. (Because I keep forgetting what they all are.)
1450
1451 11. Add PCRE_INFO_MAXLOOKBEHIND.
1452
1453 12. Applied a (slightly modified) user-supplied patch that improves performance
1454 when the heap is used for recursion (compiled with --disable-stack-for-
1455 recursion). Instead of malloc and free for each heap frame each time a
1456 logical recursion happens, frames are retained on a chain and re-used where
1457 possible. This sometimes gives as much as 30% improvement.
1458
1459 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1460 call.
1461
1462 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1463
1464 15. It is now possible to link pcretest with libedit as an alternative to
1465 libreadline.
1466
1467 16. (*COMMIT) control verb is now supported by the JIT compiler.
1468
1469 17. The Unicode data tables have been updated to Unicode 6.1.0.
1470
1471 18. Added --file-list option to pcregrep.
1472
1473 19. Added binary file support to pcregrep, including the -a, --binary-files,
1474 -I, and --text options.
1475
1476 20. The madvise function is renamed for posix_madvise for QNX compatibility
1477 reasons. Fixed by Giuseppe D'Angelo.
1478
1479 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1480
1481 22. Changed the option for creating symbolic links for 16-bit man pages from
1482 -s to -sf so that re-installing does not cause issues.
1483
1484 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1485
1486 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1487 matches in certain environments (the workspace was not being correctly
1488 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1489 some of the workspace data at the beginning of a restart.
1490
1491 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1492 was not doing so when it should - probably a typo introduced by SVN 528
1493 (change 8.10/14).
1494
1495 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1496 \w+ when the character tables indicated that \x{c4} was a word character.
1497 There were several related cases, all because the tests for doing a table
1498 lookup were testing for characters less than 127 instead of 255.
1499
1500 27. If a pattern contains capturing parentheses that are not used in a match,
1501 their slots in the ovector are set to -1. For those that are higher than
1502 any matched groups, this happens at the end of processing. In the case when
1503 there were back references that the ovector was too small to contain
1504 (causing temporary malloc'd memory to be used during matching), and the
1505 highest capturing number was not used, memory off the end of the ovector
1506 was incorrectly being set to -1. (It was using the size of the temporary
1507 memory instead of the true size.)
1508
1509 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1510 ovector size, it uses memory at the end of the block that it has got.
1511
1512 29. Check for an overlong MARK name and give an error at compile time. The
1513 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1514
1515 30. JIT compiler update.
1516
1517 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1518 Rill for the patch.
1519
1520 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1521
1522 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1523
1524 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1525 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1526 was enabled.
1527
1528 35. Improve JIT code generation for greedy plus quantifier.
1529
1530 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1531 1 to "aa" instead of to an empty string. The bug affected repeated groups
1532 that could potentially match an empty string.
1533
1534 37. Optimizing single character iterators in JIT.
1535
1536 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1537 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1538 codepoints that are too big for the mode are faulted, and in a UTF mode,
1539 disallowed codepoints are also faulted.
1540
1541 39. If PCRE was compiled with UTF support, in three places in the DFA
1542 matcher there was code that should only have been obeyed in UTF mode, but
1543 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1544 processing when bytes with values greater than 127 were present. In 16-bit
1545 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1546 both cases the values are those that cannot be the first data item in a UTF
1547 character. The three items that might have provoked this were recursions,
1548 possessively repeated groups, and atomic groups.
1549
1550 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1551 and pcregrep, because some OS require shared objects to be explicitly
1552 passed to ld, causing the link step to fail if they are not.
1553
1554 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1555 mode, patterns that started with \h* or \R* might be incorrectly matched.
1556
1557
1558 Version 8.30 04-February-2012
1559 -----------------------------
1560
1561 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1562 name is defined in ctype.h.
1563
1564 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1565 only in quite long subpatterns.
1566
1567 3. Removed the function pcre_info(), which has been obsolete and deprecated
1568 since it was replaced by pcre_fullinfo() in February 2000.
1569
1570 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1571 match a (*MARK), and the match failed at the start of the subject, a
1572 reference to memory before the start of the subject could occur. This bug
1573 was introduced by fix 17 of release 8.21.
1574
1575 5. A reference to an unset group with zero minimum repetition was giving
1576 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1577 /(another)?(\1?)test/ matched against "hello world test". This bug was
1578 introduced in release 8.13.
1579
1580 6. Add support for 16-bit character strings (a large amount of work involving
1581 many changes and refactorings).
1582
1583 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1584 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1585 from a file.
1586
1587 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1588 rounding is not applied in this particular case).
1589
1590 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1591 if they appear, or are escaped, in patterns.
1592
1593 10. Get rid of a number of -Wunused-but-set-variable warnings.
1594
1595 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1596 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1597 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1598 also returns the mark "x". This bug applied to capturing parentheses,
1599 non-capturing parentheses, and atomic parentheses. It also applied to some
1600 assertions.
1601
1602 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1603 information out of configure.ac instead of relying on pcre.h.generic, which
1604 is not stored in the repository.
1605
1606 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1607 -lreadline.
1608
1609 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1610
1611 15. Applied Graycode's patch to put the top-level frame on the stack rather
1612 than the heap when not using the stack for recursion. This gives a
1613 performance improvement in many cases when recursion is not deep.
1614
1615 16. Experimental code added to "pcretest -C" to output the stack frame size.
1616
1617
1618 Version 8.21 12-Dec-2011
1619 ------------------------
1620
1621 1. Updating the JIT compiler.
1622
1623 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1624 are added as well.
1625
1626 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1627 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1628 calling _pcre_jit_exec. Some extra comments are added.
1629
1630 4. (*MARK) settings inside atomic groups that do not contain any capturing
1631 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1632 was introduced by change 18 for 8.20.
1633
1634 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1635 ECMA-262 standard.
1636
1637 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1638 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1639 This bug was probably introduced by change 9 of 8.13.
1640
1641 7. While fixing 6 above, I noticed that a number of other items were being
1642 incorrectly rejected as "not fixed length". This arose partly because newer
1643 opcodes had not been added to the fixed-length checking code. I have (a)
1644 corrected the bug and added tests for these items, and (b) arranged for an
1645 error to occur if an unknown opcode is encountered while checking for fixed
1646 length instead of just assuming "not fixed length". The items that were
1647 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1648 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1649 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1650
1651 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1652 being incorrectly compiled and would have given unpredicatble results.
1653
1654 9. A possessively repeated subpattern with minimum repeat count greater than
1655 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1656 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1657 the first (A) could occur when it should not.
1658
1659 10. Add a cast and remove a redundant test from the code.
1660
1661 11. JIT should use pcre_malloc/pcre_free for allocation.
1662
1663 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1664 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1665 is anything other than /usr, -L is still shown).
1666
1667 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1668
1669 14. Perl does not support \N without a following name in a [] class; PCRE now
1670 also gives an error.
1671
1672 15. If a forward reference was repeated with an upper limit of around 2000,
1673 it caused the error "internal error: overran compiling workspace". The
1674 maximum number of forward references (including repeats) was limited by the
1675 internal workspace, and dependent on the LINK_SIZE. The code has been
1676 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1677 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1678 of around 200,000 forward references. While doing this, I also speeded up
1679 the filling in of repeated forward references.
1680
1681 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1682 incorrectly expecting the subject to contain another "a" after the start.
1683
1684 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1685 in the match, the SKIP should be ignored. This was not happening; instead
1686 the SKIP was being treated as NOMATCH. For patterns such as
1687 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1688 tested.
1689
1690 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1691 now much more compatible with Perl, in particular in cases where the result
1692 is a non-match for a non-anchored pattern. For example, if
1693 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1694 "m", where previously it did not return a name. A side effect of this
1695 change is that for partial matches, the last encountered mark name is
1696 returned, as for non matches. A number of tests that were previously not
1697 Perl-compatible have been moved into the Perl-compatible test files. The
1698 refactoring has had the pleasing side effect of removing one argument from
1699 the match() function, thus reducing its stack requirements.
1700
1701 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1702 subsequent uses of /S (without +) incorrectly behaved like /S+.
1703
1704 21. Retrieve executable code size support for the JIT compiler and fixing
1705 some warnings.
1706
1707 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1708 not work when the shorter character appeared right at the end of the
1709 subject string.
1710
1711 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1712 systems.
1713
1714 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1715 output it when the /M option is used in pcretest.
1716
1717 25. The CheckMan script was not being included in the distribution. Also, added
1718 an explicit "perl" to run Perl scripts from the PrepareRelease script
1719 because this is reportedly needed in Windows.
1720
1721 26. If study data was being save in a file and studying had not found a set of
1722 "starts with" bytes for the pattern, the data written to the file (though
1723 never used) was taken from uninitialized memory and so caused valgrind to
1724 complain.
1725
1726 27. Updated RunTest.bat as provided by Sheri Pierce.
1727
1728 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1729
1730 29. Computation of memory usage for the table of capturing group names was
1731 giving an unnecessarily large value.
1732
1733
1734 Version 8.20 21-Oct-2011
1735 ------------------------
1736
1737 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1738 a POSIX class. After further experiments with Perl, which convinced me that
1739 Perl has bugs and confusions, a closing square bracket is no longer allowed
1740 in a POSIX name. This bug also affected patterns with classes that started
1741 with full stops.
1742
1743 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1744 captured substring, but while checking the failing first alternative,
1745 substring 1 is temporarily captured. If the output vector supplied to
1746 pcre_exec() was not big enough for this capture, the yield of the function
1747 was still zero ("insufficient space for captured substrings"). This cannot
1748 be totally fixed without adding another stack variable, which seems a lot
1749 of expense for a edge case. However, I have improved the situation in cases
1750 such as /(a)(b)x|abc/ matched against "abc", where the return code
1751 indicates that fewer than the maximum number of slots in the ovector have
1752 been set.
1753
1754 3. Related to (2) above: when there are more back references in a pattern than
1755 slots in the output vector, pcre_exec() uses temporary memory during
1756 matching, and copies in the captures as far as possible afterwards. It was
1757 using the entire output vector, but this conflicts with the specification
1758 that only 2/3 is used for passing back captured substrings. Now it uses
1759 only the first 2/3, for compatibility. This is, of course, another edge
1760 case.
1761
1762 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1763 main code base, and can be used by building with --enable-jit. When this is
1764 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1765 runtime --no-jit option is given.
1766
1767 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1768 ovector, the return from the function was zero, implying that there were
1769 other matches that did not fit. The correct "exactly full" value is now
1770 returned.
1771
1772 6. If a subpattern that was called recursively or as a subroutine contained
1773 (*PRUNE) or any other control that caused it to give a non-standard return,
1774 invalid errors such as "Error -26 (nested recursion at the same subject
1775 position)" or even infinite loops could occur.
1776
1777 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1778 computing the minimum length on reaching *ACCEPT, and so ended up with the
1779 wrong value of 1 rather than 0. Further investigation indicates that
1780 computing a minimum subject length in the presence of *ACCEPT is difficult
1781 (think back references, subroutine calls), and so I have changed the code
1782 so that no minimum is registered for a pattern that contains *ACCEPT.
1783
1784 8. If (*THEN) was present in the first (true) branch of a conditional group,
1785 it was not handled as intended. [But see 16 below.]
1786
1787 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1788 Sheri Pierce.
1789
1790 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1791 the first byte in a match must be "a".
1792
1793 11. Change 17 for 8.13 increased the recursion depth for patterns like
1794 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1795 pattern contains any instances of (*THEN). If it does not, the old
1796 optimizations are restored. It would be nice to do this on a per-group
1797 basis, but at the moment that is not feasible.
1798
1799 12. In some environments, the output of pcretest -C is CRLF terminated. This
1800 broke RunTest's code that checks for the link size. A single white space
1801 character after the value is now allowed for.
1802
1803 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1804 For "fr", it uses the Windows-specific input and output files.
1805
1806 14. If (*THEN) appeared in a group that was called recursively or as a
1807 subroutine, it did not work as intended. [But see next item.]
1808
1809 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1810 pattern fragments (but not containing any | characters). If A and B are
1811 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1812 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1813 D. In other words, Perl considers parentheses that do not contain any |
1814 characters to be part of a surrounding alternative, whereas PCRE was
1815 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1816 differently. PCRE now behaves in the same way as Perl, except in the case
1817 of subroutine/recursion calls such as (?1) which have in any case always
1818 been different (but PCRE had them first :-).
1819
1820 16. Related to 15 above: Perl does not treat the | in a conditional group as
1821 creating alternatives. Such a group is treated in the same way as an
1822 ordinary group without any | characters when processing (*THEN). PCRE has
1823 been changed to match Perl's behaviour.
1824
1825 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1826 RunGrepTest script failed.
1827
1828 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1829 inevitable for groups that contain captures, but it can lead to a lot of
1830 stack use in large patterns. The old behaviour has been restored for atomic
1831 groups that do not contain any capturing parentheses.
1832
1833 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1834 suppress the check for a minimum subject length at run time. (If it was
1835 given to pcre_exec() or pcre_dfa_exec() it did work.)
1836
1837 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1838 fail to work when decoding hex characters in data strings in EBCDIC
1839 environments.
1840
1841 21. It appears that in at least one Mac OS environment, the isxdigit() function
1842 is implemented as a macro that evaluates to its argument more than once,
1843 contravening the C 90 Standard (I haven't checked a later standard). There
1844 was an instance in pcretest which caused it to go wrong when processing
1845 \x{...} escapes in subject strings. The has been rewritten to avoid using
1846 things like p++ in the argument of isxdigit().
1847
1848
1849 Version 8.13 16-Aug-2011
1850 ------------------------
1851
1852 1. The Unicode data tables have been updated to Unicode 6.0.0.
1853
1854 2. Two minor typos in pcre_internal.h have been fixed.
1855
1856 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1857 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1858 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1859
1860 4. There were a number of related bugs in the code for matching backrefences
1861 caselessly in UTF-8 mode when codes for the characters concerned were
1862 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1863 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1864 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1865 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1866 2-byte code at the end of the subject (it thought there wasn't enough data
1867 left).
1868
1869 5. Comprehensive information about what went wrong is now returned by
1870 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1871 as the output vector has at least 2 elements. The offset of the start of
1872 the failing character and a reason code are placed in the vector.
1873
1874 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1875 now returned is for the first byte of the failing character, instead of the
1876 last byte inspected. This is an incompatible change, but I hope it is small
1877 enough not to be a problem. It makes the returned offset consistent with
1878 pcre_exec() and pcre_dfa_exec().
1879
1880 7. pcretest now gives a text phrase as well as the error number when
1881 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1882 failure, the offset and reason code are output.
1883
1884 8. When \R was used with a maximizing quantifier it failed to skip backwards
1885 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1886 back over a single character (\n). This seems wrong (because it treated the
1887 two characters as a single entity when going forwards), conflicts with the
1888 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1889 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1890 has been changed.
1891
1892 9. Some internal refactoring has changed the processing so that the handling
1893 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1894 time (the PCRE_DOTALL option was changed this way some time ago: version
1895 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1896 which was always a bit of a fudge. It also means that there is one less
1897 argument for the match() function, which reduces its stack requirements
1898 slightly. This change also fixes an incompatibility with Perl: the pattern
1899 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1900
1901 10. More internal refactoring has drastically reduced the number of recursive
1902 calls to match() for possessively repeated groups such as (abc)++ when
1903 using pcre_exec().
1904
1905 11. While implementing 10, a number of bugs in the handling of groups were
1906 discovered and fixed:
1907
1908 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1909 (a|)*(?1) gave a compile-time internal error.
1910 ((a|)+)+ did not notice that the outer group could match an empty string.
1911 (^a|^)+ was not marked as anchored.
1912 (.*a|.*)+ was not marked as matching at start or after a newline.
1913
1914 12. Yet more internal refactoring has removed another argument from the match()
1915 function. Special calls to this function are now indicated by setting a
1916 value in a variable in the "match data" data block.
1917
1918 13. Be more explicit in pcre_study() instead of relying on "default" for
1919 opcodes that mean there is no starting character; this means that when new
1920 ones are added and accidentally left out of pcre_study(), testing should
1921 pick them up.
1922
1923 14. The -s option of pcretest has been documented for ages as being an old
1924 synonym of -m (show memory usage). I have changed it to mean "force study
1925 for every regex", that is, assume /S for every regex. This is similar to -i
1926 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1927 using it. It makes it easier to run collections of tests with and without
1928 study enabled, and thereby test pcre_study() more easily. All the standard
1929 tests are now run with and without -s (but some patterns can be marked as
1930 "never study" - see 20 below).
1931
1932 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1933 restoration of the capturing data to the outer values was not happening
1934 correctly.
1935
1936 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1937 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1938 pattern had matched an empty string, and so incorrectly returned a no
1939 match.
1940
1941 17. There was optimizing code for the last branch of non-capturing parentheses,
1942 and also for the obeyed branch of a conditional subexpression, which used
1943 tail recursion to cut down on stack usage. Unfortunately, now that there is
1944 the possibility of (*THEN) occurring in these branches, tail recursion is
1945 no longer possible because the return has to be checked for (*THEN). These
1946 two optimizations have therefore been removed. [But see 8.20/11 above.]
1947
1948 18. If a pattern containing \R was studied, it was assumed that \R always
1949 matched two bytes, thus causing the minimum subject length to be
1950 incorrectly computed because \R can also match just one byte.
1951
1952 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1953 was incorrectly computed.
1954
1955 20. If /S is present twice on a test pattern in pcretest input, it now
1956 *disables* studying, thereby overriding the use of -s on the command line
1957 (see 14 above). This is necessary for one or two tests to keep the output
1958 identical in both cases.
1959
1960 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1961 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1962
1963 22. When an atomic group that contained a capturing parenthesis was
1964 successfully matched, but the branch in which it appeared failed, the
1965 capturing was not being forgotten if a higher numbered group was later
1966 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1967 group 1 to "a", when in fact it should be unset. This applied to multi-
1968 branched capturing and non-capturing groups, repeated or not, and also to
1969 positive assertions (capturing in negative assertions does not happen
1970 in PCRE) and also to nested atomic groups.
1971
1972 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1973 subject after a captured substring, to make it easier to tell which of a
1974 number of identical substrings has been captured.
1975
1976 24. The way atomic groups are processed by pcre_exec() has been changed so that
1977 if they are repeated, backtracking one repetition now resets captured
1978 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1979 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1980 "aaa". Previously, it would have been "a". As part of this code
1981 refactoring, the way recursive calls are handled has also been changed.
1982
1983 25. If an assertion condition captured any substrings, they were not passed
1984 back unless some other capturing happened later. For example, if
1985 (?(?=(a))a) was matched against "a", no capturing was returned.
1986
1987 26. When studying a pattern that contained subroutine calls or assertions,
1988 the code for finding the minimum length of a possible match was handling
1989 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1990 group 1 called group 2 while simultaneously a separate group 2 called group
1991 1). A stack overflow occurred in this case. I have fixed this by limiting
1992 the recursion depth to 10.
1993
1994 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1995 Fortmann. This supports explicit test numbers on the command line, and has
1996 argument validation and error reporting.
1997
1998 28. An instance of \X with an unlimited repeat could fail if at any point the
1999 first character it looked at was a mark character.
2000
2001 29. Some minor code refactoring concerning Unicode properties and scripts
2002 should reduce the stack requirement of match() slightly.
2003
2004 30. Added the '=' option to pcretest to check the setting of unused capturing
2005 slots at the end of the pattern, which are documented as being -1, but are
2006 not included in the return count.
2007
2008 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
2009 compiled something random. Now it gives a compile-time error (as does
2010 Perl).
2011
2012 32. A *MARK encountered during the processing of a positive assertion is now
2013 recorded and passed back (compatible with Perl).
2014
2015 33. If --only-matching or --colour was set on a pcregrep call whose pattern
2016 had alternative anchored branches, the search for a second match in a line
2017 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
2018 matched the line "0102" twice. The same bug affected patterns that started
2019 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
2020 twice.
2021
2022 34. Previously, PCRE did not allow quantification of assertions. However, Perl
2023 does, and because of capturing effects, quantifying parenthesized
2024 assertions may at times be useful. Quantifiers are now allowed for
2025 parenthesized assertions.
2026
2027 35. A minor code tidy in pcre_compile() when checking options for \R usage.
2028
2029 36. \g was being checked for fancy things in a character class, when it should
2030 just be a literal "g".
2031
2032 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
2033 appearance of a nested POSIX class supersedes an apparent external class.
2034 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
2035 unescaped square brackets may also appear as part of class names. For
2036 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
2037 more like Perl. (But see 8.20/1 above.)
2038
2039 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
2040 was because it thought it was \N{name}, which is not supported).
2041
2042 39. Add minix to OS list not supporting the -S option in pcretest.
2043
2044 40. PCRE tries to detect cases of infinite recursion at compile time, but it
2045 cannot analyze patterns in sufficient detail to catch mutual recursions
2046 such as ((?1))((?2)). There is now a runtime test that gives an error if a
2047 subgroup is called recursively as a subpattern for a second time at the
2048 same position in the subject string. In previous releases this might have
2049 been caught by the recursion limit, or it might have run out of stack.
2050
2051 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
2052 happen only once. PCRE was, however incorrectly giving a compile time error
2053 "recursive call could loop indefinitely" because it cannot analyze the
2054 pattern in sufficient detail. The compile time test no longer happens when
2055 PCRE is compiling a conditional subpattern, but actual runaway loops are
2056 now caught at runtime (see 40 above).
2057
2058 42. It seems that Perl allows any characters other than a closing parenthesis
2059 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
2060 has been changed to be the same.
2061
2062 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
2063 as not to get warnings when autogen.sh is called. Also changed
2064 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
2065
2066 44. To help people who use pcregrep to scan files containing exceedingly long
2067 lines, the following changes have been made:
2068
2069 (a) The default value of the buffer size parameter has been increased from
2070 8K to 20K. (The actual buffer used is three times this size.)
2071
2072 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
2073 PCRE is built.
2074
2075 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
2076 to be set at run time.
2077
2078 (d) Numerical values in pcregrep options can be followed by K or M, for
2079 example --buffer-size=50K.
2080
2081 (e) If a line being scanned overflows pcregrep's buffer, an error is now
2082 given and the return code is set to 2.
2083
2084 45. Add a pointer to the latest mark to the callout data block.
2085
2086 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
2087 partial match of an empty string instead of no match. This was specific to
2088 the use of ".".
2089
2090 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
2091 complete match instead of a partial match. This bug was dependent on both
2092 the PCRE_UTF8 and PCRE_DOTALL options being set.
2093
2094 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
2095 starting byte set, because \b was not being ignored.
2096
2097
2098 Version 8.12 15-Jan-2011
2099 ------------------------
2100
2101 1. Fixed some typos in the markup of the man pages, and wrote a script that
2102 checks for such things as part of the documentation building process.
2103
2104 2. On a big-endian 64-bit system, pcregrep did not correctly process the
2105 --match-limit and --recursion-limit options (added for 8.11). In
2106 particular, this made one of the standard tests fail. (The integer value
2107 went into the wrong half of a long int.)
2108
2109 3. If the --colour option was given to pcregrep with -v (invert match), it
2110 did strange things, either producing crazy output, or crashing. It should,
2111 of course, ignore a request for colour when reporting lines that do not
2112 match.
2113
2114 4. Another pcregrep bug caused similar problems if --colour was specified with
2115 -M (multiline) and the pattern match finished with a line ending.
2116
2117 5. In pcregrep, when a pattern that ended with a literal newline sequence was
2118 matched in multiline mode, the following line was shown as part of the
2119 match. This seems wrong, so I have changed it.
2120
2121 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
2122 the check for further matches in the same line (so they could be coloured)
2123 to overrun the end of the current line. If another match was found, it was
2124 incorrectly shown (and then shown again when found in the next line).
2125
2126 7. If pcregrep was compiled under Windows, there was a reference to the
2127 function pcregrep_exit() before it was defined. I am assuming this was
2128 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
2129 reported by a user. I've moved the definition above the reference.
2130
2131
2132 Version 8.11 10-Dec-2010
2133 ------------------------
2134
2135 1. (*THEN) was not working properly if there were untried alternatives prior
2136 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
2137 backtracked to try for "b" instead of moving to the next alternative branch
2138 at the same level (in this case, to look for "c"). The Perl documentation
2139 is clear that when (*THEN) is backtracked onto, it goes to the "next
2140 alternative in the innermost enclosing group".
2141
2142 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
2143 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
2144 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
2145 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
2146 (*THEN).
2147
2148 3. If \s appeared in a character class, it removed the VT character from
2149 the class, even if it had been included by some previous item, for example
2150 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
2151 of \s, but is part of the POSIX "space" class.)
2152
2153 4. A partial match never returns an empty string (because you can always
2154 match an empty string at the end of the subject); however the checking for
2155 an empty string was starting at the "start of match" point. This has been
2156 changed to the "earliest inspected character" point, because the returned
2157 data for a partial match starts at this character. This means that, for
2158 example, /(?<=abc)def/ gives a partial match for the subject "abc"
2159 (previously it gave "no match").
2160
2161 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2162 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2163 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2164 has an implication that the given string is incomplete (because a partial
2165 match is preferred over a full match). For this reason, these items now
2166 give a partial match in this situation. [Aside: previously, the one case
2167 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2168 match rather than a full match, which was wrong by the old rules, but is
2169 now correct.]
2170
2171 6. There was a bug in the handling of #-introduced comments, recognized when
2172 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2173 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2174 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2175 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2176 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2177 places in pcre_compile().
2178
2179 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2180 comments when looking ahead for named forward references to subpatterns,
2181 the only newline sequence it recognized was NL. It now handles newlines
2182 according to the set newline convention.
2183
2184 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2185 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2186 cater for a lack of strerror(). These oversights have been fixed.
2187
2188 9. Added --match-limit and --recursion-limit to pcregrep.
2189
2190 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2191
2192 11. When the -o option was used, pcregrep was setting a return code of 1, even
2193 when matches were found, and --line-buffered was not being honoured.
2194
2195 12. Added an optional parentheses number to the -o and --only-matching options
2196 of pcregrep.
2197
2198 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2199 can match an empty string. The code to do it in pcretest and pcredemo
2200 needed fixing:
2201
2202 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2203 only one byte after an empty string match just before CRLF (this case
2204 just got forgotten; "any" and "anycrlf" were OK).
2205
2206 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2207 mode when an empty string match preceded an ASCII character followed by
2208 a non-ASCII character. (The code for advancing by one character rather
2209 than one byte was nonsense.)
2210
2211 (c) The pcredemo.c sample program did not have any code at all to handle
2212 the cases when CRLF is a valid newline sequence.
2213
2214 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2215 as a starting offset was within the subject string. There is now a new
2216 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2217 negative or greater than the length of the string. In order to test this,
2218 pcretest is extended to allow the setting of negative starting offsets.
2219
2220 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2221 starting offset points to the beginning of a UTF-8 character was
2222 unnecessarily clumsy. I tidied it up.
2223
2224 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2225 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2226
2227 17. Nobody had reported that the --include_dir option, which was added in
2228 release 7.7 should have been called --include-dir (hyphen, not underscore)
2229 for compatibility with GNU grep. I have changed it to --include-dir, but
2230 left --include_dir as an undocumented synonym, and the same for
2231 --exclude-dir, though that is not available in GNU grep, at least as of
2232 release 2.5.4.
2233
2234 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2235 characters from a string of bytes) have been redefined so as not to use
2236 loops, in order to improve performance in some environments. At the same
2237 time, I abstracted some of the common code into auxiliary macros to save
2238 repetition (this should not affect the compiled code).
2239
2240 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2241 compile-time error is now given if \c is not followed by an ASCII
2242 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2243 different, and any byte value is allowed.)
2244
2245 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2246 START_OPTIMIZE option, which is now allowed at compile time - but just
2247 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2248 to pcregrep and other applications that have no direct access to PCRE
2249 options. The new /Y option in pcretest sets this option when calling
2250 pcre_compile().
2251
2252 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2253 back references. Groups containing recursive back references were forced to
2254 be atomic by that change, but in the case of named groups, the amount of
2255 memory required was incorrectly computed, leading to "Failed: internal
2256 error: code overflow". This has been fixed.
2257
2258 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2259 pcretest.c, to avoid build problems in some Borland environments.
2260
2261
2262 Version 8.10 25-Jun-2010
2263 ------------------------
2264
2265 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2266 THEN.
2267
2268 2. (*ACCEPT) was not working when inside an atomic group.
2269
2270 3. Inside a character class, \B is treated as a literal by default, but
2271 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2272 causes the error). The code is unchanged, but I tidied the documentation.
2273
2274 4. Inside a character class, PCRE always treated \R and \X as literals,
2275 whereas Perl faults them if its -w option is set. I have changed PCRE so
2276 that it faults them when PCRE_EXTRA is set.
2277
2278 5. Added support for \N, which always matches any character other than
2279 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2280
2281 6. When compiling pcregrep with newer versions of gcc which may have
2282 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2283 declared with attribute warn_unused_result" were given. Just casting the
2284 result to (void) does not stop the warnings; a more elaborate fudge is
2285 needed. I've used a macro to implement this.
2286
2287 7. Minor change to pcretest.c to avoid a compiler warning.
2288
2289 8. Added four artifical Unicode properties to help with an option to make
2290 \s etc use properties (see next item). The new properties are: Xan
2291 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2292
2293 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2294 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2295 this option. Modified pcretest to add /W to test this facility. Added
2296 REG_UCP to make it available via the POSIX interface.
2297
2298 10. Added --line-buffered to pcregrep.
2299
2300 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2301 studied, and the match started with a letter with a code point greater than
2302 127 whose first byte was different to the first byte of the other case of
2303 the letter, the other case of this starting letter was not recognized
2304 (#976).
2305
2306 12. If a pattern that was studied started with a repeated Unicode property
2307 test, for example, \p{Nd}+, there was the theoretical possibility of
2308 setting up an incorrect bitmap of starting bytes, but fortunately it could
2309 not have actually happened in practice until change 8 above was made (it
2310 added property types that matched character-matching opcodes).
2311
2312 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2313 possible starting bytes for non-anchored patterns.
2314
2315 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2316 \R, and also a number of cases that involve Unicode properties, both
2317 explicit and implicit when PCRE_UCP is set.
2318
2319 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2320 input, it could crash or give wrong results if characters with values
2321 greater than 0xc0 were present in the subject string. (Detail: it assumed
2322 UTF-8 input when processing these items.)
2323
2324 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2325 size_t is 64-bit (#991).
2326
2327 17. Added a check for running out of memory when PCRE is compiled with
2328 --disable-stack-for-recursion (#990).
2329
2330 18. If the last data line in a file for pcretest does not have a newline on
2331 the end, a newline was missing in the output.
2332
2333 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2334 less than 128) in its various bitmaps. However, there is a facility for
2335 generating tables according to the current locale when PCRE is compiled. It
2336 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2337 characters, are recognized by isspace() and therefore were getting set in
2338 these tables, and indeed these tables seem to approximate to ISO 8859. This
2339 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2340 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2341 which of course cannot start UTF-8 characters. I have changed the code so
2342 that only real ASCII characters (less than 128) and the correct starting
2343 bytes for UTF-8 encodings are set for characters greater than 127 when in
2344 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2345 altogether.)
2346
2347 20. Added the /T option to pcretest so as to be able to run tests with non-
2348 standard character tables, thus making it possible to include the tests
2349 used for 19 above in the standard set of tests.
2350
2351 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2352 reference to a subpattern the other side of a comment that contains an
2353 opening parenthesis caused either an internal compiling error, or a
2354 reference to the wrong subpattern.
2355
2356
2357 Version 8.02 19-Mar-2010
2358 ------------------------
2359
2360 1. The Unicode data tables have been updated to Unicode 5.2.0.
2361
2362 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2363 configured.
2364
2365 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2366 original author of that file, following a query about its status.
2367
2368 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2369 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2370
2371 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2372 quantifier applied to a forward-referencing subroutine call, could compile
2373 incorrect code or give the error "internal error: previously-checked
2374 referenced subpattern not found".
2375
2376 6. Both MS Visual Studio and Symbian OS have problems with initializing
2377 variables to point to external functions. For these systems, therefore,
2378 pcre_malloc etc. are now initialized to local functions that call the
2379 relevant global functions.
2380
2381 7. There were two entries missing in the vectors called coptable and poptable
2382 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2383 I've fixed the data, and added a kludgy way of testing at compile time that
2384 the lengths are correct (equal to the number of opcodes).
2385
2386 8. Following on from 7, I added a similar kludge to check the length of the
2387 eint vector in pcreposix.c.
2388
2389 9. Error texts for pcre_compile() are held as one long string to avoid too
2390 much relocation at load time. To find a text, the string is searched,
2391 counting zeros. There was no check for running off the end of the string,
2392 which could happen if a new error number was added without updating the
2393 string.
2394
2395 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2396
2397 11. \K was not working if it appeared in an atomic group or in a group that
2398 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2399 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2400 the assertion is positive, but not if it is negative.
2401
2402 12. Change 11 fortuitously reduced the size of the stack frame used in the
2403 "match()" function of pcre_exec.c by one pointer. Forthcoming
2404 implementation of support for (*MARK) will need an extra pointer on the
2405 stack; I have reserved it now, so that the stack frame size does not
2406 decrease.
2407
2408 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2409 item in branch that calls a recursion is a subroutine call - as in the
2410 second branch in the above example - was incorrectly given the compile-
2411 time error "recursive call could loop indefinitely" because pcre_compile()
2412 was not correctly checking the subroutine for matching a non-empty string.
2413
2414 14. The checks for overrunning compiling workspace could trigger after an
2415 overrun had occurred. This is a "should never occur" error, but it can be
2416 triggered by pathological patterns such as hundreds of nested parentheses.
2417 The checks now trigger 100 bytes before the end of the workspace.
2418
2419 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2420
2421
2422 Version 8.01 19-Jan-2010
2423 ------------------------
2424
2425 1. If a pattern contained a conditional subpattern with only one branch (in
2426 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2427 computed the wrong minimum data length (which is of course zero for such
2428 subpatterns). This could cause incorrect "no match" results.
2429
2430 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2431 the pattern is reset in the first branch, pcre_compile() failed with
2432 "internal error: code overflow at offset...". This happened only when
2433 the reset was to the original external option setting. (An optimization
2434 abstracts leading options settings into an external setting, which was the
2435 cause of this.)
2436
2437 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2438 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2439 assertion pattern did not match (meaning that the assertion was true), it
2440 was incorrectly treated as false if the SKIP had been reached during the
2441 matching. This also applied to assertions used as conditions.
2442
2443 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2444 assertion subpattern, including such a pattern used as a condition,
2445 unpredictable results occurred, instead of the error return
2446 PCRE_ERROR_DFA_UITEM.
2447
2448 5. The C++ GlobalReplace function was not working like Perl for the special
2449 situation when an empty string is matched. It now does the fancy magic
2450 stuff that is necessary.
2451
2452 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2453 removed. (These were left over from very, very early versions of PCRE.)
2454
2455 7. Some cosmetic changes to the code to make life easier when compiling it
2456 as part of something else:
2457
2458 (a) Change DEBUG to PCRE_DEBUG.
2459
2460 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2461 called "current" as "current_branch", to prevent a collision with the
2462 Linux macro when compiled as a kernel module.
2463
2464 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2465 prevent a collision with the Linux macro when compiled as a kernel
2466 module.
2467
2468 8. In pcre_compile() there are some checks for integer overflows that used to
2469 cast potentially large values to (double). This has been changed to that
2470 when building, a check for int64_t is made, and if it is found, it is used
2471 instead, thus avoiding the use of floating point arithmetic. (There is no
2472 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2473 double.
2474
2475 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2476 2005 (difference between two addresses compared to an unsigned value).
2477
2478 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2479 custom one, because of the following reported problem in Windows:
2480
2481 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2482 under Win32.
2483 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2484 therefore missing the function definition.
2485 - The compiler thus generates a "C" signature for the test function.
2486 - The linker fails to find the "C" function.
2487 - PCRE fails to configure if asked to do so against libbz2.
2488
2489 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2490 messages were output:
2491
2492 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2493 rerunning libtoolize, to keep the correct libtool macros in-tree.
2494 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2495
2496 I have done both of these things.
2497
2498 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2499 most of the time, it *can* run out if it is given a pattern that contains a
2500 runaway infinite recursion. I updated the discussion in the pcrestack man
2501 page.
2502
2503 13. Now that we have gone to the x.xx style of version numbers, the minor
2504 version may start with zero. Using 08 or 09 is a bad idea because users
2505 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2506 interpreted as invalid octal numbers. I've updated the previous comment in
2507 configure.ac, and also added a check that gives an error if 08 or 09 are
2508 used.
2509
2510 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2511 causing partial matching to fail when the end of the subject matched \W
2512 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2513
2514 15. There were some discrepancies between the declarations in pcre_internal.h
2515 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2516 their definitions. The declarations used "const uschar *" and the
2517 definitions used USPTR. Even though USPTR is normally defined as "const
2518 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2519 reported that: "This difference in casting confuses some C++ compilers, for
2520 example, SunCC recognizes above declarations as different functions and
2521 generates broken code for hbpcre." I have changed the declarations to use
2522 USPTR.
2523
2524 16. GNU libtool is named differently on some systems. The autogen.sh script now
2525 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2526 (FreeBSD).
2527
2528 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2529 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2530 comment: "Figure out how to create a longlong from a string: strtoll and
2531 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2532 instance, but it only takes 2 args instead of 3!"
2533
2534 18. A subtle bug concerned with back references has been fixed by a change of
2535 specification, with a corresponding code fix. A pattern such as
2536 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2537 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2538 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2539 same bug. Such groups have to be quantified to be useful, or contained
2540 inside another quantified group. (If there's no repetition, the reference
2541 can never match.) The problem arises because, having left the group and
2542 moved on to the rest of the pattern, a later failure that backtracks into
2543 the group uses the captured value from the final iteration of the group
2544 rather than the correct earlier one. I have fixed this in PCRE by forcing
2545 any group that contains a reference to itself to be an atomic group; that
2546 is, there cannot be any backtracking into it once it has completed. This is
2547 similar to recursive and subroutine calls.
2548
2549
2550 Version 8.00 19-Oct-09
2551 ----------------------
2552
2553 1. The table for translating pcre_compile() error codes into POSIX error codes
2554 was out-of-date, and there was no check on the pcre_compile() error code
2555 being within the table. This could lead to an OK return being given in
2556 error.
2557
2558 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2559 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2560 in a Windows environment.
2561
2562 3. The pcregrep --count option prints the count for each file even when it is
2563 zero, as does GNU grep. However, pcregrep was also printing all files when
2564 --files-with-matches was added. Now, when both options are given, it prints
2565 counts only for those files that have at least one match. (GNU grep just
2566 prints the file name in this circumstance, but including the count seems
2567 more useful - otherwise, why use --count?) Also ensured that the
2568 combination -clh just lists non-zero counts, with no names.
2569
2570 4. The long form of the pcregrep -F option was incorrectly implemented as
2571 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2572 but it seems right to fix it, and I didn't think it was worth preserving
2573 the old behaviour.
2574
2575 5. The command line items --regex=pattern and --regexp=pattern were not
2576 recognized by pcregrep, which required --regex pattern or --regexp pattern
2577 (with a space rather than an '='). The man page documented the '=' forms,
2578 which are compatible with GNU grep; these now work.
2579
2580 6. No libpcreposix.pc file was created for pkg-config; there was just
2581 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2582
2583 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2584 when UCP support is not needed, by modifying the Python script that
2585 generates it from Unicode data files. This should not matter if the module
2586 is correctly used as a library, but I received one complaint about 50K of
2587 unwanted data. My guess is that the person linked everything into his
2588 program rather than using a library. Anyway, it does no harm.
2589
2590 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2591 was a minimum greater than 1 for a wide character in a possessive
2592 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2593 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2594 character. Chaos in the form of incorrect output or a compiling loop could
2595 result.
2596
2597 9. The restrictions on what a pattern can contain when partial matching is
2598 requested for pcre_exec() have been removed. All patterns can now be
2599 partially matched by this function. In addition, if there are at least two
2600 slots in the offset vector, the offset of the earliest inspected character
2601 for the match and the offset of the end of the subject are set in them when
2602 PCRE_ERROR_PARTIAL is returned.
2603
2604 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2605 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2606 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2607 and may be more useful for multi-segment matching.
2608
2609 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2610 used to be given if ever the end of the subject was reached; now it is
2611 given only if matching could not proceed because another character was
2612 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2613 string "Z", which now yields "no match" instead of "partial match". In the
2614 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2615 final character ended with (*FAIL).
2616
2617 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2618 if the pattern had a "must contain" character that was already found in the
2619 earlier partial match, unless partial matching was again requested. For
2620 example, with the pattern /dog.(body)?/, the "must contain" character is
2621 "g". If the first part-match was for the string "dog", restarting with
2622 "sbody" failed. This bug has been fixed.
2623
2624 13. The string returned by pcre_dfa_exec() after a partial match has been
2625 changed so that it starts at the first inspected character rather than the
2626 first character of the match. This makes a difference only if the pattern
2627 starts with a lookbehind assertion or \b or \B (\K is not supported by
2628 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2629 matching functions compatible, and I think it's the right thing to do.
2630
2631 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2632 so that the demonstration program is easily available in environments where
2633 PCRE has not been installed from source.
2634
2635 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2636 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2637 library.
2638
2639 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2640 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2641 is not the first non-POSIX option to be added. Clearly some people find
2642 these options useful.
2643
2644 17. If a caller to the POSIX matching function regexec() passes a non-zero
2645 value for nmatch with a NULL value for pmatch, the value of
2646 nmatch is forced to zero.
2647
2648 18. RunGrepTest did not have a test for the availability of the -u option of
2649 the diff command, as RunTest does. It now checks in the same way as
2650 RunTest, and also checks for the -b option.
2651
2652 19. If an odd number of negated classes containing just a single character
2653 interposed, within parentheses, between a forward reference to a named
2654 subpattern and the definition of the subpattern, compilation crashed with
2655 an internal error, complaining that it could not find the referenced
2656 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2657 [The bug was that it was starting one character too far in when skipping
2658 over the character class, thus treating the ] as data rather than
2659 terminating the class. This meant it could skip too much.]
2660
2661 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2662 /g option in pcretest when the pattern contains \K, which makes it possible
2663 to have an empty string match not at the start, even when the pattern is
2664 anchored. Updated pcretest and pcredemo to use this option.
2665
2666 21. If the maximum number of capturing subpatterns in a recursion was greater
2667 than the maximum at the outer level, the higher number was returned, but
2668 with unset values at the outer level. The correct (outer level) value is
2669 now given.
2670
2671 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2672 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2673 make it do so. The string so far is captured, making this feature
2674 compatible with Perl.
2675
2676 23. The tests have been re-organized, adding tests 11 and 12, to make it
2677 possible to check the Perl 5.10 features against Perl 5.10.
2678
2679 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2680 pattern matches a fixed length string. PCRE did not allow this; now it
2681 does. Neither allows recursion.
2682
2683 25. I finally figured out how to implement a request to provide the minimum
2684 length of subject string that was needed in order to match a given pattern.
2685 (It was back references and recursion that I had previously got hung up
2686 on.) This code has now been added to pcre_study(); it finds a lower bound
2687 to the length of subject needed. It is not necessarily the greatest lower
2688 bound, but using it to avoid searching strings that are too short does give
2689 some useful speed-ups. The value is available to calling programs via
2690 pcre_fullinfo().
2691
2692 26. While implementing 25, I discovered to my embarrassment that pcretest had
2693 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2694 study optimizations had never been tested with that matching function.
2695 Oops. What is worse, even when it was passed study data, there was a bug in
2696 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2697 were also very few tests of studied patterns with pcre_dfa_exec().
2698
2699 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2700 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2701 on the other side of the coin, they are no longer allowed to have different
2702 names, because these cannot be distinguished in PCRE, and this has caused
2703 confusion. (This is a difference from Perl.)
2704
2705 28. When duplicate subpattern names are present (necessarily with different
2706 numbers, as required by 27 above), and a test is made by name in a
2707 conditional pattern, either for a subpattern having been matched, or for
2708 recursion in such a pattern, all the associated numbered subpatterns are
2709 tested, and the overall condition is true if the condition is true for any
2710 one of them. This is the way Perl works, and is also more like the way
2711 testing by number works.
2712
2713
2714 Version 7.9 11-Apr-09
2715 ---------------------
2716
2717 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2718 (pcretest), all targets were linked against these libraries. This included
2719 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2720 libraries. This caused unwanted dependencies to be created. This problem
2721 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2722 pcretest is linked with readline.
2723
2724 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2725 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2726 moved outside it again, because FALSE and TRUE are already defined in AIX,
2727 but BOOL is not.
2728
2729 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2730 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2731
2732 4. The pcregrep documentation said spaces were inserted as well as colons (or
2733 hyphens) following file names and line numbers when outputting matching
2734 lines. This is not true; no spaces are inserted. I have also clarified the
2735 wording for the --colour (or --color) option.
2736
2737 5. In pcregrep, when --colour was used with -o, the list of matching strings
2738 was not coloured; this is different to GNU grep, so I have changed it to be
2739 the same.
2740
2741 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2742 each matching line was coloured. Now it goes on to look for further matches
2743 of any of the test patterns, which is the same behaviour as GNU grep.
2744
2745 7. A pattern that could match an empty string could cause pcregrep to loop; it
2746 doesn't make sense to accept an empty string match in pcregrep, so I have
2747 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2748 seems to be how GNU grep behaves. [But see later change 40 for release
2749 8.33.]
2750
2751 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2752 start or after a newline", because the conditional assertion was not being
2753 correctly handled. The rule now is that both the assertion and what follows
2754 in the first alternative must satisfy the test.
2755
2756 9. If auto-callout was enabled in a pattern with a conditional group whose
2757 condition was an assertion, PCRE could crash during matching, both with
2758 pcre_exec() and pcre_dfa_exec().
2759
2760 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2761 used for matching.
2762
2763 11. Unicode property support in character classes was not working for
2764 characters (bytes) greater than 127 when not in UTF-8 mode.
2765
2766 12. Added the -M command line option to pcretest.
2767
2768 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2769
2770 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2771
2772 16. Added comments and documentation about mis-use of no_arg in the C++
2773 wrapper.
2774
2775 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2776 from Martin Jerabek that uses macro names for all relevant character and
2777 string constants.
2778
2779 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2780 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2781 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2782 these, but not everybody uses configure.
2783
2784 19. A conditional group that had only one branch was not being correctly
2785 recognized as an item that could match an empty string. This meant that an
2786 enclosing group might also not be so recognized, causing infinite looping
2787 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2788 with the subject "ab", where knowledge that the repeated group can match
2789 nothing is needed in order to break the loop.
2790
2791 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2792 exec(), but without supplying a callout function, matching went wrong.
2793
2794 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2795 leak if the size of the offset vector was greater than 30. When the vector
2796 is smaller, the saved offsets during recursion go onto a local stack
2797 vector, but for larger vectors malloc() is used. It was failing to free
2798 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2799 error, in fact).
2800
2801 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2802 heapframe that is used only when UTF-8 support is enabled. This caused no
2803 problem, but was untidy.
2804
2805 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2806 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2807 included within another project.
2808
2809 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2810 slightly modified by me:
2811
2812 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2813 not building pcregrep.
2814
2815 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2816 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2817
2818 25. Forward references, both numeric and by name, in patterns that made use of
2819 duplicate group numbers, could behave incorrectly or give incorrect errors,
2820 because when scanning forward to find the reference group, PCRE was not
2821 taking into account the duplicate group numbers. A pattern such as
2822 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2823
2824 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2825 the feature of a custom pointer more persuasive (as requested by a user).
2826
2827 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2828 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2829 pre-defined.
2830
2831 28. Added support for (*UTF8) at the start of a pattern.
2832
2833 29. Arrange for flags added by the "release type" setting in CMake to be shown
2834 in the configuration summary.
2835
2836
2837 Version 7.8 05-Sep-08
2838 ---------------------
2839
2840 1. Replaced UCP searching code with optimized version as implemented for Ad
2841 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2842 stage table and inline lookup instead of a function, giving speed ups of 2
2843 to 5 times on some simple patterns that I tested. Permission was given to
2844 distribute the MultiStage2.py script that generates the tables (it's not in
2845 the tarball, but is in the Subversion repository).
2846
2847 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2848 scripts.
2849
2850 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2851 a group with a zero qualifier. The result of the study could be incorrect,
2852 or the function might crash, depending on the pattern.
2853
2854 4. Caseless matching was not working for non-ASCII characters in back
2855 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2856 It now works when Unicode Property Support is available.
2857
2858 5. In pcretest, an escape such as \x{de} in the data was always generating
2859 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2860 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2861 truncation.
2862
2863 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2864
2865 7. Added two (int) casts to pcregrep when printing the difference of two
2866 pointers, in case they are 64-bit values.
2867
2868 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2869 test 2 if it fails.
2870
2871 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2872 and a #define of that name to empty if it is not externally set. This is to
2873 allow users of MSVC to set it if necessary.
2874
2875 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2876 the convenience functions in the pcre_get.c source file.
2877
2878 11. An option change at the start of a pattern that had top-level alternatives
2879 could cause overwriting and/or a crash. This command provoked a crash in
2880 some environments:
2881
2882 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2883
2884 This potential security problem was recorded as CVE-2008-2371.
2885
2886 12. For a pattern where the match had to start at the beginning or immediately
2887 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2888 pcre_dfa_exec() could read past the end of the passed subject if there was
2889 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2890 pcretest so that it places the subject at the end of its malloc-ed buffer.
2891
2892 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2893 exec() might read past the end of the data buffer in UTF-8 mode.
2894
2895 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2896 the data contained the byte 0x85 as part of a UTF-8 character within its
2897 first line. This applied both to normal and DFA matching.
2898
2899 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2900 /^[^d]*?$/8 failed to match "abc".
2901
2902 16. Added a missing copyright notice to pcrecpp_internal.h.
2903
2904 17. Make it more clear in the documentation that values returned from
2905 pcre_exec() in ovector are byte offsets, not character counts.
2906
2907 18. Tidied a few places to stop certain compilers from issuing warnings.
2908
2909 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2910 supplied by Stefan Weber. I made a further small update for 7.8 because
2911 there is a change of source arrangements: the pcre_searchfuncs.c module is
2912 replaced by pcre_ucd.c.
2913
2914
2915 Version 7.7 07-May-08
2916 ---------------------
2917
2918 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2919 a string to a long long, pretend we don't even have a long long." This is
2920 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2921
2922 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2923 pre-7.6 versions, which defined a global no_arg variable instead of putting
2924 it in the RE class. (See also #8 below.)
2925
2926 3. Remove a line of dead code, identified by coverity and reported by Nuno
2927 Lopes.
2928
2929 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2930
2931 (1) The include/exclude patterns were being applied to the whole pathnames
2932 of files, instead of just to the final components.
2933
2934 (2) If there was more than one level of directory, the subdirectories were
2935 skipped unless they satisfied the include/exclude conditions. This is
2936 inconsistent with GNU grep (and could even be seen as contrary to the
2937 pcregrep specification - which I improved to make it absolutely clear).
2938 The action now is always to scan all levels of directory, and just
2939 apply the include/exclude patterns to regular files.
2940
2941 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2942 --exclude_dir in the tests to avoid scanning .svn directories.
2943
2944 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2945 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2946 doesn't support NULs in patterns.
2947
2948 7. Added some missing "const"s to declarations of static tables in
2949 pcre_compile.c and pcre_dfa_exec.c.
2950
2951 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2952 caused by fix #2 above. (Subsequently also a second patch to fix the
2953 first patch. And a third patch - this was a messy problem.)
2954
2955 9. Applied Craig's patch to remove the use of push_back().
2956
2957 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2958 matching function regexec().
2959
2960 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2961 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2962 references. PCRE supports relative numbers with this syntax (I don't think
2963 Oniguruma does).
2964
2965 12. Previously, a group with a zero repeat such as (...){0} was completely
2966 omitted from the compiled regex. However, this means that if the group
2967 was called as a subroutine from elsewhere in the pattern, things went wrong
2968 (an internal error was given). Such groups are now left in the compiled
2969 pattern, with a new opcode that causes them to be skipped at execution
2970 time.
2971
2972 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2973 to the way PCRE behaves:
2974
2975 (a) A lone ] character is dis-allowed (Perl treats it as data).
2976
2977 (b) A back reference to an unmatched subpattern matches an empty string
2978 (Perl fails the current match path).
2979
2980 (c) A data ] in a character class must be notated as \] because if the
2981 first data character in a class is ], it defines an empty class. (In
2982 Perl it is not possible to have an empty class.) The empty class []
2983 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2984 The negative empty class [^] matches any one character, independently
2985 of the DOTALL setting.
2986
2987 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2988 non-existent subpattern following a character class starting with ']' and
2989 containing () gave an internal compiling error instead of "reference to
2990 non-existent subpattern". Fortunately, when the pattern did exist, the
2991 compiled code was correct. (When scanning forwards to check for the
2992 existence of the subpattern, it was treating the data ']' as terminating
2993 the class, so got the count wrong. When actually compiling, the reference
2994 was subsequently set up correctly.)
2995
2996 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2997 it was being rejected as not supported by pcre_dfa_exec(), even though
2998 other assertions are supported. I have made pcre_dfa_exec() support
2999 (*FAIL).
3000
3001 16. The implementation of 13c above involved the invention of a new opcode,
3002 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
3003 cannot be changed at match time, I realized I could make a small
3004 improvement to matching performance by compiling OP_ALLANY instead of
3005 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
3006 on the OP_ANY path.
3007
3008 17. Compiling pcretest on Windows with readline support failed without the
3009 following two fixes: (1) Make the unistd.h include conditional on
3010 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
3011
3012 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
3013 ncurses library to be included for pcretest when ReadLine support is
3014 requested, but also to allow for it to be overridden. This patch came from
3015 Daniel Bergström.
3016
3017 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
3018 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
3019 any errors with the current Unicode tables. Thanks to Peter Kankowski for
3020 spotting this.
3021
3022
3023 Version 7.6 28-Jan-08
3024 ---------------------
3025
3026 1. A character class containing a very large number of characters with
3027 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
3028 overflow.
3029
3030 2. Patch to cut out the "long long" test in pcrecpp_unittest when
3031 HAVE_LONG_LONG is not defined.
3032
3033 3. Applied Christian Ehrlicher's patch to update the CMake build files to
3034 bring them up to date and include new features. This patch includes:
3035
3036 - Fixed PH's badly added libz and libbz2 support.
3037 - Fixed a problem with static linking.
3038 - Added pcredemo. [But later removed - see 7 below.]
3039 - Fixed dftables problem and added an option.
3040 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
3041 HAVE_LONG_LONG.
3042 - Added readline support for pcretest.
3043 - Added an listing of the option settings after cmake has run.
3044
3045 4. A user submitted a patch to Makefile that makes it easy to create
3046 "pcre.dll" under mingw when using Configure/Make. I added stuff to
3047 Makefile.am that cause it to include this special target, without
3048 affecting anything else. Note that the same mingw target plus all
3049 the other distribution libraries and programs are now supported
3050 when configuring with CMake (see 6 below) instead of with
3051 Configure/Make.
3052
3053 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
3054 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
3055 exported in the Windows port". It has not yet been confirmed that the patch
3056 solves the problem, but it does no harm.
3057
3058 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
3059 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
3060 with CMake, and also correct the comment about stack recursion.
3061
3062 7. Remove the automatic building of pcredemo from the ./configure system and
3063 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
3064 of a program that users should build themselves after PCRE is installed, so
3065 building it automatically is not really right. What is more, it gave
3066 trouble in some build environments.
3067
3068 8. Further tidies to CMakeLists.txt from Sheri and Christian.
3069
3070
3071 Version 7.5 10-Jan-08
3072 ---------------------
3073
3074 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
3075 values in parens when parsing an RE using the C++ wrapper."
3076
3077 2. Negative specials like \S did not work in character classes in UTF-8 mode.
3078 Characters greater than 255 were excluded from the class instead of being
3079 included.
3080
3081 3. The same bug as (2) above applied to negated POSIX classes such as
3082 [:^space:].
3083
3084 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
3085 defined or documented. It seems to have been a typo for PCRE_STATIC, so
3086 I have changed it.
3087
3088 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
3089 first named subpattern) and a construct such as (?&a) would reference the
3090 first named subpattern whose name started with "a" (in other words, the
3091 length check was missing). Both these problems are fixed. "Subpattern name
3092 expected" is now given for (?&) (a zero-length name), and this patch also
3093 makes it give the same error for \k'' (previously it complained that that
3094 was a reference to a non-existent subpattern).
3095
3096 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
3097 this is right because (?- can be followed by option settings as well as by
3098 digits. I have, however, made the messages clearer.
3099
3100 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
3101 than the number used in the conditional) now cause a compile-time error.
3102 This is actually not compatible with Perl, which accepts such patterns, but
3103 treats the conditional as always being FALSE (as PCRE used to), but it
3104 seems to me that giving a diagnostic is better.
3105
3106 8. Change "alphameric" to the more common word "alphanumeric" in comments
3107 and messages.
3108
3109 9. Fix two occurrences of "backslash" in comments that should have been
3110 "backspace".
3111
3112 10. Remove two redundant lines of code that can never be obeyed (their function
3113 was moved elsewhere).
3114
3115 11. The program that makes PCRE's Unicode character property table had a bug
3116 which caused it to generate incorrect table entries for sequences of
3117 characters that have the same character type, but are in different scripts.
3118 It amalgamated them into a single range, with the script of the first of
3119 them. In other words, some characters were in the wrong script. There were
3120 thirteen such cases, affecting characters in the following ranges:
3121
3122 U+002b0 - U+002c1
3123 U+0060c - U+0060d
3124 U+0061e - U+00612
3125 U+0064b - U+0065e
3126 U+0074d - U+0076d
3127 U+01800 - U+01805
3128 U+01d00 - U+01d77
3129 U+01d9b - U+01dbf
3130 U+0200b - U+0200f
3131 U+030fc - U+030fe
3132 U+03260 - U+0327f
3133 U+0fb46 - U+0fbb1
3134 U+10450 - U+1049d
3135
3136 12. The -o option (show only the matching part of a line) for pcregrep was not
3137 compatible with GNU grep in that, if there was more than one match in a
3138 line, it showed only the first of them. It now behaves in the same way as
3139 GNU grep.
3140
3141 13. If the -o and -v options were combined for pcregrep, it printed a blank
3142 line for every non-matching line. GNU grep prints nothing, and pcregrep now
3143 does the same. The return code can be used to tell if there were any
3144 non-matching lines.
3145
3146 14. Added --file-offsets and --line-offsets to pcregrep.
3147
3148 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
3149 infinitely looping recursion. The bug was that positive lookaheads were not
3150 being skipped when checking for a possible empty match (negative lookaheads
3151 and both kinds of lookbehind were skipped).
3152
3153 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
3154 inclusion of <windows.h> to before rather than after the definition of
3155 INVALID_FILE_ATTRIBUTES (patch from David Byron).
3156
3157 17. Specifying a possessive quantifier with a specific limit for a Unicode
3158 character property caused pcre_compile() to compile bad code, which led at
3159 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
3160 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
3161 caused the error; without that there was no problem.
3162
3163 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3164
3165 19. Added --enable-pcretest-libreadline.
3166
3167 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3168 RE::GlobalReplace(). As a result, the number of replacements returned was
3169 double what it should be. I removed one of the increments, but Craig sent a
3170 later patch that removed the other one (the right fix) and added unit tests
3171 that check the return values (which was not done before).
3172
3173 21. Several CMake things:
3174
3175 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3176 the names libpcre and libpcreposix, not just pcre and pcreposix.
3177
3178 (2) The above change means that pcretest and pcregrep are now correctly
3179 linked with the newly-built libraries, not previously installed ones.
3180
3181 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
3182
3183 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3184 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3185 UTF-8 newline character). The key issue is that the pattern starts .*;
3186 this means that the match must be either at the beginning, or after a
3187 newline. The bug was in the code for advancing after a failed match and
3188 checking that the new position followed a newline. It was not taking
3189 account of UTF-8 characters correctly.
3190
3191 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3192 character classes. PCRE was not treating the sequence [:...:] as a
3193 character class unless the ... were all letters. Perl, however, seems to
3194 allow any characters between [: and :], though of course it rejects as
3195 unknown any "names" that contain non-letters, because all the known class
3196 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3197 for example, whereas PCRE did not - it did not recognize a POSIX character
3198 class. This seemed a bit dangerous, so the code has been changed to be
3199 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3200 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3201 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3202 Perl does, and where it didn't before.
3203
3204 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3205 Windows environments %n is disabled by default.
3206
3207
3208 Version 7.4 21-Sep-07
3209 ---------------------
3210
3211 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3212 means that a class such as [\s] counted as "explicit reference to CR or
3213 LF". That isn't really right - the whole point of the change was to try to
3214 help when there was an actual mention of one of the two characters. So now
3215 the change happens only if \r or \n (or a literal CR or LF) character is
3216 encountered.
3217
3218 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3219 of both had grown to the point where there were only 3 bits left.
3220 Fortunately, there was spare space in the data structure, and so I have
3221 moved the internal flags into a new 16-bit field to free up more option
3222 bits.
3223
3224 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3225 but did not set the internal JCHANGED flag - either of these is enough to
3226 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3227 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3228 start sets both bits.
3229
3230 4. Added options (at build time, compile time, exec time) to change \R from
3231 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3232
3233 5. doc/pcresyntax.html was missing from the distribution.
3234
3235 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3236 compatibility, even though it is no longer used.
3237
3238 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3239 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3240 windows.h file is present (where different names are used). [This was
3241 reversed later after testing - see 16 below.]
3242
3243 8. Changed all #include <config.h> to #include "config.h". There were also
3244 some further <pcre.h> cases that I changed to "pcre.h".
3245
3246 9. When pcregrep was used with the --colour option, it missed the line ending
3247 sequence off the lines that it output.
3248
3249 10. It was pointed out to me that arrays of string pointers cause lots of
3250 relocations when a shared library is dynamically loaded. A technique of
3251 using a single long string with a table of offsets can drastically reduce
3252 these. I have refactored PCRE in four places to do this. The result is
3253 dramatic:
3254
3255 Originally: 290
3256 After changing UCP table: 187
3257 After changing error message table: 43
3258 After changing table of "verbs" 36
3259 After changing table of Posix names 22
3260
3261 Thanks to the folks working on Gregex for glib for this insight.
3262
3263 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3264 unicode-properties was also set.
3265
3266 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3267
3268 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3269 checked only for CRLF.
3270
3271 14. Added casts to pcretest.c to avoid compiler warnings.
3272
3273 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3274
3275 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3276 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3277 entirely. This removes changes made in 7 above.
3278
3279 17. The CMake files have been updated, and there is now more information about
3280 building with CMake in the NON-UNIX-USE document.
3281
3282
3283 Version 7.3 28-Aug-07
3284 ---------------------
3285
3286 1. In the rejigging of the build system that eventually resulted in 7.1, the
3287 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3288 brackets there is not right, since it causes compilers to look for an
3289 installed pcre.h, not the version that is in the source that is being
3290 compiled (which of course may be different). I have changed it back to:
3291
3292 #include "pcre.h"
3293
3294 I have a vague recollection that the change was concerned with compiling in
3295 different directories, but in the new build system, that is taken care of
3296 by the VPATH setting the Makefile.
3297
3298 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3299 when the subject happened to end in the byte 0x85 (e.g. if the last
3300 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3301 characters but of course it shouldn't be taken as a newline when it is part
3302 of another character. The bug was that, for an unlimited repeat of . in
3303 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3304 characters when looking for a newline.
3305
3306 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3307
3308 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3309 in debug output.
3310
3311 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3312 long printing in the pcrecpp unittest when running under MinGW.
3313
3314 6. ESC_K was left out of the EBCDIC table.
3315
3316 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3317 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3318 limit also applies to "virtual nesting" when a pattern is recursive, and in
3319 this case 1000 isn't so big. I have been able to remove this limit at the
3320 expense of backing off one optimization in certain circumstances. Normally,
3321 when pcre_exec() would call its internal match() function recursively and
3322 immediately return the result unconditionally, it uses a "tail recursion"
3323 feature to save stack. However, when a subpattern that can match an empty
3324 string has an unlimited repetition quantifier, it no longer makes this
3325 optimization. That gives it a stack frame in which to save the data for
3326 checking that an empty string has been matched. Previously this was taken
3327 from the 1000-entry workspace that had been reserved. So now there is no
3328 explicit limit, but more stack is used.
3329
3330 8. Applied Daniel's patches to solve problems with the import/export magic
3331 syntax that is required for Windows, and which was going wrong for the
3332 pcreposix and pcrecpp parts of the library. These were overlooked when this
3333 problem was solved for the main library.
3334
3335 9. There were some crude static tests to avoid integer overflow when computing
3336 the size of patterns that contain repeated groups with explicit upper
3337 limits. As the maximum quantifier is 65535, the maximum group length was
3338 set at 30,000 so that the product of these two numbers did not overflow a
3339 32-bit integer. However, it turns out that people want to use groups that
3340 are longer than 30,000 bytes (though not repeat them that many times).
3341 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3342 made it possible to implement the integer overflow checks in a much more
3343 dynamic way, which I have now done. The artificial limitation on group
3344 length has been removed - we now have only the limit on the total length of
3345 the compiled pattern, which depends on the LINK_SIZE setting.
3346
3347 10. Fixed a bug in the documentation for get/copy named substring when
3348 duplicate names are permitted. If none of the named substrings are set, the
3349 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3350 empty string.
3351
3352 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3353 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3354 because the ] is interpreted as the first data character and the
3355 terminating ] is not found. PCRE has been made compatible with Perl in this
3356 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3357 cause memory overwriting.
3358
3359 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3360 string has been matched (to stop an infinite loop). It was not recognizing
3361 a conditional subpattern that could match an empty string if that
3362 subpattern was within another subpattern. For example, it looped when
3363 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3364 condition was not nested. This bug has been fixed.
3365
3366 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3367 past the start of the subject in the presence of bytes with the top bit
3368 set, for example "\x8aBCD".
3369
3370 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3371 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3372
3373 14. Optimized (?!) to (*FAIL).
3374
3375 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3376 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3377 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3378 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3379 does: it's just the validity check that is more restrictive.
3380
3381 16. Inserted checks for integer overflows during escape sequence (backslash)
3382 processing, and also fixed erroneous offset values for syntax errors during
3383 backslash processing.
3384
3385 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3386 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3387
3388 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3389 caused an overrun.
3390
3391 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3392 something other than just ASCII characters) inside a group that had an
3393 unlimited repeat caused a loop at compile time (while checking to see
3394 whether the group could match an empty string).
3395
3396 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3397 [\P{Any}] did so. (Error in the code for printing property names.)
3398
3399 21. An orphan \E inside a character class could cause a crash.
3400
3401 22. A repeated capturing bracket such as (A)? could cause a wild memory
3402 reference during compilation.
3403
3404 23. There are several functions in pcre_compile() that scan along a compiled
3405 expression for various reasons (e.g. to see if it's fixed length for look
3406 behind). There were bugs in these functions when a repeated \p or \P was
3407 present in the pattern. These operators have additional parameters compared
3408 with \d, etc, and these were not being taken into account when moving along
3409 the compiled data. Specifically:
3410
3411 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3412 length.
3413
3414 (b) An item such as \pL+ within a repeated group could cause crashes or
3415 loops.
3416
3417 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3418 "reference to non-existent subpattern" error.
3419
3420 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3421
3422 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3423 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3424
3425 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3426
3427 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3428 character were causing crashes (broken optimization).
3429
3430 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3431 \p or \P) caused a compile-time loop.
3432
3433 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3434 break. For example, the unstudied pattern [\r\n]A does not match the string
3435 "\r\nA" because change 7.0/46 below moves the current point on by two
3436 characters after failing to match at the start. However, the pattern \nA
3437 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3438 the same is true. There doesn't seem any very clean way out of this, but
3439 what I have chosen to do makes the common cases work: PCRE now takes note
3440 of whether there can be an explicit match for \r or \n anywhere in the
3441 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3442 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3443 pattern has explicit CR or LF references.
3444
3445 29. Added (*CR) etc for changing newline setting at start of pattern.
3446
3447
3448 Version 7.2 19-Jun-07
3449 ---------------------
3450
3451 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3452 which is apparently normally available under Windows.
3453
3454 2. Re-jig the pcregrep tests with different newline settings in an attempt
3455 to make them independent of the local environment's newline setting.
3456
3457 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3458
3459 4. Some of the "internals" tests were previously cut out when the link size
3460 was not 2, because the output contained actual offsets. The recent new
3461 "Z" feature of pcretest means that these can be cut out, making the tests
3462 usable with all link sizes.
3463
3464 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3465 stack recursion. This gives a massive performance boost under BSD, but just
3466 a small improvement under Linux. However, it saves one field in the frame
3467 in all cases.
3468
3469 6. Added more features from the forthcoming Perl 5.10:
3470
3471 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3472 recursion call. It refers to the nth most recently opened parentheses.
3473
3474 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3475 to be opened parentheses.
3476
3477 (c) Conditions that refer to capturing parentheses can be specified
3478 relatively, for example, (?(-2)... or (?(+3)...
3479
3480 (d) \K resets the start of the current match so that everything before
3481 is not part of it.
3482
3483 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3484
3485 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3486 reference syntax.
3487
3488 (g) (?| introduces a group in which the numbering of parentheses in each
3489 alternative starts with the same number.
3490
3491 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3492
3493 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3494 PCRE_INFO_JCHANGED.
3495
3496 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3497 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3498 for detecting groups that can match an empty string.
3499
3500 9. A pattern with a very large number of alternatives (more than several
3501 hundred) was running out of internal workspace during the pre-compile
3502 phase, where pcre_compile() figures out how much memory will be needed. A
3503 bit of new cunning has reduced the workspace needed for groups with
3504 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3505 workspace instead of running out of the 4096 that are available.
3506
3507 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3508
3509 11. Applied patch from Google to remove an optimization that didn't quite work.
3510 The report of the bug said:
3511
3512 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3513 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3514 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3515
3516 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3517 it matched the wrong number of bytes.
3518
3519
3520 Version 7.1 24-Apr-07
3521 ---------------------
3522
3523 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3524 that is more "standard", making use of automake and other Autotools. There
3525 is some re-arrangement of the files and adjustment of comments consequent
3526 on this.
3527
3528 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3529 for recursive directory scanning broke on some systems because the files
3530 are not scanned in any specific order and on different systems the order
3531 was different. A call to "sort" has been inserted into RunGrepTest for the
3532 approprate test as a short-term fix. In the longer term there may be an
3533 alternative.
3534
3535 3. I had an email from Eric Raymond about problems translating some of PCRE's
3536 man pages to HTML (despite the fact that I distribute HTML pages, some
3537 people do their own conversions for various reasons). The problems
3538 concerned the use of low-level troff macros .br and .in. I have therefore
3539 removed all such uses from the man pages (some were redundant, some could
3540 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3541 HTML has been updated to handle .nf/.fi and to complain if it encounters
3542 .br or .in.
3543
3544 4. Updated comments in configure.ac that get placed in config.h.in and also
3545 arranged for config.h to be included in the distribution, with the name
3546 config.h.generic, for the benefit of those who have to compile without
3547 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3548
3549 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3550 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3551 makevp.bat for the current PCRE, using the additional files
3552 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3553
3554 6. A Windows user reported a minor discrepancy with test 2, which turned out
3555 to be caused by a trailing space on an input line that had got lost in his
3556 copy. The trailing space was an accident, so I've just removed it.
3557
3558 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3559 that is needed.
3560
3561 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3562 as "const" (a) because they are and (b) because it helps the PHP
3563 maintainers who have recently made a script to detect big data structures
3564 in the php code that should be moved to the .rodata section. I remembered
3565 to update Builducptable as well, so it won't revert if ucptable.h is ever
3566 re-created.
3567
3568 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3569 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3570 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3571 support is not required. This saves 1.5-2K of code, which is important in
3572 some applications.
3573
3574 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3575 so as not to refer to the tables, even though these functions will never be
3576 called when UTF-8 support is disabled. Otherwise there are problems with a
3577 shared library.
3578
3579 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3580
3581 (a) It was defining its arguments as char * instead of void *.
3582
3583 (b) It was assuming that all moves were upwards in memory; this was true
3584 a long time ago when I wrote it, but is no longer the case.
3585
3586 The emulated memove() is provided for those environments that have neither
3587 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3588 is clearly not the case, as these two bugs were recently reported.
3589
3590 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3591 and Detrail to create the HTML documentation, the .txt form of the man
3592 pages, and it removes trailing spaces from listed files. It also creates
3593 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3594 case, it wraps all the #defines with #ifndefs. This script should be run
3595 before "make dist".
3596
3597 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3598 with Unicode property support.
3599
3600 (a) For a maximizing quantifier, if the two different cases of the
3601 character were of different lengths in their UTF-8 codings (there are
3602 some cases like this - I found 11), and the matching function had to
3603 back up over a mixture of the two cases, it incorrectly assumed they
3604 were both the same length.
3605
3606 (b) When PCRE was configured to use the heap rather than the stack for
3607 recursion during matching, it was not correctly preserving the data for
3608 the other case of a UTF-8 character when checking ahead for a match
3609 while processing a minimizing repeat. If the check also involved
3610 matching a wide character, but failed, corruption could cause an
3611 erroneous result when trying to check for a repeat of the original
3612 character.
3613
3614 13. Some tidying changes to the testing mechanism:
3615
3616 (a) The RunTest script now detects the internal link size and whether there
3617 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3618 values substituted by "configure". (The RunGrepTest script already did
3619 this for UTF-8.) The configure.ac script no longer substitutes the
3620 relevant variables.
3621
3622 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3623 with length and offset values. This means that the output is different
3624 for different internal link sizes. Test 2 is skipped for link sizes
3625 other than 2 because of this, bypassing the problem. Unfortunately,
3626 there was also a test in test 3 (the locale tests) that used /B and
3627 failed for link sizes other than 2. Rather than cut the whole test out,
3628 I have added a new /Z option to pcretest that replaces the length and
3629 offset values with spaces. This is now used to make test 3 independent
3630 of link size. (Test 2 will be tidied up later.)
3631
3632 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3633 segmentation fault instead of returning the appropriate error message.
3634
3635 15. In multiline mode when the newline sequence was set to "any", the pattern
3636 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3637 This doesn't seem right; it now treats the CRLF combination as the line
3638 ending, and so does not match in that case. It's only a pattern such as ^$
3639 that would hit this one: something like ^ABC$ would have failed after \r
3640 and then tried again after \r\n.
3641
3642 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3643 in an attempt to make files that differ only in their line terminators
3644 compare equal. This works on Linux.
3645
3646 17. Under certain error circumstances pcregrep might try to free random memory
3647 as it exited. This is now fixed, thanks to valgrind.
3648
3649 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3650 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3651 was because its rules for how to advance for /g after matching an empty
3652 string at the end of a line did not allow for this case. They now check for
3653 it specially.
3654
3655 20. pcretest is supposed to handle patterns and data of any length, by
3656 extending its buffers when necessary. It was getting this wrong when the
3657 buffer for a data line had to be extended.
3658
3659 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3660 CRLF as a newline sequence.
3661
3662 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3663 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3664 I have nevertheless tidied it up.
3665
3666 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3667
3668 24. Added a man page for pcre-config.
3669
3670
3671 Version 7.0 19-Dec-06
3672 ---------------------
3673
3674 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3675 moving to gcc 4.1.1.
3676
3677 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3678 sys/time.h, which is documented as needed for this function. It doesn't
3679 seem to matter on Linux, but it showed up on some releases of OS X.
3680
3681 3. It seems that there are systems where bytes whose values are greater than
3682 127 match isprint() in the "C" locale. The "C" locale should be the
3683 default when a C program starts up. In most systems, only ASCII printing
3684 characters match isprint(). This difference caused the output from pcretest
3685 to vary, making some of the tests fail. I have changed pcretest so that:
3686
3687 (a) When it is outputting text in the compiled version of a pattern, bytes
3688 other than 32-126 are always shown as hex escapes.
3689
3690 (b) When it is outputting text that is a matched part of a subject string,
3691 it does the same, unless a different locale has been set for the match
3692 (using the /L modifier). In this case, it uses isprint() to decide.
3693
3694 4. Fixed a major bug that caused incorrect computation of the amount of memory
3695 required for a compiled pattern when options that changed within the
3696 pattern affected the logic of the preliminary scan that determines the
3697 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3698 that the computed length was too small. The symptoms of this bug were
3699 either the PCRE error "internal error: code overflow" from pcre_compile(),
3700 or a glibc crash with a message such as "pcretest: free(): invalid next
3701 size (fast)". Examples of patterns that provoked this bug (shown in
3702 pcretest format) are:
3703
3704 /(?-x: )/x
3705 /(?x)(?-x: \s*#\s*)/
3706 /((?i)[\x{c0}])/8
3707 /(?i:[\x{c0}])/8
3708
3709 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3710 is now done differently.
3711
3712 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3713 wrapper classes; (b) implement a new function in the C++ scanner that is
3714 more efficient than the old way of doing things because it avoids levels of
3715 recursion in the regex matching; (c) add a paragraph to the documentation
3716 for the FullMatch() function.
3717
3718 6. The escape sequence \n was being treated as whatever was defined as
3719 "newline". Not only was this contrary to the documentation, which states
3720 that \n is character 10 (hex 0A), but it also went horribly wrong when
3721 "newline" was defined as CRLF. This has been fixed.
3722
3723 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3724 was being set to -1 for the "end of line" case (supposedly a value that no
3725 character can have). Though this value is never used (the check for end of
3726 line is "zero bytes in current character"), it caused compiler complaints.
3727 I've changed it to 0xffffffff.
3728
3729 8. In pcre_version.c, the version string was being built by a sequence of
3730 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3731 string (as it is for production releases) called a macro with an empty
3732 argument. The C standard says the result of this is undefined. The gcc
3733 compiler treats it as an empty string (which was what was wanted) but it is
3734 reported that Visual C gives an error. The source has been hacked around to
3735 avoid this problem.
3736
3737 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3738 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3739 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3740 of them did).
3741
3742 10. Originally, pcretest opened its input and output without "b"; then I was
3743 told that "b" was needed in some environments, so it was added for release
3744 5.0 to both the input and output. (It makes no difference on Unix-like
3745 systems.) Later I was told that it is wrong for the input on Windows. I've
3746 now abstracted the modes into two macros, to make it easier to fiddle with
3747 them, and removed "b" from the input mode under Windows.
3748
3749 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3750
3751 12. Added -help and --help to pcretest as an official way of being reminded
3752 of the options.
3753
3754 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3755 and pcrecpp.cc because they annoy compilers at high warning levels.
3756
3757 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3758
3759 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3760 scripts are not C programs :-) and which was not noticed because it works
3761 on Linux.
3762
3763 16. pcretest is supposed to handle any length of pattern and data line (as one
3764 line or as a continued sequence of lines) by extending its input buffer if
3765 necessary. This feature was broken for very long pattern lines, leading to
3766 a string of junk being passed to pcre_compile() if the pattern was longer
3767 than about 50K.
3768
3769 17. I have done a major re-factoring of the way pcre_compile() computes the
3770 amount of memory needed for a compiled pattern. Previously, there was code
3771 that made a preliminary scan of the pattern in order to do this. That was
3772 OK when PCRE was new, but as the facilities have expanded, it has become
3773 harder and harder to keep it in step with the real compile phase, and there
3774 have been a number of bugs (see for example, 4 above). I have now found a
3775 cunning way of running the real compile function in a "fake" mode that
3776 enables it to compute how much memory it would need, while actually only
3777 ever using a few hundred bytes of working memory and without too many
3778 tests of the mode. This should make future maintenance and development
3779 easier. A side effect of this work is that the limit of 200 on the nesting
3780 depth of parentheses has been removed (though this was never a serious
3781 limitation, I suspect). However, there is a downside: pcre_compile() now
3782 runs more slowly than before (30% or more, depending on the pattern). I
3783 hope this isn't a big issue. There is no effect on runtime performance.
3784
3785 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3786 newline (only possible for the last line of a file) and it was a
3787 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3788
3789 19. Added additional timing features to pcretest. (1) The -tm option now times
3790 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3791 separate command line item, by a number that specifies the number of
3792 repeats to use when timing. The default is 50000; this gives better
3793 precision, but takes uncomfortably long for very large patterns.
3794
3795 20. Extended pcre_study() to be more clever in cases where a branch of a
3796 subpattern has no definite first character. For example, (a*|b*)[cd] would
3797 previously give no result from pcre_study(). Now it recognizes that the
3798 first character must be a, b, c, or d.
3799
3800 21. There was an incorrect error "recursive call could loop indefinitely" if
3801 a subpattern (or the entire pattern) that was being tested for matching an
3802 empty string contained only one non-empty item after a nested subpattern.
3803 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3804 incorrectly, because the \d was being skipped in the check.
3805
3806 22. The pcretest program now has a new pattern option /B and a command line
3807 option -b, which is equivalent to adding /B to every pattern. This causes
3808 it to show the compiled bytecode, without the additional information that
3809 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3810 is the same as /B/I).
3811
3812 23. A new optimization is now able automatically to treat some sequences such
3813 as a*b as a*+b. More specifically, if something simple (such as a character
3814 or a simple class like \d) has an unlimited quantifier, and is followed by
3815 something that cannot possibly match the quantified thing, the quantifier
3816 is automatically "possessified".
3817
3818 24. A recursive reference to a subpattern whose number was greater than 39
3819 went wrong under certain circumstances in UTF-8 mode. This bug could also
3820 have affected the operation of pcre_study().
3821
3822 25. Realized that a little bit of performance could be had by replacing
3823 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3824
3825 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3826
3827 27. Possessive quantifiers such as a++ were previously implemented by turning
3828 them into atomic groups such as ($>a+). Now they have their own opcodes,
3829 which improves performance. This includes the automatically created ones
3830 from 23 above.
3831
3832 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3833 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3834 the first matched character to be a colon. This applied both to named and
3835 numbered groups.
3836
3837 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3838
3839 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3840 building PCRE on Windows easier, so I have included it in the distribution.
3841
3842 31. There is now a check in pcretest against a ridiculously large number being
3843 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3844 loop, the loop is abandoned.
3845
3846 32. Forward references to subpatterns in conditions such as (?(2)...) where
3847 subpattern 2 is defined later cause pcre_compile() to search forwards in
3848 the pattern for the relevant set of parentheses. This search went wrong
3849 when there were unescaped parentheses in a character class, parentheses
3850 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3851
3852 33. "Subroutine" calls and backreferences were previously restricted to
3853 referencing subpatterns earlier in the regex. This restriction has now
3854 been removed.
3855
3856 34. Added a number of extra features that are going to be in Perl 5.10. On the
3857 whole, these are just syntactic alternatives for features that PCRE had
3858 previously implemented using the Python syntax or my own invention. The
3859 other formats are all retained for compatibility.
3860
3861 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3862 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3863 also .NET compatible.
3864
3865 (b) A recursion or subroutine call to a named group can now be defined as
3866 (?&name) as well as (?P>name).
3867
3868 (c) A backreference to a named group can now be defined as \k<name> or
3869 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3870 5.10, are also .NET compatible.
3871
3872 (d) A conditional reference to a named group can now use the syntax
3873 (?(<name>) or (?('name') as well as (?(name).
3874
3875 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3876 groups (named and numbered) that are never evaluated inline, but can be
3877 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3878 is always false. There may be only one alternative in such a group.
3879
3880 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3881 as the simple (?(R). The condition is true only if the most recent
3882 recursion is that of the given number or name. It does not search out
3883 through the entire recursion stack.
3884
3885 (g) The escape \gN or \g{N} has been added, where N is a positive or
3886 negative number, specifying an absolute or relative reference.
3887
3888 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3889 some "unreachable code" warnings.
3890
3891 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3892 things, this adds five new scripts.
3893
3894 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3895 There were also incompatibilities regarding the handling of \Q..\E inside
3896 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3897 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3898
3899 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3900 matches an empty string, and forcibly breaks the loop. There were bugs in
3901 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3902 against aaaa the result was just "a" rather than "aaaa", for example. Two
3903 separate and independent bugs (that affected different cases) have been
3904 fixed.
3905
3906 39. Refactored the code to abolish the use of different opcodes for small
3907 capturing bracket numbers. This is a tidy that I avoided doing when I
3908 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3909 The new approach is not only tidier, it makes it possible to reduce the
3910 memory needed to fix the previous bug (38).
3911
3912 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3913 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3914 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3915 mode.
3916
3917 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3918 report.
3919
3920 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3921 copy construction and assignment in the C++ wrapper.
3922
3923 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3924 couple of bugs that could have given wrong results in the "--newline=crlf"
3925 case.
3926
3927 44. Added a number of casts and did some reorganization of signed/unsigned int
3928 variables following suggestions from Dair Grant. Also renamed the variable
3929 "this" as "item" because it is a C++ keyword.
3930
3931 45. Arranged for dftables to add
3932
3933 #include "pcre_internal.h"
3934
3935 to pcre_chartables.c because without it, gcc 4.x may remove the array
3936 definition from the final binary if PCRE is built into a static library and
3937 dead code stripping is activated.
3938
3939 46. For an unanchored pattern, if a match attempt fails at the start of a
3940 newline sequence, and the newline setting is CRLF or ANY, and the next two
3941 characters are CRLF, advance by two characters instead of one.
3942
3943
3944 Version 6.7 04-Jul-06
3945 ---------------------
3946
3947 1. In order to handle tests when input lines are enormously long, pcretest has
3948 been re-factored so that it automatically extends its buffers when
3949 necessary. The code is crude, but this _is_ just a test program. The
3950 default size has been increased from 32K to 50K.
3951
3952 2. The code in pcre_study() was using the value of the re argument before
3953 testing it for NULL. (Of course, in any sensible call of the function, it
3954 won't be NULL.)
3955
3956 3. The memmove() emulation function in pcre_internal.h, which is used on
3957 systems that lack both memmove() and bcopy() - that is, hardly ever -
3958 was missing a "static" storage class specifier.
3959
3960 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3961 containing an extended class (one that cannot be represented by a bitmap
3962 because it contains high-valued characters or Unicode property items, e.g.
3963 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3964 pattern, but PCRE should not loop if you do not (it no longer does).
3965 [Detail: two cases were found: (a) a repeated subpattern containing an
3966 extended class; (b) a recursive reference to a subpattern that followed a
3967 previous extended class. It wasn't skipping over the extended class
3968 correctly when UTF-8 mode was not set.]
3969
3970 5. A negated single-character class was not being recognized as fixed-length
3971 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3972 compile error "lookbehind assertion is not fixed length".
3973
3974 6. The RunPerlTest auxiliary script was showing an unexpected difference
3975 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3976 write a Perl script that can interpret lines of an input file either as
3977 byte characters or as UTF-8, which is what "perltest" was being required to
3978 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3979 can't do is switch easily at run time between having the "use utf8;" pragma
3980 or not. In the end, I fudged it by using the RunPerlTest script to insert
3981 "use utf8;" explicitly for the UTF-8 tests.
3982
3983 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3984 the end of the subject string, contrary to the documentation and to what
3985 Perl does. This was true of both matching functions. Now it matches only at
3986 the start of the subject and immediately after *internal* newlines.
3987
3988 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3989 a pointer to an int instead of a pointer to an unsigned long int. This
3990 caused problems on 64-bit systems.
3991
3992 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3993 instance of the 'standard' template library not being so standard".
3994
3995 10. There was no check on the number of named subpatterns nor the maximum
3996 length of a subpattern name. The product of these values is used to compute
3997 the size of the memory block for a compiled pattern. By supplying a very
3998 long subpattern name and a large number of named subpatterns, the size
3999 computation could be caused to overflow. This is now prevented by limiting
4000 the length of names to 32 characters, and the number of named subpatterns
4001 to 10,000.
4002
4003 11. Subpatterns that are repeated with specific counts have to be replicated in
4004 the compiled pattern. The size of memory for this was computed from the
4005 length of the subpattern and the repeat count. The latter is limited to
4006 65535, but there was no limit on the former, meaning that integer overflow
4007 could in principle occur. The compiled length of a repeated subpattern is
4008 now limited to 30,000 bytes in order to prevent this.
4009
4010 12. Added the optional facility to have named substrings with the same name.
4011
4012 13. Added the ability to use a named substring as a condition, using the
4013 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
4014 are numbers (not recommended). Forward references are permitted.
4015
4016 14. Added forward references in named backreferences (if you see what I mean).
4017
4018 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
4019 pattern could run off the end of the subject. For example, the pattern
4020 "(?s)(.{1,5})"8 did this with the subject "ab".
4021
4022 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
4023 PCRE_CASELESS was set when matching characters that were quantified with ?
4024 or *.
4025
4026 17. A character class other than a single negated character that had a minimum
4027 but no maximum quantifier - for example [ab]{6,} - was not handled
4028 correctly by pce_dfa_exec(). It would match only one character.
4029
4030 18. A valid (though odd) pattern that looked like a POSIX character
4031 class but used an invalid character after [ (for example [[,abc,]]) caused
4032 pcre_compile() to give the error "Failed: internal error: code overflow" or
4033 in some cases to crash with a glibc free() error. This could even happen if
4034 the pattern terminated after [[ but there just happened to be a sequence of
4035 letters, a binary zero, and a closing ] in the memory that followed.
4036
4037 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
4038 over the years. Originally (before any Unicode support), just the bottom 8
4039 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
4040 output from "man perlunicode" includes this:
4041
4042 The regular expression compiler produces polymorphic opcodes. That
4043 is, the pattern adapts to the data and automatically switches to
4044 the Unicode character scheme when presented with Unicode data--or
4045 instead uses a traditional byte scheme when presented with byte
4046 data.
4047
4048 Sadly, a wide octal escape does not cause a switch, and in a string with
4049 no other multibyte characters, these octal escapes are treated as before.
4050 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
4051 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
4052 Unicode string.
4053
4054 I have not perpetrated such confusion in PCRE. Up till now, it took just
4055 the bottom 8 bits, as in old Perl. I have now made octal escapes with
4056 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
4057 translate to the appropriate multibyte character.
4058
4059 29. Applied some refactoring to reduce the number of warnings from Microsoft
4060 and Borland compilers. This has included removing the fudge introduced
4061 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
4062 a warning about an unused variable.
4063
4064 21. PCRE has not included VT (character 0x0b) in the set of whitespace
4065 characters since release 4.0, because Perl (from release 5.004) does not.
4066 [Or at least, is documented not to: some releases seem to be in conflict
4067 with the documentation.] However, when a pattern was studied with
4068 pcre_study() and all its branches started with \s, PCRE still included VT
4069 as a possible starting character. Of course, this did no harm; it just
4070 caused an unnecessary match attempt.
4071
4072 22. Removed a now-redundant internal flag bit that recorded the fact that case
4073 dependency changed within the pattern. This was once needed for "required
4074 byte" processing, but is no longer used. This recovers a now-scarce options
4075 bit. Also moved the least significant internal flag bit to the most-
4076 significant bit of the word, which was not previously used (hangover from
4077 the days when it was an int rather than a uint) to free up another bit for
4078 the future.
4079
4080 23. Added support for CRLF line endings as well as CR and LF. As well as the
4081 default being selectable at build time, it can now be changed at runtime
4082 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
4083 specify that it is scanning data with non-default line endings.
4084
4085 24. Changed the definition of CXXLINK to make it agree with the definition of
4086 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
4087
4088 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
4089 recursions. This makes a big different to stack usage for some patterns.
4090
4091 26. If a subpattern containing a named recursion or subroutine reference such
4092 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
4093 the space required for the compiled pattern went wrong and gave too small a
4094 value. Depending on the environment, this could lead to "Failed: internal
4095 error: code overflow at offset 49" or "glibc detected double free or
4096 corruption" errors.
4097
4098 27. Applied patches from Google (a) to support the new newline modes and (b) to
4099 advance over multibyte UTF-8 characters in GlobalReplace.
4100
4101 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
4102 difference for some implementation of PCRE in some Windows version.
4103
4104 29. Added some extra testing facilities to pcretest:
4105
4106 \q<number> in a data line sets the "match limit" value
4107 \Q<number> in a data line sets the "match recursion limt" value
4108 -S <number> sets the stack size, where <number> is in megabytes
4109
4110 The -S option isn't available for Windows.
4111
4112
4113 Version 6.6 06-Feb-06
4114 ---------------------
4115
4116 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
4117 in pcreposix.h. I have copied the definition from pcre.h.
4118
4119 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
4120 because pcre.h is no longer a built file.
4121
4122 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
4123 not normally included in the compiled code.
4124
4125
4126 Version 6.5 01-Feb-06
4127 ---------------------
4128
4129 1. When using the partial match feature with pcre_dfa_exec(), it was not
4130 anchoring the second and subsequent partial matches at the new starting
4131 point. This could lead to incorrect results. For example, with the pattern
4132 /1234/, partially matching against "123" and then "a4" gave a match.
4133
4134 2. Changes to pcregrep:
4135
4136 (a) All non-match returns from pcre_exec() were being treated as failures
4137 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
4138 error message is output. Some extra information is given for the
4139 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
4140 probably the only errors that are likely to be caused by users (by
4141 specifying a regex that has nested indefinite repeats, for instance).
4142 If there are more than 20 of these errors, pcregrep is abandoned.
4143
4144 (b) A binary zero was treated as data while matching, but terminated the
4145 output line if it was written out. This has been fixed: binary zeroes
4146 are now no different to any other data bytes.
4147
4148 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
4149 used to set a locale for matching. The --locale=xxxx long option has
4150 been added (no short equivalent) to specify a locale explicitly on the
4151 pcregrep command, overriding the environment variables.
4152
4153 (d) When -B was used with -n, some line numbers in the output were one less
4154 than they should have been.
4155
4156 (e) Added the -o (--only-matching) option.
4157
4158 (f) If -A or -C was used with -c (count only), some lines of context were
4159 accidentally printed for the final match.
4160
4161 (g) Added the -H (--with-filename) option.
4162
4163 (h) The combination of options -rh failed to suppress file names for files
4164 that were found from directory arguments.
4165
4166 (i) Added the -D (--devices) and -d (--directories) options.
4167
4168 (j) Added the -F (--fixed-strings) option.
4169
4170 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4171
4172 (l) Added the --colo(u)r option.
4173
4174 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4175 is not present by default.
4176
4177 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4178 items such as (?R) or (?1), when the recursion could match a number of
4179 alternatives. If it matched one of the alternatives, but subsequently,
4180 outside the recursion, there was a failure, the code tried to back up into
4181 the recursion. However, because of the way PCRE is implemented, this is not
4182 possible, and the result was an incorrect result from the match.
4183
4184 In order to prevent this happening, the specification of recursion has
4185 been changed so that all such subpatterns are automatically treated as
4186 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4187
4188 4. I had overlooked the fact that, in some locales, there are characters for
4189 which isalpha() is true but neither isupper() nor islower() are true. In
4190 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4191 and ordfeminine) are like this. This affected the treatment of \w and \W
4192 when they appeared in character classes, but not when they appeared outside
4193 a character class. The bit map for "word" characters is now created
4194 separately from the results of isalnum() instead of just taking it from the
4195 upper, lower, and digit maps. (Plus the underscore character, of course.)
4196
4197 5. The above bug also affected the handling of POSIX character classes such as
4198 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4199 permanent tables. Instead, the bit maps for such a class were previously
4200 created as the appropriate unions of the upper, lower, and digit bitmaps.
4201 Now they are created by subtraction from the [[:word:]] class, which has
4202 its own bitmap.
4203
4204 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4205 It is created by subtracting the vertical space characters (\x09, \x0a,
4206 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4207 subtraction was done in the overall bitmap for a character class, meaning
4208 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4209 be recognized. This bug has been fixed.
4210
4211 7. Patches from the folks at Google:
4212
4213 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4214 real life, but is still worth protecting against".
4215
4216 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4217 regular expressions".
4218
4219 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4220 have it.
4221
4222 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4223 "configure" and the latter not, in order to fix a problem somebody had
4224 with compiling the Arg class on HP-UX.
4225
4226 (e) Improve the error-handling of the C++ wrapper a little bit.
4227
4228 (f) New tests for checking recursion limiting.
4229
4230 8. The pcre_memmove() function, which is used only if the environment does not
4231 have a standard memmove() function (and is therefore rarely compiled),
4232 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4233 returning a result (though PCRE never actually uses the result).
4234
4235 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4236 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4237 returned instead of calling malloc() with an overflowing number that would
4238 most likely cause subsequent chaos.
4239
4240 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4241
4242 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4243 with this option is matched, the nmatch and pmatch options of regexec() are
4244 ignored.
4245
4246 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4247 provided in case anyone wants to the the POSIX interface with UTF-8
4248 strings.
4249
4250 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4251 C++ linking (needed for some HP-UX environments).
4252
4253 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4254 (unused parameter) and in the pcre_printint() function (omitted "default"
4255 switch label when the default is to do nothing).
4256
4257 15. Added some code to make it possible, when PCRE is compiled as a C++
4258 library, to replace subject pointers for pcre_exec() with a smart pointer
4259 class, thus making it possible to process discontinuous strings.
4260
4261 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4262 much the same function. They were added by different people who were trying
4263 to make PCRE easy to compile on non-Unix systems. It has been suggested
4264 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4265 for compiling on Windows systems. I have therefore replaced it with
4266 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4267 defaults to "extern" for C or "extern C" for C++, which works fine on
4268 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4269 SCOPE with something explicit in config.h. In addition:
4270
4271 (a) pcreposix.h still had just "extern" instead of either of these macros;
4272 I have replaced it with PCRE_DATA_SCOPE.
4273
4274 (b) Functions such as _pcre_xclass(), which are internal to the library,
4275 but external in the C sense, all had PCRE_EXPORT in their definitions.
4276 This is apparently wrong for the Windows case, so I have removed it.
4277 (It makes no difference on Unix-like systems.)
4278
4279 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4280 of recursive calls to match(). This is different to MATCH_LIMIT because
4281 that limits the total number of calls to match(), not all of which increase
4282 the depth of recursion. Limiting the recursion depth limits the amount of
4283 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4284 when PCRE is compiled, and changed at run time. A patch from Google adds
4285 this functionality to the C++ interface.
4286
4287 18. Changes to the handling of Unicode character properties:
4288
4289 (a) Updated the table to Unicode 4.1.0.
4290
4291 (b) Recognize characters that are not in the table as "Cn" (undefined).
4292
4293 (c) I revised the way the table is implemented to a much improved format
4294 which includes recognition of ranges. It now supports the ranges that
4295 are defined in UnicodeData.txt, and it also amalgamates other
4296 characters into ranges. This has reduced the number of entries in the
4297 table from around 16,000 to around 3,000, thus reducing its size
4298 considerably. I realized I did not need to use a tree structure after
4299 all - a binary chop search is just as efficient. Having reduced the
4300 number of entries, I extended their size from 6 bytes to 8 bytes to
4301 allow for more data.
4302
4303 (d) Added support for Unicode script names via properties such as \p{Han}.
4304
4305 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4306 matching that character.
4307
4308 20. When matching a repeated Unicode property with a minimum greater than zero,
4309 (for example \pL{2,}), PCRE could look past the end of the subject if it
4310 reached it while seeking the minimum number of characters. This could
4311 happen only if some of the characters were more than one byte long, because
4312 there is a check for at least the minimum number of bytes.
4313
4314 21. Refactored the implementation of \p and \P so as to be more general, to
4315 allow for more different types of property in future. This has changed the
4316 compiled form incompatibly. Anybody with saved compiled patterns that use
4317 \p or \P will have to recompile them.
4318
4319 22. Added "Any" and "L&" to the supported property types.
4320
4321 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4322 but give a compile time error if the value is greater than 0xff.
4323
4324 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4325 accidentally not being installed or uninstalled.
4326
4327 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4328 made were to insert the current release number. This seemed silly, because
4329 it made things harder for people building PCRE on systems that don't run
4330 "configure". I have turned pcre.h into a distributed file, no longer built
4331 by "configure", with the version identification directly included. There is
4332 no longer a pcre.h.in file.
4333
4334 However, this change necessitated a change to the pcre-config script as
4335 well. It is built from pcre-config.in, and one of the substitutions was the
4336 release number. I have updated configure.ac so that ./configure now finds
4337 the release number by grepping pcre.h.
4338
4339 26. Added the ability to run the tests under valgrind.
4340
4341
4342 Version 6.4 05-Sep-05
4343 ---------------------
4344
4345 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4346 "--" to be printed when multiple files were scanned, even when none of the
4347 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4348 consider it to be a bug, and have restored the previous behaviour.
4349
4350 2. A couple of code tidies to get rid of compiler warnings.
4351
4352 3. The pcretest program used to cheat by referring to symbols in the library
4353 whose names begin with _pcre_. These are internal symbols that are not
4354 really supposed to be visible externally, and in some environments it is
4355 possible to suppress them. The cheating is now confined to including
4356 certain files from the library's source, which is a bit cleaner.
4357
4358 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4359 file's purpose clearer.
4360
4361 5. Reorganized pcre_ucp_findchar().
4362
4363
4364 Version 6.3 15-Aug-05
4365 ---------------------
4366
4367 1. The file libpcre.pc.in did not have general read permission in the tarball.
4368
4369 2. There were some problems when building without C++ support:
4370
4371 (a) If C++ support was not built, "make install" and "make test" still
4372 tried to test it.
4373
4374 (b) There were problems when the value of CXX was explicitly set. Some
4375 changes have been made to try to fix these, and ...
4376
4377 (c) --disable-cpp can now be used to explicitly disable C++ support.
4378
4379 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4380 backslash in a target when C++ was disabled. This confuses some
4381 versions of "make", apparently. Using an intermediate variable solves
4382 this. (Same for CPP_LOBJ.)
4383
4384 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4385 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4386 necessary on certain architectures.
4387
4388 4. Added a setting of -export-symbols-regex to the link command to remove
4389 those symbols that are exported in the C sense, but actually are local
4390 within the library, and not documented. Their names all begin with
4391 "_pcre_". This is not a perfect job, because (a) we have to except some
4392 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4393 available (and never for static libraries). I have made a note to try to
4394 find a way round (a) in the future.
4395
4396
4397 Version 6.2 01-Aug-05
4398 ---------------------
4399
4400 1. There was no test for integer overflow of quantifier values. A construction
4401 such as {1111111111111111} would give undefined results. What is worse, if
4402 a minimum quantifier for a parenthesized subpattern overflowed and became
4403 negative, the calculation of the memory size went wrong. This could have
4404 led to memory overwriting.
4405
4406 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4407
4408 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4409 operating environments where this matters.
4410
4411 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4412 PCRE options from within the C++ wrapper.
4413
4414 5. Named capturing subpatterns were not being correctly counted when a pattern
4415 was compiled. This caused two problems: (a) If there were more than 100
4416 such subpatterns, the calculation of the memory needed for the whole
4417 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4418 back references of the form \12, where the number was greater than 9, were
4419 not recognized as back references, even though there were sufficient
4420 previous subpatterns.
4421
4422 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4423 versions of gcc, e.g. 2.95.4.
4424
4425
4426 Version 6.1 21-Jun-05
4427 ---------------------
4428
4429 1. There was one reference to the variable "posix" in pcretest.c that was not
4430 surrounded by "#if !defined NOPOSIX".
4431
4432 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4433 the cross-check on the old pcre_info() function, for the benefit of the
4434 cut-down version of PCRE that is currently imported into Exim.
4435
4436 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4437 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4438 patterns that start (?i) but I don't think that matters. The use of (?i) is
4439 just an example; this all applies to the other options as well.
4440
4441 4. Since libtool seems to echo the compile commands it is issuing, the output
4442 from "make" can be reduced a bit by putting "@" in front of each libtool
4443 compile command.
4444
4445 5. Patch from the folks at Google for configure.in to be a bit more thorough
4446 in checking for a suitable C++ installation before trying to compile the
4447 C++ stuff. This should fix a reported problem when a compiler was present,
4448 but no suitable headers.
4449
4450 6. The man pages all had just "PCRE" as their title. I have changed them to
4451 be the relevant file name. I have also arranged that these names are
4452 retained in the file doc/pcre.txt, which is a concatenation in text format
4453 of all the man pages except the little individual ones for each function.
4454
4455 7. The NON-UNIX-USE file had not been updated for the different set of source
4456 files that come with release 6. I also added a few comments about the C++
4457 wrapper.
4458
4459
4460 Version 6.0 07-Jun-05
4461 ---------------------
4462
4463 1. Some minor internal re-organization to help with my DFA experiments.
4464
4465 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4466 didn't matter for the library itself when fully configured, but did matter
4467 when compiling without UCP support, or within Exim, where the ucp files are
4468 not imported.
4469
4470 3. Refactoring of the library code to split up the various functions into
4471 different source modules. The addition of the new DFA matching code (see
4472 below) to a single monolithic source would have made it really too
4473 unwieldy, quite apart from causing all the code to be include in a
4474 statically linked application, when only some functions are used. This is
4475 relevant even without the DFA addition now that patterns can be compiled in
4476 one application and matched in another.
4477
4478 The downside of splitting up is that there have to be some external
4479 functions and data tables that are used internally in different modules of
4480 the library but which are not part of the API. These have all had their
4481 names changed to start with "_pcre_" so that they are unlikely to clash
4482 with other external names.
4483
4484 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4485 a different (DFA) algorithm. Although it is slower than the original
4486 function, it does have some advantages for certain types of matching
4487 problem.
4488
4489 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4490 including restarting after a partial match.
4491
4492 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4493 defined when compiling for Windows was sent to me. I have put it into the
4494 code, though I have no means of testing or verifying it.
4495
4496 7. Added the pcre_refcount() auxiliary function.
4497
4498 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4499 match before or at the first newline in the subject string. In pcretest,
4500 the /f option on a pattern can be used to set this.
4501
4502 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4503 would behave wrongly. This has been present in PCRE since release 4.0.
4504
4505 10. A number of changes to the pcregrep command:
4506
4507 (a) Refactored how -x works; insert ^(...)$ instead of setting
4508 PCRE_ANCHORED and checking the length, in preparation for adding
4509 something similar for -w.
4510
4511 (b) Added the -w (match as a word) option.
4512
4513 (c) Refactored the way lines are read and buffered so as to have more
4514 than one at a time available.
4515
4516 (d) Implemented a pcregrep test script.
4517
4518 (e) Added the -M (multiline match) option. This allows patterns to match
4519 over several lines of the subject. The buffering ensures that at least
4520 8K, or the rest of the document (whichever is the shorter) is available
4521 for matching (and similarly the previous 8K for lookbehind assertions).
4522
4523 (f) Changed the --help output so that it now says
4524
4525 -w, --word-regex(p)
4526
4527 instead of two lines, one with "regex" and the other with "regexp"
4528 because that confused at least one person since the short forms are the
4529 same. (This required a bit of code, as the output is generated
4530 automatically from a table. It wasn't just a text change.)
4531
4532 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4533 option but starts with a hyphen. Could be a pattern or a path name
4534 starting with a hyphen, for instance.
4535
4536 (h) "-" can be given as a file name to represent stdin.
4537
4538 (i) When file names are being printed, "(standard input)" is used for
4539 the standard input, for compatibility with GNU grep. Previously
4540 "<stdin>" was used.
4541
4542 (j) The option --label=xxx can be used to supply a name to be used for
4543 stdin when file names are being printed. There is no short form.
4544
4545 (k) Re-factored the options decoding logic because we are going to add
4546 two more options that take data. Such options can now be given in four
4547 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4548
4549 (l) Added the -A, -B, and -C options for requesting that lines of context
4550 around matches be printed.
4551
4552 (m) Added the -L option to print the names of files that do not contain
4553 any matching lines, that is, the complement of -l.
4554
4555 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4556 continue to scan other files.
4557
4558 (o) The -s option was incorrectly implemented. For compatibility with other
4559 greps, it now suppresses the error message for a non-existent or non-
4560 accessible file (but not the return code). There is a new option called
4561 -q that suppresses the output of matching lines, which was what -s was
4562 previously doing.
4563
4564 (p) Added --include and --exclude options to specify files for inclusion
4565 and exclusion when recursing.
4566
4567 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4568 Hopefully, it now does.
4569
4570 12. Missing cast in pcre_study().
4571
4572 13. Added an "uninstall" target to the makefile.
4573
4574 14. Replaced "extern" in the function prototypes in Makefile.in with
4575 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4576 world, but is set differently for Windows.
4577
4578 15. Added a second compiling function called pcre_compile2(). The only
4579 difference is that it has an extra argument, which is a pointer to an
4580 integer error code. When there is a compile-time failure, this is set
4581 non-zero, in addition to the error test pointer being set to point to an
4582 error message. The new argument may be NULL if no error number is required
4583 (but then you may as well call pcre_compile(), which is now just a
4584 wrapper). This facility is provided because some applications need a
4585 numeric error indication, but it has also enabled me to tidy up the way
4586 compile-time errors are handled in the POSIX wrapper.
4587
4588 16. Added VPATH=.libs to the makefile; this should help when building with one
4589 prefix path and installing with another. (Or so I'm told by someone who
4590 knows more about this stuff than I do.)
4591
4592 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4593 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4594 match everything, including newlines. This is not POSIX-compatible, but
4595 somebody wanted the feature. From pcretest it can be activated by using
4596 both the P and the s flags.
4597
4598 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4599
4600 19. libpcre.pc was being incorrectly installed as executable.
4601
4602 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4603 it now also looks for '\r' so that it will work unmodified on Windows.
4604
4605 21. Added Google's contributed C++ wrapper to the distribution.
4606
4607 22. Added some untidy missing memory free() calls in pcretest, to keep
4608 Electric Fence happy when testing.
4609
4610
4611
4612 Version 5.0 13-Sep-04
4613 ---------------------
4614
4615 1. Internal change: literal characters are no longer packed up into items
4616 containing multiple characters in a single byte-string. Each character
4617 is now matched using a separate opcode. However, there may be more than one
4618 byte in the character in UTF-8 mode.
4619
4620 2. The pcre_callout_block structure has two new fields: pattern_position and
4621 next_item_length. These contain the offset in the pattern to the next match
4622 item, and its length, respectively.
4623
4624 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4625 insertion of callouts before each pattern item. Added the /C option to
4626 pcretest to make use of this.
4627
4628 4. On the advice of a Windows user, the lines
4629
4630 #if defined(_WIN32) || defined(WIN32)
4631 _setmode( _fileno( stdout ), 0x8000 );
4632 #endif /* defined(_WIN32) || defined(WIN32) */
4633
4634 have been added to the source of pcretest. This apparently does useful
4635 magic in relation to line terminators.
4636
4637 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4638 for the benefit of those environments where the "b" makes a difference.
4639
4640 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4641 to know about it. I have put a hack into configure.in that adds in code
4642 to set GCC=yes if CC=icc. This seems to end up at a point in the
4643 generated configure script that is early enough to affect the setting of
4644 compiler options, which is what is needed, but I have no means of testing
4645 whether it really works. (The user who reported this had patched the
4646 generated configure script, which of course I cannot do.)
4647
4648 LATER: After change 22 below (new libtool files), the configure script
4649 seems to know about icc (and also ecc). Therefore, I have commented out
4650 this hack in configure.in.
4651
4652 7. Added support for pkg-config (2 patches were sent in).
4653
4654 8. Negated POSIX character classes that used a combination of internal tables
4655 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4656 [[:^ascii]]. Typically, they would match almost any characters. The other
4657 POSIX classes were not broken in this way.
4658
4659 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4660 to find the match, as PCRE was deluded into thinking that the match had to
4661 start at the start point or following a newline. The same bug applied to
4662 patterns with negative forward assertions or any backward assertions
4663 preceding ".*" at the start, unless the pattern required a fixed first
4664 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4665
4666 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4667 starting at the last subject character, bytes beyond the end of the subject
4668 string were read.
4669
4670 11. Renamed the variable "class" as "classbits" to make life easier for C++
4671 users. (Previously there was a macro definition, but it apparently wasn't
4672 enough.)
4673
4674 12. Added the new field "tables" to the extra data so that tables can be passed
4675 in at exec time, or the internal tables can be re-selected. This allows
4676 a compiled regex to be saved and re-used at a later time by a different
4677 program that might have everything at different addresses.
4678
4679 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4680 -R library as well as a -L library.
4681
4682 14. The debugging options of pcretest (-d on the command line or D on a
4683 pattern) showed incorrect output for anything following an extended class
4684 that contained multibyte characters and which was followed by a quantifier.
4685
4686 15. Added optional support for general category Unicode character properties
4687 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4688 support. It adds about 90K to the size of the library. The meanings of the
4689 inbuilt class escapes such as \d and \s have NOT been changed.
4690
4691 16. Updated pcredemo.c to include calls to free() to release the memory for the
4692 compiled pattern.
4693
4694 17. The generated file chartables.c was being created in the source directory
4695 instead of in the building directory. This caused the build to fail if the
4696 source directory was different from the building directory, and was
4697 read-only.
4698
4699 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4700 file. No doubt somebody will tell me if they don't make sense... Also added
4701 Dan Mooney's comments about building on OpenVMS.
4702
4703 19. Added support for partial matching via the PCRE_PARTIAL option for
4704 pcre_exec() and the \P data escape in pcretest.
4705
4706 20. Extended pcretest with 3 new pattern features:
4707
4708 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4709 write the compiled pattern to the file whose name is "rest-of-line".
4710 This is a straight binary dump of the data, with the saved pointer to
4711 the character tables forced to be NULL. The study data, if any, is
4712 written too. After writing, pcretest reads a new pattern.
4713
4714 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4715 compiled pattern from the given file. There must not be any
4716 occurrences of "<" in the file name (pretty unlikely); if there are,
4717 pcretest will instead treat the initial "<" as a pattern delimiter.
4718 After reading in the pattern, pcretest goes on to read data lines as
4719 usual.
4720
4721 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4722 and 16-bit fields in a compiled pattern, to simulate a pattern that
4723 was compiled on a host of opposite endianness.
4724
4725 21. The pcre-exec() function can now cope with patterns that were compiled on
4726 hosts of opposite endianness, with this restriction:
4727
4728 As for any compiled expression that is saved and used later, the tables
4729 pointer field cannot be preserved; the extra_data field in the arguments
4730 to pcre_exec() should be used to pass in a tables address if a value
4731 other than the default internal tables were used at compile time.
4732
4733 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4734 now diagnosed as an error. Previously, most of the time, a negative number
4735 would have been treated as zero, but if in addition "ovector" was passed as
4736 NULL, a crash could occur.
4737
4738 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4739 new versions from the libtool 1.5 distribution (the last one is a copy of
4740 a file called libtool.m4). This seems to have fixed the need to patch
4741 "configure" to support Darwin 1.3 (which I used to do). However, I still
4742 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4743 workstation).
4744
4745 24. Changed the PCRE licence to be the more standard "BSD" licence.
4746
4747
4748 Version 4.5 01-Dec-03
4749 ---------------------
4750
4751 1. There has been some re-arrangement of the code for the match() function so
4752 that it can be compiled in a version that does not call itself recursively.
4753 Instead, it keeps those local variables that need separate instances for
4754 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4755 needs to "recurse". Keeping track of where control must go is done by means
4756 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4757 hide most of the details from the main code, and operates only if
4758 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4759 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4760 operating.
4761
4762 To make it easier for callers to provide specially tailored get/free
4763 functions for this usage, two new functions, pcre_stack_malloc, and
4764 pcre_stack_free, are used. They are always called in strict stacking order,
4765 and the size of block requested is always the same.
4766
4767 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4768 PCRE has been compiled to use the stack or the heap for recursion. The
4769 -C option of pcretest uses this to show which version is compiled.
4770
4771 A new data escape \S, is added to pcretest; it causes the amounts of store
4772 obtained and freed by both kinds of malloc/free at match time to be added
4773 to the output.
4774
4775 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4776 what's available on my current Linux desktop machine.
4777
4778 3. When matching a UTF-8 string, the test for a valid string at the start has
4779 been extended. If start_offset is not zero, PCRE now checks that it points
4780 to a byte that is the start of a UTF-8 character. If not, it returns
4781 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4782 this is necessary because there may be backward assertions in the pattern.
4783 When matching the same subject several times, it may save resources to use
4784 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4785
4786 4. The code for checking the validity of UTF-8 strings has been tightened so
4787 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4788 containing "overlong sequences".
4789
4790 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4791 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4792 should have been "&", but it just so happened that all the cases this let
4793 through by mistake were picked up later in the function.
4794
4795 6. I had used a variable called "isblank" - this is a C99 function, causing
4796 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4797
4798 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4799 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4800 so the version is shown; (c) stop "make test" echoing "./RunTest".
4801
4802 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4803
4804 9. The prototype for memmove() for systems that don't have it was using
4805 size_t, but the inclusion of the header that defines size_t was later. I've
4806 moved the #includes for the C headers earlier to avoid this.
4807
4808 10. Added some adjustments to the code to make it easier to compiler on certain
4809 special systems:
4810
4811 (a) Some "const" qualifiers were missing.
4812 (b) Added the macro EXPORT before all exported functions; by default this
4813 is defined to be empty.
4814 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4815 that it reads its output file name as an argument instead of writing
4816 to the standard output and assuming this can be redirected.
4817
4818 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4819 class containing characters with values greater than 255, PCRE compilation
4820 went into a loop.
4821
4822 12. A recursive reference to a subpattern that was within another subpattern
4823 that had a minimum quantifier of zero caused PCRE to crash. For example,
4824 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4825 recursion. If the recursively-called subpattern itself had a zero repeat,
4826 that was OK.
4827
4828 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4829 buffer into which it was copied (for escape processing) was still set at
4830 1024, so long lines caused crashes.
4831
4832 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4833 "internal error: code overflow...". This applied to any character class
4834 that was followed by a possessive quantifier.
4835
4836 15. Modified the Makefile to add libpcre.la as a prerequisite for
4837 libpcreposix.la because I was told this is needed for a parallel build to
4838 work.
4839
4840 16. If a pattern that contained .* following optional items at the start was
4841 studied, the wrong optimizing data was generated, leading to matching
4842 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4843 matching string must start with a or b or c. The correct conclusion for
4844 this pattern is that a match can start with any character.
4845
4846
4847 Version 4.4 13-Aug-03
4848 ---------------------
4849
4850 1. In UTF-8 mode, a character class containing characters with values between
4851 127 and 255 was not handled correctly if the compiled pattern was studied.
4852 In fixing this, I have also improved the studying algorithm for such
4853 classes (slightly).
4854
4855 2. Three internal functions had redundant arguments passed to them. Removal
4856 might give a very teeny performance improvement.
4857
4858 3. Documentation bug: the value of the capture_top field in a callout is *one
4859 more than* the number of the hightest numbered captured substring.
4860
4861 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4862 in incorrectly linking with a previously installed version. They now link
4863 explicitly with libpcre.la.
4864
4865 5. configure.in no longer needs to recognize Cygwin specially.
4866
4867 6. A problem in pcre.in for Windows platforms is fixed.
4868
4869 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4870 pcretest, it used to include the size of the study block as part of its
4871 output. Unfortunately, the structure contains a field that has a different
4872 size on different hardware architectures. This meant that the tests that
4873 showed this size failed. As the block is currently always of a fixed size,
4874 this information isn't actually particularly useful in pcretest output, so
4875 I have just removed it.
4876
4877 8. Three pre-processor statements accidentally did not start in column 1.
4878 Sadly, there are *still* compilers around that complain, even though
4879 standard C has not required this for well over a decade. Sigh.
4880
4881 9. In pcretest, the code for checking callouts passed small integers in the
4882 callout_data field, which is a void * field. However, some picky compilers
4883 complained about the casts involved for this on 64-bit systems. Now
4884 pcretest passes the address of the small integer instead, which should get
4885 rid of the warnings.
4886
4887 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4888 both compile and run time, and gives an error if an invalid UTF-8 sequence
4889 is found. There is a option for disabling this check in cases where the
4890 string is known to be correct and/or the maximum performance is wanted.
4891
4892 11. In response to a bug report, I changed one line in Makefile.in from
4893
4894 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4895 to
4896 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4897
4898 to look similar to other lines, but I have no way of telling whether this
4899 is the right thing to do, as I do not use Windows. No doubt I'll get told
4900 if it's wrong...
4901
4902
4903 Version 4.3 21-May-03
4904 ---------------------
4905
4906 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4907 Makefile.
4908
4909 2. Some refactoring to improve the quality of the code:
4910
4911 (i) The utf8_table... variables are now declared "const".
4912
4913 (ii) The code for \cx, which used the "case flipping" table to upper case
4914 lower case letters, now just substracts 32. This is ASCII-specific,
4915 but the whole concept of \cx is ASCII-specific, so it seems
4916 reasonable.
4917
4918 (iii) PCRE was using its character types table to recognize decimal and
4919 hexadecimal digits in the pattern. This is silly, because it handles
4920 only 0-9, a-f, and A-F, but the character types table is locale-
4921 specific, which means strange things might happen. A private
4922 table is now used for this - though it costs 256 bytes, a table is
4923 much faster than multiple explicit tests. Of course, the standard
4924 character types table is still used for matching digits in subject
4925 strings against \d.
4926
4927 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4928 ending in _t are). So I've renamed it as ESC_tee.
4929
4930 3. The first argument for regexec() in the POSIX wrapper should have been
4931 defined as "const".
4932
4933 4. Changed pcretest to use malloc() for its buffers so that they can be
4934 Electric Fenced for debugging.
4935
4936 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4937 to read one or more bytes before the start of the subject string. Often this
4938 had no effect on PCRE's behaviour, but in some circumstances it could
4939 provoke a segmentation fault.
4940
4941 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4942 to try to read one or more bytes before the start of the subject string.
4943
4944 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4945 UTF-8 support could misbehave in various ways if the subject string
4946 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4947 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4948 back over UTF-8 characters.)
4949
4950
4951 Version 4.2 14-Apr-03
4952 ---------------------
4953
4954 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4955
4956 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4957 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4958 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4959 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4960 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4961 and BUILD_EXEEXT
4962 Note: automatic setting of the BUILD variables is not yet working
4963 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4964 compile-time but not at link-time
4965 [LINK]: use for linking executables only
4966 make different versions for Windows and non-Windows
4967 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4968 libraries
4969 [LINK_FOR_BUILD]: new variable
4970 [OBJEXT]: use throughout
4971 [EXEEXT]: use throughout
4972 <winshared>: new target
4973 <wininstall>: new target
4974 <dftables.o>: use native compiler
4975 <dftables>: use native linker
4976 <install>: handle Windows platform correctly
4977 <clean>: ditto
4978 <check>: ditto
4979 copy DLL to top builddir before testing
4980
4981 As part of these changes, -no-undefined was removed again. This was reported
4982 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4983 in any case.
4984
4985 3. Some tidies to get rid of compiler warnings:
4986
4987 . In the match_data structure, match_limit was an unsigned long int, whereas
4988 match_call_count was an int. I've made them both unsigned long ints.
4989
4990 . In pcretest the fact that a const uschar * doesn't automatically cast to
4991 a void * provoked a warning.
4992
4993 . Turning on some more compiler warnings threw up some "shadow" variables
4994 and a few more missing casts.
4995
4996 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4997 option, a class that contained a single character with a value between 128
4998 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4999
5000 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
5001 option, a class that contained several characters, but with at least one
5002 whose value was between 128 and 255 caused PCRE to crash.
5003
5004
5005 Version 4.1 12-Mar-03
5006 ---------------------
5007
5008 1. Compiling with gcc -pedantic found a couple of places where casts were
5009 needed, and a string in dftables.c that was longer than standard compilers are
5010 required to support.
5011
5012 2. Compiling with Sun's compiler found a few more places where the code could
5013 be tidied up in order to avoid warnings.
5014
5015 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
5016 first of these names is deprecated in the latest Autoconf in favour of the name
5017 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
5018 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
5019 analogy I have changed it to CFLAGS_FOR_BUILD.
5020
5021 4. Added -no-undefined to the linking command in the Makefile, because this is
5022 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
5023 linking step for the pcreposix library.
5024
5025 5. PCRE was failing to diagnose the case of two named groups with the same
5026 name.
5027
5028 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
5029 literal character that is needed in the subject for a match, and scans along to
5030 ensure that it is present before embarking on the full matching process. This
5031 saves time in cases of nested unlimited repeats that are never going to match.
5032 Problem: the scan can take a lot of time if the subject is very long (e.g.
5033 megabytes), thus penalizing straightforward matches. It is now done only if the
5034 amount of subject to be scanned is less than 1000 bytes.
5035
5036 7. A lesser problem with the same optimization is that it was recording the
5037 first character of an anchored pattern as "needed", thus provoking a search
5038 right along the subject, even when the first match of the pattern was going to
5039 fail. The "needed" character is now not set for anchored patterns, unless it
5040 follows something in the pattern that is of non-fixed length. Thus, it still
5041 fulfils its original purpose of finding quick non-matches in cases of nested
5042 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
5043
5044
5045 Version 4.0 17-Feb-03
5046 ---------------------
5047
5048 1. If a comment in an extended regex that started immediately after a meta-item
5049 extended to the end of string, PCRE compiled incorrect data. This could lead to
5050 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
5051
5052 2. Moved to autoconf 2.53 and libtool 1.4.2.
5053
5054 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
5055 the special perltest8 script is no longer needed - all the tests can be run
5056 from a single perltest script.
5057
5058 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
5059 by \s. It has now been removed in PCRE. This means it isn't recognized as
5060 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
5061 class [:space:] *does* include VT, thereby creating a mess.
5062
5063 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
5064 space and tab.
5065
5066 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
5067 its new features into the main test script, reducing the number of scripts.
5068
5069 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
5070 were backward compatible, and made the (?i) apply to the whole pattern, as if
5071 /i were given. Now it behaves more logically, and applies the option setting
5072 only to what follows. PCRE has been changed to follow suit. However, if it
5073 finds options settings right at the start of the pattern, it extracts them into
5074 the global options, as before. Thus, they show up in the info data.
5075
5076 8. Added support for the \Q...\E escape sequence. Characters in between are
5077 treated as literals. This is slightly different from Perl in that $ and @ are
5078 also handled as literals inside the quotes. In Perl, they will cause variable
5079 interpolation. Note the following examples:
5080
5081 Pattern PCRE matches Perl matches
5082
5083 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
5084 \Qabc\$xyz\E abc\$xyz abc\$xyz
5085 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
5086
5087 For compatibility with Perl, \Q...\E sequences are recognized inside character
5088 classes as well as outside them.
5089
5090 9. Re-organized 3 code statements in pcretest to avoid "overflow in
5091 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
5092 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
5093 signed/unsigned warnings.
5094
5095 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
5096 option for pcretest, so I've replaced it by a simple function that does just
5097 that job.
5098
5099 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
5100 "pcregrep -".
5101
5102 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
5103 Java package. This provides some syntactic sugar for simple cases of what my
5104 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
5105 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
5106 item, you can use this simplified notation. Note that only makes sense with
5107 greedy quantifiers. Consequently, the use of the possessive quantifier forces
5108 greediness, whatever the setting of the PCRE_UNGREEDY option.
5109
5110 13. A change of greediness default within a pattern was not taking effect at
5111 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
5112 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
5113 was abstracted outside.
5114
5115 14. PCRE now supports the \G assertion. It is true when the current matching
5116 position is at the start point of the match. This differs from \A when the
5117 starting offset is non-zero. Used with the /g option of pcretest (or similar
5118 code), it works in the same way as it does for Perl's /g option. If all
5119 alternatives of a regex begin with \G, the expression is anchored to the start
5120 match position, and the "anchored" flag is set in the compiled expression.
5121
5122 15. Some bugs concerning the handling of certain option changes within patterns
5123 have been fixed. These applied to options other than (?ims). For example,
5124 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
5125 been the other way round. Some of this was related to change 7 above.
5126
5127 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
5128 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
5129 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
5130 POSIX classes only within a class (e.g. /[[:alpha:]]/).
5131
5132 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
5133 mode. Unlike ".", it always matches newline, whatever the setting of
5134 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
5135 assertions. Perl allows it, but it doesn't (in general) work because it can't
5136 calculate the length of the lookbehind. At least, that's the case for Perl
5137 5.8.0 - I've been told they are going to document that it doesn't work in
5138 future.
5139
5140 18. Added an error diagnosis for escapes that PCRE does not support: these are
5141 \L, \l, \N, \P, \p, \U, \u, and \X.
5142
5143 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
5144 reading past the end of the pattern in cases such as /[abcd/.
5145
5146 20. PCRE was getting more memory than necessary for patterns with classes that
5147 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
5148
5149 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
5150 compiling PCRE for use with Virtual Pascal.
5151
5152 22. Small fix to the Makefile to make it work properly if the build is done
5153 outside the source tree.
5154
5155 23. Added a new extension: a condition to go with recursion. If a conditional
5156 subpattern starts with (?(R) the "true" branch is used if recursion has
5157 happened, whereas the "false" branch is used only at the top level.
5158
5159 24. When there was a very long string of literal characters (over 255 bytes
5160 without UTF support, over 250 bytes with UTF support), the computation of how
5161 much memory was required could be incorrect, leading to segfaults or other
5162 strange effects.
5163
5164 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5165 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5166 there was a subsequent back reference to those brackets. This meant that, for
5167 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5168 possible to check for precisely this case. All we can do is abandon the
5169 optimization if .* occurs inside capturing brackets when there are any back
5170 references whatsoever. (See below for a better fix that came later.)
5171
5172 26. The handling of the optimization for finding the first character of a
5173 non-anchored pattern, and for finding a character that is required later in the
5174 match were failing in some cases. This didn't break the matching; it just
5175 failed to optimize when it could. The way this is done has been re-implemented.
5176
5177 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5178
5179 28. Added a new feature that provides some of the functionality that Perl
5180 provides with (?{...}). The facility is termed a "callout". The way it is done
5181 in PCRE is for the caller to provide an optional function, by setting
5182 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5183 global variable. By default it is unset, which disables all calling out. To get
5184 the function called, the regex must include (?C) at appropriate points. This
5185 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5186 This provides a means of identifying different callout points. When PCRE
5187 reaches such a point in the regex, if pcre_callout has been set, the external
5188 function is called. It is provided with data in a structure called
5189 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5190 matching continues; if it returns a non-zero value, the match at the current
5191 point fails. However, backtracking will occur if possible. [This was changed
5192 later and other features added - see item 49 below.]
5193
5194 29. pcretest is upgraded to test the callout functionality. It provides a
5195 callout function that displays information. By default, it shows the start of
5196 the match and the current position in the text. There are some new data escapes
5197 to vary what happens:
5198
5199 \C+ in addition, show current contents of captured substrings
5200 \C- do not supply a callout function
5201 \C!n return 1 when callout number n is reached
5202 \C!n!m return 1 when callout number n is reached for the mth time
5203
5204 30. If pcregrep was called with the -l option and just a single file name, it
5205 output "<stdin>" if a match was found, instead of the file name.
5206
5207 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5208 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5209 pcre_exec(). This saves a malloc/free per call. The default value of
5210 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5211 when configuring.
5212
5213 32. The default maximum size of a compiled pattern is 64K. There have been a
5214 few cases of people hitting this limit. The code now uses macros to handle the
5215 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5216 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5217 configuring. Tests 2 and 5 work only with 2-byte links because they output
5218 debugging information about compiled patterns.
5219
5220 33. Internal code re-arrangements:
5221
5222 (a) Moved the debugging function for printing out a compiled regex into
5223 its own source file (printint.c) and used #include to pull it into
5224 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5225 separate copies.
5226
5227 (b) Defined the list of op-code names for debugging as a macro in
5228 internal.h so that it is next to the definition of the opcodes.
5229
5230 (c) Defined a table of op-code lengths for simpler skipping along compiled
5231 code. This is again a macro in internal.h so that it is next to the
5232 definition of the opcodes.
5233
5234 34. Added support for recursive calls to individual subpatterns, along the
5235 lines of Robin Houston's patch (but implemented somewhat differently).
5236
5237 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5238 allow it to read and process whole directories in Win32. This code was
5239 contributed by Lionel Fourquaux; it has not been tested by me.
5240
5241 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5242 used to name a group. Names consist of alphanumerics and underscores, and must
5243 be unique. Back references use the syntax (?P=name) and recursive calls use
5244 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5245 numbers. The function pcre_fullinfo() can be used after compilation to extract
5246 a name/number map. There are three relevant calls:
5247
5248 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5249 PCRE_INFO_NAMECOUNT yields the number of entries
5250 PCRE_INFO_NAMETABLE yields a pointer to the map.
5251
5252 The map is a vector of fixed-size entries. The size of each entry depends on
5253 the length of the longest name used. The first two bytes of each entry are the
5254 group number, most significant byte first. There follows the corresponding
5255 name, zero terminated. The names are in alphabetical order.
5256
5257 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5258 case instead of 255. Making it the same both with and without UTF-8 support
5259 means that the same test output works with both.
5260
5261 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5262 calling malloc() with a zero argument.
5263
5264 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5265 optimization. I've improved things by keeping a bitmap of backreferences with
5266 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5267 fact referenced, the optimization can be applied. It is unlikely that a
5268 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5269 the match to follow \n) will appear inside brackets with a number greater than
5270 31, but if it does, any back reference > 31 suppresses the optimization.
5271
5272 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5273 of disabling numbered capturing parentheses. Any opening parenthesis that is
5274 not followed by ? behaves as if it were followed by ?: but named parentheses
5275 can still be used for capturing (and they will acquire numbers in the usual
5276 way).
5277
5278 41. Redesigned the return codes from the match() function into yes/no/error so
5279 that errors can be passed back from deep inside the nested calls. A malloc
5280 failure while inside a recursive subpattern call now causes the
5281 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5282
5283 42. It is now possible to set a limit on the number of times the match()
5284 function is called in a call to pcre_exec(). This facility makes it possible to
5285 limit the amount of recursion and backtracking, though not in a directly
5286 obvious way, because the match() function is used in a number of different
5287 circumstances. The count starts from zero for each position in the subject
5288 string (for non-anchored patterns). The default limit is, for compatibility, a
5289 large number, namely 10 000 000. You can change this in two ways:
5290
5291 (a) When configuring PCRE before making, you can use --with-match-limit=n
5292 to set a default value for the compiled library.
5293
5294 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5295 a different value is set. See 45 below.
5296
5297 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5298
5299 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5300 of things that can be changed at compile time. The first argument specifies
5301 what is wanted and the second points to where the information is to be placed.
5302 The current list of available information is:
5303
5304 PCRE_CONFIG_UTF8
5305
5306 The output is an integer that is set to one if UTF-8 support is available;
5307 otherwise it is set to zero.
5308
5309 PCRE_CONFIG_NEWLINE
5310
5311 The output is an integer that it set to the value of the code that is used for
5312 newline. It is either LF (10) or CR (13).
5313
5314 PCRE_CONFIG_LINK_SIZE
5315
5316 The output is an integer that contains the number of bytes used for internal
5317 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5318
5319 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5320
5321 The output is an integer that contains the threshold above which the POSIX
5322 interface uses malloc() for output vectors. See item 31 above.
5323
5324 PCRE_CONFIG_MATCH_LIMIT
5325
5326 The output is an unsigned integer that contains the default limit of the number
5327 of match() calls in a pcre_exec() execution. See 42 above.
5328
5329 44. pcretest has been upgraded by the addition of the -C option. This causes it
5330 to extract all the available output from the new pcre_config() function, and to
5331 output it. The program then exits immediately.
5332
5333 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5334 order to support additional features. One way would have been to define
5335 pcre_exec2() (for example) with extra arguments, but this would not have been
5336 extensible, and would also have required all calls to the original function to
5337 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5338 is used for passing in "extra" data from pcre_study().
5339
5340 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5341 contains the following fields:
5342
5343 flags a bitmap indicating which of the following fields are set
5344 study_data opaque data from pcre_study()
5345 match_limit a way of specifying a limit on match() calls for a specific
5346 call to pcre_exec()
5347 callout_data data for callouts (see 49 below)
5348
5349 The flag bits are also defined in pcre.h, and are
5350
5351 PCRE_EXTRA_STUDY_DATA
5352 PCRE_EXTRA_MATCH_LIMIT
5353 PCRE_EXTRA_CALLOUT_DATA
5354
5355 The pcre_study() function now returns one of these new pcre_extra blocks, with
5356 the actual study data pointed to by the study_data field, and the
5357 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5358 before. That is, this change is entirely upwards-compatible and requires no
5359 change to existing code.
5360
5361 If you want to pass in additional data to pcre_exec(), you can either place it
5362 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5363 block.
5364
5365 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5366 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5367 times with different match limits, until it finds the minimum value needed for
5368 pcre_exec() to complete. The value is then output. This can be instructive; for
5369 most simple matches the number is quite small, but for pathological cases it
5370 gets very large very quickly.
5371
5372 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5373 returns the size of the data block pointed to by the study_data field in a
5374 pcre_extra block, that is, the value that was passed as the argument to
5375 pcre_malloc() when PCRE was getting memory in which to place the information
5376 created by pcre_study(). The fourth argument should point to a size_t variable.
5377 pcretest has been extended so that this information is shown after a successful
5378 pcre_study() call when information about the compiled regex is being displayed.
5379
5380 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5381 because what follows is always an absolute path. (Later: it turns out that this
5382 is more than cosmetic for MinGW, because it doesn't like empty path
5383 components.)
5384
5385 49. Some changes have been made to the callout feature (see 28 above):
5386
5387 (i) A callout function now has three choices for what it returns:
5388
5389 0 => success, carry on matching
5390 > 0 => failure at this point, but backtrack if possible
5391 < 0 => serious error, return this value from pcre_exec()
5392
5393 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5394 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5395 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5396 use by callout functions. It will never be used by PCRE itself.
5397
5398 (ii) The pcre_extra structure (see 45 above) has a void * field called
5399 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5400 pcre_callout_block structure has a field of the same name. The contents of
5401 the field passed in the pcre_extra structure are passed to the callout
5402 function in the corresponding field in the callout block. This makes it
5403 easier to use the same callout-containing regex from multiple threads. For
5404 testing, the pcretest program has a new data escape
5405
5406 \C*n pass the number n (may be negative) as callout_data
5407
5408 If the callout function in pcretest receives a non-zero value as
5409 callout_data, it returns that value.
5410
5411 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5412 there were some redundant $(CFLAGS) in commands that are now specified as
5413 $(LINK), which already includes $(CFLAGS).
5414
5415 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5416 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5417 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5418 one-byte characters throughout. Note that case-insensitive matching applies
5419 only to characters whose values are less than 256. PCRE doesn't support the
5420 notion of cases for higher-valued characters.
5421
5422 (i) A character class whose characters are all within 0-255 is handled as
5423 a bit map, and the map is inverted for negative classes. Previously, a
5424 character > 255 always failed to match such a class; however it should
5425 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5426
5427 (ii) A negated character class with a single character < 255 is coded as
5428 "not this character" (OP_NOT). This wasn't working properly when the test
5429 character was multibyte, either singly or repeated.
5430
5431 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5432 mode, for example: \x{100}{2,3}.
5433
5434 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5435 singly or repeated) now correctly test multibyte characters. However,
5436 PCRE doesn't recognize any characters with values greater than 255 as
5437 digits, spaces, or word characters. Such characters always match \D, \S,
5438 and \W, and never match \d, \s, or \w.
5439
5440 (v) Classes may now contain characters and character ranges with values
5441 greater than 255. For example: [ab\x{100}-\x{400}].
5442
5443 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5444 PCRE in UTF-8 mode.
5445
5446 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5447 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5448 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5449 value.)
5450
5451 53. The single man page has become too large. I have therefore split it up into
5452 a number of separate man pages. These also give rise to individual HTML pages;
5453 these are now put in a separate directory, and there is an index.html page that
5454 lists them all. Some hyperlinking between the pages has been installed.
5455
5456 54. Added convenience functions for handling named capturing parentheses.
5457
5458 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5459 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5460 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5461 are faulted.
5462
5463 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5464 calling configure. These values are used when compiling the dftables.c program
5465 which is run to generate the source of the default character tables. They
5466 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5467 you will need to set these values.
5468
5469 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5470
5471
5472 Version 3.9 02-Jan-02
5473 ---------------------
5474
5475 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5476
5477 2. If --disable-static was given, the building process failed when trying to
5478 build pcretest and pcregrep. (For some reason it was using libtool to compile
5479 them, which is not right, as they aren't part of the library.)
5480
5481
5482 Version 3.8 18-Dec-01
5483 ---------------------
5484
5485 1. The experimental UTF-8 code was completely screwed up. It was packing the
5486 bytes in the wrong order. How dumb can you get?
5487
5488
5489 Version 3.7 29-Oct-01
5490 ---------------------
5491
5492 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5493 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5494 this didn't happen under Solaris 8, where I normally test things.
5495
5496 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5497 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5498 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5499 this fix an hour or so after the initial 3.7 release.)
5500
5501
5502 Version 3.6 23-Oct-01
5503 ---------------------
5504
5505 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5506 offsets passed as NULL with zero offset count.
5507
5508 2. The config.guess and config.sub files had not been updated when I moved to
5509 the latest autoconf.
5510
5511
5512 Version 3.5 15-Aug-01
5513 ---------------------
5514
5515 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5516 had been forgotten.
5517
5518 2. By using declared but undefined structures, we can avoid using "void"
5519 definitions in pcre.h while keeping the internal definitions of the structures
5520 private.
5521
5522 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5523 user point of view, this means that both static and shared libraries are built
5524 by default, but this can be individually controlled. More of the work of
5525 handling this static/shared cases is now inside libtool instead of PCRE's make
5526 file.
5527
5528 4. The pcretest utility is now installed along with pcregrep because it is
5529 useful for users (to test regexs) and by doing this, it automatically gets
5530 relinked by libtool. The documentation has been turned into a man page, so
5531 there are now .1, .txt, and .html versions in /doc.
5532
5533 5. Upgrades to pcregrep:
5534 (i) Added long-form option names like gnu grep.
5535 (ii) Added --help to list all options with an explanatory phrase.
5536 (iii) Added -r, --recursive to recurse into sub-directories.
5537 (iv) Added -f, --file to read patterns from a file.
5538
5539 6. pcre_exec() was referring to its "code" argument before testing that
5540 argument for NULL (and giving an error if it was NULL).
5541
5542 7. Upgraded Makefile.in to allow for compiling in a different directory from
5543 the source directory.
5544
5545 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5546 options bits, the pointer it was passed was to an int instead of to an unsigned
5547 long int. This mattered only on 64-bit systems.
5548
5549 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5550 generated) instead of pcre.in, which it its source. Also made the same change
5551 in several of the .c files.
5552
5553 10. A new release of gcc defines printf() as a macro, which broke pcretest
5554 because it had an ifdef in the middle of a string argument for printf(). Fixed
5555 by using separate calls to printf().
5556
5557 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5558 script, to force use of CR or LF instead of \n in the source. On non-Unix
5559 systems, the value can be set in config.h.
5560
5561 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5562 absolute limit. Changed the text of the error message to make this clear, and
5563 likewise updated the man page.
5564
5565 13. The limit of 99 on the number of capturing subpatterns has been removed.
5566 The new limit is 65535, which I hope will not be a "real" limit.
5567
5568
5569 Version 3.4 22-Aug-00
5570 ---------------------
5571
5572 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5573
5574 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5575
5576
5577 Version 3.3 01-Aug-00
5578 ---------------------
5579
5580 1. If an octal character was given, but the value was greater than \377, it
5581 was not getting masked to the least significant bits, as documented. This could
5582 lead to crashes in some systems.
5583
5584 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5585 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5586
5587 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5588 These just pass their arguments on to (pcre_free)(), but they are provided
5589 because some uses of PCRE bind it to non-C systems that can call its functions,
5590 but cannot call free() or pcre_free() directly.
5591
5592 4. Add "make test" as a synonym for "make check". Corrected some comments in
5593 the Makefile.
5594
5595 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5596 Makefile.
5597
5598 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5599 command called pgrep for grepping around the active processes.
5600
5601 7. Added the beginnings of support for UTF-8 character strings.
5602
5603 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5604 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5605 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5606 out for the ar command.)
5607
5608
5609 Version 3.2 12-May-00
5610 ---------------------
5611
5612 This is purely a bug fixing release.
5613
5614 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5615 of ZA. This was just one example of several cases that could provoke this bug,
5616 which was introduced by change 9 of version 2.00. The code for breaking
5617 infinite loops after an iteration that matches an empty string was't working
5618 correctly.
5619
5620 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5621 when matched against abbab (for example). After matching an empty string, it
5622 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5623 caused it to match further down the string than it should.
5624
5625 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5626 was there because it doesn't seem to be needed, and it causes trouble on some
5627 systems, as it is not a Standard C header. It has been removed.
5628
5629 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5630 were reported on the Macintosh. The changes were from
5631
5632 while ((c = *(++ptr)) != 0 && c != '\n');
5633 to
5634 while ((c = *(++ptr)) != 0 && c != '\n') ;
5635
5636 Totally extraordinary, but if that's what it takes...
5637
5638 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5639 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5640 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5641 assumes the way PCRE uses memmove() (always moving upwards).
5642
5643 6. PCRE is being used in one environment where strchr() is not available. There
5644 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5645 faster code anyway.
5646
5647
5648 Version 3.1 09-Feb-00
5649 ---------------------
5650
5651 The only change in this release is the fixing of some bugs in Makefile.in for
5652 the "install" target:
5653
5654 (1) It was failing to install pcreposix.h.
5655
5656 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5657
5658
5659 Version 3.0 01-Feb-00
5660 ---------------------
5661
5662 1. Add support for the /+ modifier to perltest (to output $` like it does in
5663 pcretest).
5664
5665 2. Add support for the /g modifier to perltest.
5666
5667 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5668 matches null strings.
5669
5670 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5671 pattern. Perl treats empty patterns specially - it reuses the most recent
5672 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5673 effect.
5674
5675 5. The POSIX interface was broken in that it was just handing over the POSIX
5676 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5677 required a bigger vector, with some working space on the end. This means that
5678 the POSIX wrapper now has to get and free some memory, and copy the results.
5679
5680 6. Added some simple autoconf support, placing the test data and the
5681 documentation in separate directories, re-organizing some of the
5682 information files, and making it build pcre-config (a GNU standard). Also added
5683 libtool support for building PCRE as a shared library, which is now the
5684 default.
5685
5686 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5687 09 are not valid octal constants. Single digits will be used for minor values
5688 less than 10.
5689
5690 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5691 existing programs that set these in the POSIX interface can use PCRE without
5692 modification.
5693
5694 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5695 return all that pcre_info() returns, plus additional data. The pcre_info()
5696 function is retained for compatibility, but is considered to be obsolete.
5697
5698 10. Added experimental recursion feature (?R) to handle one common case that
5699 Perl 5.6 will be able to do with (?p{...}).
5700
5701 11. Added support for POSIX character classes like [:alpha:], which Perl is
5702 adopting.
5703
5704
5705 Version 2.08 31-Aug-99
5706 ----------------------
5707
5708 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5709 trying to match at the startoffset position, but instead was moving forward to
5710 the next newline as if a previous match had failed.
5711
5712 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5713 and could get into a loop if a null string was matched other than at the start
5714 of the subject.
5715
5716 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5717 be distinguished at compile time, and for completeness also added PCRE_DATE.
5718
5719 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5720 in GnuWin32 environments.
5721
5722
5723 Version 2.07 29-Jul-99
5724 ----------------------
5725
5726 1. The documentation is now supplied in plain text form and HTML as well as in
5727 the form of man page sources.
5728
5729 2. C++ compilers don't like assigning (void *) values to other pointer types.
5730 In particular this affects malloc(). Although there is no problem in Standard
5731 C, I've put in casts to keep C++ compilers happy.
5732
5733 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5734 should be (const char *).
5735
5736 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5737 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5738 However, I haven't made this a standard facility. The documentation doesn't
5739 mention it, and the Makefile doesn't support it.
5740
5741 5. The Makefile now contains an "install" target, with editable destinations at
5742 the top of the file. The pcretest program is not installed.
5743
5744 6. pgrep -V now gives the PCRE version number and date.
5745
5746 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5747 causing the entire string to be ignored, instead of just the last character.
5748
5749 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5750 non-matching string, it can take a very, very long time, even for strings of
5751 quite modest length, because of the nested recursion. PCRE now does better in
5752 some of these cases. It does this by remembering the last required literal
5753 character in the pattern, and pre-searching the subject to ensure it is present
5754 before running the real match. In other words, it applies a heuristic to detect
5755 some types of certain failure quickly, and in the above example, if presented
5756 with a string that has no trailing " it gives "no match" very quickly.
5757
5758 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5759 other alternatives are tried instead.
5760
5761
5762 Version 2.06 09-Jun-99
5763 ----------------------
5764
5765 1. Change pcretest's output for amount of store used to show just the code
5766 space, because the remainder (the data block) varies in size between 32-bit and
5767 64-bit systems.
5768
5769 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5770 start matching at. This allows lookbehinds to work when searching for multiple
5771 occurrences in a string.
5772
5773 3. Added additional options to pcretest for testing multiple occurrences:
5774
5775 /+ outputs the rest of the string that follows a match
5776 /g loops for multiple occurrences, using the new startoffset argument
5777 /G loops for multiple occurrences by passing an incremented pointer
5778
5779 4. PCRE wasn't doing the "first character" optimization for patterns starting
5780 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5781 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5782 the letter 'x'. On long subject strings, this gives a significant speed-up.
5783
5784
5785 Version 2.05 21-Apr-99
5786 ----------------------
5787
5788 1. Changed the type of magic_number from int to long int so that it works
5789 properly on 16-bit systems.
5790
5791 2. Fixed a bug which caused patterns starting with .* not to work correctly
5792 when the subject string contained newline characters. PCRE was assuming
5793 anchoring for such patterns in all cases, which is not correct because .* will
5794 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5795 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5796 must be retried after every newline in the subject.
5797
5798
5799 Version 2.04 18-Feb-99
5800 ----------------------
5801
5802 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5803 computation of the store needed to hold the pattern was incorrect (too large).
5804 If such patterns were nested a few deep, this could multiply and become a real
5805 problem.
5806
5807 2. Added /M option to pcretest to show the memory requirement of a specific
5808 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5809
5810 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5811 compiled in such a way that the backtracking after subsequent failure was
5812 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5813 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5814
5815
5816 Version 2.03 02-Feb-99
5817 ----------------------
5818
5819 1. Fixed typo and small mistake in man page.
5820
5821 2. Added 4th condition (GPL supersedes if conflict) and created separate
5822 LICENCE file containing the conditions.
5823
5824 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5825 Perl, that is the internal \ allows the delimiter to be included in the
5826 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5827 the final delimiter, add \ to the end of the pattern (to test the error).
5828
5829 4. Added the convenience functions for extracting substrings after a successful
5830 match. Updated pcretest to make it able to test these functions.
5831
5832
5833 Version 2.02 14-Jan-99
5834 ----------------------
5835
5836 1. Initialized the working variables associated with each extraction so that
5837 their saving and restoring doesn't refer to uninitialized store.
5838
5839 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5840 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5841 fix the problem.
5842
5843 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5844 calls, and wasn't printing the correct value for compiling calls. Increased the
5845 default value of LOOPREPEAT, and the number of significant figures in the
5846 times.
5847
5848 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5849
5850 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5851 a building problem on Windows NT with a FAT file system.
5852
5853
5854 Version 2.01 21-Oct-98
5855 ----------------------
5856
5857 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5858 to character tables built by pcre_maketables() in the current locale. If NULL
5859 is passed, the default tables are used.
5860
5861
5862 Version 2.00 24-Sep-98
5863 ----------------------
5864
5865 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5866 it any more.
5867
5868 2. Allow quantification of (?>) groups, and make it work correctly.
5869
5870 3. The first character computation wasn't working for (?>) groups.
5871
5872 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5873 end of the subject) and add 5.005's \z, which really does match only at the
5874 very end of the subject.
5875
5876 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5877
5878 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5879 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5880 localized options. All options to pcre_study() were also removed.
5881
5882 7. Add other new features from 5.005:
5883
5884 $(?<= positive lookbehind
5885 $(?<! negative lookbehind
5886 (?imsx-imsx) added the unsetting capability
5887 such a setting is global if at outer level; local otherwise
5888 (?imsx-imsx:) non-capturing groups with option setting
5889 (?(cond)re|re) conditional pattern matching
5890
5891 A backreference to itself in a repeated group matches the previous
5892 captured string.
5893
5894 8. General tidying up of studying (both automatic and via "study")
5895 consequential on the addition of new assertions.
5896
5897 9. As in 5.005, unlimited repeated groups that could match an empty substring
5898 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5899 runtime if any iteration does actually match an empty substring.
5900
5901 10. Include the RunTest script in the distribution.
5902
5903 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5904 discrepancies, some of which were old and were also with respect to 5.004. They
5905 have now been fixed.
5906
5907
5908 Version 1.09 28-Apr-98
5909 ----------------------
5910
5911 1. A negated single character class followed by a quantifier with a minimum
5912 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5913 program crashes, or just wrong answers. This did not apply to negated classes
5914 containing more than one character, or to minima other than one.
5915
5916
5917 Version 1.08 27-Mar-98
5918 ----------------------
5919
5920 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5921
5922 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5923 latter must appear before anything that relies on it in the pattern.
5924
5925
5926 Version 1.07 16-Feb-98
5927 ----------------------
5928
5929 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5930 repeat of a potentially empty string).
5931
5932
5933 Version 1.06 23-Jan-98
5934 ----------------------
5935
5936 1. Added Markus Oberhumer's little patches for C++.
5937
5938 2. Literal strings longer than 255 characters were broken.
5939
5940
5941 Version 1.05 23-Dec-97
5942 ----------------------
5943
5944 1. Negated character classes containing more than one character were failing if
5945 PCRE_CASELESS was set at run time.
5946
5947
5948 Version 1.04 19-Dec-97
5949 ----------------------
5950
5951 1. Corrected the man page, where some "const" qualifiers had been omitted.
5952
5953 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5954 input syntax.
5955
5956 3. Fixed memory leak which occurred when a regex with back references was
5957 matched with an offsets vector that wasn't big enough. The temporary memory
5958 that is used in this case wasn't being freed if the match failed.
5959
5960 4. Tidied pcretest to ensure it frees memory that it gets.
5961
5962 5. Temporary memory was being obtained in the case where the passed offsets
5963 vector was exactly big enough.
5964
5965 6. Corrected definition of offsetof() from change 5 below.
5966
5967 7. I had screwed up change 6 below and broken the rules for the use of
5968 setjmp(). Now fixed.
5969
5970
5971 Version 1.03 18-Dec-97
5972 ----------------------
5973
5974 1. A erroneous regex with a missing opening parenthesis was correctly
5975 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5976 on some systems.
5977
5978 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5979 it was reported that one broken compiler failed on the former because "code" is
5980 also an independent variable.
5981
5982 3. The erroneous regex a[]b caused an array overrun reference.
5983
5984 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5985 fail on data ending with that character. (It was going on too far, and checking
5986 the next character, typically a binary zero.) This was specific to the
5987 optimized code for single-character negative classes.
5988
5989 5. Added a contributed patch from the TIN world which does the following:
5990
5991 + Add an undef for memmove, in case the the system defines a macro for it.
5992
5993 + Add a definition of offsetof(), in case there isn't one. (I don't know
5994 the reason behind this - offsetof() is part of the ANSI standard - but
5995 it does no harm).
5996
5997 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5998 most of the places where whitespace preceded '#'. I have given up and
5999 allowed the remaining 2 cases to be at the margin.
6000
6001 + Rename some variables in pcre to eliminate shadowing. This seems very
6002 pedantic, but does no harm, of course.
6003
6004 6. Moved the call to setjmp() into its own function, to get rid of warnings
6005 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
6006
6007 7. Constructs such as \d{8,} were compiling into the equivalent of
6008 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
6009 outcome, but in this particular case used more store than had been allocated,
6010 which caused the bug to be discovered because it threw up an internal error.
6011
6012 8. The debugging code in both pcre and pcretest for outputting the compiled
6013 form of a regex was going wrong in the case of back references followed by
6014 curly-bracketed repeats.
6015
6016
6017 Version 1.02 12-Dec-97
6018 ----------------------
6019
6020 1. Typos in pcre.3 and comments in the source fixed.
6021
6022 2. Applied a contributed patch to get rid of places where it used to remove
6023 'const' from variables, and fixed some signed/unsigned and uninitialized
6024 variable warnings.
6025
6026 3. Added the "runtest" target to Makefile.
6027
6028 4. Set default compiler flag to -O2 rather than just -O.
6029
6030
6031 Version 1.01 19-Nov-97
6032 ----------------------
6033
6034 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
6035 like /([ab]*)*/, that is, for classes with more than one character in them.
6036
6037 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
6038 as /((?>a*))*/ (a PCRE_EXTRA facility).
6039
6040
6041 Version 1.00 18-Nov-97
6042 ----------------------
6043
6044 1. Added compile-time macros to support systems such as SunOS4 which don't have
6045 memmove() or strerror() but have other things that can be used instead.
6046
6047 2. Arranged that "make clean" removes the executables.
6048
6049
6050 Version 0.99 27-Oct-97
6051 ----------------------
6052
6053 1. Fixed bug in code for optimizing classes with only one character. It was
6054 initializing a 32-byte map regardless, which could cause it to run off the end
6055 of the memory it had got.