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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1741 - (show annotations)
Sat Oct 20 09:38:44 2018 UTC (2 years, 8 months ago) by ph10
File size: 298042 byte(s)
Fix zero-repeat leading subroutine call first character error.
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 8.43 25-June-2018
9 -------------------------
10
11 1. Some time ago the config macro SUPPORT_UTF8 was changed to SUPPORT_UTF
12 because it also applies to UTF-16 and UTF-32. However, this change was not made
13 in the pcre2cpp files; consequently the C++ wrapper has from then been compiled
14 with a bug in it, which would have been picked up by the unit test except that
15 it also had its UTF8 code cut out. The bug was in a global replace when moving
16 forward after matching an empty string.
17
18 2. The C++ wrapper got broken a long time ago (version 7.3, August 2007) when
19 (*CR) was invented (assuming it was the first such start-of-pattern option).
20 The wrapper could never handle such patterns because it wraps patterns in
21 (?:...)\z in order to support end anchoring. I have hacked in some code to fix
22 this, that is, move the wrapping till after any existing start-of-pattern
23 special settings.
24
25 3. "pcre2grep" (sic) was accidentally mentioned in an error message (fix was
26 ported from PCRE2).
27
28 4. Typo LCC_ALL for LC_ALL fixed in pcregrep.
29
30 5. In a pattern such as /[^\x{100}-\x{ffff}]*[\x80-\xff]/ which has a repeated
31 negative class with no characters less than 0x100 followed by a positive class
32 with only characters less than 0x100, the first class was incorrectly being
33 auto-possessified, causing incorrect match failures.
34
35 6. If the only branch in a conditional subpattern was anchored, the whole
36 subpattern was treated as anchored, when it should not have been, since the
37 assumed empty second branch cannot be anchored. Demonstrated by test patterns
38 such as /(?(1)^())b/ or /(?(?=^))b/.
39
40 7. Fix subject buffer overread in JIT when UTF is disabled and \X or \R has
41 a greater than 1 fixed quantifier. This issue was found by Yunho Kim.
42
43 8. If a pattern started with a subroutine call that had a quantifier with a
44 minimum of zero, an incorrect "match must start with this character" could be
45 recorded. Example: /(?&xxx)*ABC(?<xxx>XYZ)/ would (incorrectly) expect 'A' to
46 be the first character of a match.
47
48
49 Version 8.42 20-March-2018
50 --------------------------
51
52 1. Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
53
54 2. Fixed outdated real_pcre definitions in pcre.h.in (patch by Evgeny Kotkov).
55
56 3. pcregrep was truncating components of file names to 128 characters when
57 processing files with the -r option, and also (some very odd code) truncating
58 path names to 512 characters. There is now a check on the absolute length of
59 full path file names, which may be up to 2047 characters long.
60
61 4. Using pcre_dfa_exec(), in UTF mode when UCP support was not defined, there
62 was the possibility of a false positive match when caselessly matching a "not
63 this character" item such as [^\x{1234}] (with a code point greater than 127)
64 because the "other case" variable was not being initialized.
65
66 5. Although pcre_jit_exec checks whether the pattern is compiled
67 in a given mode, it was also expected that at least one mode is available.
68 This is fixed and pcre_jit_exec returns with PCRE_ERROR_JIT_BADOPTION
69 when the pattern is not optimized by JIT at all.
70
71 6. The line number and related variables such as match counts in pcregrep
72 were all int variables, causing overflow when files with more than 2147483647
73 lines were processed (assuming 32-bit ints). They have all been changed to
74 unsigned long ints.
75
76 7. If a backreference with a minimum repeat count of zero was first in a
77 pattern, apart from assertions, an incorrect first matching character could be
78 recorded. For example, for the pattern /(?=(a))\1?b/, "b" was incorrectly set
79 as the first character of a match.
80
81 8. Fix out-of-bounds read for partial matching of /./ against an empty string
82 when the newline type is CRLF.
83
84 9. When matching using the the REG_STARTEND feature of the POSIX API with a
85 non-zero starting offset, unset capturing groups with lower numbers than a
86 group that did capture something were not being correctly returned as "unset"
87 (that is, with offset values of -1).
88
89 10. Matching the pattern /(*UTF)\C[^\v]+\x80/ against an 8-bit string
90 containing multi-code-unit characters caused bad behaviour and possibly a
91 crash. This issue was fixed for other kinds of repeat in release 8.37 by change
92 38, but repeating character classes were overlooked.
93
94 11. A small fix to pcregrep to avoid compiler warnings for -Wformat-overflow=2.
95
96 12. Added --enable-jit=auto support to configure.ac.
97
98 13. Fix misleading error message in configure.ac.
99
100
101 Version 8.41 05-July-2017
102 -------------------------
103
104 1. Fixed typo in CMakeLists.txt (wrong number of arguments for
105 PCRE_STATIC_RUNTIME (affects MSVC only).
106
107 2. Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
108 mode with --only-matching matched several lines, it restarted scanning at the
109 next line instead of moving on to the end of the matched string, which can be
110 several lines after the start.
111
112 3. Fix a missing else in the JIT compiler reported by 'idaifish'.
113
114 4. A (?# style comment is now ignored between a basic quantifier and a
115 following '+' or '?' (example: /X+(?#comment)?Y/.
116
117 5. Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
118 Pisar).
119
120 6. Fuzzers have reported issues in pcretest. These are NOT serious (it is,
121 after all, just a test program). However, to stop the reports, some easy ones
122 are fixed:
123
124 (a) Check for values < 256 when calling isprint() in pcretest.
125 (b) Give an error for too big a number after \O.
126
127 7. In the 32-bit library in non-UTF mode, an attempt to find a Unicode
128 property for a character with a code point greater than 0x10ffff (the Unicode
129 maximum) caused a crash.
130
131 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
132 encountered a character class with a possessive repeat, for example [a-f]{3}+.
133
134 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
135 length incorrectly, which could result in buffer overflow.
136
137 10. Remove redundant line of code (accidentally left in ages ago).
138
139 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
140 with namespace pcrecpp (Bugzilla #2084).
141
142 12. Remove a duplication typo in pcre_tables.c.
143
144 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
145 starting offset greater than zero.
146
147
148 Version 8.40 11-January-2017
149 ----------------------------
150
151 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
152 the match extended over a line boundary.
153
154 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
155 MSVC static compilation, putting the first patch under a new option.
156
157 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
158
159 4. Ignore "show all captures" (/=) for DFA matching.
160
161 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
162
163 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
164 without PCRE_UCP set, a negative character type such as \D in a positive
165 class should cause all characters greater than 255 to match, whatever else
166 is in the class. There was a bug that caused this not to happen if a
167 Unicode property item was added to such a class, for example [\D\P{Nd}] or
168 [\W\pL].
169
170 7. When pcretest was outputing information from a callout, the caret indicator
171 for the current position in the subject line was incorrect if it was after
172 an escape sequence for a character whose code point was greater than
173 \x{ff}.
174
175 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
176 the conditional was interpreted as a reference to capturing group 1 instead
177 of a test for recursion. Any group whose name began with R was
178 misinterpreted in this way. (The reference interpretation should only
179 happen if the group's name is precisely "R".)
180
181 9. A number of bugs have been mended relating to match start-up optimizations
182 when the first thing in a pattern is a positive lookahead. These all
183 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
184
185 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
186 both an initial 'X' and a following 'X'.
187 (b) Some patterns starting with an assertion that started with .* were
188 incorrectly optimized as having to match at the start of the subject or
189 after a newline. There are cases where this is not true, for example,
190 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
191 start with spaces. Starting .* in an assertion is no longer taken as an
192 indication of matching at the start (or after a newline).
193
194
195 Version 8.39 14-June-2016
196 -------------------------
197
198 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
199 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
200 misbehaved. This bug was found by the LLVM fuzzer.
201
202 2. Similar to the above, if an isolated \E was present between an item and its
203 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
204 bug was found by the LLVM fuzzer.
205
206 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
207 working correctly in UCP mode.
208
209 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
210 was set when the pmatch argument was NULL. It now returns REG_INVARG.
211
212 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
213
214 6. An empty \Q\E sequence between an item and its qualifier caused
215 pcre_compile() to misbehave when auto callouts were enabled. This bug was
216 found by the LLVM fuzzer.
217
218 7. If a pattern that was compiled with PCRE_EXTENDED started with white
219 space or a #-type comment that was followed by (?-x), which turns off
220 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
221 pcre_compile() assumed that (?-x) applied to the whole pattern and
222 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
223
224 8. A call of pcre_copy_named_substring() for a named substring whose number
225 was greater than the space in the ovector could cause a crash.
226
227 9. Yet another buffer overflow bug involved duplicate named groups with a
228 group that reset capture numbers (compare 8.38/7 below). Once again, I have
229 just allowed for more memory, even if not needed. (A proper fix is
230 implemented in PCRE2, but it involves a lot of refactoring.)
231
232 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
233 start of the match to be earlier than the end.
234
235 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
236
237 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
238 assertion, caused pcretest to generate incorrect output, and also to read
239 uninitialized memory (detected by ASAN or valgrind).
240
241 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
242 nested set of parentheses of sufficient size caused an overflow of the
243 compiling workspace (which was diagnosed, but of course is not desirable).
244
245 14. And yet another buffer overflow bug involving duplicate named groups, this
246 time nested, with a nested back reference. Yet again, I have just allowed
247 for more memory, because anything more needs all the refactoring that has
248 been done for PCRE2. An example pattern that provoked this bug is:
249 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
250 registered as CVE-2016-1283.
251
252 15. pcretest went into a loop if global matching was requested with an ovector
253 size less than 2. It now gives an error message. This bug was found by
254 afl-fuzz.
255
256 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
257 ("assertion expected") when (?(?C) was not followed by an opening
258 parenthesis.
259
260 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
261 actually affect anything, by sheer luck.
262
263 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
264 static compilation.
265
266 19. Modified the RunTest script to incorporate a valgrind suppressions file so
267 that certain errors, provoked by the SSE2 instruction set when JIT is used,
268 are ignored.
269
270 20. A racing condition is fixed in JIT reported by Mozilla.
271
272 21. Minor code refactor to avoid "array subscript is below array bounds"
273 compiler warning.
274
275 22. Minor code refactor to avoid "left shift of negative number" warning.
276
277 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
278 UCP support.
279
280 24. Refactor to avoid compiler warnings in pcrecpp.cc.
281
282 25. Refactor to fix a typo in pcre_jit_test.c
283
284 26. Patch to support compiling pcrecpp.cc with Intel compiler.
285
286
287 Version 8.38 23-November-2015
288 -----------------------------
289
290 1. If a group that contained a recursive back reference also contained a
291 forward reference subroutine call followed by a non-forward-reference
292 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
293 compile correct code, leading to undefined behaviour or an internally
294 detected error. This bug was discovered by the LLVM fuzzer.
295
296 2. Quantification of certain items (e.g. atomic back references) could cause
297 incorrect code to be compiled when recursive forward references were
298 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
299 This bug was discovered by the LLVM fuzzer.
300
301 3. A repeated conditional group whose condition was a reference by name caused
302 a buffer overflow if there was more than one group with the given name.
303 This bug was discovered by the LLVM fuzzer.
304
305 4. A recursive back reference by name within a group that had the same name as
306 another group caused a buffer overflow. For example:
307 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
308
309 5. A forward reference by name to a group whose number is the same as the
310 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
311 a buffer overflow at compile time. This bug was discovered by the LLVM
312 fuzzer.
313
314 6. A lookbehind assertion within a set of mutually recursive subpatterns could
315 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
316
317 7. Another buffer overflow bug involved duplicate named groups with a
318 reference between their definition, with a group that reset capture
319 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
320 fixed by always allowing for more memory, even if not needed. (A proper fix
321 is implemented in PCRE2, but it involves more refactoring.)
322
323 8. There was no check for integer overflow in subroutine calls such as (?123).
324
325 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
326 being treated as a literal 'l' instead of causing an error.
327
328 10. There was a buffer overflow if pcre_exec() was called with an ovector of
329 size 1. This bug was found by american fuzzy lop.
330
331 11. If a non-capturing group containing a conditional group that could match
332 an empty string was repeated, it was not identified as matching an empty
333 string itself. For example: /^(?:(?(1)x|)+)+$()/.
334
335 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
336 \a and \e in test subject lines.
337
338 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
339 instead of the EBCDIC value.
340
341 14. The handling of \c in an EBCDIC environment has been revised so that it is
342 now compatible with the specification in Perl's perlebcdic page.
343
344 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
345 ASCII/Unicode. This has now been added to the list of characters that are
346 recognized as white space in EBCDIC.
347
348 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
349 error (correctly) when used outside a class, but did not give an error
350 within a class.
351
352 17. \h within a class was incorrectly compiled in EBCDIC environments.
353
354 18. A pattern with an unmatched closing parenthesis that contained a backward
355 assertion which itself contained a forward reference caused buffer
356 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
357
358 19. JIT should return with error when the compiled pattern requires more stack
359 space than the maximum.
360
361 20. A possessively repeated conditional group that could match an empty string,
362 for example, /(?(R))*+/, was incorrectly compiled.
363
364 21. Fix infinite recursion in the JIT compiler when certain patterns such as
365 /(?:|a|){100}x/ are analysed.
366
367 22. Some patterns with character classes involving [: and \\ were incorrectly
368 compiled and could cause reading from uninitialized memory or an incorrect
369 error diagnosis.
370
371 23. Pathological patterns containing many nested occurrences of [: caused
372 pcre_compile() to run for a very long time.
373
374 24. A conditional group with only one branch has an implicit empty alternative
375 branch and must therefore be treated as potentially matching an empty
376 string.
377
378 25. If (?R was followed by - or + incorrect behaviour happened instead of a
379 diagnostic.
380
381 26. Arrange to give up on finding the minimum matching length for overly
382 complex patterns.
383
384 27. Similar to (4) above: in a pattern with duplicated named groups and an
385 occurrence of (?| it is possible for an apparently non-recursive back
386 reference to become recursive if a later named group with the relevant
387 number is encountered. This could lead to a buffer overflow. Wen Guanxing
388 from Venustech ADLAB discovered this bug.
389
390 28. If pcregrep was given the -q option with -c or -l, or when handling a
391 binary file, it incorrectly wrote output to stdout.
392
393 29. The JIT compiler did not restore the control verb head in case of *THEN
394 control verbs. This issue was found by Karl Skomski with a custom LLVM
395 fuzzer.
396
397 30. Error messages for syntax errors following \g and \k were giving inaccurate
398 offsets in the pattern.
399
400 31. Added a check for integer overflow in conditions (?(<digits>) and
401 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
402 fuzzer.
403
404 32. Handling recursive references such as (?2) when the reference is to a group
405 later in the pattern uses code that is very hacked about and error-prone.
406 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
407 give an internal error if it is obvious that compiling has gone wrong.
408
409 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
410 This issue was found by Karl Skomski with a custom LLVM fuzzer.
411
412 34. The JIT compiler should restore the control chain for empty possessive
413 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
414
415 35. Match limit check added to JIT recursion. This issue was found by Karl
416 Skomski with a custom LLVM fuzzer.
417
418 36. Yet another case similar to 27 above has been circumvented by an
419 unconditional allocation of extra memory. This issue is fixed "properly" in
420 PCRE2 by refactoring the way references are handled. Wen Guanxing
421 from Venustech ADLAB discovered this bug.
422
423 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
424 with a custom LLVM fuzzer.
425
426 38. Fixed a corner case of range optimization in JIT.
427
428 39. An incorrect error "overran compiling workspace" was given if there were
429 exactly enough group forward references such that the last one extended
430 into the workspace safety margin. The next one would have expanded the
431 workspace. The test for overflow was not including the safety margin.
432
433 40. A match limit issue is fixed in JIT which was found by Karl Skomski
434 with a custom LLVM fuzzer.
435
436 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
437 work under Windows. (Why has it taken so long for anyone to notice?)
438
439 42. In a character class such as [\W\p{Any}] where both a negative-type escape
440 ("not a word character") and a property escape were present, the property
441 escape was being ignored.
442
443 43. Fix crash caused by very long (*MARK) or (*THEN) names.
444
445 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
446 by a single ASCII character in a class item, was incorrectly compiled in
447 UCP mode. The POSIX class got lost, but only if the single character
448 followed it.
449
450 45. [:punct:] in UCP mode was matching some characters in the range 128-255
451 that should not have been matched.
452
453 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
454 class, all characters with code points greater than 255 are in the class.
455 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
456 such as \w are turned into Unicode properties), wide characters were not
457 correctly handled, and could fail to match.
458
459
460 Version 8.37 28-April-2015
461 --------------------------
462
463 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
464 for those parentheses to be closed with whatever has been captured so far.
465 However, it was failing to mark any other groups between the hightest
466 capture so far and the currrent group as "unset". Thus, the ovector for
467 those groups contained whatever was previously there. An example is the
468 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
469
470 2. If an assertion condition was quantified with a minimum of zero (an odd
471 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
472
473 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
474 unrecognized modifier, a crash could occur.
475
476 4. An attempt to do global matching in pcretest with a zero-length ovector
477 caused a crash.
478
479 5. Fixed a memory leak during matching that could occur for a subpattern
480 subroutine call (recursive or otherwise) if the number of captured groups
481 that had to be saved was greater than ten.
482
483 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
484 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
485 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
486
487 7. A UTF pattern containing a "not" match of a non-ASCII character and a
488 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
489
490 8. When a pattern is compiled, it remembers the highest back reference so that
491 when matching, if the ovector is too small, extra memory can be obtained to
492 use instead. A conditional subpattern whose condition is a check on a
493 capture having happened, such as, for example in the pattern
494 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
495 setting the highest backreference number. This mattered only if pcre_exec()
496 was called with an ovector that was too small to hold the capture, and there
497 was no other kind of back reference (a situation which is probably quite
498 rare). The effect of the bug was that the condition was always treated as
499 FALSE when the capture could not be consulted, leading to a incorrect
500 behaviour by pcre_exec(). This bug has been fixed.
501
502 9. A reference to a duplicated named group (either a back reference or a test
503 for being set in a conditional) that occurred in a part of the pattern where
504 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
505 to be incorrectly calculated, leading to overwriting.
506
507 10. A mutually recursive set of back references such as (\2)(\1) caused a
508 segfault at study time (while trying to find the minimum matching length).
509 The infinite loop is now broken (with the minimum length unset, that is,
510 zero).
511
512 11. If an assertion that was used as a condition was quantified with a minimum
513 of zero, matching went wrong. In particular, if the whole group had
514 unlimited repetition and could match an empty string, a segfault was
515 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
516 assertions to be quantified, but not if they are being used as conditions,
517 so the above pattern is faulted by Perl. PCRE has now been changed so that
518 it also rejects such patterns.
519
520 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
521 failed to allow the zero-repeat case if pcre2_exec() was called with an
522 ovector too small to capture the group.
523
524 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
525 Red Hat Product Security:
526
527 (a) A crash if /K and /F were both set with the option to save the compiled
528 pattern.
529
530 (b) Another crash if the option to print captured substrings in a callout
531 was combined with setting a null ovector, for example \O\C+ as a subject
532 string.
533
534 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
535 forward reference repeated a large (but limited) number of times within a
536 repeated outer group that has a zero minimum quantifier, caused incorrect
537 code to be compiled, leading to the error "internal error:
538 previously-checked referenced subpattern not found" when an incorrect
539 memory address was read. This bug was reported as "heap overflow",
540 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
541 CVE-2015-2325.
542
543 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
544 call within a group that also contained a recursive back reference caused
545 incorrect code to be compiled. This bug was reported as "heap overflow",
546 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
547 number CVE-2015-2326.
548
549 24. Computing the size of the JIT read-only data in advance has been a source
550 of various issues, and new ones are still appear unfortunately. To fix
551 existing and future issues, size computation is eliminated from the code,
552 and replaced by on-demand memory allocation.
553
554 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
555 adjacent to the end of the range, and the range contained characters with
556 more than one other case, caused incorrect behaviour when compiled in UTF
557 mode. In that example, the range a-j was left out of the class.
558
559 26. Fix JIT compilation of conditional blocks, which assertion
560 is converted to (*FAIL). E.g: /(?(?!))/.
561
562 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
563 discovered by the LLVM fuzzer.
564
565 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
566 when this assertion was used as a condition, for example (?(?!)a|b). In
567 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
568 error about an unsupported item.
569
570 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
571 possessification code could take exponential time to complete. A recursion
572 depth limit of 1000 has been imposed to limit the resources used by this
573 optimization.
574
575 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
576 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
577 because \S ensures they are all in the class. The code for doing this was
578 interacting badly with the code for computing the amount of space needed to
579 compile the pattern, leading to a buffer overflow. This bug was discovered
580 by the LLVM fuzzer.
581
582 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
583 other kinds of group caused stack overflow at compile time. This bug was
584 discovered by the LLVM fuzzer.
585
586 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
587 between a subroutine call and its quantifier was incorrectly compiled,
588 leading to buffer overflow or other errors. This bug was discovered by the
589 LLVM fuzzer.
590
591 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
592 assertion after (?(. The code was failing to check the character after
593 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
594 was discovered by the LLVM fuzzer.
595
596 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
597 a fixed maximum following a group that contains a subroutine reference was
598 incorrectly compiled and could trigger buffer overflow. This bug was
599 discovered by the LLVM fuzzer.
600
601 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
602 caused a stack overflow instead of the diagnosis of a non-fixed length
603 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
604
605 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
606 (e.g. /(?<=\Ka)/) could make pcregrep loop.
607
608 37. There was a similar problem to 36 in pcretest for global matches.
609
610 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
611 and a subsequent item in the pattern caused a non-match, backtracking over
612 the repeated \X did not stop, but carried on past the start of the subject,
613 causing reference to random memory and/or a segfault. There were also some
614 other cases where backtracking after \C could crash. This set of bugs was
615 discovered by the LLVM fuzzer.
616
617 39. The function for finding the minimum length of a matching string could take
618 a very long time if mutual recursion was present many times in a pattern,
619 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
620 method has been implemented. This infelicity was discovered by the LLVM
621 fuzzer.
622
623 40. Static linking against the PCRE library using the pkg-config module was
624 failing on missing pthread symbols.
625
626
627 Version 8.36 26-September-2014
628 ------------------------------
629
630 1. Got rid of some compiler warnings in the C++ modules that were shown up by
631 -Wmissing-field-initializers and -Wunused-parameter.
632
633 2. The tests for quantifiers being too big (greater than 65535) were being
634 applied after reading the number, and stupidly assuming that integer
635 overflow would give a negative number. The tests are now applied as the
636 numbers are read.
637
638 3. Tidy code in pcre_exec.c where two branches that used to be different are
639 now the same.
640
641 4. The JIT compiler did not generate match limit checks for certain
642 bracketed expressions with quantifiers. This may lead to exponential
643 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
644 issue should be resolved now.
645
646 5. Fixed an issue, which occures when nested alternatives are optimized
647 with table jumps.
648
649 6. Inserted two casts and changed some ints to size_t in the light of some
650 reported 64-bit compiler warnings (Bugzilla 1477).
651
652 7. Fixed a bug concerned with zero-minimum possessive groups that could match
653 an empty string, which sometimes were behaving incorrectly in the
654 interpreter (though correctly in the JIT matcher). This pcretest input is
655 an example:
656
657 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
658 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
659
660 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
661 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
662 for an empty string was breaking the inner loop and carrying on at a lower
663 level, when possessive repeated groups should always return to a higher
664 level as they have no backtrack points in them. The empty string test now
665 occurs at the outer level.
666
667 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
668 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
669
670 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
671 doesn't).
672
673 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
674 to hold a pattern to be incorrectly computed (too small) when there were
675 named back references to duplicated names. This could cause "internal
676 error: code overflow" or "double free or corruption" or other memory
677 handling errors.
678
679 11. When named subpatterns had the same prefixes, back references could be
680 confused. For example, in this pattern:
681
682 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
683
684 the reference to 'Name' was incorrectly treated as a reference to a
685 duplicate name.
686
687 12. A pattern such as /^s?c/mi8 where the optional character has more than
688 one "other case" was incorrectly compiled such that it would only try to
689 match starting at "c".
690
691 13. When a pattern starting with \s was studied, VT was not included in the
692 list of possible starting characters; this should have been part of the
693 8.34/18 patch.
694
695 14. If a character class started [\Qx]... where x is any character, the class
696 was incorrectly terminated at the ].
697
698 15. If a pattern that started with a caseless match for a character with more
699 than one "other case" was studied, PCRE did not set up the starting code
700 unit bit map for the list of possible characters. Now it does. This is an
701 optimization improvement, not a bug fix.
702
703 16. The Unicode data tables have been updated to Unicode 7.0.0.
704
705 17. Fixed a number of memory leaks in pcregrep.
706
707 18. Avoid a compiler warning (from some compilers) for a function call with
708 a cast that removes "const" from an lvalue by using an intermediate
709 variable (to which the compiler does not object).
710
711 19. Incorrect code was compiled if a group that contained an internal recursive
712 back reference was optional (had quantifier with a minimum of zero). This
713 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
714 caused segmentation faults because of stack overflows at compile time.
715
716 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
717 group that is quantified with an indefinite repeat, caused a compile-time
718 loop which used up all the system stack and provoked a segmentation fault.
719 This was not the same bug as 19 above.
720
721 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
722 Patch by Mike Frysinger.
723
724
725 Version 8.35 04-April-2014
726 --------------------------
727
728 1. A new flag is set, when property checks are present in an XCLASS.
729 When this flag is not set, PCRE can perform certain optimizations
730 such as studying these XCLASS-es.
731
732 2. The auto-possessification of character sets were improved: a normal
733 and an extended character set can be compared now. Furthermore
734 the JIT compiler optimizes more character set checks.
735
736 3. Got rid of some compiler warnings for potentially uninitialized variables
737 that show up only when compiled with -O2.
738
739 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
740 of a match later then the end of the match. The pcretest program was not
741 handling the case sensibly - it was outputting from the start to the next
742 binary zero. It now reports this situation in a message, and outputs the
743 text from the end to the start.
744
745 5. Fast forward search is improved in JIT. Instead of the first three
746 characters, any three characters with fixed position can be searched.
747 Search order: first, last, middle.
748
749 6. Improve character range checks in JIT. Characters are read by an inprecise
750 function now, which returns with an unknown value if the character code is
751 above a certain threshold (e.g: 256). The only limitation is that the value
752 must be bigger than the threshold as well. This function is useful when
753 the characters above the threshold are handled in the same way.
754
755 7. The macros whose names start with RAWUCHAR are placeholders for a future
756 mode in which only the bottom 21 bits of 32-bit data items are used. To
757 make this more memorable for those maintaining the code, the names have
758 been changed to start with UCHAR21, and an extensive comment has been added
759 to their definition.
760
761 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
762 to the export list in Makefile.am (they were accidentally omitted from the
763 8.34 tarball).
764
765 9. The informational output from pcretest used the phrase "starting byte set"
766 which is inappropriate for the 16-bit and 32-bit libraries. As the output
767 for "first char" and "need char" really means "non-UTF-char", I've changed
768 "byte" to "char", and slightly reworded the output. The documentation about
769 these values has also been (I hope) clarified.
770
771 10. Another JIT related optimization: use table jumps for selecting the correct
772 backtracking path, when more than four alternatives are present inside a
773 bracket.
774
775 11. Empty match is not possible, when the minimum length is greater than zero,
776 and there is no \K in the pattern. JIT should avoid empty match checks in
777 such cases.
778
779 12. In a caseless character class with UCP support, when a character with more
780 than one alternative case was not the first character of a range, not all
781 the alternative cases were added to the class. For example, s and \x{17f}
782 are both alternative cases for S: the class [RST] was handled correctly,
783 but [R-T] was not.
784
785 13. The configure.ac file always checked for pthread support when JIT was
786 enabled. This is not used in Windows, so I have put this test inside a
787 check for the presence of windows.h (which was already tested for).
788
789 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
790 The algorithm provides a way to skip certain starting offsets, and usually
791 faster than linear prefix searches.
792
793 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
794 as for 'fr_FR' and 'french'. For some reason, however, it then used the
795 Windows-specific input and output files, which have 'french' screwed in.
796 So this could never have worked. One of the problems with locales is that
797 they aren't always the same. I have now updated RunTest so that it checks
798 the output of the locale test (test 3) against three different output
799 files, and it allows the test to pass if any one of them matches. With luck
800 this should make the test pass on some versions of Solaris where it was
801 failing. Because of the uncertainty, the script did not used to stop if
802 test 3 failed; it now does. If further versions of a French locale ever
803 come to light, they can now easily be added.
804
805 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
806 there was a message during ./configure, but it did not stop. This now
807 provokes an error. The invalid example in README has been corrected.
808 If a value less than the minimum is given, the minimum value has always
809 been used, but now a warning is given.
810
811 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
812 was a bug in the test system, which is now fixed. Also, the list of various
813 configurations that are tested for each release did not have one with both
814 16/32 bits and --enable-bar-anycrlf. It now does.
815
816 18. pcretest was missing "-C bsr" for displaying the \R default setting.
817
818 19. Little endian PowerPC systems are supported now by the JIT compiler.
819
820 20. The fast forward newline mechanism could enter to an infinite loop on
821 certain invalid UTF-8 input. Although we don't support these cases
822 this issue can be fixed by a performance optimization.
823
824 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
825 not take account if existing stack usage. There is now a new global
826 variable called pcre_stack_guard that can be set to point to an external
827 function to check stack availability. It is called at the start of
828 processing every parenthesized group.
829
830 22. A typo in the code meant that in ungreedy mode the max/min qualifier
831 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
832 not match "ab".
833
834 23. When UTF was disabled, the JIT program reported some incorrect compile
835 errors. These messages are silenced now.
836
837 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
838 compiler.
839
840 25. Change all the temporary files used in RunGrepTest to be different to those
841 used by RunTest so that the tests can be run simultaneously, for example by
842 "make -j check".
843
844
845 Version 8.34 15-December-2013
846 -----------------------------
847
848 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
849 executable memory. Patch inspired by Carsten Klein.
850
851 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
852 this macro is never tested and has no effect, because the work to support
853 coverage involves only compiling and linking options and special targets in
854 the Makefile. The comment in config.h implied that defining the macro would
855 enable coverage support, which is totally false. There was also support for
856 setting this macro in the CMake files (my fault, I just copied it from
857 configure). SUPPORT_GCOV has now been removed.
858
859 3. Make a small performance improvement in strlen16() and strlen32() in
860 pcretest.
861
862 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
863 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
864 these cases). There was also one in pcretest.c.
865
866 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
867
868 6. In UTF mode, the code for checking whether a group could match an empty
869 string (which is used for indefinitely repeated groups to allow for
870 breaking an infinite loop) was broken when the group contained a repeated
871 negated single-character class with a character that occupied more than one
872 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
873 UTF-8 mode). The effect was undefined: the group might or might not be
874 deemed as matching an empty string, or the program might have crashed.
875
876 7. The code for checking whether a group could match an empty string was not
877 recognizing that \h, \H, \v, \V, and \R must match a character.
878
879 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
880 an empty string. If it can, pcretest shows this in its information output.
881
882 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
883 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
884 matched by pcre_exec() without using JIT:
885
886 (a) If the rest of the pattern did not match after a maximal run of
887 grapheme clusters, the code for backing up to try with fewer of them
888 did not always back up over a full grapheme when characters that do not
889 have the modifier quality were involved, e.g. Hangul syllables.
890
891 (b) If the match point in a subject started with modifier character, and
892 there was no match, the code could incorrectly back up beyond the match
893 point, and potentially beyond the first character in the subject,
894 leading to a segfault or an incorrect match result.
895
896 10. A conditional group with an assertion condition could lead to PCRE
897 recording an incorrect first data item for a match if no other first data
898 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
899 first data item, and therefore matched "ca" after "c" instead of at the
900 start.
901
902 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
903 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
904
905 12. The source of pcregrep now includes z/OS-specific code so that it can be
906 compiled for z/OS as part of the special z/OS distribution.
907
908 13. Added the -T and -TM options to pcretest.
909
910 14. The code in pcre_compile.c for creating the table of named capturing groups
911 has been refactored. Instead of creating the table dynamically during the
912 actual compiling pass, the information is remembered during the pre-compile
913 pass (on the stack unless there are more than 20 named groups, in which
914 case malloc() is used) and the whole table is created before the actual
915 compile happens. This has simplified the code (it is now nearly 150 lines
916 shorter) and prepared the way for better handling of references to groups
917 with duplicate names.
918
919 15. A back reference to a named subpattern when there is more than one of the
920 same name now checks them in the order in which they appear in the pattern.
921 The first one that is set is used for the reference. Previously only the
922 first one was inspected. This change makes PCRE more compatible with Perl.
923
924 16. Unicode character properties were updated from Unicode 6.3.0.
925
926 17. The compile-time code for auto-possessification has been refactored, based
927 on a patch by Zoltan Herczeg. It now happens after instead of during
928 compilation. The code is cleaner, and more cases are handled. The option
929 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
930 options in pcretest are provided to set it. It can also be set by
931 (*NO_AUTO_POSSESS) at the start of a pattern.
932
933 18. The character VT has been added to the default ("C" locale) set of
934 characters that match \s and are generally treated as white space,
935 following this same change in Perl 5.18. There is now no difference between
936 "Perl space" and "POSIX space". Whether VT is treated as white space in
937 other locales depends on the locale.
938
939 19. The code for checking named groups as conditions, either for being set or
940 for being recursed, has been refactored (this is related to 14 and 15
941 above). Processing unduplicated named groups should now be as fast at
942 numerical groups, and processing duplicated groups should be faster than
943 before.
944
945 20. Two patches to the CMake build system, by Alexander Barkov:
946
947 (1) Replace the "source" command by "." in CMakeLists.txt because
948 "source" is a bash-ism.
949
950 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
951 without these the CMake build does not work on Solaris.
952
953 21. Perl has changed its handling of \8 and \9. If there is no previously
954 encountered capturing group of those numbers, they are treated as the
955 literal characters 8 and 9 instead of a binary zero followed by the
956 literals. PCRE now does the same.
957
958 22. Following Perl, added \o{} to specify codepoints in octal, making it
959 possible to specify values greater than 0777 and also making them
960 unambiguous.
961
962 23. Perl now gives an error for missing closing braces after \x{... instead of
963 treating the string as literal. PCRE now does the same.
964
965 24. RunTest used to grumble if an inappropriate test was selected explicitly,
966 but just skip it when running all tests. This make it awkward to run ranges
967 of tests when one of them was inappropriate. Now it just skips any
968 inappropriate tests, as it always did when running all tests.
969
970 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
971 character types such as \d or \w, too many callouts were inserted, and the
972 data that they returned was rubbish.
973
974 26. In UCP mode, \s was not matching two of the characters that Perl matches,
975 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
976 were matched by \h. The code has now been refactored so that the lists of
977 the horizontal and vertical whitespace characters used for \h and \v (which
978 are defined only in one place) are now also used for \s.
979
980 27. Add JIT support for the 64 bit TileGX architecture.
981 Patch by Jiong Wang (Tilera Corporation).
982
983 28. Possessive quantifiers for classes (both explicit and automatically
984 generated) now use special opcodes instead of wrapping in ONCE brackets.
985
986 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
987 (because it's meaningless), this was not happening when PCRE_CASELESS was
988 set. Not wrong, but inefficient.
989
990 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
991 properties for \w, \d, etc) is present in a test regex. Otherwise if the
992 test contains no characters greater than 255, Perl doesn't realise it
993 should be using Unicode semantics.
994
995 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
996 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
997 does in Unicode mode.
998
999 32. Added the "forbid" facility to pcretest so that putting tests into the
1000 wrong test files can sometimes be quickly detected.
1001
1002 33. There is now a limit (default 250) on the depth of nesting of parentheses.
1003 This limit is imposed to control the amount of system stack used at compile
1004 time. It can be changed at build time by --with-parens-nest-limit=xxx or
1005 the equivalent in CMake.
1006
1007 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
1008 errors. Perl warns for these when in warning mode, but PCRE has no facility
1009 for giving warnings.
1010
1011 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
1012 However, this was not working for (?!) because it is optimized to (*FAIL),
1013 for which PCRE does not allow quantifiers. The optimization is now disabled
1014 when a quantifier follows (?!). I can't see any use for this, but it makes
1015 things uniform.
1016
1017 36. Perl no longer allows group names to start with digits, so I have made this
1018 change also in PCRE. It simplifies the code a bit.
1019
1020 37. In extended mode, Perl ignores spaces before a + that indicates a
1021 possessive quantifier. PCRE allowed a space before the quantifier, but not
1022 before the possessive +. It now does.
1023
1024 38. The use of \K (reset reported match start) within a repeated possessive
1025 group such as (a\Kb)*+ was not working.
1026
1027 40. Document that the same character tables must be used at compile time and
1028 run time, and that the facility to pass tables to pcre_exec() and
1029 pcre_dfa_exec() is for use only with saved/restored patterns.
1030
1031 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
1032 features for Builds with MSVC:
1033
1034 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
1035 builds). The .rc files can be used to set FileDescription and many other
1036 attributes.
1037
1038 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
1039 This allows higher-level build scripts which want .pdb files to avoid
1040 hard-coding the exact files needed."
1041
1042 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
1043 mean "start of word" and "end of word", respectively, as a transition aid.
1044
1045 43. A minimizing repeat of a class containing codepoints greater than 255 in
1046 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
1047 compiled to use the heap for recursion.
1048
1049 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
1050 is configured.
1051
1052
1053 Version 8.33 28-May-2013
1054 ------------------------
1055
1056 1. Added 'U' to some constants that are compared to unsigned integers, to
1057 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
1058 variables that are added to signed variables, to ensure the result is
1059 signed and can be negated.
1060
1061 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
1062 CMake config files.
1063
1064 3. Revise the creation of config.h.generic so that all boolean macros are
1065 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
1066 overriding via -D on the command line possible.
1067
1068 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
1069 to unsigned int is reported to make a quite noticeable speed difference in
1070 a specific Windows environment. Testing on Linux did also appear to show
1071 some benefit (and it is clearly not harmful). Also fixed the definition of
1072 Xop which should be unsigned.
1073
1074 5. Related to (4), changing the definition of the intermediate variable cc
1075 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
1076 improvements.
1077
1078 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
1079 unnecessary spaces.
1080
1081 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
1082 and later.
1083
1084 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
1085
1086 9. Optimizing fast_forward_start_bits in JIT.
1087
1088 10. Adding support for callouts in JIT, and fixing some issues revealed
1089 during this work. Namely:
1090
1091 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
1092
1093 (b) Minimum length was not checked before the matching is started.
1094
1095 11. The value of capture_last that is passed to callouts was incorrect in some
1096 cases when there was a capture on one path that was subsequently abandoned
1097 after a backtrack. Also, the capture_last value is now reset after a
1098 recursion, since all captures are also reset in this case.
1099
1100 12. The interpreter no longer returns the "too many substrings" error in the
1101 case when an overflowing capture is in a branch that is subsequently
1102 abandoned after a backtrack.
1103
1104 13. In the pathological case when an offset vector of size 2 is used, pcretest
1105 now prints out the matched string after a yield of 0 or 1.
1106
1107 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
1108 Only supported by the JIT compiler at the moment.
1109
1110 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
1111
1112 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
1113 offset of the starting point of the matching process, provided the offsets
1114 vector is large enough.
1115
1116 17. The \A escape now records a lookbehind value of 1, though its execution
1117 does not actually inspect the previous character. This is to ensure that,
1118 in partial multi-segment matching, at least one character from the old
1119 segment is retained when a new segment is processed. Otherwise, if there
1120 are no lookbehinds in the pattern, \A might match incorrectly at the start
1121 of a new segment.
1122
1123 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
1124
1125 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
1126 gives some modest performance improvement in 8-bit mode.
1127
1128 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
1129 be expressed in certain programming languages using Universal Character
1130 Names.
1131
1132 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
1133 which points out that "non characters" are not "characters that may not
1134 appear in Unicode strings" but rather "characters that are reserved for
1135 internal use and have only local meaning".
1136
1137 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
1138 there was a conditional group that depended on an assertion, if the
1139 assertion was false, the callout that immediately followed the alternation
1140 in the condition was skipped when pcre_exec() was used for matching.
1141
1142 23. Allow an explicit callout to be inserted before an assertion that is the
1143 condition for a conditional group, for compatibility with automatic
1144 callouts, which always insert a callout at this point.
1145
1146 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1147 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1148
1149 25. (*PRUNE) is now supported by the JIT compiler.
1150
1151 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1152
1153 27. Fix the case where there are two or more SKIPs with arguments that may be
1154 ignored.
1155
1156 28. (*SKIP) is now supported by the JIT compiler.
1157
1158 29. (*THEN) is now supported by the JIT compiler.
1159
1160 30. Update RunTest with additional test selector options.
1161
1162 31. The way PCRE handles backtracking verbs has been changed in two ways.
1163
1164 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1165 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1166 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1167 rather obscure rules do not always do the same thing.
1168
1169 (2) Previously, backtracking verbs were confined within assertions. This is
1170 no longer the case for positive assertions, except for (*ACCEPT). Again,
1171 this sometimes improves Perl compatibility, and sometimes does not.
1172
1173 32. A number of tests that were in test 2 because Perl did things differently
1174 have been moved to test 1, because either Perl or PCRE has changed, and
1175 these tests are now compatible.
1176
1177 32. Backtracking control verbs are now handled in the same way in JIT and
1178 interpreter.
1179
1180 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1181 contained a forward subroutine reference caused a compile error.
1182
1183 34. Auto-detect and optimize limited repetitions in JIT.
1184
1185 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1186 blocking (*UTF) etc.
1187
1188 36. In the interpreter, maximizing pattern repetitions for characters and
1189 character types now use tail recursion, which reduces stack usage.
1190
1191 37. The value of the max lookbehind was not correctly preserved if a compiled
1192 and saved regex was reloaded on a host of different endianness.
1193
1194 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1195 of the compiled pattern block, expand the flags field from 16 to 32 bits
1196 because it was almost full.
1197
1198 39. Try madvise first before posix_madvise.
1199
1200 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1201 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1202 original change locked out all matches of empty strings. This has been
1203 changed so that one match of an empty string per line is recognized.
1204 Subsequent searches on the same line (for colouring or for --only-matching,
1205 for example) do not recognize empty strings.
1206
1207 41. Applied a user patch to fix a number of spelling mistakes in comments.
1208
1209 42. Data lines longer than 65536 caused pcretest to crash.
1210
1211 43. Clarified the data type for length and startoffset arguments for pcre_exec
1212 and pcre_dfa_exec in the function-specific man pages, where they were
1213 explicitly stated to be in bytes, never having been updated. I also added
1214 some clarification to the pcreapi man page.
1215
1216 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1217 a segmentation fault.
1218
1219
1220 Version 8.32 30-November-2012
1221 -----------------------------
1222
1223 1. Improved JIT compiler optimizations for first character search and single
1224 character iterators.
1225
1226 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1227 Patch by Daniel Richard G.
1228
1229 3. Single character iterator optimizations in the JIT compiler.
1230
1231 4. Improved JIT compiler optimizations for character ranges.
1232
1233 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1234 Reported by Giuseppe D'Angelo.
1235
1236 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1237 of a line, was being set incorrectly in cases where .* appeared inside
1238 atomic brackets at the start of a pattern, or where there was a subsequent
1239 *PRUNE or *SKIP.
1240
1241 7. Improved instruction cache flush for POWER/PowerPC.
1242 Patch by Daniel Richard G.
1243
1244 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1245 grep:
1246
1247 (a) There is now no limit to the number of patterns to be matched.
1248
1249 (b) An error is given if a pattern is too long.
1250
1251 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1252 are now supported.
1253
1254 (d) --exclude-from and --include-from (multiple use) have been added.
1255
1256 (e) Exclusions and inclusions now apply to all files and directories, not
1257 just to those obtained from scanning a directory recursively.
1258
1259 (f) Multiple uses of -f and --file-list are now supported.
1260
1261 (g) In a Windows environment, the default for -d has been changed from
1262 "read" (the GNU grep default) to "skip", because otherwise the presence
1263 of a directory in the file list provokes an error.
1264
1265 (h) The documentation has been revised and clarified in places.
1266
1267 9. Improve the matching speed of capturing brackets.
1268
1269 10. Changed the meaning of \X so that it now matches a Unicode extended
1270 grapheme cluster.
1271
1272 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1273 out POSIX threads when JIT support is configured.
1274
1275 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1276
1277 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1278 structure could go wrong in environments where size_t is not the same size
1279 as int.
1280
1281 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1282 set.
1283
1284 15. The EBCDIC support had decayed; later updates to the code had included
1285 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1286 general tidy up of EBCDIC-related issues, and the documentation was also
1287 not quite right. There is now a test that can be run on ASCII systems to
1288 check some of the EBCDIC-related things (but is it not a full test).
1289
1290 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1291 in a small tidy to the code.
1292
1293 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1294
1295 18. If the --only-matching (-o) option in pcregrep is specified multiple
1296 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1297 substrings matched by the 1st and 2nd capturing parentheses. A separating
1298 string can be specified by --om-separator (default empty).
1299
1300 19. Improving the first n character searches.
1301
1302 20. Turn case lists for horizontal and vertical white space into macros so that
1303 they are defined only once.
1304
1305 21. This set of changes together give more compatible Unicode case-folding
1306 behaviour for characters that have more than one other case when UCP
1307 support is available.
1308
1309 (a) The Unicode property table now has offsets into a new table of sets of
1310 three or more characters that are case-equivalent. The MultiStage2.py
1311 script that generates these tables (the pcre_ucd.c file) now scans
1312 CaseFolding.txt instead of UnicodeData.txt for character case
1313 information.
1314
1315 (b) The code for adding characters or ranges of characters to a character
1316 class has been abstracted into a generalized function that also handles
1317 case-independence. In UTF-mode with UCP support, this uses the new data
1318 to handle characters with more than one other case.
1319
1320 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1321 whose other case is greater than 256 are now correctly matched
1322 caselessly. Previously, the high codepoint matched the low one, but not
1323 vice versa.
1324
1325 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1326 of the new class addition function, using character lists defined as
1327 macros alongside the case definitions of 20 above.
1328
1329 (e) Caseless back references now work with characters that have more than
1330 one other case.
1331
1332 (f) General caseless matching of characters with more than one other case
1333 is supported.
1334
1335 22. Unicode character properties were updated from Unicode 6.2.0
1336
1337 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1338
1339 24. Add support for 32-bit character strings, and UTF-32
1340
1341 25. Major JIT compiler update (code refactoring and bugfixing).
1342 Experimental Sparc 32 support is added.
1343
1344 26. Applied a modified version of Daniel Richard G's patch to create
1345 pcre.h.generic and config.h.generic by "make" instead of in the
1346 PrepareRelease script.
1347
1348 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1349 pcre_compile.c when checking for a zero character.
1350
1351 28. Introducing a native interface for JIT. Through this interface, the compiled
1352 machine code can be directly executed. The purpose of this interface is to
1353 provide fast pattern matching, so several sanity checks are not performed.
1354 However, feature tests are still performed. The new interface provides
1355 1.4x speedup compared to the old one.
1356
1357 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1358 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1359 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1360
1361 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1362 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1363 the "old" RFC 2279). Instead, it ended up passing a negative length to
1364 pcre_exec().
1365
1366 31. Add support for GCC's visibility feature to hide internal functions.
1367
1368 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1369 "unknown -C option" after outputting 0 or 1.
1370
1371 33. There is now support for generating a code coverage report for the test
1372 suite in environments where gcc is the compiler and lcov is installed. This
1373 is mainly for the benefit of the developers.
1374
1375 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1376 unaddressable using valgrind annotations, allowing valgrind to detect
1377 invalid memory accesses. This is mainly for the benefit of the developers.
1378
1379 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1380
1381 26. Give configure error if --enable-cpp but no C++ compiler found.
1382
1383
1384 Version 8.31 06-July-2012
1385 -------------------------
1386
1387 1. Fixing a wrong JIT test case and some compiler warnings.
1388
1389 2. Removed a bashism from the RunTest script.
1390
1391 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1392 to unsigned type, result still unsigned" that was given by an MS compiler
1393 on encountering the code "-sizeof(xxx)".
1394
1395 4. Partial matching support is added to the JIT compiler.
1396
1397 5. Fixed several bugs concerned with partial matching of items that consist
1398 of more than one character:
1399
1400 (a) /^(..)\1/ did not partially match "aba" because checking references was
1401 done on an "all or nothing" basis. This also applied to repeated
1402 references.
1403
1404 (b) \R did not give a hard partial match if \r was found at the end of the
1405 subject.
1406
1407 (c) \X did not give a hard partial match after matching one or more
1408 characters at the end of the subject.
1409
1410 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1411 a partial match for the string "\r".
1412
1413 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1414 a partial match for a CR character at the end of the subject string.
1415
1416 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1417 running pcretest, the text "(JIT)" added to the output whenever JIT is
1418 actually used to run the match.
1419
1420 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1421 or /S+[+] with a digit between 1 and 7.
1422
1423 8. OP_NOT now supports any UTF character not just single-byte ones.
1424
1425 9. (*MARK) control verb is now supported by the JIT compiler.
1426
1427 10. The command "./RunTest list" lists the available tests without actually
1428 running any of them. (Because I keep forgetting what they all are.)
1429
1430 11. Add PCRE_INFO_MAXLOOKBEHIND.
1431
1432 12. Applied a (slightly modified) user-supplied patch that improves performance
1433 when the heap is used for recursion (compiled with --disable-stack-for-
1434 recursion). Instead of malloc and free for each heap frame each time a
1435 logical recursion happens, frames are retained on a chain and re-used where
1436 possible. This sometimes gives as much as 30% improvement.
1437
1438 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1439 call.
1440
1441 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1442
1443 15. It is now possible to link pcretest with libedit as an alternative to
1444 libreadline.
1445
1446 16. (*COMMIT) control verb is now supported by the JIT compiler.
1447
1448 17. The Unicode data tables have been updated to Unicode 6.1.0.
1449
1450 18. Added --file-list option to pcregrep.
1451
1452 19. Added binary file support to pcregrep, including the -a, --binary-files,
1453 -I, and --text options.
1454
1455 20. The madvise function is renamed for posix_madvise for QNX compatibility
1456 reasons. Fixed by Giuseppe D'Angelo.
1457
1458 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1459
1460 22. Changed the option for creating symbolic links for 16-bit man pages from
1461 -s to -sf so that re-installing does not cause issues.
1462
1463 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1464
1465 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1466 matches in certain environments (the workspace was not being correctly
1467 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1468 some of the workspace data at the beginning of a restart.
1469
1470 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1471 was not doing so when it should - probably a typo introduced by SVN 528
1472 (change 8.10/14).
1473
1474 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1475 \w+ when the character tables indicated that \x{c4} was a word character.
1476 There were several related cases, all because the tests for doing a table
1477 lookup were testing for characters less than 127 instead of 255.
1478
1479 27. If a pattern contains capturing parentheses that are not used in a match,
1480 their slots in the ovector are set to -1. For those that are higher than
1481 any matched groups, this happens at the end of processing. In the case when
1482 there were back references that the ovector was too small to contain
1483 (causing temporary malloc'd memory to be used during matching), and the
1484 highest capturing number was not used, memory off the end of the ovector
1485 was incorrectly being set to -1. (It was using the size of the temporary
1486 memory instead of the true size.)
1487
1488 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1489 ovector size, it uses memory at the end of the block that it has got.
1490
1491 29. Check for an overlong MARK name and give an error at compile time. The
1492 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1493
1494 30. JIT compiler update.
1495
1496 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1497 Rill for the patch.
1498
1499 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1500
1501 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1502
1503 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1504 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1505 was enabled.
1506
1507 35. Improve JIT code generation for greedy plus quantifier.
1508
1509 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1510 1 to "aa" instead of to an empty string. The bug affected repeated groups
1511 that could potentially match an empty string.
1512
1513 37. Optimizing single character iterators in JIT.
1514
1515 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1516 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1517 codepoints that are too big for the mode are faulted, and in a UTF mode,
1518 disallowed codepoints are also faulted.
1519
1520 39. If PCRE was compiled with UTF support, in three places in the DFA
1521 matcher there was code that should only have been obeyed in UTF mode, but
1522 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1523 processing when bytes with values greater than 127 were present. In 16-bit
1524 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1525 both cases the values are those that cannot be the first data item in a UTF
1526 character. The three items that might have provoked this were recursions,
1527 possessively repeated groups, and atomic groups.
1528
1529 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1530 and pcregrep, because some OS require shared objects to be explicitly
1531 passed to ld, causing the link step to fail if they are not.
1532
1533 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1534 mode, patterns that started with \h* or \R* might be incorrectly matched.
1535
1536
1537 Version 8.30 04-February-2012
1538 -----------------------------
1539
1540 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1541 name is defined in ctype.h.
1542
1543 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1544 only in quite long subpatterns.
1545
1546 3. Removed the function pcre_info(), which has been obsolete and deprecated
1547 since it was replaced by pcre_fullinfo() in February 2000.
1548
1549 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1550 match a (*MARK), and the match failed at the start of the subject, a
1551 reference to memory before the start of the subject could occur. This bug
1552 was introduced by fix 17 of release 8.21.
1553
1554 5. A reference to an unset group with zero minimum repetition was giving
1555 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1556 /(another)?(\1?)test/ matched against "hello world test". This bug was
1557 introduced in release 8.13.
1558
1559 6. Add support for 16-bit character strings (a large amount of work involving
1560 many changes and refactorings).
1561
1562 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1563 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1564 from a file.
1565
1566 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1567 rounding is not applied in this particular case).
1568
1569 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1570 if they appear, or are escaped, in patterns.
1571
1572 10. Get rid of a number of -Wunused-but-set-variable warnings.
1573
1574 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1575 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1576 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1577 also returns the mark "x". This bug applied to capturing parentheses,
1578 non-capturing parentheses, and atomic parentheses. It also applied to some
1579 assertions.
1580
1581 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1582 information out of configure.ac instead of relying on pcre.h.generic, which
1583 is not stored in the repository.
1584
1585 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1586 -lreadline.
1587
1588 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1589
1590 15. Applied Graycode's patch to put the top-level frame on the stack rather
1591 than the heap when not using the stack for recursion. This gives a
1592 performance improvement in many cases when recursion is not deep.
1593
1594 16. Experimental code added to "pcretest -C" to output the stack frame size.
1595
1596
1597 Version 8.21 12-Dec-2011
1598 ------------------------
1599
1600 1. Updating the JIT compiler.
1601
1602 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1603 are added as well.
1604
1605 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1606 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1607 calling _pcre_jit_exec. Some extra comments are added.
1608
1609 4. (*MARK) settings inside atomic groups that do not contain any capturing
1610 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1611 was introduced by change 18 for 8.20.
1612
1613 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1614 ECMA-262 standard.
1615
1616 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1617 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1618 This bug was probably introduced by change 9 of 8.13.
1619
1620 7. While fixing 6 above, I noticed that a number of other items were being
1621 incorrectly rejected as "not fixed length". This arose partly because newer
1622 opcodes had not been added to the fixed-length checking code. I have (a)
1623 corrected the bug and added tests for these items, and (b) arranged for an
1624 error to occur if an unknown opcode is encountered while checking for fixed
1625 length instead of just assuming "not fixed length". The items that were
1626 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1627 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1628 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1629
1630 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1631 being incorrectly compiled and would have given unpredicatble results.
1632
1633 9. A possessively repeated subpattern with minimum repeat count greater than
1634 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1635 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1636 the first (A) could occur when it should not.
1637
1638 10. Add a cast and remove a redundant test from the code.
1639
1640 11. JIT should use pcre_malloc/pcre_free for allocation.
1641
1642 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1643 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1644 is anything other than /usr, -L is still shown).
1645
1646 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1647
1648 14. Perl does not support \N without a following name in a [] class; PCRE now
1649 also gives an error.
1650
1651 15. If a forward reference was repeated with an upper limit of around 2000,
1652 it caused the error "internal error: overran compiling workspace". The
1653 maximum number of forward references (including repeats) was limited by the
1654 internal workspace, and dependent on the LINK_SIZE. The code has been
1655 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1656 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1657 of around 200,000 forward references. While doing this, I also speeded up
1658 the filling in of repeated forward references.
1659
1660 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1661 incorrectly expecting the subject to contain another "a" after the start.
1662
1663 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1664 in the match, the SKIP should be ignored. This was not happening; instead
1665 the SKIP was being treated as NOMATCH. For patterns such as
1666 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1667 tested.
1668
1669 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1670 now much more compatible with Perl, in particular in cases where the result
1671 is a non-match for a non-anchored pattern. For example, if
1672 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1673 "m", where previously it did not return a name. A side effect of this
1674 change is that for partial matches, the last encountered mark name is
1675 returned, as for non matches. A number of tests that were previously not
1676 Perl-compatible have been moved into the Perl-compatible test files. The
1677 refactoring has had the pleasing side effect of removing one argument from
1678 the match() function, thus reducing its stack requirements.
1679
1680 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1681 subsequent uses of /S (without +) incorrectly behaved like /S+.
1682
1683 21. Retrieve executable code size support for the JIT compiler and fixing
1684 some warnings.
1685
1686 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1687 not work when the shorter character appeared right at the end of the
1688 subject string.
1689
1690 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1691 systems.
1692
1693 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1694 output it when the /M option is used in pcretest.
1695
1696 25. The CheckMan script was not being included in the distribution. Also, added
1697 an explicit "perl" to run Perl scripts from the PrepareRelease script
1698 because this is reportedly needed in Windows.
1699
1700 26. If study data was being save in a file and studying had not found a set of
1701 "starts with" bytes for the pattern, the data written to the file (though
1702 never used) was taken from uninitialized memory and so caused valgrind to
1703 complain.
1704
1705 27. Updated RunTest.bat as provided by Sheri Pierce.
1706
1707 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1708
1709 29. Computation of memory usage for the table of capturing group names was
1710 giving an unnecessarily large value.
1711
1712
1713 Version 8.20 21-Oct-2011
1714 ------------------------
1715
1716 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1717 a POSIX class. After further experiments with Perl, which convinced me that
1718 Perl has bugs and confusions, a closing square bracket is no longer allowed
1719 in a POSIX name. This bug also affected patterns with classes that started
1720 with full stops.
1721
1722 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1723 captured substring, but while checking the failing first alternative,
1724 substring 1 is temporarily captured. If the output vector supplied to
1725 pcre_exec() was not big enough for this capture, the yield of the function
1726 was still zero ("insufficient space for captured substrings"). This cannot
1727 be totally fixed without adding another stack variable, which seems a lot
1728 of expense for a edge case. However, I have improved the situation in cases
1729 such as /(a)(b)x|abc/ matched against "abc", where the return code
1730 indicates that fewer than the maximum number of slots in the ovector have
1731 been set.
1732
1733 3. Related to (2) above: when there are more back references in a pattern than
1734 slots in the output vector, pcre_exec() uses temporary memory during
1735 matching, and copies in the captures as far as possible afterwards. It was
1736 using the entire output vector, but this conflicts with the specification
1737 that only 2/3 is used for passing back captured substrings. Now it uses
1738 only the first 2/3, for compatibility. This is, of course, another edge
1739 case.
1740
1741 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1742 main code base, and can be used by building with --enable-jit. When this is
1743 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1744 runtime --no-jit option is given.
1745
1746 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1747 ovector, the return from the function was zero, implying that there were
1748 other matches that did not fit. The correct "exactly full" value is now
1749 returned.
1750
1751 6. If a subpattern that was called recursively or as a subroutine contained
1752 (*PRUNE) or any other control that caused it to give a non-standard return,
1753 invalid errors such as "Error -26 (nested recursion at the same subject
1754 position)" or even infinite loops could occur.
1755
1756 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1757 computing the minimum length on reaching *ACCEPT, and so ended up with the
1758 wrong value of 1 rather than 0. Further investigation indicates that
1759 computing a minimum subject length in the presence of *ACCEPT is difficult
1760 (think back references, subroutine calls), and so I have changed the code
1761 so that no minimum is registered for a pattern that contains *ACCEPT.
1762
1763 8. If (*THEN) was present in the first (true) branch of a conditional group,
1764 it was not handled as intended. [But see 16 below.]
1765
1766 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1767 Sheri Pierce.
1768
1769 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1770 the first byte in a match must be "a".
1771
1772 11. Change 17 for 8.13 increased the recursion depth for patterns like
1773 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1774 pattern contains any instances of (*THEN). If it does not, the old
1775 optimizations are restored. It would be nice to do this on a per-group
1776 basis, but at the moment that is not feasible.
1777
1778 12. In some environments, the output of pcretest -C is CRLF terminated. This
1779 broke RunTest's code that checks for the link size. A single white space
1780 character after the value is now allowed for.
1781
1782 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1783 For "fr", it uses the Windows-specific input and output files.
1784
1785 14. If (*THEN) appeared in a group that was called recursively or as a
1786 subroutine, it did not work as intended. [But see next item.]
1787
1788 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1789 pattern fragments (but not containing any | characters). If A and B are
1790 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1791 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1792 D. In other words, Perl considers parentheses that do not contain any |
1793 characters to be part of a surrounding alternative, whereas PCRE was
1794 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1795 differently. PCRE now behaves in the same way as Perl, except in the case
1796 of subroutine/recursion calls such as (?1) which have in any case always
1797 been different (but PCRE had them first :-).
1798
1799 16. Related to 15 above: Perl does not treat the | in a conditional group as
1800 creating alternatives. Such a group is treated in the same way as an
1801 ordinary group without any | characters when processing (*THEN). PCRE has
1802 been changed to match Perl's behaviour.
1803
1804 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1805 RunGrepTest script failed.
1806
1807 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1808 inevitable for groups that contain captures, but it can lead to a lot of
1809 stack use in large patterns. The old behaviour has been restored for atomic
1810 groups that do not contain any capturing parentheses.
1811
1812 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1813 suppress the check for a minimum subject length at run time. (If it was
1814 given to pcre_exec() or pcre_dfa_exec() it did work.)
1815
1816 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1817 fail to work when decoding hex characters in data strings in EBCDIC
1818 environments.
1819
1820 21. It appears that in at least one Mac OS environment, the isxdigit() function
1821 is implemented as a macro that evaluates to its argument more than once,
1822 contravening the C 90 Standard (I haven't checked a later standard). There
1823 was an instance in pcretest which caused it to go wrong when processing
1824 \x{...} escapes in subject strings. The has been rewritten to avoid using
1825 things like p++ in the argument of isxdigit().
1826
1827
1828 Version 8.13 16-Aug-2011
1829 ------------------------
1830
1831 1. The Unicode data tables have been updated to Unicode 6.0.0.
1832
1833 2. Two minor typos in pcre_internal.h have been fixed.
1834
1835 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1836 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1837 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1838
1839 4. There were a number of related bugs in the code for matching backrefences
1840 caselessly in UTF-8 mode when codes for the characters concerned were
1841 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1842 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1843 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1844 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1845 2-byte code at the end of the subject (it thought there wasn't enough data
1846 left).
1847
1848 5. Comprehensive information about what went wrong is now returned by
1849 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1850 as the output vector has at least 2 elements. The offset of the start of
1851 the failing character and a reason code are placed in the vector.
1852
1853 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1854 now returned is for the first byte of the failing character, instead of the
1855 last byte inspected. This is an incompatible change, but I hope it is small
1856 enough not to be a problem. It makes the returned offset consistent with
1857 pcre_exec() and pcre_dfa_exec().
1858
1859 7. pcretest now gives a text phrase as well as the error number when
1860 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1861 failure, the offset and reason code are output.
1862
1863 8. When \R was used with a maximizing quantifier it failed to skip backwards
1864 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1865 back over a single character (\n). This seems wrong (because it treated the
1866 two characters as a single entity when going forwards), conflicts with the
1867 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1868 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1869 has been changed.
1870
1871 9. Some internal refactoring has changed the processing so that the handling
1872 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1873 time (the PCRE_DOTALL option was changed this way some time ago: version
1874 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1875 which was always a bit of a fudge. It also means that there is one less
1876 argument for the match() function, which reduces its stack requirements
1877 slightly. This change also fixes an incompatibility with Perl: the pattern
1878 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1879
1880 10. More internal refactoring has drastically reduced the number of recursive
1881 calls to match() for possessively repeated groups such as (abc)++ when
1882 using pcre_exec().
1883
1884 11. While implementing 10, a number of bugs in the handling of groups were
1885 discovered and fixed:
1886
1887 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1888 (a|)*(?1) gave a compile-time internal error.
1889 ((a|)+)+ did not notice that the outer group could match an empty string.
1890 (^a|^)+ was not marked as anchored.
1891 (.*a|.*)+ was not marked as matching at start or after a newline.
1892
1893 12. Yet more internal refactoring has removed another argument from the match()
1894 function. Special calls to this function are now indicated by setting a
1895 value in a variable in the "match data" data block.
1896
1897 13. Be more explicit in pcre_study() instead of relying on "default" for
1898 opcodes that mean there is no starting character; this means that when new
1899 ones are added and accidentally left out of pcre_study(), testing should
1900 pick them up.
1901
1902 14. The -s option of pcretest has been documented for ages as being an old
1903 synonym of -m (show memory usage). I have changed it to mean "force study
1904 for every regex", that is, assume /S for every regex. This is similar to -i
1905 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1906 using it. It makes it easier to run collections of tests with and without
1907 study enabled, and thereby test pcre_study() more easily. All the standard
1908 tests are now run with and without -s (but some patterns can be marked as
1909 "never study" - see 20 below).
1910
1911 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1912 restoration of the capturing data to the outer values was not happening
1913 correctly.
1914
1915 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1916 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1917 pattern had matched an empty string, and so incorrectly returned a no
1918 match.
1919
1920 17. There was optimizing code for the last branch of non-capturing parentheses,
1921 and also for the obeyed branch of a conditional subexpression, which used
1922 tail recursion to cut down on stack usage. Unfortunately, now that there is
1923 the possibility of (*THEN) occurring in these branches, tail recursion is
1924 no longer possible because the return has to be checked for (*THEN). These
1925 two optimizations have therefore been removed. [But see 8.20/11 above.]
1926
1927 18. If a pattern containing \R was studied, it was assumed that \R always
1928 matched two bytes, thus causing the minimum subject length to be
1929 incorrectly computed because \R can also match just one byte.
1930
1931 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1932 was incorrectly computed.
1933
1934 20. If /S is present twice on a test pattern in pcretest input, it now
1935 *disables* studying, thereby overriding the use of -s on the command line
1936 (see 14 above). This is necessary for one or two tests to keep the output
1937 identical in both cases.
1938
1939 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1940 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1941
1942 22. When an atomic group that contained a capturing parenthesis was
1943 successfully matched, but the branch in which it appeared failed, the
1944 capturing was not being forgotten if a higher numbered group was later
1945 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1946 group 1 to "a", when in fact it should be unset. This applied to multi-
1947 branched capturing and non-capturing groups, repeated or not, and also to
1948 positive assertions (capturing in negative assertions does not happen
1949 in PCRE) and also to nested atomic groups.
1950
1951 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1952 subject after a captured substring, to make it easier to tell which of a
1953 number of identical substrings has been captured.
1954
1955 24. The way atomic groups are processed by pcre_exec() has been changed so that
1956 if they are repeated, backtracking one repetition now resets captured
1957 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1958 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1959 "aaa". Previously, it would have been "a". As part of this code
1960 refactoring, the way recursive calls are handled has also been changed.
1961
1962 25. If an assertion condition captured any substrings, they were not passed
1963 back unless some other capturing happened later. For example, if
1964 (?(?=(a))a) was matched against "a", no capturing was returned.
1965
1966 26. When studying a pattern that contained subroutine calls or assertions,
1967 the code for finding the minimum length of a possible match was handling
1968 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1969 group 1 called group 2 while simultaneously a separate group 2 called group
1970 1). A stack overflow occurred in this case. I have fixed this by limiting
1971 the recursion depth to 10.
1972
1973 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1974 Fortmann. This supports explicit test numbers on the command line, and has
1975 argument validation and error reporting.
1976
1977 28. An instance of \X with an unlimited repeat could fail if at any point the
1978 first character it looked at was a mark character.
1979
1980 29. Some minor code refactoring concerning Unicode properties and scripts
1981 should reduce the stack requirement of match() slightly.
1982
1983 30. Added the '=' option to pcretest to check the setting of unused capturing
1984 slots at the end of the pattern, which are documented as being -1, but are
1985 not included in the return count.
1986
1987 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1988 compiled something random. Now it gives a compile-time error (as does
1989 Perl).
1990
1991 32. A *MARK encountered during the processing of a positive assertion is now
1992 recorded and passed back (compatible with Perl).
1993
1994 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1995 had alternative anchored branches, the search for a second match in a line
1996 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1997 matched the line "0102" twice. The same bug affected patterns that started
1998 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1999 twice.
2000
2001 34. Previously, PCRE did not allow quantification of assertions. However, Perl
2002 does, and because of capturing effects, quantifying parenthesized
2003 assertions may at times be useful. Quantifiers are now allowed for
2004 parenthesized assertions.
2005
2006 35. A minor code tidy in pcre_compile() when checking options for \R usage.
2007
2008 36. \g was being checked for fancy things in a character class, when it should
2009 just be a literal "g".
2010
2011 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
2012 appearance of a nested POSIX class supersedes an apparent external class.
2013 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
2014 unescaped square brackets may also appear as part of class names. For
2015 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
2016 more like Perl. (But see 8.20/1 above.)
2017
2018 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
2019 was because it thought it was \N{name}, which is not supported).
2020
2021 39. Add minix to OS list not supporting the -S option in pcretest.
2022
2023 40. PCRE tries to detect cases of infinite recursion at compile time, but it
2024 cannot analyze patterns in sufficient detail to catch mutual recursions
2025 such as ((?1))((?2)). There is now a runtime test that gives an error if a
2026 subgroup is called recursively as a subpattern for a second time at the
2027 same position in the subject string. In previous releases this might have
2028 been caught by the recursion limit, or it might have run out of stack.
2029
2030 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
2031 happen only once. PCRE was, however incorrectly giving a compile time error
2032 "recursive call could loop indefinitely" because it cannot analyze the
2033 pattern in sufficient detail. The compile time test no longer happens when
2034 PCRE is compiling a conditional subpattern, but actual runaway loops are
2035 now caught at runtime (see 40 above).
2036
2037 42. It seems that Perl allows any characters other than a closing parenthesis
2038 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
2039 has been changed to be the same.
2040
2041 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
2042 as not to get warnings when autogen.sh is called. Also changed
2043 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
2044
2045 44. To help people who use pcregrep to scan files containing exceedingly long
2046 lines, the following changes have been made:
2047
2048 (a) The default value of the buffer size parameter has been increased from
2049 8K to 20K. (The actual buffer used is three times this size.)
2050
2051 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
2052 PCRE is built.
2053
2054 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
2055 to be set at run time.
2056
2057 (d) Numerical values in pcregrep options can be followed by K or M, for
2058 example --buffer-size=50K.
2059
2060 (e) If a line being scanned overflows pcregrep's buffer, an error is now
2061 given and the return code is set to 2.
2062
2063 45. Add a pointer to the latest mark to the callout data block.
2064
2065 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
2066 partial match of an empty string instead of no match. This was specific to
2067 the use of ".".
2068
2069 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
2070 complete match instead of a partial match. This bug was dependent on both
2071 the PCRE_UTF8 and PCRE_DOTALL options being set.
2072
2073 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
2074 starting byte set, because \b was not being ignored.
2075
2076
2077 Version 8.12 15-Jan-2011
2078 ------------------------
2079
2080 1. Fixed some typos in the markup of the man pages, and wrote a script that
2081 checks for such things as part of the documentation building process.
2082
2083 2. On a big-endian 64-bit system, pcregrep did not correctly process the
2084 --match-limit and --recursion-limit options (added for 8.11). In
2085 particular, this made one of the standard tests fail. (The integer value
2086 went into the wrong half of a long int.)
2087
2088 3. If the --colour option was given to pcregrep with -v (invert match), it
2089 did strange things, either producing crazy output, or crashing. It should,
2090 of course, ignore a request for colour when reporting lines that do not
2091 match.
2092
2093 4. Another pcregrep bug caused similar problems if --colour was specified with
2094 -M (multiline) and the pattern match finished with a line ending.
2095
2096 5. In pcregrep, when a pattern that ended with a literal newline sequence was
2097 matched in multiline mode, the following line was shown as part of the
2098 match. This seems wrong, so I have changed it.
2099
2100 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
2101 the check for further matches in the same line (so they could be coloured)
2102 to overrun the end of the current line. If another match was found, it was
2103 incorrectly shown (and then shown again when found in the next line).
2104
2105 7. If pcregrep was compiled under Windows, there was a reference to the
2106 function pcregrep_exit() before it was defined. I am assuming this was
2107 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
2108 reported by a user. I've moved the definition above the reference.
2109
2110
2111 Version 8.11 10-Dec-2010
2112 ------------------------
2113
2114 1. (*THEN) was not working properly if there were untried alternatives prior
2115 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
2116 backtracked to try for "b" instead of moving to the next alternative branch
2117 at the same level (in this case, to look for "c"). The Perl documentation
2118 is clear that when (*THEN) is backtracked onto, it goes to the "next
2119 alternative in the innermost enclosing group".
2120
2121 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
2122 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
2123 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
2124 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
2125 (*THEN).
2126
2127 3. If \s appeared in a character class, it removed the VT character from
2128 the class, even if it had been included by some previous item, for example
2129 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
2130 of \s, but is part of the POSIX "space" class.)
2131
2132 4. A partial match never returns an empty string (because you can always
2133 match an empty string at the end of the subject); however the checking for
2134 an empty string was starting at the "start of match" point. This has been
2135 changed to the "earliest inspected character" point, because the returned
2136 data for a partial match starts at this character. This means that, for
2137 example, /(?<=abc)def/ gives a partial match for the subject "abc"
2138 (previously it gave "no match").
2139
2140 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2141 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2142 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2143 has an implication that the given string is incomplete (because a partial
2144 match is preferred over a full match). For this reason, these items now
2145 give a partial match in this situation. [Aside: previously, the one case
2146 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2147 match rather than a full match, which was wrong by the old rules, but is
2148 now correct.]
2149
2150 6. There was a bug in the handling of #-introduced comments, recognized when
2151 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2152 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2153 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2154 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2155 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2156 places in pcre_compile().
2157
2158 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2159 comments when looking ahead for named forward references to subpatterns,
2160 the only newline sequence it recognized was NL. It now handles newlines
2161 according to the set newline convention.
2162
2163 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2164 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2165 cater for a lack of strerror(). These oversights have been fixed.
2166
2167 9. Added --match-limit and --recursion-limit to pcregrep.
2168
2169 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2170
2171 11. When the -o option was used, pcregrep was setting a return code of 1, even
2172 when matches were found, and --line-buffered was not being honoured.
2173
2174 12. Added an optional parentheses number to the -o and --only-matching options
2175 of pcregrep.
2176
2177 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2178 can match an empty string. The code to do it in pcretest and pcredemo
2179 needed fixing:
2180
2181 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2182 only one byte after an empty string match just before CRLF (this case
2183 just got forgotten; "any" and "anycrlf" were OK).
2184
2185 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2186 mode when an empty string match preceded an ASCII character followed by
2187 a non-ASCII character. (The code for advancing by one character rather
2188 than one byte was nonsense.)
2189
2190 (c) The pcredemo.c sample program did not have any code at all to handle
2191 the cases when CRLF is a valid newline sequence.
2192
2193 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2194 as a starting offset was within the subject string. There is now a new
2195 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2196 negative or greater than the length of the string. In order to test this,
2197 pcretest is extended to allow the setting of negative starting offsets.
2198
2199 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2200 starting offset points to the beginning of a UTF-8 character was
2201 unnecessarily clumsy. I tidied it up.
2202
2203 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2204 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2205
2206 17. Nobody had reported that the --include_dir option, which was added in
2207 release 7.7 should have been called --include-dir (hyphen, not underscore)
2208 for compatibility with GNU grep. I have changed it to --include-dir, but
2209 left --include_dir as an undocumented synonym, and the same for
2210 --exclude-dir, though that is not available in GNU grep, at least as of
2211 release 2.5.4.
2212
2213 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2214 characters from a string of bytes) have been redefined so as not to use
2215 loops, in order to improve performance in some environments. At the same
2216 time, I abstracted some of the common code into auxiliary macros to save
2217 repetition (this should not affect the compiled code).
2218
2219 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2220 compile-time error is now given if \c is not followed by an ASCII
2221 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2222 different, and any byte value is allowed.)
2223
2224 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2225 START_OPTIMIZE option, which is now allowed at compile time - but just
2226 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2227 to pcregrep and other applications that have no direct access to PCRE
2228 options. The new /Y option in pcretest sets this option when calling
2229 pcre_compile().
2230
2231 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2232 back references. Groups containing recursive back references were forced to
2233 be atomic by that change, but in the case of named groups, the amount of
2234 memory required was incorrectly computed, leading to "Failed: internal
2235 error: code overflow". This has been fixed.
2236
2237 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2238 pcretest.c, to avoid build problems in some Borland environments.
2239
2240
2241 Version 8.10 25-Jun-2010
2242 ------------------------
2243
2244 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2245 THEN.
2246
2247 2. (*ACCEPT) was not working when inside an atomic group.
2248
2249 3. Inside a character class, \B is treated as a literal by default, but
2250 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2251 causes the error). The code is unchanged, but I tidied the documentation.
2252
2253 4. Inside a character class, PCRE always treated \R and \X as literals,
2254 whereas Perl faults them if its -w option is set. I have changed PCRE so
2255 that it faults them when PCRE_EXTRA is set.
2256
2257 5. Added support for \N, which always matches any character other than
2258 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2259
2260 6. When compiling pcregrep with newer versions of gcc which may have
2261 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2262 declared with attribute warn_unused_result" were given. Just casting the
2263 result to (void) does not stop the warnings; a more elaborate fudge is
2264 needed. I've used a macro to implement this.
2265
2266 7. Minor change to pcretest.c to avoid a compiler warning.
2267
2268 8. Added four artifical Unicode properties to help with an option to make
2269 \s etc use properties (see next item). The new properties are: Xan
2270 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2271
2272 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2273 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2274 this option. Modified pcretest to add /W to test this facility. Added
2275 REG_UCP to make it available via the POSIX interface.
2276
2277 10. Added --line-buffered to pcregrep.
2278
2279 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2280 studied, and the match started with a letter with a code point greater than
2281 127 whose first byte was different to the first byte of the other case of
2282 the letter, the other case of this starting letter was not recognized
2283 (#976).
2284
2285 12. If a pattern that was studied started with a repeated Unicode property
2286 test, for example, \p{Nd}+, there was the theoretical possibility of
2287 setting up an incorrect bitmap of starting bytes, but fortunately it could
2288 not have actually happened in practice until change 8 above was made (it
2289 added property types that matched character-matching opcodes).
2290
2291 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2292 possible starting bytes for non-anchored patterns.
2293
2294 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2295 \R, and also a number of cases that involve Unicode properties, both
2296 explicit and implicit when PCRE_UCP is set.
2297
2298 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2299 input, it could crash or give wrong results if characters with values
2300 greater than 0xc0 were present in the subject string. (Detail: it assumed
2301 UTF-8 input when processing these items.)
2302
2303 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2304 size_t is 64-bit (#991).
2305
2306 17. Added a check for running out of memory when PCRE is compiled with
2307 --disable-stack-for-recursion (#990).
2308
2309 18. If the last data line in a file for pcretest does not have a newline on
2310 the end, a newline was missing in the output.
2311
2312 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2313 less than 128) in its various bitmaps. However, there is a facility for
2314 generating tables according to the current locale when PCRE is compiled. It
2315 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2316 characters, are recognized by isspace() and therefore were getting set in
2317 these tables, and indeed these tables seem to approximate to ISO 8859. This
2318 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2319 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2320 which of course cannot start UTF-8 characters. I have changed the code so
2321 that only real ASCII characters (less than 128) and the correct starting
2322 bytes for UTF-8 encodings are set for characters greater than 127 when in
2323 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2324 altogether.)
2325
2326 20. Added the /T option to pcretest so as to be able to run tests with non-
2327 standard character tables, thus making it possible to include the tests
2328 used for 19 above in the standard set of tests.
2329
2330 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2331 reference to a subpattern the other side of a comment that contains an
2332 opening parenthesis caused either an internal compiling error, or a
2333 reference to the wrong subpattern.
2334
2335
2336 Version 8.02 19-Mar-2010
2337 ------------------------
2338
2339 1. The Unicode data tables have been updated to Unicode 5.2.0.
2340
2341 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2342 configured.
2343
2344 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2345 original author of that file, following a query about its status.
2346
2347 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2348 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2349
2350 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2351 quantifier applied to a forward-referencing subroutine call, could compile
2352 incorrect code or give the error "internal error: previously-checked
2353 referenced subpattern not found".
2354
2355 6. Both MS Visual Studio and Symbian OS have problems with initializing
2356 variables to point to external functions. For these systems, therefore,
2357 pcre_malloc etc. are now initialized to local functions that call the
2358 relevant global functions.
2359
2360 7. There were two entries missing in the vectors called coptable and poptable
2361 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2362 I've fixed the data, and added a kludgy way of testing at compile time that
2363 the lengths are correct (equal to the number of opcodes).
2364
2365 8. Following on from 7, I added a similar kludge to check the length of the
2366 eint vector in pcreposix.c.
2367
2368 9. Error texts for pcre_compile() are held as one long string to avoid too
2369 much relocation at load time. To find a text, the string is searched,
2370 counting zeros. There was no check for running off the end of the string,
2371 which could happen if a new error number was added without updating the
2372 string.
2373
2374 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2375
2376 11. \K was not working if it appeared in an atomic group or in a group that
2377 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2378 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2379 the assertion is positive, but not if it is negative.
2380
2381 12. Change 11 fortuitously reduced the size of the stack frame used in the
2382 "match()" function of pcre_exec.c by one pointer. Forthcoming
2383 implementation of support for (*MARK) will need an extra pointer on the
2384 stack; I have reserved it now, so that the stack frame size does not
2385 decrease.
2386
2387 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2388 item in branch that calls a recursion is a subroutine call - as in the
2389 second branch in the above example - was incorrectly given the compile-
2390 time error "recursive call could loop indefinitely" because pcre_compile()
2391 was not correctly checking the subroutine for matching a non-empty string.
2392
2393 14. The checks for overrunning compiling workspace could trigger after an
2394 overrun had occurred. This is a "should never occur" error, but it can be
2395 triggered by pathological patterns such as hundreds of nested parentheses.
2396 The checks now trigger 100 bytes before the end of the workspace.
2397
2398 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2399
2400
2401 Version 8.01 19-Jan-2010
2402 ------------------------
2403
2404 1. If a pattern contained a conditional subpattern with only one branch (in
2405 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2406 computed the wrong minimum data length (which is of course zero for such
2407 subpatterns). This could cause incorrect "no match" results.
2408
2409 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2410 the pattern is reset in the first branch, pcre_compile() failed with
2411 "internal error: code overflow at offset...". This happened only when
2412 the reset was to the original external option setting. (An optimization
2413 abstracts leading options settings into an external setting, which was the
2414 cause of this.)
2415
2416 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2417 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2418 assertion pattern did not match (meaning that the assertion was true), it
2419 was incorrectly treated as false if the SKIP had been reached during the
2420 matching. This also applied to assertions used as conditions.
2421
2422 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2423 assertion subpattern, including such a pattern used as a condition,
2424 unpredictable results occurred, instead of the error return
2425 PCRE_ERROR_DFA_UITEM.
2426
2427 5. The C++ GlobalReplace function was not working like Perl for the special
2428 situation when an empty string is matched. It now does the fancy magic
2429 stuff that is necessary.
2430
2431 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2432 removed. (These were left over from very, very early versions of PCRE.)
2433
2434 7. Some cosmetic changes to the code to make life easier when compiling it
2435 as part of something else:
2436
2437 (a) Change DEBUG to PCRE_DEBUG.
2438
2439 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2440 called "current" as "current_branch", to prevent a collision with the
2441 Linux macro when compiled as a kernel module.
2442
2443 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2444 prevent a collision with the Linux macro when compiled as a kernel
2445 module.
2446
2447 8. In pcre_compile() there are some checks for integer overflows that used to
2448 cast potentially large values to (double). This has been changed to that
2449 when building, a check for int64_t is made, and if it is found, it is used
2450 instead, thus avoiding the use of floating point arithmetic. (There is no
2451 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2452 double.
2453
2454 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2455 2005 (difference between two addresses compared to an unsigned value).
2456
2457 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2458 custom one, because of the following reported problem in Windows:
2459
2460 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2461 under Win32.
2462 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2463 therefore missing the function definition.
2464 - The compiler thus generates a "C" signature for the test function.
2465 - The linker fails to find the "C" function.
2466 - PCRE fails to configure if asked to do so against libbz2.
2467
2468 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2469 messages were output:
2470
2471 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2472 rerunning libtoolize, to keep the correct libtool macros in-tree.
2473 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2474
2475 I have done both of these things.
2476
2477 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2478 most of the time, it *can* run out if it is given a pattern that contains a
2479 runaway infinite recursion. I updated the discussion in the pcrestack man
2480 page.
2481
2482 13. Now that we have gone to the x.xx style of version numbers, the minor
2483 version may start with zero. Using 08 or 09 is a bad idea because users
2484 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2485 interpreted as invalid octal numbers. I've updated the previous comment in
2486 configure.ac, and also added a check that gives an error if 08 or 09 are
2487 used.
2488
2489 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2490 causing partial matching to fail when the end of the subject matched \W
2491 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2492
2493 15. There were some discrepancies between the declarations in pcre_internal.h
2494 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2495 their definitions. The declarations used "const uschar *" and the
2496 definitions used USPTR. Even though USPTR is normally defined as "const
2497 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2498 reported that: "This difference in casting confuses some C++ compilers, for
2499 example, SunCC recognizes above declarations as different functions and
2500 generates broken code for hbpcre." I have changed the declarations to use
2501 USPTR.
2502
2503 16. GNU libtool is named differently on some systems. The autogen.sh script now
2504 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2505 (FreeBSD).
2506
2507 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2508 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2509 comment: "Figure out how to create a longlong from a string: strtoll and
2510 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2511 instance, but it only takes 2 args instead of 3!"
2512
2513 18. A subtle bug concerned with back references has been fixed by a change of
2514 specification, with a corresponding code fix. A pattern such as
2515 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2516 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2517 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2518 same bug. Such groups have to be quantified to be useful, or contained
2519 inside another quantified group. (If there's no repetition, the reference
2520 can never match.) The problem arises because, having left the group and
2521 moved on to the rest of the pattern, a later failure that backtracks into
2522 the group uses the captured value from the final iteration of the group
2523 rather than the correct earlier one. I have fixed this in PCRE by forcing
2524 any group that contains a reference to itself to be an atomic group; that
2525 is, there cannot be any backtracking into it once it has completed. This is
2526 similar to recursive and subroutine calls.
2527
2528
2529 Version 8.00 19-Oct-09
2530 ----------------------
2531
2532 1. The table for translating pcre_compile() error codes into POSIX error codes
2533 was out-of-date, and there was no check on the pcre_compile() error code
2534 being within the table. This could lead to an OK return being given in
2535 error.
2536
2537 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2538 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2539 in a Windows environment.
2540
2541 3. The pcregrep --count option prints the count for each file even when it is
2542 zero, as does GNU grep. However, pcregrep was also printing all files when
2543 --files-with-matches was added. Now, when both options are given, it prints
2544 counts only for those files that have at least one match. (GNU grep just
2545 prints the file name in this circumstance, but including the count seems
2546 more useful - otherwise, why use --count?) Also ensured that the
2547 combination -clh just lists non-zero counts, with no names.
2548
2549 4. The long form of the pcregrep -F option was incorrectly implemented as
2550 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2551 but it seems right to fix it, and I didn't think it was worth preserving
2552 the old behaviour.
2553
2554 5. The command line items --regex=pattern and --regexp=pattern were not
2555 recognized by pcregrep, which required --regex pattern or --regexp pattern
2556 (with a space rather than an '='). The man page documented the '=' forms,
2557 which are compatible with GNU grep; these now work.
2558
2559 6. No libpcreposix.pc file was created for pkg-config; there was just
2560 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2561
2562 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2563 when UCP support is not needed, by modifying the Python script that
2564 generates it from Unicode data files. This should not matter if the module
2565 is correctly used as a library, but I received one complaint about 50K of
2566 unwanted data. My guess is that the person linked everything into his
2567 program rather than using a library. Anyway, it does no harm.
2568
2569 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2570 was a minimum greater than 1 for a wide character in a possessive
2571 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2572 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2573 character. Chaos in the form of incorrect output or a compiling loop could
2574 result.
2575
2576 9. The restrictions on what a pattern can contain when partial matching is
2577 requested for pcre_exec() have been removed. All patterns can now be
2578 partially matched by this function. In addition, if there are at least two
2579 slots in the offset vector, the offset of the earliest inspected character
2580 for the match and the offset of the end of the subject are set in them when
2581 PCRE_ERROR_PARTIAL is returned.
2582
2583 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2584 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2585 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2586 and may be more useful for multi-segment matching.
2587
2588 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2589 used to be given if ever the end of the subject was reached; now it is
2590 given only if matching could not proceed because another character was
2591 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2592 string "Z", which now yields "no match" instead of "partial match". In the
2593 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2594 final character ended with (*FAIL).
2595
2596 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2597 if the pattern had a "must contain" character that was already found in the
2598 earlier partial match, unless partial matching was again requested. For
2599 example, with the pattern /dog.(body)?/, the "must contain" character is
2600 "g". If the first part-match was for the string "dog", restarting with
2601 "sbody" failed. This bug has been fixed.
2602
2603 13. The string returned by pcre_dfa_exec() after a partial match has been
2604 changed so that it starts at the first inspected character rather than the
2605 first character of the match. This makes a difference only if the pattern
2606 starts with a lookbehind assertion or \b or \B (\K is not supported by
2607 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2608 matching functions compatible, and I think it's the right thing to do.
2609
2610 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2611 so that the demonstration program is easily available in environments where
2612 PCRE has not been installed from source.
2613
2614 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2615 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2616 library.
2617
2618 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2619 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2620 is not the first non-POSIX option to be added. Clearly some people find
2621 these options useful.
2622
2623 17. If a caller to the POSIX matching function regexec() passes a non-zero
2624 value for nmatch with a NULL value for pmatch, the value of
2625 nmatch is forced to zero.
2626
2627 18. RunGrepTest did not have a test for the availability of the -u option of
2628 the diff command, as RunTest does. It now checks in the same way as
2629 RunTest, and also checks for the -b option.
2630
2631 19. If an odd number of negated classes containing just a single character
2632 interposed, within parentheses, between a forward reference to a named
2633 subpattern and the definition of the subpattern, compilation crashed with
2634 an internal error, complaining that it could not find the referenced
2635 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2636 [The bug was that it was starting one character too far in when skipping
2637 over the character class, thus treating the ] as data rather than
2638 terminating the class. This meant it could skip too much.]
2639
2640 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2641 /g option in pcretest when the pattern contains \K, which makes it possible
2642 to have an empty string match not at the start, even when the pattern is
2643 anchored. Updated pcretest and pcredemo to use this option.
2644
2645 21. If the maximum number of capturing subpatterns in a recursion was greater
2646 than the maximum at the outer level, the higher number was returned, but
2647 with unset values at the outer level. The correct (outer level) value is
2648 now given.
2649
2650 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2651 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2652 make it do so. The string so far is captured, making this feature
2653 compatible with Perl.
2654
2655 23. The tests have been re-organized, adding tests 11 and 12, to make it
2656 possible to check the Perl 5.10 features against Perl 5.10.
2657
2658 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2659 pattern matches a fixed length string. PCRE did not allow this; now it
2660 does. Neither allows recursion.
2661
2662 25. I finally figured out how to implement a request to provide the minimum
2663 length of subject string that was needed in order to match a given pattern.
2664 (It was back references and recursion that I had previously got hung up
2665 on.) This code has now been added to pcre_study(); it finds a lower bound
2666 to the length of subject needed. It is not necessarily the greatest lower
2667 bound, but using it to avoid searching strings that are too short does give
2668 some useful speed-ups. The value is available to calling programs via
2669 pcre_fullinfo().
2670
2671 26. While implementing 25, I discovered to my embarrassment that pcretest had
2672 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2673 study optimizations had never been tested with that matching function.
2674 Oops. What is worse, even when it was passed study data, there was a bug in
2675 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2676 were also very few tests of studied patterns with pcre_dfa_exec().
2677
2678 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2679 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2680 on the other side of the coin, they are no longer allowed to have different
2681 names, because these cannot be distinguished in PCRE, and this has caused
2682 confusion. (This is a difference from Perl.)
2683
2684 28. When duplicate subpattern names are present (necessarily with different
2685 numbers, as required by 27 above), and a test is made by name in a
2686 conditional pattern, either for a subpattern having been matched, or for
2687 recursion in such a pattern, all the associated numbered subpatterns are
2688 tested, and the overall condition is true if the condition is true for any
2689 one of them. This is the way Perl works, and is also more like the way
2690 testing by number works.
2691
2692
2693 Version 7.9 11-Apr-09
2694 ---------------------
2695
2696 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2697 (pcretest), all targets were linked against these libraries. This included
2698 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2699 libraries. This caused unwanted dependencies to be created. This problem
2700 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2701 pcretest is linked with readline.
2702
2703 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2704 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2705 moved outside it again, because FALSE and TRUE are already defined in AIX,
2706 but BOOL is not.
2707
2708 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2709 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2710
2711 4. The pcregrep documentation said spaces were inserted as well as colons (or
2712 hyphens) following file names and line numbers when outputting matching
2713 lines. This is not true; no spaces are inserted. I have also clarified the
2714 wording for the --colour (or --color) option.
2715
2716 5. In pcregrep, when --colour was used with -o, the list of matching strings
2717 was not coloured; this is different to GNU grep, so I have changed it to be
2718 the same.
2719
2720 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2721 each matching line was coloured. Now it goes on to look for further matches
2722 of any of the test patterns, which is the same behaviour as GNU grep.
2723
2724 7. A pattern that could match an empty string could cause pcregrep to loop; it
2725 doesn't make sense to accept an empty string match in pcregrep, so I have
2726 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2727 seems to be how GNU grep behaves. [But see later change 40 for release
2728 8.33.]
2729
2730 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2731 start or after a newline", because the conditional assertion was not being
2732 correctly handled. The rule now is that both the assertion and what follows
2733 in the first alternative must satisfy the test.
2734
2735 9. If auto-callout was enabled in a pattern with a conditional group whose
2736 condition was an assertion, PCRE could crash during matching, both with
2737 pcre_exec() and pcre_dfa_exec().
2738
2739 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2740 used for matching.
2741
2742 11. Unicode property support in character classes was not working for
2743 characters (bytes) greater than 127 when not in UTF-8 mode.
2744
2745 12. Added the -M command line option to pcretest.
2746
2747 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2748
2749 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2750
2751 16. Added comments and documentation about mis-use of no_arg in the C++
2752 wrapper.
2753
2754 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2755 from Martin Jerabek that uses macro names for all relevant character and
2756 string constants.
2757
2758 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2759 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2760 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2761 these, but not everybody uses configure.
2762
2763 19. A conditional group that had only one branch was not being correctly
2764 recognized as an item that could match an empty string. This meant that an
2765 enclosing group might also not be so recognized, causing infinite looping
2766 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2767 with the subject "ab", where knowledge that the repeated group can match
2768 nothing is needed in order to break the loop.
2769
2770 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2771 exec(), but without supplying a callout function, matching went wrong.
2772
2773 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2774 leak if the size of the offset vector was greater than 30. When the vector
2775 is smaller, the saved offsets during recursion go onto a local stack
2776 vector, but for larger vectors malloc() is used. It was failing to free
2777 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2778 error, in fact).
2779
2780 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2781 heapframe that is used only when UTF-8 support is enabled. This caused no
2782 problem, but was untidy.
2783
2784 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2785 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2786 included within another project.
2787
2788 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2789 slightly modified by me:
2790
2791 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2792 not building pcregrep.
2793
2794 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2795 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2796
2797 25. Forward references, both numeric and by name, in patterns that made use of
2798 duplicate group numbers, could behave incorrectly or give incorrect errors,
2799 because when scanning forward to find the reference group, PCRE was not
2800 taking into account the duplicate group numbers. A pattern such as
2801 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2802
2803 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2804 the feature of a custom pointer more persuasive (as requested by a user).
2805
2806 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2807 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2808 pre-defined.
2809
2810 28. Added support for (*UTF8) at the start of a pattern.
2811
2812 29. Arrange for flags added by the "release type" setting in CMake to be shown
2813 in the configuration summary.
2814
2815
2816 Version 7.8 05-Sep-08
2817 ---------------------
2818
2819 1. Replaced UCP searching code with optimized version as implemented for Ad
2820 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2821 stage table and inline lookup instead of a function, giving speed ups of 2
2822 to 5 times on some simple patterns that I tested. Permission was given to
2823 distribute the MultiStage2.py script that generates the tables (it's not in
2824 the tarball, but is in the Subversion repository).
2825
2826 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2827 scripts.
2828
2829 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2830 a group with a zero qualifier. The result of the study could be incorrect,
2831 or the function might crash, depending on the pattern.
2832
2833 4. Caseless matching was not working for non-ASCII characters in back
2834 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2835 It now works when Unicode Property Support is available.
2836
2837 5. In pcretest, an escape such as \x{de} in the data was always generating
2838 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2839 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2840 truncation.
2841
2842 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2843
2844 7. Added two (int) casts to pcregrep when printing the difference of two
2845 pointers, in case they are 64-bit values.
2846
2847 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2848 test 2 if it fails.
2849
2850 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2851 and a #define of that name to empty if it is not externally set. This is to
2852 allow users of MSVC to set it if necessary.
2853
2854 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2855 the convenience functions in the pcre_get.c source file.
2856
2857 11. An option change at the start of a pattern that had top-level alternatives
2858 could cause overwriting and/or a crash. This command provoked a crash in
2859 some environments:
2860
2861 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2862
2863 This potential security problem was recorded as CVE-2008-2371.
2864
2865 12. For a pattern where the match had to start at the beginning or immediately
2866 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2867 pcre_dfa_exec() could read past the end of the passed subject if there was
2868 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2869 pcretest so that it places the subject at the end of its malloc-ed buffer.
2870
2871 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2872 exec() might read past the end of the data buffer in UTF-8 mode.
2873
2874 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2875 the data contained the byte 0x85 as part of a UTF-8 character within its
2876 first line. This applied both to normal and DFA matching.
2877
2878 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2879 /^[^d]*?$/8 failed to match "abc".
2880
2881 16. Added a missing copyright notice to pcrecpp_internal.h.
2882
2883 17. Make it more clear in the documentation that values returned from
2884 pcre_exec() in ovector are byte offsets, not character counts.
2885
2886 18. Tidied a few places to stop certain compilers from issuing warnings.
2887
2888 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2889 supplied by Stefan Weber. I made a further small update for 7.8 because
2890 there is a change of source arrangements: the pcre_searchfuncs.c module is
2891 replaced by pcre_ucd.c.
2892
2893
2894 Version 7.7 07-May-08
2895 ---------------------
2896
2897 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2898 a string to a long long, pretend we don't even have a long long." This is
2899 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2900
2901 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2902 pre-7.6 versions, which defined a global no_arg variable instead of putting
2903 it in the RE class. (See also #8 below.)
2904
2905 3. Remove a line of dead code, identified by coverity and reported by Nuno
2906 Lopes.
2907
2908 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2909
2910 (1) The include/exclude patterns were being applied to the whole pathnames
2911 of files, instead of just to the final components.
2912
2913 (2) If there was more than one level of directory, the subdirectories were
2914 skipped unless they satisfied the include/exclude conditions. This is
2915 inconsistent with GNU grep (and could even be seen as contrary to the
2916 pcregrep specification - which I improved to make it absolutely clear).
2917 The action now is always to scan all levels of directory, and just
2918 apply the include/exclude patterns to regular files.
2919
2920 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2921 --exclude_dir in the tests to avoid scanning .svn directories.
2922
2923 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2924 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2925 doesn't support NULs in patterns.
2926
2927 7. Added some missing "const"s to declarations of static tables in
2928 pcre_compile.c and pcre_dfa_exec.c.
2929
2930 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2931 caused by fix #2 above. (Subsequently also a second patch to fix the
2932 first patch. And a third patch - this was a messy problem.)
2933
2934 9. Applied Craig's patch to remove the use of push_back().
2935
2936 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2937 matching function regexec().
2938
2939 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2940 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2941 references. PCRE supports relative numbers with this syntax (I don't think
2942 Oniguruma does).
2943
2944 12. Previously, a group with a zero repeat such as (...){0} was completely
2945 omitted from the compiled regex. However, this means that if the group
2946 was called as a subroutine from elsewhere in the pattern, things went wrong
2947 (an internal error was given). Such groups are now left in the compiled
2948 pattern, with a new opcode that causes them to be skipped at execution
2949 time.
2950
2951 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2952 to the way PCRE behaves:
2953
2954 (a) A lone ] character is dis-allowed (Perl treats it as data).
2955
2956 (b) A back reference to an unmatched subpattern matches an empty string
2957 (Perl fails the current match path).
2958
2959 (c) A data ] in a character class must be notated as \] because if the
2960 first data character in a class is ], it defines an empty class. (In
2961 Perl it is not possible to have an empty class.) The empty class []
2962 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2963 The negative empty class [^] matches any one character, independently
2964 of the DOTALL setting.
2965
2966 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2967 non-existent subpattern following a character class starting with ']' and
2968 containing () gave an internal compiling error instead of "reference to
2969 non-existent subpattern". Fortunately, when the pattern did exist, the
2970 compiled code was correct. (When scanning forwards to check for the
2971 existence of the subpattern, it was treating the data ']' as terminating
2972 the class, so got the count wrong. When actually compiling, the reference
2973 was subsequently set up correctly.)
2974
2975 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2976 it was being rejected as not supported by pcre_dfa_exec(), even though
2977 other assertions are supported. I have made pcre_dfa_exec() support
2978 (*FAIL).
2979
2980 16. The implementation of 13c above involved the invention of a new opcode,
2981 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2982 cannot be changed at match time, I realized I could make a small
2983 improvement to matching performance by compiling OP_ALLANY instead of
2984 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2985 on the OP_ANY path.
2986
2987 17. Compiling pcretest on Windows with readline support failed without the
2988 following two fixes: (1) Make the unistd.h include conditional on
2989 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2990
2991 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2992 ncurses library to be included for pcretest when ReadLine support is
2993 requested, but also to allow for it to be overridden. This patch came from
2994 Daniel Bergström.
2995
2996 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2997 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2998 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2999 spotting this.
3000
3001
3002 Version 7.6 28-Jan-08
3003 ---------------------
3004
3005 1. A character class containing a very large number of characters with
3006 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
3007 overflow.
3008
3009 2. Patch to cut out the "long long" test in pcrecpp_unittest when
3010 HAVE_LONG_LONG is not defined.
3011
3012 3. Applied Christian Ehrlicher's patch to update the CMake build files to
3013 bring them up to date and include new features. This patch includes:
3014
3015 - Fixed PH's badly added libz and libbz2 support.
3016 - Fixed a problem with static linking.
3017 - Added pcredemo. [But later removed - see 7 below.]
3018 - Fixed dftables problem and added an option.
3019 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
3020 HAVE_LONG_LONG.
3021 - Added readline support for pcretest.
3022 - Added an listing of the option settings after cmake has run.
3023
3024 4. A user submitted a patch to Makefile that makes it easy to create
3025 "pcre.dll" under mingw when using Configure/Make. I added stuff to
3026 Makefile.am that cause it to include this special target, without
3027 affecting anything else. Note that the same mingw target plus all
3028 the other distribution libraries and programs are now supported
3029 when configuring with CMake (see 6 below) instead of with
3030 Configure/Make.
3031
3032 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
3033 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
3034 exported in the Windows port". It has not yet been confirmed that the patch
3035 solves the problem, but it does no harm.
3036
3037 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
3038 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
3039 with CMake, and also correct the comment about stack recursion.
3040
3041 7. Remove the automatic building of pcredemo from the ./configure system and
3042 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
3043 of a program that users should build themselves after PCRE is installed, so
3044 building it automatically is not really right. What is more, it gave
3045 trouble in some build environments.
3046
3047 8. Further tidies to CMakeLists.txt from Sheri and Christian.
3048
3049
3050 Version 7.5 10-Jan-08
3051 ---------------------
3052
3053 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
3054 values in parens when parsing an RE using the C++ wrapper."
3055
3056 2. Negative specials like \S did not work in character classes in UTF-8 mode.
3057 Characters greater than 255 were excluded from the class instead of being
3058 included.
3059
3060 3. The same bug as (2) above applied to negated POSIX classes such as
3061 [:^space:].
3062
3063 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
3064 defined or documented. It seems to have been a typo for PCRE_STATIC, so
3065 I have changed it.
3066
3067 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
3068 first named subpattern) and a construct such as (?&a) would reference the
3069 first named subpattern whose name started with "a" (in other words, the
3070 length check was missing). Both these problems are fixed. "Subpattern name
3071 expected" is now given for (?&) (a zero-length name), and this patch also
3072 makes it give the same error for \k'' (previously it complained that that
3073 was a reference to a non-existent subpattern).
3074
3075 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
3076 this is right because (?- can be followed by option settings as well as by
3077 digits. I have, however, made the messages clearer.
3078
3079 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
3080 than the number used in the conditional) now cause a compile-time error.
3081 This is actually not compatible with Perl, which accepts such patterns, but
3082 treats the conditional as always being FALSE (as PCRE used to), but it
3083 seems to me that giving a diagnostic is better.
3084
3085 8. Change "alphameric" to the more common word "alphanumeric" in comments
3086 and messages.
3087
3088 9. Fix two occurrences of "backslash" in comments that should have been
3089 "backspace".
3090
3091 10. Remove two redundant lines of code that can never be obeyed (their function
3092 was moved elsewhere).
3093
3094 11. The program that makes PCRE's Unicode character property table had a bug
3095 which caused it to generate incorrect table entries for sequences of
3096 characters that have the same character type, but are in different scripts.
3097 It amalgamated them into a single range, with the script of the first of
3098 them. In other words, some characters were in the wrong script. There were
3099 thirteen such cases, affecting characters in the following ranges:
3100
3101 U+002b0 - U+002c1
3102 U+0060c - U+0060d
3103 U+0061e - U+00612
3104 U+0064b - U+0065e
3105 U+0074d - U+0076d
3106 U+01800 - U+01805
3107 U+01d00 - U+01d77
3108 U+01d9b - U+01dbf
3109 U+0200b - U+0200f
3110 U+030fc - U+030fe
3111 U+03260 - U+0327f
3112 U+0fb46 - U+0fbb1
3113 U+10450 - U+1049d
3114
3115 12. The -o option (show only the matching part of a line) for pcregrep was not
3116 compatible with GNU grep in that, if there was more than one match in a
3117 line, it showed only the first of them. It now behaves in the same way as
3118 GNU grep.
3119
3120 13. If the -o and -v options were combined for pcregrep, it printed a blank
3121 line for every non-matching line. GNU grep prints nothing, and pcregrep now
3122 does the same. The return code can be used to tell if there were any
3123 non-matching lines.
3124
3125 14. Added --file-offsets and --line-offsets to pcregrep.
3126
3127 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
3128 infinitely looping recursion. The bug was that positive lookaheads were not
3129 being skipped when checking for a possible empty match (negative lookaheads
3130 and both kinds of lookbehind were skipped).
3131
3132 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
3133 inclusion of <windows.h> to before rather than after the definition of
3134 INVALID_FILE_ATTRIBUTES (patch from David Byron).
3135
3136 17. Specifying a possessive quantifier with a specific limit for a Unicode
3137 character property caused pcre_compile() to compile bad code, which led at
3138 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
3139 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
3140 caused the error; without that there was no problem.
3141
3142 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3143
3144 19. Added --enable-pcretest-libreadline.
3145
3146 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3147 RE::GlobalReplace(). As a result, the number of replacements returned was
3148 double what it should be. I removed one of the increments, but Craig sent a
3149 later patch that removed the other one (the right fix) and added unit tests
3150 that check the return values (which was not done before).
3151
3152 21. Several CMake things:
3153
3154 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3155 the names libpcre and libpcreposix, not just pcre and pcreposix.
3156
3157 (2) The above change means that pcretest and pcregrep are now correctly
3158 linked with the newly-built libraries, not previously installed ones.
3159
3160 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
3161
3162 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3163 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3164 UTF-8 newline character). The key issue is that the pattern starts .*;
3165 this means that the match must be either at the beginning, or after a
3166 newline. The bug was in the code for advancing after a failed match and
3167 checking that the new position followed a newline. It was not taking
3168 account of UTF-8 characters correctly.
3169
3170 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3171 character classes. PCRE was not treating the sequence [:...:] as a
3172 character class unless the ... were all letters. Perl, however, seems to
3173 allow any characters between [: and :], though of course it rejects as
3174 unknown any "names" that contain non-letters, because all the known class
3175 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3176 for example, whereas PCRE did not - it did not recognize a POSIX character
3177 class. This seemed a bit dangerous, so the code has been changed to be
3178 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3179 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3180 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3181 Perl does, and where it didn't before.
3182
3183 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3184 Windows environments %n is disabled by default.
3185
3186
3187 Version 7.4 21-Sep-07
3188 ---------------------
3189
3190 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3191 means that a class such as [\s] counted as "explicit reference to CR or
3192 LF". That isn't really right - the whole point of the change was to try to
3193 help when there was an actual mention of one of the two characters. So now
3194 the change happens only if \r or \n (or a literal CR or LF) character is
3195 encountered.
3196
3197 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3198 of both had grown to the point where there were only 3 bits left.
3199 Fortunately, there was spare space in the data structure, and so I have
3200 moved the internal flags into a new 16-bit field to free up more option
3201 bits.
3202
3203 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3204 but did not set the internal JCHANGED flag - either of these is enough to
3205 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3206 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3207 start sets both bits.
3208
3209 4. Added options (at build time, compile time, exec time) to change \R from
3210 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3211
3212 5. doc/pcresyntax.html was missing from the distribution.
3213
3214 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3215 compatibility, even though it is no longer used.
3216
3217 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3218 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3219 windows.h file is present (where different names are used). [This was
3220 reversed later after testing - see 16 below.]
3221
3222 8. Changed all #include <config.h> to #include "config.h". There were also
3223 some further <pcre.h> cases that I changed to "pcre.h".
3224
3225 9. When pcregrep was used with the --colour option, it missed the line ending
3226 sequence off the lines that it output.
3227
3228 10. It was pointed out to me that arrays of string pointers cause lots of
3229 relocations when a shared library is dynamically loaded. A technique of
3230 using a single long string with a table of offsets can drastically reduce
3231 these. I have refactored PCRE in four places to do this. The result is
3232 dramatic:
3233
3234 Originally: 290
3235 After changing UCP table: 187
3236 After changing error message table: 43
3237 After changing table of "verbs" 36
3238 After changing table of Posix names 22
3239
3240 Thanks to the folks working on Gregex for glib for this insight.
3241
3242 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3243 unicode-properties was also set.
3244
3245 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3246
3247 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3248 checked only for CRLF.
3249
3250 14. Added casts to pcretest.c to avoid compiler warnings.
3251
3252 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3253
3254 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3255 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3256 entirely. This removes changes made in 7 above.
3257
3258 17. The CMake files have been updated, and there is now more information about
3259 building with CMake in the NON-UNIX-USE document.
3260
3261
3262 Version 7.3 28-Aug-07
3263 ---------------------
3264
3265 1. In the rejigging of the build system that eventually resulted in 7.1, the
3266 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3267 brackets there is not right, since it causes compilers to look for an
3268 installed pcre.h, not the version that is in the source that is being
3269 compiled (which of course may be different). I have changed it back to:
3270
3271 #include "pcre.h"
3272
3273 I have a vague recollection that the change was concerned with compiling in
3274 different directories, but in the new build system, that is taken care of
3275 by the VPATH setting the Makefile.
3276
3277 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3278 when the subject happened to end in the byte 0x85 (e.g. if the last
3279 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3280 characters but of course it shouldn't be taken as a newline when it is part
3281 of another character. The bug was that, for an unlimited repeat of . in
3282 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3283 characters when looking for a newline.
3284
3285 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3286
3287 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3288 in debug output.
3289
3290 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3291 long printing in the pcrecpp unittest when running under MinGW.
3292
3293 6. ESC_K was left out of the EBCDIC table.
3294
3295 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3296 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3297 limit also applies to "virtual nesting" when a pattern is recursive, and in
3298 this case 1000 isn't so big. I have been able to remove this limit at the
3299 expense of backing off one optimization in certain circumstances. Normally,
3300 when pcre_exec() would call its internal match() function recursively and
3301 immediately return the result unconditionally, it uses a "tail recursion"
3302 feature to save stack. However, when a subpattern that can match an empty
3303 string has an unlimited repetition quantifier, it no longer makes this
3304 optimization. That gives it a stack frame in which to save the data for
3305 checking that an empty string has been matched. Previously this was taken
3306 from the 1000-entry workspace that had been reserved. So now there is no
3307 explicit limit, but more stack is used.
3308
3309 8. Applied Daniel's patches to solve problems with the import/export magic
3310 syntax that is required for Windows, and which was going wrong for the
3311 pcreposix and pcrecpp parts of the library. These were overlooked when this
3312 problem was solved for the main library.
3313
3314 9. There were some crude static tests to avoid integer overflow when computing
3315 the size of patterns that contain repeated groups with explicit upper
3316 limits. As the maximum quantifier is 65535, the maximum group length was
3317 set at 30,000 so that the product of these two numbers did not overflow a
3318 32-bit integer. However, it turns out that people want to use groups that
3319 are longer than 30,000 bytes (though not repeat them that many times).
3320 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3321 made it possible to implement the integer overflow checks in a much more
3322 dynamic way, which I have now done. The artificial limitation on group
3323 length has been removed - we now have only the limit on the total length of
3324 the compiled pattern, which depends on the LINK_SIZE setting.
3325
3326 10. Fixed a bug in the documentation for get/copy named substring when
3327 duplicate names are permitted. If none of the named substrings are set, the
3328 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3329 empty string.
3330
3331 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3332 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3333 because the ] is interpreted as the first data character and the
3334 terminating ] is not found. PCRE has been made compatible with Perl in this
3335 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3336 cause memory overwriting.
3337
3338 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3339 string has been matched (to stop an infinite loop). It was not recognizing
3340 a conditional subpattern that could match an empty string if that
3341 subpattern was within another subpattern. For example, it looped when
3342 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3343 condition was not nested. This bug has been fixed.
3344
3345 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3346 past the start of the subject in the presence of bytes with the top bit
3347 set, for example "\x8aBCD".
3348
3349 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3350 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3351
3352 14. Optimized (?!) to (*FAIL).
3353
3354 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3355 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3356 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3357 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3358 does: it's just the validity check that is more restrictive.
3359
3360 16. Inserted checks for integer overflows during escape sequence (backslash)
3361 processing, and also fixed erroneous offset values for syntax errors during
3362 backslash processing.
3363
3364 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3365 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3366
3367 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3368 caused an overrun.
3369
3370 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3371 something other than just ASCII characters) inside a group that had an
3372 unlimited repeat caused a loop at compile time (while checking to see
3373 whether the group could match an empty string).
3374
3375 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3376 [\P{Any}] did so. (Error in the code for printing property names.)
3377
3378 21. An orphan \E inside a character class could cause a crash.
3379
3380 22. A repeated capturing bracket such as (A)? could cause a wild memory
3381 reference during compilation.
3382
3383 23. There are several functions in pcre_compile() that scan along a compiled
3384 expression for various reasons (e.g. to see if it's fixed length for look
3385 behind). There were bugs in these functions when a repeated \p or \P was
3386 present in the pattern. These operators have additional parameters compared
3387 with \d, etc, and these were not being taken into account when moving along
3388 the compiled data. Specifically:
3389
3390 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3391 length.
3392
3393 (b) An item such as \pL+ within a repeated group could cause crashes or
3394 loops.
3395
3396 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3397 "reference to non-existent subpattern" error.
3398
3399 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3400
3401 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3402 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3403
3404 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3405
3406 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3407 character were causing crashes (broken optimization).
3408
3409 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3410 \p or \P) caused a compile-time loop.
3411
3412 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3413 break. For example, the unstudied pattern [\r\n]A does not match the string
3414 "\r\nA" because change 7.0/46 below moves the current point on by two
3415 characters after failing to match at the start. However, the pattern \nA
3416 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3417 the same is true. There doesn't seem any very clean way out of this, but
3418 what I have chosen to do makes the common cases work: PCRE now takes note
3419 of whether there can be an explicit match for \r or \n anywhere in the
3420 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3421 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3422 pattern has explicit CR or LF references.
3423
3424 29. Added (*CR) etc for changing newline setting at start of pattern.
3425
3426
3427 Version 7.2 19-Jun-07
3428 ---------------------
3429
3430 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3431 which is apparently normally available under Windows.
3432
3433 2. Re-jig the pcregrep tests with different newline settings in an attempt
3434 to make them independent of the local environment's newline setting.
3435
3436 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3437
3438 4. Some of the "internals" tests were previously cut out when the link size
3439 was not 2, because the output contained actual offsets. The recent new
3440 "Z" feature of pcretest means that these can be cut out, making the tests
3441 usable with all link sizes.
3442
3443 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3444 stack recursion. This gives a massive performance boost under BSD, but just
3445 a small improvement under Linux. However, it saves one field in the frame
3446 in all cases.
3447
3448 6. Added more features from the forthcoming Perl 5.10:
3449
3450 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3451 recursion call. It refers to the nth most recently opened parentheses.
3452
3453 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3454 to be opened parentheses.
3455
3456 (c) Conditions that refer to capturing parentheses can be specified
3457 relatively, for example, (?(-2)... or (?(+3)...
3458
3459 (d) \K resets the start of the current match so that everything before
3460 is not part of it.
3461
3462 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3463
3464 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3465 reference syntax.
3466
3467 (g) (?| introduces a group in which the numbering of parentheses in each
3468 alternative starts with the same number.
3469
3470 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3471
3472 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3473 PCRE_INFO_JCHANGED.
3474
3475 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3476 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3477 for detecting groups that can match an empty string.
3478
3479 9. A pattern with a very large number of alternatives (more than several
3480 hundred) was running out of internal workspace during the pre-compile
3481 phase, where pcre_compile() figures out how much memory will be needed. A
3482 bit of new cunning has reduced the workspace needed for groups with
3483 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3484 workspace instead of running out of the 4096 that are available.
3485
3486 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3487
3488 11. Applied patch from Google to remove an optimization that didn't quite work.
3489 The report of the bug said:
3490
3491 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3492 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3493 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3494
3495 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3496 it matched the wrong number of bytes.
3497
3498
3499 Version 7.1 24-Apr-07
3500 ---------------------
3501
3502 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3503 that is more "standard", making use of automake and other Autotools. There
3504 is some re-arrangement of the files and adjustment of comments consequent
3505 on this.
3506
3507 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3508 for recursive directory scanning broke on some systems because the files
3509 are not scanned in any specific order and on different systems the order
3510 was different. A call to "sort" has been inserted into RunGrepTest for the
3511 approprate test as a short-term fix. In the longer term there may be an
3512 alternative.
3513
3514 3. I had an email from Eric Raymond about problems translating some of PCRE's
3515 man pages to HTML (despite the fact that I distribute HTML pages, some
3516 people do their own conversions for various reasons). The problems
3517 concerned the use of low-level troff macros .br and .in. I have therefore
3518 removed all such uses from the man pages (some were redundant, some could
3519 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3520 HTML has been updated to handle .nf/.fi and to complain if it encounters
3521 .br or .in.
3522
3523 4. Updated comments in configure.ac that get placed in config.h.in and also
3524 arranged for config.h to be included in the distribution, with the name
3525 config.h.generic, for the benefit of those who have to compile without
3526 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3527
3528 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3529 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3530 makevp.bat for the current PCRE, using the additional files
3531 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3532
3533 6. A Windows user reported a minor discrepancy with test 2, which turned out
3534 to be caused by a trailing space on an input line that had got lost in his
3535 copy. The trailing space was an accident, so I've just removed it.
3536
3537 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3538 that is needed.
3539
3540 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3541 as "const" (a) because they are and (b) because it helps the PHP
3542 maintainers who have recently made a script to detect big data structures
3543 in the php code that should be moved to the .rodata section. I remembered
3544 to update Builducptable as well, so it won't revert if ucptable.h is ever
3545 re-created.
3546
3547 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3548 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3549 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3550 support is not required. This saves 1.5-2K of code, which is important in
3551 some applications.
3552
3553 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3554 so as not to refer to the tables, even though these functions will never be
3555 called when UTF-8 support is disabled. Otherwise there are problems with a
3556 shared library.
3557
3558 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3559
3560 (a) It was defining its arguments as char * instead of void *.
3561
3562 (b) It was assuming that all moves were upwards in memory; this was true
3563 a long time ago when I wrote it, but is no longer the case.
3564
3565 The emulated memove() is provided for those environments that have neither
3566 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3567 is clearly not the case, as these two bugs were recently reported.
3568
3569 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3570 and Detrail to create the HTML documentation, the .txt form of the man
3571 pages, and it removes trailing spaces from listed files. It also creates
3572 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3573 case, it wraps all the #defines with #ifndefs. This script should be run
3574 before "make dist".
3575
3576 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3577 with Unicode property support.
3578
3579 (a) For a maximizing quantifier, if the two different cases of the
3580 character were of different lengths in their UTF-8 codings (there are
3581 some cases like this - I found 11), and the matching function had to
3582 back up over a mixture of the two cases, it incorrectly assumed they
3583 were both the same length.
3584
3585 (b) When PCRE was configured to use the heap rather than the stack for
3586 recursion during matching, it was not correctly preserving the data for
3587 the other case of a UTF-8 character when checking ahead for a match
3588 while processing a minimizing repeat. If the check also involved
3589 matching a wide character, but failed, corruption could cause an
3590 erroneous result when trying to check for a repeat of the original
3591 character.
3592
3593 13. Some tidying changes to the testing mechanism:
3594
3595 (a) The RunTest script now detects the internal link size and whether there
3596 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3597 values substituted by "configure". (The RunGrepTest script already did
3598 this for UTF-8.) The configure.ac script no longer substitutes the
3599 relevant variables.
3600
3601 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3602 with length and offset values. This means that the output is different
3603 for different internal link sizes. Test 2 is skipped for link sizes
3604 other than 2 because of this, bypassing the problem. Unfortunately,
3605 there was also a test in test 3 (the locale tests) that used /B and
3606 failed for link sizes other than 2. Rather than cut the whole test out,
3607 I have added a new /Z option to pcretest that replaces the length and
3608 offset values with spaces. This is now used to make test 3 independent
3609 of link size. (Test 2 will be tidied up later.)
3610
3611 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3612 segmentation fault instead of returning the appropriate error message.
3613
3614 15. In multiline mode when the newline sequence was set to "any", the pattern
3615 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3616 This doesn't seem right; it now treats the CRLF combination as the line
3617 ending, and so does not match in that case. It's only a pattern such as ^$
3618 that would hit this one: something like ^ABC$ would have failed after \r
3619 and then tried again after \r\n.
3620
3621 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3622 in an attempt to make files that differ only in their line terminators
3623 compare equal. This works on Linux.
3624
3625 17. Under certain error circumstances pcregrep might try to free random memory
3626 as it exited. This is now fixed, thanks to valgrind.
3627
3628 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3629 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3630 was because its rules for how to advance for /g after matching an empty
3631 string at the end of a line did not allow for this case. They now check for
3632 it specially.
3633
3634 20. pcretest is supposed to handle patterns and data of any length, by
3635 extending its buffers when necessary. It was getting this wrong when the
3636 buffer for a data line had to be extended.
3637
3638 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3639 CRLF as a newline sequence.
3640
3641 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3642 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3643 I have nevertheless tidied it up.
3644
3645 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3646
3647 24. Added a man page for pcre-config.
3648
3649
3650 Version 7.0 19-Dec-06
3651 ---------------------
3652
3653 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3654 moving to gcc 4.1.1.
3655
3656 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3657 sys/time.h, which is documented as needed for this function. It doesn't
3658 seem to matter on Linux, but it showed up on some releases of OS X.
3659
3660 3. It seems that there are systems where bytes whose values are greater than
3661 127 match isprint() in the "C" locale. The "C" locale should be the
3662 default when a C program starts up. In most systems, only ASCII printing
3663 characters match isprint(). This difference caused the output from pcretest
3664 to vary, making some of the tests fail. I have changed pcretest so that:
3665
3666 (a) When it is outputting text in the compiled version of a pattern, bytes
3667 other than 32-126 are always shown as hex escapes.
3668
3669 (b) When it is outputting text that is a matched part of a subject string,
3670 it does the same, unless a different locale has been set for the match
3671 (using the /L modifier). In this case, it uses isprint() to decide.
3672
3673 4. Fixed a major bug that caused incorrect computation of the amount of memory
3674 required for a compiled pattern when options that changed within the
3675 pattern affected the logic of the preliminary scan that determines the
3676 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3677 that the computed length was too small. The symptoms of this bug were
3678 either the PCRE error "internal error: code overflow" from pcre_compile(),
3679 or a glibc crash with a message such as "pcretest: free(): invalid next
3680 size (fast)". Examples of patterns that provoked this bug (shown in
3681 pcretest format) are:
3682
3683 /(?-x: )/x
3684 /(?x)(?-x: \s*#\s*)/
3685 /((?i)[\x{c0}])/8
3686 /(?i:[\x{c0}])/8
3687
3688 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3689 is now done differently.
3690
3691 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3692 wrapper classes; (b) implement a new function in the C++ scanner that is
3693 more efficient than the old way of doing things because it avoids levels of
3694 recursion in the regex matching; (c) add a paragraph to the documentation
3695 for the FullMatch() function.
3696
3697 6. The escape sequence \n was being treated as whatever was defined as
3698 "newline". Not only was this contrary to the documentation, which states
3699 that \n is character 10 (hex 0A), but it also went horribly wrong when
3700 "newline" was defined as CRLF. This has been fixed.
3701
3702 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3703 was being set to -1 for the "end of line" case (supposedly a value that no
3704 character can have). Though this value is never used (the check for end of
3705 line is "zero bytes in current character"), it caused compiler complaints.
3706 I've changed it to 0xffffffff.
3707
3708 8. In pcre_version.c, the version string was being built by a sequence of
3709 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3710 string (as it is for production releases) called a macro with an empty
3711 argument. The C standard says the result of this is undefined. The gcc
3712 compiler treats it as an empty string (which was what was wanted) but it is
3713 reported that Visual C gives an error. The source has been hacked around to
3714 avoid this problem.
3715
3716 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3717 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3718 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3719 of them did).
3720
3721 10. Originally, pcretest opened its input and output without "b"; then I was
3722 told that "b" was needed in some environments, so it was added for release
3723 5.0 to both the input and output. (It makes no difference on Unix-like
3724 systems.) Later I was told that it is wrong for the input on Windows. I've
3725 now abstracted the modes into two macros, to make it easier to fiddle with
3726 them, and removed "b" from the input mode under Windows.
3727
3728 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3729
3730 12. Added -help and --help to pcretest as an official way of being reminded
3731 of the options.
3732
3733 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3734 and pcrecpp.cc because they annoy compilers at high warning levels.
3735
3736 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3737
3738 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3739 scripts are not C programs :-) and which was not noticed because it works
3740 on Linux.
3741
3742 16. pcretest is supposed to handle any length of pattern and data line (as one
3743 line or as a continued sequence of lines) by extending its input buffer if
3744 necessary. This feature was broken for very long pattern lines, leading to
3745 a string of junk being passed to pcre_compile() if the pattern was longer
3746 than about 50K.
3747
3748 17. I have done a major re-factoring of the way pcre_compile() computes the
3749 amount of memory needed for a compiled pattern. Previously, there was code
3750 that made a preliminary scan of the pattern in order to do this. That was
3751 OK when PCRE was new, but as the facilities have expanded, it has become
3752 harder and harder to keep it in step with the real compile phase, and there
3753 have been a number of bugs (see for example, 4 above). I have now found a
3754 cunning way of running the real compile function in a "fake" mode that
3755 enables it to compute how much memory it would need, while actually only
3756 ever using a few hundred bytes of working memory and without too many
3757 tests of the mode. This should make future maintenance and development
3758 easier. A side effect of this work is that the limit of 200 on the nesting
3759 depth of parentheses has been removed (though this was never a serious
3760 limitation, I suspect). However, there is a downside: pcre_compile() now
3761 runs more slowly than before (30% or more, depending on the pattern). I
3762 hope this isn't a big issue. There is no effect on runtime performance.
3763
3764 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3765 newline (only possible for the last line of a file) and it was a
3766 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3767
3768 19. Added additional timing features to pcretest. (1) The -tm option now times
3769 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3770 separate command line item, by a number that specifies the number of
3771 repeats to use when timing. The default is 50000; this gives better
3772 precision, but takes uncomfortably long for very large patterns.
3773
3774 20. Extended pcre_study() to be more clever in cases where a branch of a
3775 subpattern has no definite first character. For example, (a*|b*)[cd] would
3776 previously give no result from pcre_study(). Now it recognizes that the
3777 first character must be a, b, c, or d.
3778
3779 21. There was an incorrect error "recursive call could loop indefinitely" if
3780 a subpattern (or the entire pattern) that was being tested for matching an
3781 empty string contained only one non-empty item after a nested subpattern.
3782 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3783 incorrectly, because the \d was being skipped in the check.
3784
3785 22. The pcretest program now has a new pattern option /B and a command line
3786 option -b, which is equivalent to adding /B to every pattern. This causes
3787 it to show the compiled bytecode, without the additional information that
3788 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3789 is the same as /B/I).
3790
3791 23. A new optimization is now able automatically to treat some sequences such
3792 as a*b as a*+b. More specifically, if something simple (such as a character
3793 or a simple class like \d) has an unlimited quantifier, and is followed by
3794 something that cannot possibly match the quantified thing, the quantifier
3795 is automatically "possessified".
3796
3797 24. A recursive reference to a subpattern whose number was greater than 39
3798 went wrong under certain circumstances in UTF-8 mode. This bug could also
3799 have affected the operation of pcre_study().
3800
3801 25. Realized that a little bit of performance could be had by replacing
3802 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3803
3804 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3805
3806 27. Possessive quantifiers such as a++ were previously implemented by turning
3807 them into atomic groups such as ($>a+). Now they have their own opcodes,
3808 which improves performance. This includes the automatically created ones
3809 from 23 above.
3810
3811 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3812 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3813 the first matched character to be a colon. This applied both to named and
3814 numbered groups.
3815
3816 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3817
3818 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3819 building PCRE on Windows easier, so I have included it in the distribution.
3820
3821 31. There is now a check in pcretest against a ridiculously large number being
3822 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3823 loop, the loop is abandoned.
3824
3825 32. Forward references to subpatterns in conditions such as (?(2)...) where
3826 subpattern 2 is defined later cause pcre_compile() to search forwards in
3827 the pattern for the relevant set of parentheses. This search went wrong
3828 when there were unescaped parentheses in a character class, parentheses
3829 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3830
3831 33. "Subroutine" calls and backreferences were previously restricted to
3832 referencing subpatterns earlier in the regex. This restriction has now
3833 been removed.
3834
3835 34. Added a number of extra features that are going to be in Perl 5.10. On the
3836 whole, these are just syntactic alternatives for features that PCRE had
3837 previously implemented using the Python syntax or my own invention. The
3838 other formats are all retained for compatibility.
3839
3840 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3841 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3842 also .NET compatible.
3843
3844 (b) A recursion or subroutine call to a named group can now be defined as
3845 (?&name) as well as (?P>name).
3846
3847 (c) A backreference to a named group can now be defined as \k<name> or
3848 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3849 5.10, are also .NET compatible.
3850
3851 (d) A conditional reference to a named group can now use the syntax
3852 (?(<name>) or (?('name') as well as (?(name).
3853
3854 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3855 groups (named and numbered) that are never evaluated inline, but can be
3856 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3857 is always false. There may be only one alternative in such a group.
3858
3859 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3860 as the simple (?(R). The condition is true only if the most recent
3861 recursion is that of the given number or name. It does not search out
3862 through the entire recursion stack.
3863
3864 (g) The escape \gN or \g{N} has been added, where N is a positive or
3865 negative number, specifying an absolute or relative reference.
3866
3867 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3868 some "unreachable code" warnings.
3869
3870 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3871 things, this adds five new scripts.
3872
3873 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3874 There were also incompatibilities regarding the handling of \Q..\E inside
3875 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3876 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3877
3878 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3879 matches an empty string, and forcibly breaks the loop. There were bugs in
3880 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3881 against aaaa the result was just "a" rather than "aaaa", for example. Two
3882 separate and independent bugs (that affected different cases) have been
3883 fixed.
3884
3885 39. Refactored the code to abolish the use of different opcodes for small
3886 capturing bracket numbers. This is a tidy that I avoided doing when I
3887 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3888 The new approach is not only tidier, it makes it possible to reduce the
3889 memory needed to fix the previous bug (38).
3890
3891 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3892 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3893 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3894 mode.
3895
3896 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3897 report.
3898
3899 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3900 copy construction and assignment in the C++ wrapper.
3901
3902 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3903 couple of bugs that could have given wrong results in the "--newline=crlf"
3904 case.
3905
3906 44. Added a number of casts and did some reorganization of signed/unsigned int
3907 variables following suggestions from Dair Grant. Also renamed the variable
3908 "this" as "item" because it is a C++ keyword.
3909
3910 45. Arranged for dftables to add
3911
3912 #include "pcre_internal.h"
3913
3914 to pcre_chartables.c because without it, gcc 4.x may remove the array
3915 definition from the final binary if PCRE is built into a static library and
3916 dead code stripping is activated.
3917
3918 46. For an unanchored pattern, if a match attempt fails at the start of a
3919 newline sequence, and the newline setting is CRLF or ANY, and the next two
3920 characters are CRLF, advance by two characters instead of one.
3921
3922
3923 Version 6.7 04-Jul-06
3924 ---------------------
3925
3926 1. In order to handle tests when input lines are enormously long, pcretest has
3927 been re-factored so that it automatically extends its buffers when
3928 necessary. The code is crude, but this _is_ just a test program. The
3929 default size has been increased from 32K to 50K.
3930
3931 2. The code in pcre_study() was using the value of the re argument before
3932 testing it for NULL. (Of course, in any sensible call of the function, it
3933 won't be NULL.)
3934
3935 3. The memmove() emulation function in pcre_internal.h, which is used on
3936 systems that lack both memmove() and bcopy() - that is, hardly ever -
3937 was missing a "static" storage class specifier.
3938
3939 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3940 containing an extended class (one that cannot be represented by a bitmap
3941 because it contains high-valued characters or Unicode property items, e.g.
3942 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3943 pattern, but PCRE should not loop if you do not (it no longer does).
3944 [Detail: two cases were found: (a) a repeated subpattern containing an
3945 extended class; (b) a recursive reference to a subpattern that followed a
3946 previous extended class. It wasn't skipping over the extended class
3947 correctly when UTF-8 mode was not set.]
3948
3949 5. A negated single-character class was not being recognized as fixed-length
3950 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3951 compile error "lookbehind assertion is not fixed length".
3952
3953 6. The RunPerlTest auxiliary script was showing an unexpected difference
3954 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3955 write a Perl script that can interpret lines of an input file either as
3956 byte characters or as UTF-8, which is what "perltest" was being required to
3957 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3958 can't do is switch easily at run time between having the "use utf8;" pragma
3959 or not. In the end, I fudged it by using the RunPerlTest script to insert
3960 "use utf8;" explicitly for the UTF-8 tests.
3961
3962 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3963 the end of the subject string, contrary to the documentation and to what
3964 Perl does. This was true of both matching functions. Now it matches only at
3965 the start of the subject and immediately after *internal* newlines.
3966
3967 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3968 a pointer to an int instead of a pointer to an unsigned long int. This
3969 caused problems on 64-bit systems.
3970
3971 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3972 instance of the 'standard' template library not being so standard".
3973
3974 10. There was no check on the number of named subpatterns nor the maximum
3975 length of a subpattern name. The product of these values is used to compute
3976 the size of the memory block for a compiled pattern. By supplying a very
3977 long subpattern name and a large number of named subpatterns, the size
3978 computation could be caused to overflow. This is now prevented by limiting
3979 the length of names to 32 characters, and the number of named subpatterns
3980 to 10,000.
3981
3982 11. Subpatterns that are repeated with specific counts have to be replicated in
3983 the compiled pattern. The size of memory for this was computed from the
3984 length of the subpattern and the repeat count. The latter is limited to
3985 65535, but there was no limit on the former, meaning that integer overflow
3986 could in principle occur. The compiled length of a repeated subpattern is
3987 now limited to 30,000 bytes in order to prevent this.
3988
3989 12. Added the optional facility to have named substrings with the same name.
3990
3991 13. Added the ability to use a named substring as a condition, using the
3992 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3993 are numbers (not recommended). Forward references are permitted.
3994
3995 14. Added forward references in named backreferences (if you see what I mean).
3996
3997 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3998 pattern could run off the end of the subject. For example, the pattern
3999 "(?s)(.{1,5})"8 did this with the subject "ab".
4000
4001 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
4002 PCRE_CASELESS was set when matching characters that were quantified with ?
4003 or *.
4004
4005 17. A character class other than a single negated character that had a minimum
4006 but no maximum quantifier - for example [ab]{6,} - was not handled
4007 correctly by pce_dfa_exec(). It would match only one character.
4008
4009 18. A valid (though odd) pattern that looked like a POSIX character
4010 class but used an invalid character after [ (for example [[,abc,]]) caused
4011 pcre_compile() to give the error "Failed: internal error: code overflow" or
4012 in some cases to crash with a glibc free() error. This could even happen if
4013 the pattern terminated after [[ but there just happened to be a sequence of
4014 letters, a binary zero, and a closing ] in the memory that followed.
4015
4016 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
4017 over the years. Originally (before any Unicode support), just the bottom 8
4018 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
4019 output from "man perlunicode" includes this:
4020
4021 The regular expression compiler produces polymorphic opcodes. That
4022 is, the pattern adapts to the data and automatically switches to
4023 the Unicode character scheme when presented with Unicode data--or
4024 instead uses a traditional byte scheme when presented with byte
4025 data.
4026
4027 Sadly, a wide octal escape does not cause a switch, and in a string with
4028 no other multibyte characters, these octal escapes are treated as before.
4029 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
4030 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
4031 Unicode string.
4032
4033 I have not perpetrated such confusion in PCRE. Up till now, it took just
4034 the bottom 8 bits, as in old Perl. I have now made octal escapes with
4035 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
4036 translate to the appropriate multibyte character.
4037
4038 29. Applied some refactoring to reduce the number of warnings from Microsoft
4039 and Borland compilers. This has included removing the fudge introduced
4040 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
4041 a warning about an unused variable.
4042
4043 21. PCRE has not included VT (character 0x0b) in the set of whitespace
4044 characters since release 4.0, because Perl (from release 5.004) does not.
4045 [Or at least, is documented not to: some releases seem to be in conflict
4046 with the documentation.] However, when a pattern was studied with
4047 pcre_study() and all its branches started with \s, PCRE still included VT
4048 as a possible starting character. Of course, this did no harm; it just
4049 caused an unnecessary match attempt.
4050
4051 22. Removed a now-redundant internal flag bit that recorded the fact that case
4052 dependency changed within the pattern. This was once needed for "required
4053 byte" processing, but is no longer used. This recovers a now-scarce options
4054 bit. Also moved the least significant internal flag bit to the most-
4055 significant bit of the word, which was not previously used (hangover from
4056 the days when it was an int rather than a uint) to free up another bit for
4057 the future.
4058
4059 23. Added support for CRLF line endings as well as CR and LF. As well as the
4060 default being selectable at build time, it can now be changed at runtime
4061 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
4062 specify that it is scanning data with non-default line endings.
4063
4064 24. Changed the definition of CXXLINK to make it agree with the definition of
4065 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
4066
4067 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
4068 recursions. This makes a big different to stack usage for some patterns.
4069
4070 26. If a subpattern containing a named recursion or subroutine reference such
4071 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
4072 the space required for the compiled pattern went wrong and gave too small a
4073 value. Depending on the environment, this could lead to "Failed: internal
4074 error: code overflow at offset 49" or "glibc detected double free or
4075 corruption" errors.
4076
4077 27. Applied patches from Google (a) to support the new newline modes and (b) to
4078 advance over multibyte UTF-8 characters in GlobalReplace.
4079
4080 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
4081 difference for some implementation of PCRE in some Windows version.
4082
4083 29. Added some extra testing facilities to pcretest:
4084
4085 \q<number> in a data line sets the "match limit" value
4086 \Q<number> in a data line sets the "match recursion limt" value
4087 -S <number> sets the stack size, where <number> is in megabytes
4088
4089 The -S option isn't available for Windows.
4090
4091
4092 Version 6.6 06-Feb-06
4093 ---------------------
4094
4095 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
4096 in pcreposix.h. I have copied the definition from pcre.h.
4097
4098 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
4099 because pcre.h is no longer a built file.
4100
4101 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
4102 not normally included in the compiled code.
4103
4104
4105 Version 6.5 01-Feb-06
4106 ---------------------
4107
4108 1. When using the partial match feature with pcre_dfa_exec(), it was not
4109 anchoring the second and subsequent partial matches at the new starting
4110 point. This could lead to incorrect results. For example, with the pattern
4111 /1234/, partially matching against "123" and then "a4" gave a match.
4112
4113 2. Changes to pcregrep:
4114
4115 (a) All non-match returns from pcre_exec() were being treated as failures
4116 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
4117 error message is output. Some extra information is given for the
4118 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
4119 probably the only errors that are likely to be caused by users (by
4120 specifying a regex that has nested indefinite repeats, for instance).
4121 If there are more than 20 of these errors, pcregrep is abandoned.
4122
4123 (b) A binary zero was treated as data while matching, but terminated the
4124 output line if it was written out. This has been fixed: binary zeroes
4125 are now no different to any other data bytes.
4126
4127 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
4128 used to set a locale for matching. The --locale=xxxx long option has
4129 been added (no short equivalent) to specify a locale explicitly on the
4130 pcregrep command, overriding the environment variables.
4131
4132 (d) When -B was used with -n, some line numbers in the output were one less
4133 than they should have been.
4134
4135 (e) Added the -o (--only-matching) option.
4136
4137 (f) If -A or -C was used with -c (count only), some lines of context were
4138 accidentally printed for the final match.
4139
4140 (g) Added the -H (--with-filename) option.
4141
4142 (h) The combination of options -rh failed to suppress file names for files
4143 that were found from directory arguments.
4144
4145 (i) Added the -D (--devices) and -d (--directories) options.
4146
4147 (j) Added the -F (--fixed-strings) option.
4148
4149 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4150
4151 (l) Added the --colo(u)r option.
4152
4153 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4154 is not present by default.
4155
4156 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4157 items such as (?R) or (?1), when the recursion could match a number of
4158 alternatives. If it matched one of the alternatives, but subsequently,
4159 outside the recursion, there was a failure, the code tried to back up into
4160 the recursion. However, because of the way PCRE is implemented, this is not
4161 possible, and the result was an incorrect result from the match.
4162
4163 In order to prevent this happening, the specification of recursion has
4164 been changed so that all such subpatterns are automatically treated as
4165 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4166
4167 4. I had overlooked the fact that, in some locales, there are characters for
4168 which isalpha() is true but neither isupper() nor islower() are true. In
4169 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4170 and ordfeminine) are like this. This affected the treatment of \w and \W
4171 when they appeared in character classes, but not when they appeared outside
4172 a character class. The bit map for "word" characters is now created
4173 separately from the results of isalnum() instead of just taking it from the
4174 upper, lower, and digit maps. (Plus the underscore character, of course.)
4175
4176 5. The above bug also affected the handling of POSIX character classes such as
4177 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4178 permanent tables. Instead, the bit maps for such a class were previously
4179 created as the appropriate unions of the upper, lower, and digit bitmaps.
4180 Now they are created by subtraction from the [[:word:]] class, which has
4181 its own bitmap.
4182
4183 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4184 It is created by subtracting the vertical space characters (\x09, \x0a,
4185 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4186 subtraction was done in the overall bitmap for a character class, meaning
4187 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4188 be recognized. This bug has been fixed.
4189
4190 7. Patches from the folks at Google:
4191
4192 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4193 real life, but is still worth protecting against".
4194
4195 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4196 regular expressions".
4197
4198 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4199 have it.
4200
4201 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4202 "configure" and the latter not, in order to fix a problem somebody had
4203 with compiling the Arg class on HP-UX.
4204
4205 (e) Improve the error-handling of the C++ wrapper a little bit.
4206
4207 (f) New tests for checking recursion limiting.
4208
4209 8. The pcre_memmove() function, which is used only if the environment does not
4210 have a standard memmove() function (and is therefore rarely compiled),
4211 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4212 returning a result (though PCRE never actually uses the result).
4213
4214 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4215 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4216 returned instead of calling malloc() with an overflowing number that would
4217 most likely cause subsequent chaos.
4218
4219 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4220
4221 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4222 with this option is matched, the nmatch and pmatch options of regexec() are
4223 ignored.
4224
4225 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4226 provided in case anyone wants to the the POSIX interface with UTF-8
4227 strings.
4228
4229 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4230 C++ linking (needed for some HP-UX environments).
4231
4232 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4233 (unused parameter) and in the pcre_printint() function (omitted "default"
4234 switch label when the default is to do nothing).
4235
4236 15. Added some code to make it possible, when PCRE is compiled as a C++
4237 library, to replace subject pointers for pcre_exec() with a smart pointer
4238 class, thus making it possible to process discontinuous strings.
4239
4240 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4241 much the same function. They were added by different people who were trying
4242 to make PCRE easy to compile on non-Unix systems. It has been suggested
4243 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4244 for compiling on Windows systems. I have therefore replaced it with
4245 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4246 defaults to "extern" for C or "extern C" for C++, which works fine on
4247 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4248 SCOPE with something explicit in config.h. In addition:
4249
4250 (a) pcreposix.h still had just "extern" instead of either of these macros;
4251 I have replaced it with PCRE_DATA_SCOPE.
4252
4253 (b) Functions such as _pcre_xclass(), which are internal to the library,
4254 but external in the C sense, all had PCRE_EXPORT in their definitions.
4255 This is apparently wrong for the Windows case, so I have removed it.
4256 (It makes no difference on Unix-like systems.)
4257
4258 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4259 of recursive calls to match(). This is different to MATCH_LIMIT because
4260 that limits the total number of calls to match(), not all of which increase
4261 the depth of recursion. Limiting the recursion depth limits the amount of
4262 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4263 when PCRE is compiled, and changed at run time. A patch from Google adds
4264 this functionality to the C++ interface.
4265
4266 18. Changes to the handling of Unicode character properties:
4267
4268 (a) Updated the table to Unicode 4.1.0.
4269
4270 (b) Recognize characters that are not in the table as "Cn" (undefined).
4271
4272 (c) I revised the way the table is implemented to a much improved format
4273 which includes recognition of ranges. It now supports the ranges that
4274 are defined in UnicodeData.txt, and it also amalgamates other
4275 characters into ranges. This has reduced the number of entries in the
4276 table from around 16,000 to around 3,000, thus reducing its size
4277 considerably. I realized I did not need to use a tree structure after
4278 all - a binary chop search is just as efficient. Having reduced the
4279 number of entries, I extended their size from 6 bytes to 8 bytes to
4280 allow for more data.
4281
4282 (d) Added support for Unicode script names via properties such as \p{Han}.
4283
4284 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4285 matching that character.
4286
4287 20. When matching a repeated Unicode property with a minimum greater than zero,
4288 (for example \pL{2,}), PCRE could look past the end of the subject if it
4289 reached it while seeking the minimum number of characters. This could
4290 happen only if some of the characters were more than one byte long, because
4291 there is a check for at least the minimum number of bytes.
4292
4293 21. Refactored the implementation of \p and \P so as to be more general, to
4294 allow for more different types of property in future. This has changed the
4295 compiled form incompatibly. Anybody with saved compiled patterns that use
4296 \p or \P will have to recompile them.
4297
4298 22. Added "Any" and "L&" to the supported property types.
4299
4300 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4301 but give a compile time error if the value is greater than 0xff.
4302
4303 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4304 accidentally not being installed or uninstalled.
4305
4306 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4307 made were to insert the current release number. This seemed silly, because
4308 it made things harder for people building PCRE on systems that don't run
4309 "configure". I have turned pcre.h into a distributed file, no longer built
4310 by "configure", with the version identification directly included. There is
4311 no longer a pcre.h.in file.
4312
4313 However, this change necessitated a change to the pcre-config script as
4314 well. It is built from pcre-config.in, and one of the substitutions was the
4315 release number. I have updated configure.ac so that ./configure now finds
4316 the release number by grepping pcre.h.
4317
4318 26. Added the ability to run the tests under valgrind.
4319
4320
4321 Version 6.4 05-Sep-05
4322 ---------------------
4323
4324 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4325 "--" to be printed when multiple files were scanned, even when none of the
4326 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4327 consider it to be a bug, and have restored the previous behaviour.
4328
4329 2. A couple of code tidies to get rid of compiler warnings.
4330
4331 3. The pcretest program used to cheat by referring to symbols in the library
4332 whose names begin with _pcre_. These are internal symbols that are not
4333 really supposed to be visible externally, and in some environments it is
4334 possible to suppress them. The cheating is now confined to including
4335 certain files from the library's source, which is a bit cleaner.
4336
4337 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4338 file's purpose clearer.
4339
4340 5. Reorganized pcre_ucp_findchar().
4341
4342
4343 Version 6.3 15-Aug-05
4344 ---------------------
4345
4346 1. The file libpcre.pc.in did not have general read permission in the tarball.
4347
4348 2. There were some problems when building without C++ support:
4349
4350 (a) If C++ support was not built, "make install" and "make test" still
4351 tried to test it.
4352
4353 (b) There were problems when the value of CXX was explicitly set. Some
4354 changes have been made to try to fix these, and ...
4355
4356 (c) --disable-cpp can now be used to explicitly disable C++ support.
4357
4358 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4359 backslash in a target when C++ was disabled. This confuses some
4360 versions of "make", apparently. Using an intermediate variable solves
4361 this. (Same for CPP_LOBJ.)
4362
4363 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4364 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4365 necessary on certain architectures.
4366
4367 4. Added a setting of -export-symbols-regex to the link command to remove
4368 those symbols that are exported in the C sense, but actually are local
4369 within the library, and not documented. Their names all begin with
4370 "_pcre_". This is not a perfect job, because (a) we have to except some
4371 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4372 available (and never for static libraries). I have made a note to try to
4373 find a way round (a) in the future.
4374
4375
4376 Version 6.2 01-Aug-05
4377 ---------------------
4378
4379 1. There was no test for integer overflow of quantifier values. A construction
4380 such as {1111111111111111} would give undefined results. What is worse, if
4381 a minimum quantifier for a parenthesized subpattern overflowed and became
4382 negative, the calculation of the memory size went wrong. This could have
4383 led to memory overwriting.
4384
4385 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4386
4387 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4388 operating environments where this matters.
4389
4390 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4391 PCRE options from within the C++ wrapper.
4392
4393 5. Named capturing subpatterns were not being correctly counted when a pattern
4394 was compiled. This caused two problems: (a) If there were more than 100
4395 such subpatterns, the calculation of the memory needed for the whole
4396 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4397 back references of the form \12, where the number was greater than 9, were
4398 not recognized as back references, even though there were sufficient
4399 previous subpatterns.
4400
4401 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4402 versions of gcc, e.g. 2.95.4.
4403
4404
4405 Version 6.1 21-Jun-05
4406 ---------------------
4407
4408 1. There was one reference to the variable "posix" in pcretest.c that was not
4409 surrounded by "#if !defined NOPOSIX".
4410
4411 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4412 the cross-check on the old pcre_info() function, for the benefit of the
4413 cut-down version of PCRE that is currently imported into Exim.
4414
4415 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4416 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4417 patterns that start (?i) but I don't think that matters. The use of (?i) is
4418 just an example; this all applies to the other options as well.
4419
4420 4. Since libtool seems to echo the compile commands it is issuing, the output
4421 from "make" can be reduced a bit by putting "@" in front of each libtool
4422 compile command.
4423
4424 5. Patch from the folks at Google for configure.in to be a bit more thorough
4425 in checking for a suitable C++ installation before trying to compile the
4426 C++ stuff. This should fix a reported problem when a compiler was present,
4427 but no suitable headers.
4428
4429 6. The man pages all had just "PCRE" as their title. I have changed them to
4430 be the relevant file name. I have also arranged that these names are
4431 retained in the file doc/pcre.txt, which is a concatenation in text format
4432 of all the man pages except the little individual ones for each function.
4433
4434 7. The NON-UNIX-USE file had not been updated for the different set of source
4435 files that come with release 6. I also added a few comments about the C++
4436 wrapper.
4437
4438
4439 Version 6.0 07-Jun-05
4440 ---------------------
4441
4442 1. Some minor internal re-organization to help with my DFA experiments.
4443
4444 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4445 didn't matter for the library itself when fully configured, but did matter
4446 when compiling without UCP support, or within Exim, where the ucp files are
4447 not imported.
4448
4449 3. Refactoring of the library code to split up the various functions into
4450 different source modules. The addition of the new DFA matching code (see
4451 below) to a single monolithic source would have made it really too
4452 unwieldy, quite apart from causing all the code to be include in a
4453 statically linked application, when only some functions are used. This is
4454 relevant even without the DFA addition now that patterns can be compiled in
4455 one application and matched in another.
4456
4457 The downside of splitting up is that there have to be some external
4458 functions and data tables that are used internally in different modules of
4459 the library but which are not part of the API. These have all had their
4460 names changed to start with "_pcre_" so that they are unlikely to clash
4461 with other external names.
4462
4463 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4464 a different (DFA) algorithm. Although it is slower than the original
4465 function, it does have some advantages for certain types of matching
4466 problem.
4467
4468 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4469 including restarting after a partial match.
4470
4471 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4472 defined when compiling for Windows was sent to me. I have put it into the
4473 code, though I have no means of testing or verifying it.
4474
4475 7. Added the pcre_refcount() auxiliary function.
4476
4477 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4478 match before or at the first newline in the subject string. In pcretest,
4479 the /f option on a pattern can be used to set this.
4480
4481 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4482 would behave wrongly. This has been present in PCRE since release 4.0.
4483
4484 10. A number of changes to the pcregrep command:
4485
4486 (a) Refactored how -x works; insert ^(...)$ instead of setting
4487 PCRE_ANCHORED and checking the length, in preparation for adding
4488 something similar for -w.
4489
4490 (b) Added the -w (match as a word) option.
4491
4492 (c) Refactored the way lines are read and buffered so as to have more
4493 than one at a time available.
4494
4495 (d) Implemented a pcregrep test script.
4496
4497 (e) Added the -M (multiline match) option. This allows patterns to match
4498 over several lines of the subject. The buffering ensures that at least
4499 8K, or the rest of the document (whichever is the shorter) is available
4500 for matching (and similarly the previous 8K for lookbehind assertions).
4501
4502 (f) Changed the --help output so that it now says
4503
4504 -w, --word-regex(p)
4505
4506 instead of two lines, one with "regex" and the other with "regexp"
4507 because that confused at least one person since the short forms are the
4508 same. (This required a bit of code, as the output is generated
4509 automatically from a table. It wasn't just a text change.)
4510
4511 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4512 option but starts with a hyphen. Could be a pattern or a path name
4513 starting with a hyphen, for instance.
4514
4515 (h) "-" can be given as a file name to represent stdin.
4516
4517 (i) When file names are being printed, "(standard input)" is used for
4518 the standard input, for compatibility with GNU grep. Previously
4519 "<stdin>" was used.
4520
4521 (j) The option --label=xxx can be used to supply a name to be used for
4522 stdin when file names are being printed. There is no short form.
4523
4524 (k) Re-factored the options decoding logic because we are going to add
4525 two more options that take data. Such options can now be given in four
4526 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4527
4528 (l) Added the -A, -B, and -C options for requesting that lines of context
4529 around matches be printed.
4530
4531 (m) Added the -L option to print the names of files that do not contain
4532 any matching lines, that is, the complement of -l.
4533
4534 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4535 continue to scan other files.
4536
4537 (o) The -s option was incorrectly implemented. For compatibility with other
4538 greps, it now suppresses the error message for a non-existent or non-
4539 accessible file (but not the return code). There is a new option called
4540 -q that suppresses the output of matching lines, which was what -s was
4541 previously doing.
4542
4543 (p) Added --include and --exclude options to specify files for inclusion
4544 and exclusion when recursing.
4545
4546 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4547 Hopefully, it now does.
4548
4549 12. Missing cast in pcre_study().
4550
4551 13. Added an "uninstall" target to the makefile.
4552
4553 14. Replaced "extern" in the function prototypes in Makefile.in with
4554 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4555 world, but is set differently for Windows.
4556
4557 15. Added a second compiling function called pcre_compile2(). The only
4558 difference is that it has an extra argument, which is a pointer to an
4559 integer error code. When there is a compile-time failure, this is set
4560 non-zero, in addition to the error test pointer being set to point to an
4561 error message. The new argument may be NULL if no error number is required
4562 (but then you may as well call pcre_compile(), which is now just a
4563 wrapper). This facility is provided because some applications need a
4564 numeric error indication, but it has also enabled me to tidy up the way
4565 compile-time errors are handled in the POSIX wrapper.
4566
4567 16. Added VPATH=.libs to the makefile; this should help when building with one
4568 prefix path and installing with another. (Or so I'm told by someone who
4569 knows more about this stuff than I do.)
4570
4571 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4572 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4573 match everything, including newlines. This is not POSIX-compatible, but
4574 somebody wanted the feature. From pcretest it can be activated by using
4575 both the P and the s flags.
4576
4577 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4578
4579 19. libpcre.pc was being incorrectly installed as executable.
4580
4581 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4582 it now also looks for '\r' so that it will work unmodified on Windows.
4583
4584 21. Added Google's contributed C++ wrapper to the distribution.
4585
4586 22. Added some untidy missing memory free() calls in pcretest, to keep
4587 Electric Fence happy when testing.
4588
4589
4590
4591 Version 5.0 13-Sep-04
4592 ---------------------
4593
4594 1. Internal change: literal characters are no longer packed up into items
4595 containing multiple characters in a single byte-string. Each character
4596 is now matched using a separate opcode. However, there may be more than one
4597 byte in the character in UTF-8 mode.
4598
4599 2. The pcre_callout_block structure has two new fields: pattern_position and
4600 next_item_length. These contain the offset in the pattern to the next match
4601 item, and its length, respectively.
4602
4603 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4604 insertion of callouts before each pattern item. Added the /C option to
4605 pcretest to make use of this.
4606
4607 4. On the advice of a Windows user, the lines
4608
4609 #if defined(_WIN32) || defined(WIN32)
4610 _setmode( _fileno( stdout ), 0x8000 );
4611 #endif /* defined(_WIN32) || defined(WIN32) */
4612
4613 have been added to the source of pcretest. This apparently does useful
4614 magic in relation to line terminators.
4615
4616 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4617 for the benefit of those environments where the "b" makes a difference.
4618
4619 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4620 to know about it. I have put a hack into configure.in that adds in code
4621 to set GCC=yes if CC=icc. This seems to end up at a point in the
4622 generated configure script that is early enough to affect the setting of
4623 compiler options, which is what is needed, but I have no means of testing
4624 whether it really works. (The user who reported this had patched the
4625 generated configure script, which of course I cannot do.)
4626
4627 LATER: After change 22 below (new libtool files), the configure script
4628 seems to know about icc (and also ecc). Therefore, I have commented out
4629 this hack in configure.in.
4630
4631 7. Added support for pkg-config (2 patches were sent in).
4632
4633 8. Negated POSIX character classes that used a combination of internal tables
4634 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4635 [[:^ascii]]. Typically, they would match almost any characters. The other
4636 POSIX classes were not broken in this way.
4637
4638 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4639 to find the match, as PCRE was deluded into thinking that the match had to
4640 start at the start point or following a newline. The same bug applied to
4641 patterns with negative forward assertions or any backward assertions
4642 preceding ".*" at the start, unless the pattern required a fixed first
4643 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4644
4645 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4646 starting at the last subject character, bytes beyond the end of the subject
4647 string were read.
4648
4649 11. Renamed the variable "class" as "classbits" to make life easier for C++
4650 users. (Previously there was a macro definition, but it apparently wasn't
4651 enough.)
4652
4653 12. Added the new field "tables" to the extra data so that tables can be passed
4654 in at exec time, or the internal tables can be re-selected. This allows
4655 a compiled regex to be saved and re-used at a later time by a different
4656 program that might have everything at different addresses.
4657
4658 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4659 -R library as well as a -L library.
4660
4661 14. The debugging options of pcretest (-d on the command line or D on a
4662 pattern) showed incorrect output for anything following an extended class
4663 that contained multibyte characters and which was followed by a quantifier.
4664
4665 15. Added optional support for general category Unicode character properties
4666 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4667 support. It adds about 90K to the size of the library. The meanings of the
4668 inbuilt class escapes such as \d and \s have NOT been changed.
4669
4670 16. Updated pcredemo.c to include calls to free() to release the memory for the
4671 compiled pattern.
4672
4673 17. The generated file chartables.c was being created in the source directory
4674 instead of in the building directory. This caused the build to fail if the
4675 source directory was different from the building directory, and was
4676 read-only.
4677
4678 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4679 file. No doubt somebody will tell me if they don't make sense... Also added
4680 Dan Mooney's comments about building on OpenVMS.
4681
4682 19. Added support for partial matching via the PCRE_PARTIAL option for
4683 pcre_exec() and the \P data escape in pcretest.
4684
4685 20. Extended pcretest with 3 new pattern features:
4686
4687 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4688 write the compiled pattern to the file whose name is "rest-of-line".
4689 This is a straight binary dump of the data, with the saved pointer to
4690 the character tables forced to be NULL. The study data, if any, is
4691 written too. After writing, pcretest reads a new pattern.
4692
4693 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4694 compiled pattern from the given file. There must not be any
4695 occurrences of "<" in the file name (pretty unlikely); if there are,
4696 pcretest will instead treat the initial "<" as a pattern delimiter.
4697 After reading in the pattern, pcretest goes on to read data lines as
4698 usual.
4699
4700 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4701 and 16-bit fields in a compiled pattern, to simulate a pattern that
4702 was compiled on a host of opposite endianness.
4703
4704 21. The pcre-exec() function can now cope with patterns that were compiled on
4705 hosts of opposite endianness, with this restriction:
4706
4707 As for any compiled expression that is saved and used later, the tables
4708 pointer field cannot be preserved; the extra_data field in the arguments
4709 to pcre_exec() should be used to pass in a tables address if a value
4710 other than the default internal tables were used at compile time.
4711
4712 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4713 now diagnosed as an error. Previously, most of the time, a negative number
4714 would have been treated as zero, but if in addition "ovector" was passed as
4715 NULL, a crash could occur.
4716
4717 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4718 new versions from the libtool 1.5 distribution (the last one is a copy of
4719 a file called libtool.m4). This seems to have fixed the need to patch
4720 "configure" to support Darwin 1.3 (which I used to do). However, I still
4721 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4722 workstation).
4723
4724 24. Changed the PCRE licence to be the more standard "BSD" licence.
4725
4726
4727 Version 4.5 01-Dec-03
4728 ---------------------
4729
4730 1. There has been some re-arrangement of the code for the match() function so
4731 that it can be compiled in a version that does not call itself recursively.
4732 Instead, it keeps those local variables that need separate instances for
4733 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4734 needs to "recurse". Keeping track of where control must go is done by means
4735 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4736 hide most of the details from the main code, and operates only if
4737 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4738 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4739 operating.
4740
4741 To make it easier for callers to provide specially tailored get/free
4742 functions for this usage, two new functions, pcre_stack_malloc, and
4743 pcre_stack_free, are used. They are always called in strict stacking order,
4744 and the size of block requested is always the same.
4745
4746 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4747 PCRE has been compiled to use the stack or the heap for recursion. The
4748 -C option of pcretest uses this to show which version is compiled.
4749
4750 A new data escape \S, is added to pcretest; it causes the amounts of store
4751 obtained and freed by both kinds of malloc/free at match time to be added
4752 to the output.
4753
4754 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4755 what's available on my current Linux desktop machine.
4756
4757 3. When matching a UTF-8 string, the test for a valid string at the start has
4758 been extended. If start_offset is not zero, PCRE now checks that it points
4759 to a byte that is the start of a UTF-8 character. If not, it returns
4760 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4761 this is necessary because there may be backward assertions in the pattern.
4762 When matching the same subject several times, it may save resources to use
4763 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4764
4765 4. The code for checking the validity of UTF-8 strings has been tightened so
4766 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4767 containing "overlong sequences".
4768
4769 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4770 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4771 should have been "&", but it just so happened that all the cases this let
4772 through by mistake were picked up later in the function.
4773
4774 6. I had used a variable called "isblank" - this is a C99 function, causing
4775 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4776
4777 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4778 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4779 so the version is shown; (c) stop "make test" echoing "./RunTest".
4780
4781 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4782
4783 9. The prototype for memmove() for systems that don't have it was using
4784 size_t, but the inclusion of the header that defines size_t was later. I've
4785 moved the #includes for the C headers earlier to avoid this.
4786
4787 10. Added some adjustments to the code to make it easier to compiler on certain
4788 special systems:
4789
4790 (a) Some "const" qualifiers were missing.
4791 (b) Added the macro EXPORT before all exported functions; by default this
4792 is defined to be empty.
4793 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4794 that it reads its output file name as an argument instead of writing
4795 to the standard output and assuming this can be redirected.
4796
4797 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4798 class containing characters with values greater than 255, PCRE compilation
4799 went into a loop.
4800
4801 12. A recursive reference to a subpattern that was within another subpattern
4802 that had a minimum quantifier of zero caused PCRE to crash. For example,
4803 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4804 recursion. If the recursively-called subpattern itself had a zero repeat,
4805 that was OK.
4806
4807 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4808 buffer into which it was copied (for escape processing) was still set at
4809 1024, so long lines caused crashes.
4810
4811 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4812 "internal error: code overflow...". This applied to any character class
4813 that was followed by a possessive quantifier.
4814
4815 15. Modified the Makefile to add libpcre.la as a prerequisite for
4816 libpcreposix.la because I was told this is needed for a parallel build to
4817 work.
4818
4819 16. If a pattern that contained .* following optional items at the start was
4820 studied, the wrong optimizing data was generated, leading to matching
4821 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4822 matching string must start with a or b or c. The correct conclusion for
4823 this pattern is that a match can start with any character.
4824
4825
4826 Version 4.4 13-Aug-03
4827 ---------------------
4828
4829 1. In UTF-8 mode, a character class containing characters with values between
4830 127 and 255 was not handled correctly if the compiled pattern was studied.
4831 In fixing this, I have also improved the studying algorithm for such
4832 classes (slightly).
4833
4834 2. Three internal functions had redundant arguments passed to them. Removal
4835 might give a very teeny performance improvement.
4836
4837 3. Documentation bug: the value of the capture_top field in a callout is *one
4838 more than* the number of the hightest numbered captured substring.
4839
4840 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4841 in incorrectly linking with a previously installed version. They now link
4842 explicitly with libpcre.la.
4843
4844 5. configure.in no longer needs to recognize Cygwin specially.
4845
4846 6. A problem in pcre.in for Windows platforms is fixed.
4847
4848 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4849 pcretest, it used to include the size of the study block as part of its
4850 output. Unfortunately, the structure contains a field that has a different
4851 size on different hardware architectures. This meant that the tests that
4852 showed this size failed. As the block is currently always of a fixed size,
4853 this information isn't actually particularly useful in pcretest output, so
4854 I have just removed it.
4855
4856 8. Three pre-processor statements accidentally did not start in column 1.
4857 Sadly, there are *still* compilers around that complain, even though
4858 standard C has not required this for well over a decade. Sigh.
4859
4860 9. In pcretest, the code for checking callouts passed small integers in the
4861 callout_data field, which is a void * field. However, some picky compilers
4862 complained about the casts involved for this on 64-bit systems. Now
4863 pcretest passes the address of the small integer instead, which should get
4864 rid of the warnings.
4865
4866 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4867 both compile and run time, and gives an error if an invalid UTF-8 sequence
4868 is found. There is a option for disabling this check in cases where the
4869 string is known to be correct and/or the maximum performance is wanted.
4870
4871 11. In response to a bug report, I changed one line in Makefile.in from
4872
4873 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4874 to
4875 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4876
4877 to look similar to other lines, but I have no way of telling whether this
4878 is the right thing to do, as I do not use Windows. No doubt I'll get told
4879 if it's wrong...
4880
4881
4882 Version 4.3 21-May-03
4883 ---------------------
4884
4885 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4886 Makefile.
4887
4888 2. Some refactoring to improve the quality of the code:
4889
4890 (i) The utf8_table... variables are now declared "const".
4891
4892 (ii) The code for \cx, which used the "case flipping" table to upper case
4893 lower case letters, now just substracts 32. This is ASCII-specific,
4894 but the whole concept of \cx is ASCII-specific, so it seems
4895 reasonable.
4896
4897 (iii) PCRE was using its character types table to recognize decimal and
4898 hexadecimal digits in the pattern. This is silly, because it handles
4899 only 0-9, a-f, and A-F, but the character types table is locale-
4900 specific, which means strange things might happen. A private
4901 table is now used for this - though it costs 256 bytes, a table is
4902 much faster than multiple explicit tests. Of course, the standard
4903 character types table is still used for matching digits in subject
4904 strings against \d.
4905
4906 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4907 ending in _t are). So I've renamed it as ESC_tee.
4908
4909 3. The first argument for regexec() in the POSIX wrapper should have been
4910 defined as "const".
4911
4912 4. Changed pcretest to use malloc() for its buffers so that they can be
4913 Electric Fenced for debugging.
4914
4915 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4916 to read one or more bytes before the start of the subject string. Often this
4917 had no effect on PCRE's behaviour, but in some circumstances it could
4918 provoke a segmentation fault.
4919
4920 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4921 to try to read one or more bytes before the start of the subject string.
4922
4923 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4924 UTF-8 support could misbehave in various ways if the subject string
4925 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4926 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4927 back over UTF-8 characters.)
4928
4929
4930 Version 4.2 14-Apr-03
4931 ---------------------
4932
4933 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4934
4935 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4936 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4937 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4938 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4939 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4940 and BUILD_EXEEXT
4941 Note: automatic setting of the BUILD variables is not yet working
4942 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4943 compile-time but not at link-time
4944 [LINK]: use for linking executables only
4945 make different versions for Windows and non-Windows
4946 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4947 libraries
4948 [LINK_FOR_BUILD]: new variable
4949 [OBJEXT]: use throughout
4950 [EXEEXT]: use throughout
4951 <winshared>: new target
4952 <wininstall>: new target
4953 <dftables.o>: use native compiler
4954 <dftables>: use native linker
4955 <install>: handle Windows platform correctly
4956 <clean>: ditto
4957 <check>: ditto
4958 copy DLL to top builddir before testing
4959
4960 As part of these changes, -no-undefined was removed again. This was reported
4961 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4962 in any case.
4963
4964 3. Some tidies to get rid of compiler warnings:
4965
4966 . In the match_data structure, match_limit was an unsigned long int, whereas
4967 match_call_count was an int. I've made them both unsigned long ints.
4968
4969 . In pcretest the fact that a const uschar * doesn't automatically cast to
4970 a void * provoked a warning.
4971
4972 . Turning on some more compiler warnings threw up some "shadow" variables
4973 and a few more missing casts.
4974
4975 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4976 option, a class that contained a single character with a value between 128
4977 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4978
4979 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4980 option, a class that contained several characters, but with at least one
4981 whose value was between 128 and 255 caused PCRE to crash.
4982
4983
4984 Version 4.1 12-Mar-03
4985 ---------------------
4986
4987 1. Compiling with gcc -pedantic found a couple of places where casts were
4988 needed, and a string in dftables.c that was longer than standard compilers are
4989 required to support.
4990
4991 2. Compiling with Sun's compiler found a few more places where the code could
4992 be tidied up in order to avoid warnings.
4993
4994 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4995 first of these names is deprecated in the latest Autoconf in favour of the name
4996 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4997 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4998 analogy I have changed it to CFLAGS_FOR_BUILD.
4999
5000 4. Added -no-undefined to the linking command in the Makefile, because this is
5001 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
5002 linking step for the pcreposix library.
5003
5004 5. PCRE was failing to diagnose the case of two named groups with the same
5005 name.
5006
5007 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
5008 literal character that is needed in the subject for a match, and scans along to
5009 ensure that it is present before embarking on the full matching process. This
5010 saves time in cases of nested unlimited repeats that are never going to match.
5011 Problem: the scan can take a lot of time if the subject is very long (e.g.
5012 megabytes), thus penalizing straightforward matches. It is now done only if the
5013 amount of subject to be scanned is less than 1000 bytes.
5014
5015 7. A lesser problem with the same optimization is that it was recording the
5016 first character of an anchored pattern as "needed", thus provoking a search
5017 right along the subject, even when the first match of the pattern was going to
5018 fail. The "needed" character is now not set for anchored patterns, unless it
5019 follows something in the pattern that is of non-fixed length. Thus, it still
5020 fulfils its original purpose of finding quick non-matches in cases of nested
5021 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
5022
5023
5024 Version 4.0 17-Feb-03
5025 ---------------------
5026
5027 1. If a comment in an extended regex that started immediately after a meta-item
5028 extended to the end of string, PCRE compiled incorrect data. This could lead to
5029 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
5030
5031 2. Moved to autoconf 2.53 and libtool 1.4.2.
5032
5033 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
5034 the special perltest8 script is no longer needed - all the tests can be run
5035 from a single perltest script.
5036
5037 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
5038 by \s. It has now been removed in PCRE. This means it isn't recognized as
5039 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
5040 class [:space:] *does* include VT, thereby creating a mess.
5041
5042 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
5043 space and tab.
5044
5045 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
5046 its new features into the main test script, reducing the number of scripts.
5047
5048 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
5049 were backward compatible, and made the (?i) apply to the whole pattern, as if
5050 /i were given. Now it behaves more logically, and applies the option setting
5051 only to what follows. PCRE has been changed to follow suit. However, if it
5052 finds options settings right at the start of the pattern, it extracts them into
5053 the global options, as before. Thus, they show up in the info data.
5054
5055 8. Added support for the \Q...\E escape sequence. Characters in between are
5056 treated as literals. This is slightly different from Perl in that $ and @ are
5057 also handled as literals inside the quotes. In Perl, they will cause variable
5058 interpolation. Note the following examples:
5059
5060 Pattern PCRE matches Perl matches
5061
5062 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
5063 \Qabc\$xyz\E abc\$xyz abc\$xyz
5064 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
5065
5066 For compatibility with Perl, \Q...\E sequences are recognized inside character
5067 classes as well as outside them.
5068
5069 9. Re-organized 3 code statements in pcretest to avoid "overflow in
5070 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
5071 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
5072 signed/unsigned warnings.
5073
5074 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
5075 option for pcretest, so I've replaced it by a simple function that does just
5076 that job.
5077
5078 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
5079 "pcregrep -".
5080
5081 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
5082 Java package. This provides some syntactic sugar for simple cases of what my
5083 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
5084 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
5085 item, you can use this simplified notation. Note that only makes sense with
5086 greedy quantifiers. Consequently, the use of the possessive quantifier forces
5087 greediness, whatever the setting of the PCRE_UNGREEDY option.
5088
5089 13. A change of greediness default within a pattern was not taking effect at
5090 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
5091 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
5092 was abstracted outside.
5093
5094 14. PCRE now supports the \G assertion. It is true when the current matching
5095 position is at the start point of the match. This differs from \A when the
5096 starting offset is non-zero. Used with the /g option of pcretest (or similar
5097 code), it works in the same way as it does for Perl's /g option. If all
5098 alternatives of a regex begin with \G, the expression is anchored to the start
5099 match position, and the "anchored" flag is set in the compiled expression.
5100
5101 15. Some bugs concerning the handling of certain option changes within patterns
5102 have been fixed. These applied to options other than (?ims). For example,
5103 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
5104 been the other way round. Some of this was related to change 7 above.
5105
5106 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
5107 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
5108 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
5109 POSIX classes only within a class (e.g. /[[:alpha:]]/).
5110
5111 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
5112 mode. Unlike ".", it always matches newline, whatever the setting of
5113 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
5114 assertions. Perl allows it, but it doesn't (in general) work because it can't
5115 calculate the length of the lookbehind. At least, that's the case for Perl
5116 5.8.0 - I've been told they are going to document that it doesn't work in
5117 future.
5118
5119 18. Added an error diagnosis for escapes that PCRE does not support: these are
5120 \L, \l, \N, \P, \p, \U, \u, and \X.
5121
5122 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
5123 reading past the end of the pattern in cases such as /[abcd/.
5124
5125 20. PCRE was getting more memory than necessary for patterns with classes that
5126 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
5127
5128 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
5129 compiling PCRE for use with Virtual Pascal.
5130
5131 22. Small fix to the Makefile to make it work properly if the build is done
5132 outside the source tree.
5133
5134 23. Added a new extension: a condition to go with recursion. If a conditional
5135 subpattern starts with (?(R) the "true" branch is used if recursion has
5136 happened, whereas the "false" branch is used only at the top level.
5137
5138 24. When there was a very long string of literal characters (over 255 bytes
5139 without UTF support, over 250 bytes with UTF support), the computation of how
5140 much memory was required could be incorrect, leading to segfaults or other
5141 strange effects.
5142
5143 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5144 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5145 there was a subsequent back reference to those brackets. This meant that, for
5146 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5147 possible to check for precisely this case. All we can do is abandon the
5148 optimization if .* occurs inside capturing brackets when there are any back
5149 references whatsoever. (See below for a better fix that came later.)
5150
5151 26. The handling of the optimization for finding the first character of a
5152 non-anchored pattern, and for finding a character that is required later in the
5153 match were failing in some cases. This didn't break the matching; it just
5154 failed to optimize when it could. The way this is done has been re-implemented.
5155
5156 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5157
5158 28. Added a new feature that provides some of the functionality that Perl
5159 provides with (?{...}). The facility is termed a "callout". The way it is done
5160 in PCRE is for the caller to provide an optional function, by setting
5161 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5162 global variable. By default it is unset, which disables all calling out. To get
5163 the function called, the regex must include (?C) at appropriate points. This
5164 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5165 This provides a means of identifying different callout points. When PCRE
5166 reaches such a point in the regex, if pcre_callout has been set, the external
5167 function is called. It is provided with data in a structure called
5168 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5169 matching continues; if it returns a non-zero value, the match at the current
5170 point fails. However, backtracking will occur if possible. [This was changed
5171 later and other features added - see item 49 below.]
5172
5173 29. pcretest is upgraded to test the callout functionality. It provides a
5174 callout function that displays information. By default, it shows the start of
5175 the match and the current position in the text. There are some new data escapes
5176 to vary what happens:
5177
5178 \C+ in addition, show current contents of captured substrings
5179 \C- do not supply a callout function
5180 \C!n return 1 when callout number n is reached
5181 \C!n!m return 1 when callout number n is reached for the mth time
5182
5183 30. If pcregrep was called with the -l option and just a single file name, it
5184 output "<stdin>" if a match was found, instead of the file name.
5185
5186 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5187 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5188 pcre_exec(). This saves a malloc/free per call. The default value of
5189 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5190 when configuring.
5191
5192 32. The default maximum size of a compiled pattern is 64K. There have been a
5193 few cases of people hitting this limit. The code now uses macros to handle the
5194 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5195 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5196 configuring. Tests 2 and 5 work only with 2-byte links because they output
5197 debugging information about compiled patterns.
5198
5199 33. Internal code re-arrangements:
5200
5201 (a) Moved the debugging function for printing out a compiled regex into
5202 its own source file (printint.c) and used #include to pull it into
5203 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5204 separate copies.
5205
5206 (b) Defined the list of op-code names for debugging as a macro in
5207 internal.h so that it is next to the definition of the opcodes.
5208
5209 (c) Defined a table of op-code lengths for simpler skipping along compiled
5210 code. This is again a macro in internal.h so that it is next to the
5211 definition of the opcodes.
5212
5213 34. Added support for recursive calls to individual subpatterns, along the
5214 lines of Robin Houston's patch (but implemented somewhat differently).
5215
5216 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5217 allow it to read and process whole directories in Win32. This code was
5218 contributed by Lionel Fourquaux; it has not been tested by me.
5219
5220 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5221 used to name a group. Names consist of alphanumerics and underscores, and must
5222 be unique. Back references use the syntax (?P=name) and recursive calls use
5223 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5224 numbers. The function pcre_fullinfo() can be used after compilation to extract
5225 a name/number map. There are three relevant calls:
5226
5227 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5228 PCRE_INFO_NAMECOUNT yields the number of entries
5229 PCRE_INFO_NAMETABLE yields a pointer to the map.
5230
5231 The map is a vector of fixed-size entries. The size of each entry depends on
5232 the length of the longest name used. The first two bytes of each entry are the
5233 group number, most significant byte first. There follows the corresponding
5234 name, zero terminated. The names are in alphabetical order.
5235
5236 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5237 case instead of 255. Making it the same both with and without UTF-8 support
5238 means that the same test output works with both.
5239
5240 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5241 calling malloc() with a zero argument.
5242
5243 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5244 optimization. I've improved things by keeping a bitmap of backreferences with
5245 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5246 fact referenced, the optimization can be applied. It is unlikely that a
5247 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5248 the match to follow \n) will appear inside brackets with a number greater than
5249 31, but if it does, any back reference > 31 suppresses the optimization.
5250
5251 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5252 of disabling numbered capturing parentheses. Any opening parenthesis that is
5253 not followed by ? behaves as if it were followed by ?: but named parentheses
5254 can still be used for capturing (and they will acquire numbers in the usual
5255 way).
5256
5257 41. Redesigned the return codes from the match() function into yes/no/error so
5258 that errors can be passed back from deep inside the nested calls. A malloc
5259 failure while inside a recursive subpattern call now causes the
5260 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5261
5262 42. It is now possible to set a limit on the number of times the match()
5263 function is called in a call to pcre_exec(). This facility makes it possible to
5264 limit the amount of recursion and backtracking, though not in a directly
5265 obvious way, because the match() function is used in a number of different
5266 circumstances. The count starts from zero for each position in the subject
5267 string (for non-anchored patterns). The default limit is, for compatibility, a
5268 large number, namely 10 000 000. You can change this in two ways:
5269
5270 (a) When configuring PCRE before making, you can use --with-match-limit=n
5271 to set a default value for the compiled library.
5272
5273 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5274 a different value is set. See 45 below.
5275
5276 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5277
5278 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5279 of things that can be changed at compile time. The first argument specifies
5280 what is wanted and the second points to where the information is to be placed.
5281 The current list of available information is:
5282
5283 PCRE_CONFIG_UTF8
5284
5285 The output is an integer that is set to one if UTF-8 support is available;
5286 otherwise it is set to zero.
5287
5288 PCRE_CONFIG_NEWLINE
5289
5290 The output is an integer that it set to the value of the code that is used for
5291 newline. It is either LF (10) or CR (13).
5292
5293 PCRE_CONFIG_LINK_SIZE
5294
5295 The output is an integer that contains the number of bytes used for internal
5296 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5297
5298 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5299
5300 The output is an integer that contains the threshold above which the POSIX
5301 interface uses malloc() for output vectors. See item 31 above.
5302
5303 PCRE_CONFIG_MATCH_LIMIT
5304
5305 The output is an unsigned integer that contains the default limit of the number
5306 of match() calls in a pcre_exec() execution. See 42 above.
5307
5308 44. pcretest has been upgraded by the addition of the -C option. This causes it
5309 to extract all the available output from the new pcre_config() function, and to
5310 output it. The program then exits immediately.
5311
5312 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5313 order to support additional features. One way would have been to define
5314 pcre_exec2() (for example) with extra arguments, but this would not have been
5315 extensible, and would also have required all calls to the original function to
5316 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5317 is used for passing in "extra" data from pcre_study().
5318
5319 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5320 contains the following fields:
5321
5322 flags a bitmap indicating which of the following fields are set
5323 study_data opaque data from pcre_study()
5324 match_limit a way of specifying a limit on match() calls for a specific
5325 call to pcre_exec()
5326 callout_data data for callouts (see 49 below)
5327
5328 The flag bits are also defined in pcre.h, and are
5329
5330 PCRE_EXTRA_STUDY_DATA
5331 PCRE_EXTRA_MATCH_LIMIT
5332 PCRE_EXTRA_CALLOUT_DATA
5333
5334 The pcre_study() function now returns one of these new pcre_extra blocks, with
5335 the actual study data pointed to by the study_data field, and the
5336 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5337 before. That is, this change is entirely upwards-compatible and requires no
5338 change to existing code.
5339
5340 If you want to pass in additional data to pcre_exec(), you can either place it
5341 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5342 block.
5343
5344 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5345 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5346 times with different match limits, until it finds the minimum value needed for
5347 pcre_exec() to complete. The value is then output. This can be instructive; for
5348 most simple matches the number is quite small, but for pathological cases it
5349 gets very large very quickly.
5350
5351 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5352 returns the size of the data block pointed to by the study_data field in a
5353 pcre_extra block, that is, the value that was passed as the argument to
5354 pcre_malloc() when PCRE was getting memory in which to place the information
5355 created by pcre_study(). The fourth argument should point to a size_t variable.
5356 pcretest has been extended so that this information is shown after a successful
5357 pcre_study() call when information about the compiled regex is being displayed.
5358
5359 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5360 because what follows is always an absolute path. (Later: it turns out that this
5361 is more than cosmetic for MinGW, because it doesn't like empty path
5362 components.)
5363
5364 49. Some changes have been made to the callout feature (see 28 above):
5365
5366 (i) A callout function now has three choices for what it returns:
5367
5368 0 => success, carry on matching
5369 > 0 => failure at this point, but backtrack if possible
5370 < 0 => serious error, return this value from pcre_exec()
5371
5372 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5373 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5374 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5375 use by callout functions. It will never be used by PCRE itself.
5376
5377 (ii) The pcre_extra structure (see 45 above) has a void * field called
5378 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5379 pcre_callout_block structure has a field of the same name. The contents of
5380 the field passed in the pcre_extra structure are passed to the callout
5381 function in the corresponding field in the callout block. This makes it
5382 easier to use the same callout-containing regex from multiple threads. For
5383 testing, the pcretest program has a new data escape
5384
5385 \C*n pass the number n (may be negative) as callout_data
5386
5387 If the callout function in pcretest receives a non-zero value as
5388 callout_data, it returns that value.
5389
5390 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5391 there were some redundant $(CFLAGS) in commands that are now specified as
5392 $(LINK), which already includes $(CFLAGS).
5393
5394 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5395 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5396 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5397 one-byte characters throughout. Note that case-insensitive matching applies
5398 only to characters whose values are less than 256. PCRE doesn't support the
5399 notion of cases for higher-valued characters.
5400
5401 (i) A character class whose characters are all within 0-255 is handled as
5402 a bit map, and the map is inverted for negative classes. Previously, a
5403 character > 255 always failed to match such a class; however it should
5404 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5405
5406 (ii) A negated character class with a single character < 255 is coded as
5407 "not this character" (OP_NOT). This wasn't working properly when the test
5408 character was multibyte, either singly or repeated.
5409
5410 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5411 mode, for example: \x{100}{2,3}.
5412
5413 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5414 singly or repeated) now correctly test multibyte characters. However,
5415 PCRE doesn't recognize any characters with values greater than 255 as
5416 digits, spaces, or word characters. Such characters always match \D, \S,
5417 and \W, and never match \d, \s, or \w.
5418
5419 (v) Classes may now contain characters and character ranges with values
5420 greater than 255. For example: [ab\x{100}-\x{400}].
5421
5422 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5423 PCRE in UTF-8 mode.
5424
5425 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5426 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5427 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5428 value.)
5429
5430 53. The single man page has become too large. I have therefore split it up into
5431 a number of separate man pages. These also give rise to individual HTML pages;
5432 these are now put in a separate directory, and there is an index.html page that
5433 lists them all. Some hyperlinking between the pages has been installed.
5434
5435 54. Added convenience functions for handling named capturing parentheses.
5436
5437 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5438 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5439 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5440 are faulted.
5441
5442 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5443 calling configure. These values are used when compiling the dftables.c program
5444 which is run to generate the source of the default character tables. They
5445 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5446 you will need to set these values.
5447
5448 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5449
5450
5451 Version 3.9 02-Jan-02
5452 ---------------------
5453
5454 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5455
5456 2. If --disable-static was given, the building process failed when trying to
5457 build pcretest and pcregrep. (For some reason it was using libtool to compile
5458 them, which is not right, as they aren't part of the library.)
5459
5460
5461 Version 3.8 18-Dec-01
5462 ---------------------
5463
5464 1. The experimental UTF-8 code was completely screwed up. It was packing the
5465 bytes in the wrong order. How dumb can you get?
5466
5467
5468 Version 3.7 29-Oct-01
5469 ---------------------
5470
5471 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5472 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5473 this didn't happen under Solaris 8, where I normally test things.
5474
5475 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5476 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5477 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5478 this fix an hour or so after the initial 3.7 release.)
5479
5480
5481 Version 3.6 23-Oct-01
5482 ---------------------
5483
5484 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5485 offsets passed as NULL with zero offset count.
5486
5487 2. The config.guess and config.sub files had not been updated when I moved to
5488 the latest autoconf.
5489
5490
5491 Version 3.5 15-Aug-01
5492 ---------------------
5493
5494 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5495 had been forgotten.
5496
5497 2. By using declared but undefined structures, we can avoid using "void"
5498 definitions in pcre.h while keeping the internal definitions of the structures
5499 private.
5500
5501 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5502 user point of view, this means that both static and shared libraries are built
5503 by default, but this can be individually controlled. More of the work of
5504 handling this static/shared cases is now inside libtool instead of PCRE's make
5505 file.
5506
5507 4. The pcretest utility is now installed along with pcregrep because it is
5508 useful for users (to test regexs) and by doing this, it automatically gets
5509 relinked by libtool. The documentation has been turned into a man page, so
5510 there are now .1, .txt, and .html versions in /doc.
5511
5512 5. Upgrades to pcregrep:
5513 (i) Added long-form option names like gnu grep.
5514 (ii) Added --help to list all options with an explanatory phrase.
5515 (iii) Added -r, --recursive to recurse into sub-directories.
5516 (iv) Added -f, --file to read patterns from a file.
5517
5518 6. pcre_exec() was referring to its "code" argument before testing that
5519 argument for NULL (and giving an error if it was NULL).
5520
5521 7. Upgraded Makefile.in to allow for compiling in a different directory from
5522 the source directory.
5523
5524 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5525 options bits, the pointer it was passed was to an int instead of to an unsigned
5526 long int. This mattered only on 64-bit systems.
5527
5528 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5529 generated) instead of pcre.in, which it its source. Also made the same change
5530 in several of the .c files.
5531
5532 10. A new release of gcc defines printf() as a macro, which broke pcretest
5533 because it had an ifdef in the middle of a string argument for printf(). Fixed
5534 by using separate calls to printf().
5535
5536 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5537 script, to force use of CR or LF instead of \n in the source. On non-Unix
5538 systems, the value can be set in config.h.
5539
5540 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5541 absolute limit. Changed the text of the error message to make this clear, and
5542 likewise updated the man page.
5543
5544 13. The limit of 99 on the number of capturing subpatterns has been removed.
5545 The new limit is 65535, which I hope will not be a "real" limit.
5546
5547
5548 Version 3.4 22-Aug-00
5549 ---------------------
5550
5551 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5552
5553 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5554
5555
5556 Version 3.3 01-Aug-00
5557 ---------------------
5558
5559 1. If an octal character was given, but the value was greater than \377, it
5560 was not getting masked to the least significant bits, as documented. This could
5561 lead to crashes in some systems.
5562
5563 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5564 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5565
5566 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5567 These just pass their arguments on to (pcre_free)(), but they are provided
5568 because some uses of PCRE bind it to non-C systems that can call its functions,
5569 but cannot call free() or pcre_free() directly.
5570
5571 4. Add "make test" as a synonym for "make check". Corrected some comments in
5572 the Makefile.
5573
5574 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5575 Makefile.
5576
5577 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5578 command called pgrep for grepping around the active processes.
5579
5580 7. Added the beginnings of support for UTF-8 character strings.
5581
5582 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5583 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5584 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5585 out for the ar command.)
5586
5587
5588 Version 3.2 12-May-00
5589 ---------------------
5590
5591 This is purely a bug fixing release.
5592
5593 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5594 of ZA. This was just one example of several cases that could provoke this bug,
5595 which was introduced by change 9 of version 2.00. The code for breaking
5596 infinite loops after an iteration that matches an empty string was't working
5597 correctly.
5598
5599 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5600 when matched against abbab (for example). After matching an empty string, it
5601 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5602 caused it to match further down the string than it should.
5603
5604 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5605 was there because it doesn't seem to be needed, and it causes trouble on some
5606 systems, as it is not a Standard C header. It has been removed.
5607
5608 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5609 were reported on the Macintosh. The changes were from
5610
5611 while ((c = *(++ptr)) != 0 && c != '\n');
5612 to
5613 while ((c = *(++ptr)) != 0 && c != '\n') ;
5614
5615 Totally extraordinary, but if that's what it takes...
5616
5617 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5618 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5619 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5620 assumes the way PCRE uses memmove() (always moving upwards).
5621
5622 6. PCRE is being used in one environment where strchr() is not available. There
5623 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5624 faster code anyway.
5625
5626
5627 Version 3.1 09-Feb-00
5628 ---------------------
5629
5630 The only change in this release is the fixing of some bugs in Makefile.in for
5631 the "install" target:
5632
5633 (1) It was failing to install pcreposix.h.
5634
5635 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5636
5637
5638 Version 3.0 01-Feb-00
5639 ---------------------
5640
5641 1. Add support for the /+ modifier to perltest (to output $` like it does in
5642 pcretest).
5643
5644 2. Add support for the /g modifier to perltest.
5645
5646 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5647 matches null strings.
5648
5649 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5650 pattern. Perl treats empty patterns specially - it reuses the most recent
5651 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5652 effect.
5653
5654 5. The POSIX interface was broken in that it was just handing over the POSIX
5655 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5656 required a bigger vector, with some working space on the end. This means that
5657 the POSIX wrapper now has to get and free some memory, and copy the results.
5658
5659 6. Added some simple autoconf support, placing the test data and the
5660 documentation in separate directories, re-organizing some of the
5661 information files, and making it build pcre-config (a GNU standard). Also added
5662 libtool support for building PCRE as a shared library, which is now the
5663 default.
5664
5665 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5666 09 are not valid octal constants. Single digits will be used for minor values
5667 less than 10.
5668
5669 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5670 existing programs that set these in the POSIX interface can use PCRE without
5671 modification.
5672
5673 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5674 return all that pcre_info() returns, plus additional data. The pcre_info()
5675 function is retained for compatibility, but is considered to be obsolete.
5676
5677 10. Added experimental recursion feature (?R) to handle one common case that
5678 Perl 5.6 will be able to do with (?p{...}).
5679
5680 11. Added support for POSIX character classes like [:alpha:], which Perl is
5681 adopting.
5682
5683
5684 Version 2.08 31-Aug-99
5685 ----------------------
5686
5687 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5688 trying to match at the startoffset position, but instead was moving forward to
5689 the next newline as if a previous match had failed.
5690
5691 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5692 and could get into a loop if a null string was matched other than at the start
5693 of the subject.
5694
5695 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5696 be distinguished at compile time, and for completeness also added PCRE_DATE.
5697
5698 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5699 in GnuWin32 environments.
5700
5701
5702 Version 2.07 29-Jul-99
5703 ----------------------
5704
5705 1. The documentation is now supplied in plain text form and HTML as well as in
5706 the form of man page sources.
5707
5708 2. C++ compilers don't like assigning (void *) values to other pointer types.
5709 In particular this affects malloc(). Although there is no problem in Standard
5710 C, I've put in casts to keep C++ compilers happy.
5711
5712 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5713 should be (const char *).
5714
5715 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5716 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5717 However, I haven't made this a standard facility. The documentation doesn't
5718 mention it, and the Makefile doesn't support it.
5719
5720 5. The Makefile now contains an "install" target, with editable destinations at
5721 the top of the file. The pcretest program is not installed.
5722
5723 6. pgrep -V now gives the PCRE version number and date.
5724
5725 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5726 causing the entire string to be ignored, instead of just the last character.
5727
5728 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5729 non-matching string, it can take a very, very long time, even for strings of
5730 quite modest length, because of the nested recursion. PCRE now does better in
5731 some of these cases. It does this by remembering the last required literal
5732 character in the pattern, and pre-searching the subject to ensure it is present
5733 before running the real match. In other words, it applies a heuristic to detect
5734 some types of certain failure quickly, and in the above example, if presented
5735 with a string that has no trailing " it gives "no match" very quickly.
5736
5737 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5738 other alternatives are tried instead.
5739
5740
5741 Version 2.06 09-Jun-99
5742 ----------------------
5743
5744 1. Change pcretest's output for amount of store used to show just the code
5745 space, because the remainder (the data block) varies in size between 32-bit and
5746 64-bit systems.
5747
5748 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5749 start matching at. This allows lookbehinds to work when searching for multiple
5750 occurrences in a string.
5751
5752 3. Added additional options to pcretest for testing multiple occurrences:
5753
5754 /+ outputs the rest of the string that follows a match
5755 /g loops for multiple occurrences, using the new startoffset argument
5756 /G loops for multiple occurrences by passing an incremented pointer
5757
5758 4. PCRE wasn't doing the "first character" optimization for patterns starting
5759 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5760 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5761 the letter 'x'. On long subject strings, this gives a significant speed-up.
5762
5763
5764 Version 2.05 21-Apr-99
5765 ----------------------
5766
5767 1. Changed the type of magic_number from int to long int so that it works
5768 properly on 16-bit systems.
5769
5770 2. Fixed a bug which caused patterns starting with .* not to work correctly
5771 when the subject string contained newline characters. PCRE was assuming
5772 anchoring for such patterns in all cases, which is not correct because .* will
5773 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5774 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5775 must be retried after every newline in the subject.
5776
5777
5778 Version 2.04 18-Feb-99
5779 ----------------------
5780
5781 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5782 computation of the store needed to hold the pattern was incorrect (too large).
5783 If such patterns were nested a few deep, this could multiply and become a real
5784 problem.
5785
5786 2. Added /M option to pcretest to show the memory requirement of a specific
5787 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5788
5789 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5790 compiled in such a way that the backtracking after subsequent failure was
5791 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5792 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5793
5794
5795 Version 2.03 02-Feb-99
5796 ----------------------
5797
5798 1. Fixed typo and small mistake in man page.
5799
5800 2. Added 4th condition (GPL supersedes if conflict) and created separate
5801 LICENCE file containing the conditions.
5802
5803 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5804 Perl, that is the internal \ allows the delimiter to be included in the
5805 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5806 the final delimiter, add \ to the end of the pattern (to test the error).
5807
5808 4. Added the convenience functions for extracting substrings after a successful
5809 match. Updated pcretest to make it able to test these functions.
5810
5811
5812 Version 2.02 14-Jan-99
5813 ----------------------
5814
5815 1. Initialized the working variables associated with each extraction so that
5816 their saving and restoring doesn't refer to uninitialized store.
5817
5818 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5819 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5820 fix the problem.
5821
5822 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5823 calls, and wasn't printing the correct value for compiling calls. Increased the
5824 default value of LOOPREPEAT, and the number of significant figures in the
5825 times.
5826
5827 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5828
5829 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5830 a building problem on Windows NT with a FAT file system.
5831
5832
5833 Version 2.01 21-Oct-98
5834 ----------------------
5835
5836 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5837 to character tables built by pcre_maketables() in the current locale. If NULL
5838 is passed, the default tables are used.
5839
5840
5841 Version 2.00 24-Sep-98
5842 ----------------------
5843
5844 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5845 it any more.
5846
5847 2. Allow quantification of (?>) groups, and make it work correctly.
5848
5849 3. The first character computation wasn't working for (?>) groups.
5850
5851 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5852 end of the subject) and add 5.005's \z, which really does match only at the
5853 very end of the subject.
5854
5855 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5856
5857 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5858 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5859 localized options. All options to pcre_study() were also removed.
5860
5861 7. Add other new features from 5.005:
5862
5863 $(?<= positive lookbehind
5864 $(?<! negative lookbehind
5865 (?imsx-imsx) added the unsetting capability
5866 such a setting is global if at outer level; local otherwise
5867 (?imsx-imsx:) non-capturing groups with option setting
5868 (?(cond)re|re) conditional pattern matching
5869
5870 A backreference to itself in a repeated group matches the previous
5871 captured string.
5872
5873 8. General tidying up of studying (both automatic and via "study")
5874 consequential on the addition of new assertions.
5875
5876 9. As in 5.005, unlimited repeated groups that could match an empty substring
5877 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5878 runtime if any iteration does actually match an empty substring.
5879
5880 10. Include the RunTest script in the distribution.
5881
5882 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5883 discrepancies, some of which were old and were also with respect to 5.004. They
5884 have now been fixed.
5885
5886
5887 Version 1.09 28-Apr-98
5888 ----------------------
5889
5890 1. A negated single character class followed by a quantifier with a minimum
5891 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5892 program crashes, or just wrong answers. This did not apply to negated classes
5893 containing more than one character, or to minima other than one.
5894
5895
5896 Version 1.08 27-Mar-98
5897 ----------------------
5898
5899 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5900
5901 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5902 latter must appear before anything that relies on it in the pattern.
5903
5904
5905 Version 1.07 16-Feb-98
5906 ----------------------
5907
5908 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5909 repeat of a potentially empty string).
5910
5911
5912 Version 1.06 23-Jan-98
5913 ----------------------
5914
5915 1. Added Markus Oberhumer's little patches for C++.
5916
5917 2. Literal strings longer than 255 characters were broken.
5918
5919
5920 Version 1.05 23-Dec-97
5921 ----------------------
5922
5923 1. Negated character classes containing more than one character were failing if
5924 PCRE_CASELESS was set at run time.
5925
5926
5927 Version 1.04 19-Dec-97
5928 ----------------------
5929
5930 1. Corrected the man page, where some "const" qualifiers had been omitted.
5931
5932 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5933 input syntax.
5934
5935 3. Fixed memory leak which occurred when a regex with back references was
5936 matched with an offsets vector that wasn't big enough. The temporary memory
5937 that is used in this case wasn't being freed if the match failed.
5938
5939 4. Tidied pcretest to ensure it frees memory that it gets.
5940
5941 5. Temporary memory was being obtained in the case where the passed offsets
5942 vector was exactly big enough.
5943
5944 6. Corrected definition of offsetof() from change 5 below.
5945
5946 7. I had screwed up change 6 below and broken the rules for the use of
5947 setjmp(). Now fixed.
5948
5949
5950 Version 1.03 18-Dec-97
5951 ----------------------
5952
5953 1. A erroneous regex with a missing opening parenthesis was correctly
5954 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5955 on some systems.
5956
5957 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5958 it was reported that one broken compiler failed on the former because "code" is
5959 also an independent variable.
5960
5961 3. The erroneous regex a[]b caused an array overrun reference.
5962
5963 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5964 fail on data ending with that character. (It was going on too far, and checking
5965 the next character, typically a binary zero.) This was specific to the
5966 optimized code for single-character negative classes.
5967
5968 5. Added a contributed patch from the TIN world which does the following:
5969
5970 + Add an undef for memmove, in case the the system defines a macro for it.
5971
5972 + Add a definition of offsetof(), in case there isn't one. (I don't know
5973 the reason behind this - offsetof() is part of the ANSI standard - but
5974 it does no harm).
5975
5976 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5977 most of the places where whitespace preceded '#'. I have given up and
5978 allowed the remaining 2 cases to be at the margin.
5979
5980 + Rename some variables in pcre to eliminate shadowing. This seems very
5981 pedantic, but does no harm, of course.
5982
5983 6. Moved the call to setjmp() into its own function, to get rid of warnings
5984 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5985
5986 7. Constructs such as \d{8,} were compiling into the equivalent of
5987 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5988 outcome, but in this particular case used more store than had been allocated,
5989 which caused the bug to be discovered because it threw up an internal error.
5990
5991 8. The debugging code in both pcre and pcretest for outputting the compiled
5992 form of a regex was going wrong in the case of back references followed by
5993 curly-bracketed repeats.
5994
5995
5996 Version 1.02 12-Dec-97
5997 ----------------------
5998
5999 1. Typos in pcre.3 and comments in the source fixed.
6000
6001 2. Applied a contributed patch to get rid of places where it used to remove
6002 'const' from variables, and fixed some signed/unsigned and uninitialized
6003 variable warnings.
6004
6005 3. Added the "runtest" target to Makefile.
6006
6007 4. Set default compiler flag to -O2 rather than just -O.
6008
6009
6010 Version 1.01 19-Nov-97
6011 ----------------------
6012
6013 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
6014 like /([ab]*)*/, that is, for classes with more than one character in them.
6015
6016 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
6017 as /((?>a*))*/ (a PCRE_EXTRA facility).
6018
6019
6020 Version 1.00 18-Nov-97
6021 ----------------------
6022
6023 1. Added compile-time macros to support systems such as SunOS4 which don't have
6024 memmove() or strerror() but have other things that can be used instead.
6025
6026 2. Arranged that "make clean" removes the executables.
6027
6028
6029 Version 0.99 27-Oct-97
6030 ----------------------
6031
6032 1. Fixed bug in code for optimizing classes with only one character. It was
6033 initializing a 32-byte map regardless, which could cause it to run off the end
6034 of the memory it had got.
6035
6036 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
6037
6038
6039 Version 0.98 22-Oct-97
6040 ----------------------
6041
6042 1. Fixed bug in code for handling temporary memory usage when there are more
6043 back references than supplied space in the ovector. This could cause segfaults.
6044
6045
6046 Version 0.97 21-Oct-97
6047 ----------------------
6048
6049 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
6050
6051 2. Optimized negated single characters not to use a bit map.
6052
6053 3. Brought error texts together as macro definitions; clarified some of them;
6054 fixed one that was wrong - it said "range out of order" when it meant "invalid
6055 escape sequence".
60