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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1753 - (show annotations)
Thu May 23 16:27:33 2019 UTC (3 months, 3 weeks ago) by ph10
File size: 298679 byte(s)
Fix omission of (*LF) from list in the C++ wrapper.
1 ChangeLog for PCRE
2 ------------------
3
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
6
7
8 Version 8.44-RC1 01 March-2019
9 ------------------------------
10
11 1. Setting --enable-jit=auto for an out-of-tree build failed because the
12 source directory wasn't in the search path for AC_TRY_COMPILE always. Patch
13 from Ross Burton.
14
15 2. Applied a patch from Michael Shigorin to fix 8.43 build on e2k arch
16 with lcc compiler (EDG frontend based); the problem it fixes is:
17
18 lcc: "pcrecpp.cc", line 74: error: declaration aliased to undefined entity
19 "_ZN7pcrecpp2RE6no_argE" [-Werror]
20
21 3. Change 2 for 8.43 omitted (*LF) from the list of start-of-pattern items. Now
22 added.
23
24
25 Version 8.43 23-February-2019
26 -----------------------------
27
28 1. Some time ago the config macro SUPPORT_UTF8 was changed to SUPPORT_UTF
29 because it also applies to UTF-16 and UTF-32. However, this change was not made
30 in the pcre2cpp files; consequently the C++ wrapper has from then been compiled
31 with a bug in it, which would have been picked up by the unit test except that
32 it also had its UTF8 code cut out. The bug was in a global replace when moving
33 forward after matching an empty string.
34
35 2. The C++ wrapper got broken a long time ago (version 7.3, August 2007) when
36 (*CR) was invented (assuming it was the first such start-of-pattern option).
37 The wrapper could never handle such patterns because it wraps patterns in
38 (?:...)\z in order to support end anchoring. I have hacked in some code to fix
39 this, that is, move the wrapping till after any existing start-of-pattern
40 special settings.
41
42 3. "pcre2grep" (sic) was accidentally mentioned in an error message (fix was
43 ported from PCRE2).
44
45 4. Typo LCC_ALL for LC_ALL fixed in pcregrep.
46
47 5. In a pattern such as /[^\x{100}-\x{ffff}]*[\x80-\xff]/ which has a repeated
48 negative class with no characters less than 0x100 followed by a positive class
49 with only characters less than 0x100, the first class was incorrectly being
50 auto-possessified, causing incorrect match failures.
51
52 6. If the only branch in a conditional subpattern was anchored, the whole
53 subpattern was treated as anchored, when it should not have been, since the
54 assumed empty second branch cannot be anchored. Demonstrated by test patterns
55 such as /(?(1)^())b/ or /(?(?=^))b/.
56
57 7. Fix subject buffer overread in JIT when UTF is disabled and \X or \R has
58 a greater than 1 fixed quantifier. This issue was found by Yunho Kim.
59
60 8. If a pattern started with a subroutine call that had a quantifier with a
61 minimum of zero, an incorrect "match must start with this character" could be
62 recorded. Example: /(?&xxx)*ABC(?<xxx>XYZ)/ would (incorrectly) expect 'A' to
63 be the first character of a match.
64
65 9. Improve MAP_JIT flag usage on MacOS. Patch by Rich Siegel.
66
67
68 Version 8.42 20-March-2018
69 --------------------------
70
71 1. Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
72
73 2. Fixed outdated real_pcre definitions in pcre.h.in (patch by Evgeny Kotkov).
74
75 3. pcregrep was truncating components of file names to 128 characters when
76 processing files with the -r option, and also (some very odd code) truncating
77 path names to 512 characters. There is now a check on the absolute length of
78 full path file names, which may be up to 2047 characters long.
79
80 4. Using pcre_dfa_exec(), in UTF mode when UCP support was not defined, there
81 was the possibility of a false positive match when caselessly matching a "not
82 this character" item such as [^\x{1234}] (with a code point greater than 127)
83 because the "other case" variable was not being initialized.
84
85 5. Although pcre_jit_exec checks whether the pattern is compiled
86 in a given mode, it was also expected that at least one mode is available.
87 This is fixed and pcre_jit_exec returns with PCRE_ERROR_JIT_BADOPTION
88 when the pattern is not optimized by JIT at all.
89
90 6. The line number and related variables such as match counts in pcregrep
91 were all int variables, causing overflow when files with more than 2147483647
92 lines were processed (assuming 32-bit ints). They have all been changed to
93 unsigned long ints.
94
95 7. If a backreference with a minimum repeat count of zero was first in a
96 pattern, apart from assertions, an incorrect first matching character could be
97 recorded. For example, for the pattern /(?=(a))\1?b/, "b" was incorrectly set
98 as the first character of a match.
99
100 8. Fix out-of-bounds read for partial matching of /./ against an empty string
101 when the newline type is CRLF.
102
103 9. When matching using the the REG_STARTEND feature of the POSIX API with a
104 non-zero starting offset, unset capturing groups with lower numbers than a
105 group that did capture something were not being correctly returned as "unset"
106 (that is, with offset values of -1).
107
108 10. Matching the pattern /(*UTF)\C[^\v]+\x80/ against an 8-bit string
109 containing multi-code-unit characters caused bad behaviour and possibly a
110 crash. This issue was fixed for other kinds of repeat in release 8.37 by change
111 38, but repeating character classes were overlooked.
112
113 11. A small fix to pcregrep to avoid compiler warnings for -Wformat-overflow=2.
114
115 12. Added --enable-jit=auto support to configure.ac.
116
117 13. Fix misleading error message in configure.ac.
118
119
120 Version 8.41 05-July-2017
121 -------------------------
122
123 1. Fixed typo in CMakeLists.txt (wrong number of arguments for
124 PCRE_STATIC_RUNTIME (affects MSVC only).
125
126 2. Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
127 mode with --only-matching matched several lines, it restarted scanning at the
128 next line instead of moving on to the end of the matched string, which can be
129 several lines after the start.
130
131 3. Fix a missing else in the JIT compiler reported by 'idaifish'.
132
133 4. A (?# style comment is now ignored between a basic quantifier and a
134 following '+' or '?' (example: /X+(?#comment)?Y/.
135
136 5. Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
137 Pisar).
138
139 6. Fuzzers have reported issues in pcretest. These are NOT serious (it is,
140 after all, just a test program). However, to stop the reports, some easy ones
141 are fixed:
142
143 (a) Check for values < 256 when calling isprint() in pcretest.
144 (b) Give an error for too big a number after \O.
145
146 7. In the 32-bit library in non-UTF mode, an attempt to find a Unicode
147 property for a character with a code point greater than 0x10ffff (the Unicode
148 maximum) caused a crash.
149
150 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
151 encountered a character class with a possessive repeat, for example [a-f]{3}+.
152
153 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
154 length incorrectly, which could result in buffer overflow.
155
156 10. Remove redundant line of code (accidentally left in ages ago).
157
158 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
159 with namespace pcrecpp (Bugzilla #2084).
160
161 12. Remove a duplication typo in pcre_tables.c.
162
163 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
164 starting offset greater than zero.
165
166
167 Version 8.40 11-January-2017
168 ----------------------------
169
170 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
171 the match extended over a line boundary.
172
173 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
174 MSVC static compilation, putting the first patch under a new option.
175
176 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
177
178 4. Ignore "show all captures" (/=) for DFA matching.
179
180 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
181
182 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
183 without PCRE_UCP set, a negative character type such as \D in a positive
184 class should cause all characters greater than 255 to match, whatever else
185 is in the class. There was a bug that caused this not to happen if a
186 Unicode property item was added to such a class, for example [\D\P{Nd}] or
187 [\W\pL].
188
189 7. When pcretest was outputing information from a callout, the caret indicator
190 for the current position in the subject line was incorrect if it was after
191 an escape sequence for a character whose code point was greater than
192 \x{ff}.
193
194 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
195 the conditional was interpreted as a reference to capturing group 1 instead
196 of a test for recursion. Any group whose name began with R was
197 misinterpreted in this way. (The reference interpretation should only
198 happen if the group's name is precisely "R".)
199
200 9. A number of bugs have been mended relating to match start-up optimizations
201 when the first thing in a pattern is a positive lookahead. These all
202 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
203
204 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
205 both an initial 'X' and a following 'X'.
206 (b) Some patterns starting with an assertion that started with .* were
207 incorrectly optimized as having to match at the start of the subject or
208 after a newline. There are cases where this is not true, for example,
209 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
210 start with spaces. Starting .* in an assertion is no longer taken as an
211 indication of matching at the start (or after a newline).
212
213
214 Version 8.39 14-June-2016
215 -------------------------
216
217 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
218 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
219 misbehaved. This bug was found by the LLVM fuzzer.
220
221 2. Similar to the above, if an isolated \E was present between an item and its
222 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
223 bug was found by the LLVM fuzzer.
224
225 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
226 working correctly in UCP mode.
227
228 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
229 was set when the pmatch argument was NULL. It now returns REG_INVARG.
230
231 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
232
233 6. An empty \Q\E sequence between an item and its qualifier caused
234 pcre_compile() to misbehave when auto callouts were enabled. This bug was
235 found by the LLVM fuzzer.
236
237 7. If a pattern that was compiled with PCRE_EXTENDED started with white
238 space or a #-type comment that was followed by (?-x), which turns off
239 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
240 pcre_compile() assumed that (?-x) applied to the whole pattern and
241 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
242
243 8. A call of pcre_copy_named_substring() for a named substring whose number
244 was greater than the space in the ovector could cause a crash.
245
246 9. Yet another buffer overflow bug involved duplicate named groups with a
247 group that reset capture numbers (compare 8.38/7 below). Once again, I have
248 just allowed for more memory, even if not needed. (A proper fix is
249 implemented in PCRE2, but it involves a lot of refactoring.)
250
251 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
252 start of the match to be earlier than the end.
253
254 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
255
256 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
257 assertion, caused pcretest to generate incorrect output, and also to read
258 uninitialized memory (detected by ASAN or valgrind).
259
260 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
261 nested set of parentheses of sufficient size caused an overflow of the
262 compiling workspace (which was diagnosed, but of course is not desirable).
263
264 14. And yet another buffer overflow bug involving duplicate named groups, this
265 time nested, with a nested back reference. Yet again, I have just allowed
266 for more memory, because anything more needs all the refactoring that has
267 been done for PCRE2. An example pattern that provoked this bug is:
268 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
269 registered as CVE-2016-1283.
270
271 15. pcretest went into a loop if global matching was requested with an ovector
272 size less than 2. It now gives an error message. This bug was found by
273 afl-fuzz.
274
275 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
276 ("assertion expected") when (?(?C) was not followed by an opening
277 parenthesis.
278
279 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
280 actually affect anything, by sheer luck.
281
282 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
283 static compilation.
284
285 19. Modified the RunTest script to incorporate a valgrind suppressions file so
286 that certain errors, provoked by the SSE2 instruction set when JIT is used,
287 are ignored.
288
289 20. A racing condition is fixed in JIT reported by Mozilla.
290
291 21. Minor code refactor to avoid "array subscript is below array bounds"
292 compiler warning.
293
294 22. Minor code refactor to avoid "left shift of negative number" warning.
295
296 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
297 UCP support.
298
299 24. Refactor to avoid compiler warnings in pcrecpp.cc.
300
301 25. Refactor to fix a typo in pcre_jit_test.c
302
303 26. Patch to support compiling pcrecpp.cc with Intel compiler.
304
305
306 Version 8.38 23-November-2015
307 -----------------------------
308
309 1. If a group that contained a recursive back reference also contained a
310 forward reference subroutine call followed by a non-forward-reference
311 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
312 compile correct code, leading to undefined behaviour or an internally
313 detected error. This bug was discovered by the LLVM fuzzer.
314
315 2. Quantification of certain items (e.g. atomic back references) could cause
316 incorrect code to be compiled when recursive forward references were
317 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
318 This bug was discovered by the LLVM fuzzer.
319
320 3. A repeated conditional group whose condition was a reference by name caused
321 a buffer overflow if there was more than one group with the given name.
322 This bug was discovered by the LLVM fuzzer.
323
324 4. A recursive back reference by name within a group that had the same name as
325 another group caused a buffer overflow. For example:
326 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
327
328 5. A forward reference by name to a group whose number is the same as the
329 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
330 a buffer overflow at compile time. This bug was discovered by the LLVM
331 fuzzer.
332
333 6. A lookbehind assertion within a set of mutually recursive subpatterns could
334 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
335
336 7. Another buffer overflow bug involved duplicate named groups with a
337 reference between their definition, with a group that reset capture
338 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
339 fixed by always allowing for more memory, even if not needed. (A proper fix
340 is implemented in PCRE2, but it involves more refactoring.)
341
342 8. There was no check for integer overflow in subroutine calls such as (?123).
343
344 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
345 being treated as a literal 'l' instead of causing an error.
346
347 10. There was a buffer overflow if pcre_exec() was called with an ovector of
348 size 1. This bug was found by american fuzzy lop.
349
350 11. If a non-capturing group containing a conditional group that could match
351 an empty string was repeated, it was not identified as matching an empty
352 string itself. For example: /^(?:(?(1)x|)+)+$()/.
353
354 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
355 \a and \e in test subject lines.
356
357 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
358 instead of the EBCDIC value.
359
360 14. The handling of \c in an EBCDIC environment has been revised so that it is
361 now compatible with the specification in Perl's perlebcdic page.
362
363 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
364 ASCII/Unicode. This has now been added to the list of characters that are
365 recognized as white space in EBCDIC.
366
367 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
368 error (correctly) when used outside a class, but did not give an error
369 within a class.
370
371 17. \h within a class was incorrectly compiled in EBCDIC environments.
372
373 18. A pattern with an unmatched closing parenthesis that contained a backward
374 assertion which itself contained a forward reference caused buffer
375 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
376
377 19. JIT should return with error when the compiled pattern requires more stack
378 space than the maximum.
379
380 20. A possessively repeated conditional group that could match an empty string,
381 for example, /(?(R))*+/, was incorrectly compiled.
382
383 21. Fix infinite recursion in the JIT compiler when certain patterns such as
384 /(?:|a|){100}x/ are analysed.
385
386 22. Some patterns with character classes involving [: and \\ were incorrectly
387 compiled and could cause reading from uninitialized memory or an incorrect
388 error diagnosis.
389
390 23. Pathological patterns containing many nested occurrences of [: caused
391 pcre_compile() to run for a very long time.
392
393 24. A conditional group with only one branch has an implicit empty alternative
394 branch and must therefore be treated as potentially matching an empty
395 string.
396
397 25. If (?R was followed by - or + incorrect behaviour happened instead of a
398 diagnostic.
399
400 26. Arrange to give up on finding the minimum matching length for overly
401 complex patterns.
402
403 27. Similar to (4) above: in a pattern with duplicated named groups and an
404 occurrence of (?| it is possible for an apparently non-recursive back
405 reference to become recursive if a later named group with the relevant
406 number is encountered. This could lead to a buffer overflow. Wen Guanxing
407 from Venustech ADLAB discovered this bug.
408
409 28. If pcregrep was given the -q option with -c or -l, or when handling a
410 binary file, it incorrectly wrote output to stdout.
411
412 29. The JIT compiler did not restore the control verb head in case of *THEN
413 control verbs. This issue was found by Karl Skomski with a custom LLVM
414 fuzzer.
415
416 30. Error messages for syntax errors following \g and \k were giving inaccurate
417 offsets in the pattern.
418
419 31. Added a check for integer overflow in conditions (?(<digits>) and
420 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
421 fuzzer.
422
423 32. Handling recursive references such as (?2) when the reference is to a group
424 later in the pattern uses code that is very hacked about and error-prone.
425 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
426 give an internal error if it is obvious that compiling has gone wrong.
427
428 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
429 This issue was found by Karl Skomski with a custom LLVM fuzzer.
430
431 34. The JIT compiler should restore the control chain for empty possessive
432 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
433
434 35. Match limit check added to JIT recursion. This issue was found by Karl
435 Skomski with a custom LLVM fuzzer.
436
437 36. Yet another case similar to 27 above has been circumvented by an
438 unconditional allocation of extra memory. This issue is fixed "properly" in
439 PCRE2 by refactoring the way references are handled. Wen Guanxing
440 from Venustech ADLAB discovered this bug.
441
442 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
443 with a custom LLVM fuzzer.
444
445 38. Fixed a corner case of range optimization in JIT.
446
447 39. An incorrect error "overran compiling workspace" was given if there were
448 exactly enough group forward references such that the last one extended
449 into the workspace safety margin. The next one would have expanded the
450 workspace. The test for overflow was not including the safety margin.
451
452 40. A match limit issue is fixed in JIT which was found by Karl Skomski
453 with a custom LLVM fuzzer.
454
455 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
456 work under Windows. (Why has it taken so long for anyone to notice?)
457
458 42. In a character class such as [\W\p{Any}] where both a negative-type escape
459 ("not a word character") and a property escape were present, the property
460 escape was being ignored.
461
462 43. Fix crash caused by very long (*MARK) or (*THEN) names.
463
464 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
465 by a single ASCII character in a class item, was incorrectly compiled in
466 UCP mode. The POSIX class got lost, but only if the single character
467 followed it.
468
469 45. [:punct:] in UCP mode was matching some characters in the range 128-255
470 that should not have been matched.
471
472 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
473 class, all characters with code points greater than 255 are in the class.
474 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
475 such as \w are turned into Unicode properties), wide characters were not
476 correctly handled, and could fail to match.
477
478
479 Version 8.37 28-April-2015
480 --------------------------
481
482 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
483 for those parentheses to be closed with whatever has been captured so far.
484 However, it was failing to mark any other groups between the hightest
485 capture so far and the currrent group as "unset". Thus, the ovector for
486 those groups contained whatever was previously there. An example is the
487 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
488
489 2. If an assertion condition was quantified with a minimum of zero (an odd
490 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
491
492 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
493 unrecognized modifier, a crash could occur.
494
495 4. An attempt to do global matching in pcretest with a zero-length ovector
496 caused a crash.
497
498 5. Fixed a memory leak during matching that could occur for a subpattern
499 subroutine call (recursive or otherwise) if the number of captured groups
500 that had to be saved was greater than ten.
501
502 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
503 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
504 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
505
506 7. A UTF pattern containing a "not" match of a non-ASCII character and a
507 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
508
509 8. When a pattern is compiled, it remembers the highest back reference so that
510 when matching, if the ovector is too small, extra memory can be obtained to
511 use instead. A conditional subpattern whose condition is a check on a
512 capture having happened, such as, for example in the pattern
513 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
514 setting the highest backreference number. This mattered only if pcre_exec()
515 was called with an ovector that was too small to hold the capture, and there
516 was no other kind of back reference (a situation which is probably quite
517 rare). The effect of the bug was that the condition was always treated as
518 FALSE when the capture could not be consulted, leading to a incorrect
519 behaviour by pcre_exec(). This bug has been fixed.
520
521 9. A reference to a duplicated named group (either a back reference or a test
522 for being set in a conditional) that occurred in a part of the pattern where
523 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
524 to be incorrectly calculated, leading to overwriting.
525
526 10. A mutually recursive set of back references such as (\2)(\1) caused a
527 segfault at study time (while trying to find the minimum matching length).
528 The infinite loop is now broken (with the minimum length unset, that is,
529 zero).
530
531 11. If an assertion that was used as a condition was quantified with a minimum
532 of zero, matching went wrong. In particular, if the whole group had
533 unlimited repetition and could match an empty string, a segfault was
534 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
535 assertions to be quantified, but not if they are being used as conditions,
536 so the above pattern is faulted by Perl. PCRE has now been changed so that
537 it also rejects such patterns.
538
539 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
540 failed to allow the zero-repeat case if pcre2_exec() was called with an
541 ovector too small to capture the group.
542
543 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
544 Red Hat Product Security:
545
546 (a) A crash if /K and /F were both set with the option to save the compiled
547 pattern.
548
549 (b) Another crash if the option to print captured substrings in a callout
550 was combined with setting a null ovector, for example \O\C+ as a subject
551 string.
552
553 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
554 forward reference repeated a large (but limited) number of times within a
555 repeated outer group that has a zero minimum quantifier, caused incorrect
556 code to be compiled, leading to the error "internal error:
557 previously-checked referenced subpattern not found" when an incorrect
558 memory address was read. This bug was reported as "heap overflow",
559 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
560 CVE-2015-2325.
561
562 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
563 call within a group that also contained a recursive back reference caused
564 incorrect code to be compiled. This bug was reported as "heap overflow",
565 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
566 number CVE-2015-2326.
567
568 24. Computing the size of the JIT read-only data in advance has been a source
569 of various issues, and new ones are still appear unfortunately. To fix
570 existing and future issues, size computation is eliminated from the code,
571 and replaced by on-demand memory allocation.
572
573 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
574 adjacent to the end of the range, and the range contained characters with
575 more than one other case, caused incorrect behaviour when compiled in UTF
576 mode. In that example, the range a-j was left out of the class.
577
578 26. Fix JIT compilation of conditional blocks, which assertion
579 is converted to (*FAIL). E.g: /(?(?!))/.
580
581 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
582 discovered by the LLVM fuzzer.
583
584 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
585 when this assertion was used as a condition, for example (?(?!)a|b). In
586 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
587 error about an unsupported item.
588
589 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
590 possessification code could take exponential time to complete. A recursion
591 depth limit of 1000 has been imposed to limit the resources used by this
592 optimization.
593
594 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
595 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
596 because \S ensures they are all in the class. The code for doing this was
597 interacting badly with the code for computing the amount of space needed to
598 compile the pattern, leading to a buffer overflow. This bug was discovered
599 by the LLVM fuzzer.
600
601 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
602 other kinds of group caused stack overflow at compile time. This bug was
603 discovered by the LLVM fuzzer.
604
605 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
606 between a subroutine call and its quantifier was incorrectly compiled,
607 leading to buffer overflow or other errors. This bug was discovered by the
608 LLVM fuzzer.
609
610 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
611 assertion after (?(. The code was failing to check the character after
612 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
613 was discovered by the LLVM fuzzer.
614
615 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
616 a fixed maximum following a group that contains a subroutine reference was
617 incorrectly compiled and could trigger buffer overflow. This bug was
618 discovered by the LLVM fuzzer.
619
620 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
621 caused a stack overflow instead of the diagnosis of a non-fixed length
622 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
623
624 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
625 (e.g. /(?<=\Ka)/) could make pcregrep loop.
626
627 37. There was a similar problem to 36 in pcretest for global matches.
628
629 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
630 and a subsequent item in the pattern caused a non-match, backtracking over
631 the repeated \X did not stop, but carried on past the start of the subject,
632 causing reference to random memory and/or a segfault. There were also some
633 other cases where backtracking after \C could crash. This set of bugs was
634 discovered by the LLVM fuzzer.
635
636 39. The function for finding the minimum length of a matching string could take
637 a very long time if mutual recursion was present many times in a pattern,
638 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
639 method has been implemented. This infelicity was discovered by the LLVM
640 fuzzer.
641
642 40. Static linking against the PCRE library using the pkg-config module was
643 failing on missing pthread symbols.
644
645
646 Version 8.36 26-September-2014
647 ------------------------------
648
649 1. Got rid of some compiler warnings in the C++ modules that were shown up by
650 -Wmissing-field-initializers and -Wunused-parameter.
651
652 2. The tests for quantifiers being too big (greater than 65535) were being
653 applied after reading the number, and stupidly assuming that integer
654 overflow would give a negative number. The tests are now applied as the
655 numbers are read.
656
657 3. Tidy code in pcre_exec.c where two branches that used to be different are
658 now the same.
659
660 4. The JIT compiler did not generate match limit checks for certain
661 bracketed expressions with quantifiers. This may lead to exponential
662 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
663 issue should be resolved now.
664
665 5. Fixed an issue, which occures when nested alternatives are optimized
666 with table jumps.
667
668 6. Inserted two casts and changed some ints to size_t in the light of some
669 reported 64-bit compiler warnings (Bugzilla 1477).
670
671 7. Fixed a bug concerned with zero-minimum possessive groups that could match
672 an empty string, which sometimes were behaving incorrectly in the
673 interpreter (though correctly in the JIT matcher). This pcretest input is
674 an example:
675
676 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
677 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
678
679 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
680 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
681 for an empty string was breaking the inner loop and carrying on at a lower
682 level, when possessive repeated groups should always return to a higher
683 level as they have no backtrack points in them. The empty string test now
684 occurs at the outer level.
685
686 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
687 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
688
689 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
690 doesn't).
691
692 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
693 to hold a pattern to be incorrectly computed (too small) when there were
694 named back references to duplicated names. This could cause "internal
695 error: code overflow" or "double free or corruption" or other memory
696 handling errors.
697
698 11. When named subpatterns had the same prefixes, back references could be
699 confused. For example, in this pattern:
700
701 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
702
703 the reference to 'Name' was incorrectly treated as a reference to a
704 duplicate name.
705
706 12. A pattern such as /^s?c/mi8 where the optional character has more than
707 one "other case" was incorrectly compiled such that it would only try to
708 match starting at "c".
709
710 13. When a pattern starting with \s was studied, VT was not included in the
711 list of possible starting characters; this should have been part of the
712 8.34/18 patch.
713
714 14. If a character class started [\Qx]... where x is any character, the class
715 was incorrectly terminated at the ].
716
717 15. If a pattern that started with a caseless match for a character with more
718 than one "other case" was studied, PCRE did not set up the starting code
719 unit bit map for the list of possible characters. Now it does. This is an
720 optimization improvement, not a bug fix.
721
722 16. The Unicode data tables have been updated to Unicode 7.0.0.
723
724 17. Fixed a number of memory leaks in pcregrep.
725
726 18. Avoid a compiler warning (from some compilers) for a function call with
727 a cast that removes "const" from an lvalue by using an intermediate
728 variable (to which the compiler does not object).
729
730 19. Incorrect code was compiled if a group that contained an internal recursive
731 back reference was optional (had quantifier with a minimum of zero). This
732 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
733 caused segmentation faults because of stack overflows at compile time.
734
735 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
736 group that is quantified with an indefinite repeat, caused a compile-time
737 loop which used up all the system stack and provoked a segmentation fault.
738 This was not the same bug as 19 above.
739
740 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
741 Patch by Mike Frysinger.
742
743
744 Version 8.35 04-April-2014
745 --------------------------
746
747 1. A new flag is set, when property checks are present in an XCLASS.
748 When this flag is not set, PCRE can perform certain optimizations
749 such as studying these XCLASS-es.
750
751 2. The auto-possessification of character sets were improved: a normal
752 and an extended character set can be compared now. Furthermore
753 the JIT compiler optimizes more character set checks.
754
755 3. Got rid of some compiler warnings for potentially uninitialized variables
756 that show up only when compiled with -O2.
757
758 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
759 of a match later then the end of the match. The pcretest program was not
760 handling the case sensibly - it was outputting from the start to the next
761 binary zero. It now reports this situation in a message, and outputs the
762 text from the end to the start.
763
764 5. Fast forward search is improved in JIT. Instead of the first three
765 characters, any three characters with fixed position can be searched.
766 Search order: first, last, middle.
767
768 6. Improve character range checks in JIT. Characters are read by an inprecise
769 function now, which returns with an unknown value if the character code is
770 above a certain threshold (e.g: 256). The only limitation is that the value
771 must be bigger than the threshold as well. This function is useful when
772 the characters above the threshold are handled in the same way.
773
774 7. The macros whose names start with RAWUCHAR are placeholders for a future
775 mode in which only the bottom 21 bits of 32-bit data items are used. To
776 make this more memorable for those maintaining the code, the names have
777 been changed to start with UCHAR21, and an extensive comment has been added
778 to their definition.
779
780 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
781 to the export list in Makefile.am (they were accidentally omitted from the
782 8.34 tarball).
783
784 9. The informational output from pcretest used the phrase "starting byte set"
785 which is inappropriate for the 16-bit and 32-bit libraries. As the output
786 for "first char" and "need char" really means "non-UTF-char", I've changed
787 "byte" to "char", and slightly reworded the output. The documentation about
788 these values has also been (I hope) clarified.
789
790 10. Another JIT related optimization: use table jumps for selecting the correct
791 backtracking path, when more than four alternatives are present inside a
792 bracket.
793
794 11. Empty match is not possible, when the minimum length is greater than zero,
795 and there is no \K in the pattern. JIT should avoid empty match checks in
796 such cases.
797
798 12. In a caseless character class with UCP support, when a character with more
799 than one alternative case was not the first character of a range, not all
800 the alternative cases were added to the class. For example, s and \x{17f}
801 are both alternative cases for S: the class [RST] was handled correctly,
802 but [R-T] was not.
803
804 13. The configure.ac file always checked for pthread support when JIT was
805 enabled. This is not used in Windows, so I have put this test inside a
806 check for the presence of windows.h (which was already tested for).
807
808 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
809 The algorithm provides a way to skip certain starting offsets, and usually
810 faster than linear prefix searches.
811
812 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
813 as for 'fr_FR' and 'french'. For some reason, however, it then used the
814 Windows-specific input and output files, which have 'french' screwed in.
815 So this could never have worked. One of the problems with locales is that
816 they aren't always the same. I have now updated RunTest so that it checks
817 the output of the locale test (test 3) against three different output
818 files, and it allows the test to pass if any one of them matches. With luck
819 this should make the test pass on some versions of Solaris where it was
820 failing. Because of the uncertainty, the script did not used to stop if
821 test 3 failed; it now does. If further versions of a French locale ever
822 come to light, they can now easily be added.
823
824 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
825 there was a message during ./configure, but it did not stop. This now
826 provokes an error. The invalid example in README has been corrected.
827 If a value less than the minimum is given, the minimum value has always
828 been used, but now a warning is given.
829
830 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
831 was a bug in the test system, which is now fixed. Also, the list of various
832 configurations that are tested for each release did not have one with both
833 16/32 bits and --enable-bar-anycrlf. It now does.
834
835 18. pcretest was missing "-C bsr" for displaying the \R default setting.
836
837 19. Little endian PowerPC systems are supported now by the JIT compiler.
838
839 20. The fast forward newline mechanism could enter to an infinite loop on
840 certain invalid UTF-8 input. Although we don't support these cases
841 this issue can be fixed by a performance optimization.
842
843 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
844 not take account if existing stack usage. There is now a new global
845 variable called pcre_stack_guard that can be set to point to an external
846 function to check stack availability. It is called at the start of
847 processing every parenthesized group.
848
849 22. A typo in the code meant that in ungreedy mode the max/min qualifier
850 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
851 not match "ab".
852
853 23. When UTF was disabled, the JIT program reported some incorrect compile
854 errors. These messages are silenced now.
855
856 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
857 compiler.
858
859 25. Change all the temporary files used in RunGrepTest to be different to those
860 used by RunTest so that the tests can be run simultaneously, for example by
861 "make -j check".
862
863
864 Version 8.34 15-December-2013
865 -----------------------------
866
867 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
868 executable memory. Patch inspired by Carsten Klein.
869
870 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
871 this macro is never tested and has no effect, because the work to support
872 coverage involves only compiling and linking options and special targets in
873 the Makefile. The comment in config.h implied that defining the macro would
874 enable coverage support, which is totally false. There was also support for
875 setting this macro in the CMake files (my fault, I just copied it from
876 configure). SUPPORT_GCOV has now been removed.
877
878 3. Make a small performance improvement in strlen16() and strlen32() in
879 pcretest.
880
881 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
882 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
883 these cases). There was also one in pcretest.c.
884
885 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
886
887 6. In UTF mode, the code for checking whether a group could match an empty
888 string (which is used for indefinitely repeated groups to allow for
889 breaking an infinite loop) was broken when the group contained a repeated
890 negated single-character class with a character that occupied more than one
891 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
892 UTF-8 mode). The effect was undefined: the group might or might not be
893 deemed as matching an empty string, or the program might have crashed.
894
895 7. The code for checking whether a group could match an empty string was not
896 recognizing that \h, \H, \v, \V, and \R must match a character.
897
898 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
899 an empty string. If it can, pcretest shows this in its information output.
900
901 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
902 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
903 matched by pcre_exec() without using JIT:
904
905 (a) If the rest of the pattern did not match after a maximal run of
906 grapheme clusters, the code for backing up to try with fewer of them
907 did not always back up over a full grapheme when characters that do not
908 have the modifier quality were involved, e.g. Hangul syllables.
909
910 (b) If the match point in a subject started with modifier character, and
911 there was no match, the code could incorrectly back up beyond the match
912 point, and potentially beyond the first character in the subject,
913 leading to a segfault or an incorrect match result.
914
915 10. A conditional group with an assertion condition could lead to PCRE
916 recording an incorrect first data item for a match if no other first data
917 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
918 first data item, and therefore matched "ca" after "c" instead of at the
919 start.
920
921 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
922 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
923
924 12. The source of pcregrep now includes z/OS-specific code so that it can be
925 compiled for z/OS as part of the special z/OS distribution.
926
927 13. Added the -T and -TM options to pcretest.
928
929 14. The code in pcre_compile.c for creating the table of named capturing groups
930 has been refactored. Instead of creating the table dynamically during the
931 actual compiling pass, the information is remembered during the pre-compile
932 pass (on the stack unless there are more than 20 named groups, in which
933 case malloc() is used) and the whole table is created before the actual
934 compile happens. This has simplified the code (it is now nearly 150 lines
935 shorter) and prepared the way for better handling of references to groups
936 with duplicate names.
937
938 15. A back reference to a named subpattern when there is more than one of the
939 same name now checks them in the order in which they appear in the pattern.
940 The first one that is set is used for the reference. Previously only the
941 first one was inspected. This change makes PCRE more compatible with Perl.
942
943 16. Unicode character properties were updated from Unicode 6.3.0.
944
945 17. The compile-time code for auto-possessification has been refactored, based
946 on a patch by Zoltan Herczeg. It now happens after instead of during
947 compilation. The code is cleaner, and more cases are handled. The option
948 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
949 options in pcretest are provided to set it. It can also be set by
950 (*NO_AUTO_POSSESS) at the start of a pattern.
951
952 18. The character VT has been added to the default ("C" locale) set of
953 characters that match \s and are generally treated as white space,
954 following this same change in Perl 5.18. There is now no difference between
955 "Perl space" and "POSIX space". Whether VT is treated as white space in
956 other locales depends on the locale.
957
958 19. The code for checking named groups as conditions, either for being set or
959 for being recursed, has been refactored (this is related to 14 and 15
960 above). Processing unduplicated named groups should now be as fast at
961 numerical groups, and processing duplicated groups should be faster than
962 before.
963
964 20. Two patches to the CMake build system, by Alexander Barkov:
965
966 (1) Replace the "source" command by "." in CMakeLists.txt because
967 "source" is a bash-ism.
968
969 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
970 without these the CMake build does not work on Solaris.
971
972 21. Perl has changed its handling of \8 and \9. If there is no previously
973 encountered capturing group of those numbers, they are treated as the
974 literal characters 8 and 9 instead of a binary zero followed by the
975 literals. PCRE now does the same.
976
977 22. Following Perl, added \o{} to specify codepoints in octal, making it
978 possible to specify values greater than 0777 and also making them
979 unambiguous.
980
981 23. Perl now gives an error for missing closing braces after \x{... instead of
982 treating the string as literal. PCRE now does the same.
983
984 24. RunTest used to grumble if an inappropriate test was selected explicitly,
985 but just skip it when running all tests. This make it awkward to run ranges
986 of tests when one of them was inappropriate. Now it just skips any
987 inappropriate tests, as it always did when running all tests.
988
989 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
990 character types such as \d or \w, too many callouts were inserted, and the
991 data that they returned was rubbish.
992
993 26. In UCP mode, \s was not matching two of the characters that Perl matches,
994 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
995 were matched by \h. The code has now been refactored so that the lists of
996 the horizontal and vertical whitespace characters used for \h and \v (which
997 are defined only in one place) are now also used for \s.
998
999 27. Add JIT support for the 64 bit TileGX architecture.
1000 Patch by Jiong Wang (Tilera Corporation).
1001
1002 28. Possessive quantifiers for classes (both explicit and automatically
1003 generated) now use special opcodes instead of wrapping in ONCE brackets.
1004
1005 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
1006 (because it's meaningless), this was not happening when PCRE_CASELESS was
1007 set. Not wrong, but inefficient.
1008
1009 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
1010 properties for \w, \d, etc) is present in a test regex. Otherwise if the
1011 test contains no characters greater than 255, Perl doesn't realise it
1012 should be using Unicode semantics.
1013
1014 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
1015 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
1016 does in Unicode mode.
1017
1018 32. Added the "forbid" facility to pcretest so that putting tests into the
1019 wrong test files can sometimes be quickly detected.
1020
1021 33. There is now a limit (default 250) on the depth of nesting of parentheses.
1022 This limit is imposed to control the amount of system stack used at compile
1023 time. It can be changed at build time by --with-parens-nest-limit=xxx or
1024 the equivalent in CMake.
1025
1026 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
1027 errors. Perl warns for these when in warning mode, but PCRE has no facility
1028 for giving warnings.
1029
1030 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
1031 However, this was not working for (?!) because it is optimized to (*FAIL),
1032 for which PCRE does not allow quantifiers. The optimization is now disabled
1033 when a quantifier follows (?!). I can't see any use for this, but it makes
1034 things uniform.
1035
1036 36. Perl no longer allows group names to start with digits, so I have made this
1037 change also in PCRE. It simplifies the code a bit.
1038
1039 37. In extended mode, Perl ignores spaces before a + that indicates a
1040 possessive quantifier. PCRE allowed a space before the quantifier, but not
1041 before the possessive +. It now does.
1042
1043 38. The use of \K (reset reported match start) within a repeated possessive
1044 group such as (a\Kb)*+ was not working.
1045
1046 40. Document that the same character tables must be used at compile time and
1047 run time, and that the facility to pass tables to pcre_exec() and
1048 pcre_dfa_exec() is for use only with saved/restored patterns.
1049
1050 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
1051 features for Builds with MSVC:
1052
1053 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
1054 builds). The .rc files can be used to set FileDescription and many other
1055 attributes.
1056
1057 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
1058 This allows higher-level build scripts which want .pdb files to avoid
1059 hard-coding the exact files needed."
1060
1061 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
1062 mean "start of word" and "end of word", respectively, as a transition aid.
1063
1064 43. A minimizing repeat of a class containing codepoints greater than 255 in
1065 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
1066 compiled to use the heap for recursion.
1067
1068 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
1069 is configured.
1070
1071
1072 Version 8.33 28-May-2013
1073 ------------------------
1074
1075 1. Added 'U' to some constants that are compared to unsigned integers, to
1076 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
1077 variables that are added to signed variables, to ensure the result is
1078 signed and can be negated.
1079
1080 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
1081 CMake config files.
1082
1083 3. Revise the creation of config.h.generic so that all boolean macros are
1084 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
1085 overriding via -D on the command line possible.
1086
1087 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
1088 to unsigned int is reported to make a quite noticeable speed difference in
1089 a specific Windows environment. Testing on Linux did also appear to show
1090 some benefit (and it is clearly not harmful). Also fixed the definition of
1091 Xop which should be unsigned.
1092
1093 5. Related to (4), changing the definition of the intermediate variable cc
1094 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
1095 improvements.
1096
1097 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
1098 unnecessary spaces.
1099
1100 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
1101 and later.
1102
1103 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
1104
1105 9. Optimizing fast_forward_start_bits in JIT.
1106
1107 10. Adding support for callouts in JIT, and fixing some issues revealed
1108 during this work. Namely:
1109
1110 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
1111
1112 (b) Minimum length was not checked before the matching is started.
1113
1114 11. The value of capture_last that is passed to callouts was incorrect in some
1115 cases when there was a capture on one path that was subsequently abandoned
1116 after a backtrack. Also, the capture_last value is now reset after a
1117 recursion, since all captures are also reset in this case.
1118
1119 12. The interpreter no longer returns the "too many substrings" error in the
1120 case when an overflowing capture is in a branch that is subsequently
1121 abandoned after a backtrack.
1122
1123 13. In the pathological case when an offset vector of size 2 is used, pcretest
1124 now prints out the matched string after a yield of 0 or 1.
1125
1126 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
1127 Only supported by the JIT compiler at the moment.
1128
1129 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
1130
1131 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
1132 offset of the starting point of the matching process, provided the offsets
1133 vector is large enough.
1134
1135 17. The \A escape now records a lookbehind value of 1, though its execution
1136 does not actually inspect the previous character. This is to ensure that,
1137 in partial multi-segment matching, at least one character from the old
1138 segment is retained when a new segment is processed. Otherwise, if there
1139 are no lookbehinds in the pattern, \A might match incorrectly at the start
1140 of a new segment.
1141
1142 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
1143
1144 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
1145 gives some modest performance improvement in 8-bit mode.
1146
1147 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
1148 be expressed in certain programming languages using Universal Character
1149 Names.
1150
1151 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
1152 which points out that "non characters" are not "characters that may not
1153 appear in Unicode strings" but rather "characters that are reserved for
1154 internal use and have only local meaning".
1155
1156 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
1157 there was a conditional group that depended on an assertion, if the
1158 assertion was false, the callout that immediately followed the alternation
1159 in the condition was skipped when pcre_exec() was used for matching.
1160
1161 23. Allow an explicit callout to be inserted before an assertion that is the
1162 condition for a conditional group, for compatibility with automatic
1163 callouts, which always insert a callout at this point.
1164
1165 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1166 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1167
1168 25. (*PRUNE) is now supported by the JIT compiler.
1169
1170 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1171
1172 27. Fix the case where there are two or more SKIPs with arguments that may be
1173 ignored.
1174
1175 28. (*SKIP) is now supported by the JIT compiler.
1176
1177 29. (*THEN) is now supported by the JIT compiler.
1178
1179 30. Update RunTest with additional test selector options.
1180
1181 31. The way PCRE handles backtracking verbs has been changed in two ways.
1182
1183 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1184 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1185 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1186 rather obscure rules do not always do the same thing.
1187
1188 (2) Previously, backtracking verbs were confined within assertions. This is
1189 no longer the case for positive assertions, except for (*ACCEPT). Again,
1190 this sometimes improves Perl compatibility, and sometimes does not.
1191
1192 32. A number of tests that were in test 2 because Perl did things differently
1193 have been moved to test 1, because either Perl or PCRE has changed, and
1194 these tests are now compatible.
1195
1196 32. Backtracking control verbs are now handled in the same way in JIT and
1197 interpreter.
1198
1199 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1200 contained a forward subroutine reference caused a compile error.
1201
1202 34. Auto-detect and optimize limited repetitions in JIT.
1203
1204 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1205 blocking (*UTF) etc.
1206
1207 36. In the interpreter, maximizing pattern repetitions for characters and
1208 character types now use tail recursion, which reduces stack usage.
1209
1210 37. The value of the max lookbehind was not correctly preserved if a compiled
1211 and saved regex was reloaded on a host of different endianness.
1212
1213 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1214 of the compiled pattern block, expand the flags field from 16 to 32 bits
1215 because it was almost full.
1216
1217 39. Try madvise first before posix_madvise.
1218
1219 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1220 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1221 original change locked out all matches of empty strings. This has been
1222 changed so that one match of an empty string per line is recognized.
1223 Subsequent searches on the same line (for colouring or for --only-matching,
1224 for example) do not recognize empty strings.
1225
1226 41. Applied a user patch to fix a number of spelling mistakes in comments.
1227
1228 42. Data lines longer than 65536 caused pcretest to crash.
1229
1230 43. Clarified the data type for length and startoffset arguments for pcre_exec
1231 and pcre_dfa_exec in the function-specific man pages, where they were
1232 explicitly stated to be in bytes, never having been updated. I also added
1233 some clarification to the pcreapi man page.
1234
1235 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1236 a segmentation fault.
1237
1238
1239 Version 8.32 30-November-2012
1240 -----------------------------
1241
1242 1. Improved JIT compiler optimizations for first character search and single
1243 character iterators.
1244
1245 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1246 Patch by Daniel Richard G.
1247
1248 3. Single character iterator optimizations in the JIT compiler.
1249
1250 4. Improved JIT compiler optimizations for character ranges.
1251
1252 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1253 Reported by Giuseppe D'Angelo.
1254
1255 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1256 of a line, was being set incorrectly in cases where .* appeared inside
1257 atomic brackets at the start of a pattern, or where there was a subsequent
1258 *PRUNE or *SKIP.
1259
1260 7. Improved instruction cache flush for POWER/PowerPC.
1261 Patch by Daniel Richard G.
1262
1263 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1264 grep:
1265
1266 (a) There is now no limit to the number of patterns to be matched.
1267
1268 (b) An error is given if a pattern is too long.
1269
1270 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1271 are now supported.
1272
1273 (d) --exclude-from and --include-from (multiple use) have been added.
1274
1275 (e) Exclusions and inclusions now apply to all files and directories, not
1276 just to those obtained from scanning a directory recursively.
1277
1278 (f) Multiple uses of -f and --file-list are now supported.
1279
1280 (g) In a Windows environment, the default for -d has been changed from
1281 "read" (the GNU grep default) to "skip", because otherwise the presence
1282 of a directory in the file list provokes an error.
1283
1284 (h) The documentation has been revised and clarified in places.
1285
1286 9. Improve the matching speed of capturing brackets.
1287
1288 10. Changed the meaning of \X so that it now matches a Unicode extended
1289 grapheme cluster.
1290
1291 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1292 out POSIX threads when JIT support is configured.
1293
1294 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1295
1296 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1297 structure could go wrong in environments where size_t is not the same size
1298 as int.
1299
1300 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1301 set.
1302
1303 15. The EBCDIC support had decayed; later updates to the code had included
1304 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1305 general tidy up of EBCDIC-related issues, and the documentation was also
1306 not quite right. There is now a test that can be run on ASCII systems to
1307 check some of the EBCDIC-related things (but is it not a full test).
1308
1309 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1310 in a small tidy to the code.
1311
1312 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1313
1314 18. If the --only-matching (-o) option in pcregrep is specified multiple
1315 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1316 substrings matched by the 1st and 2nd capturing parentheses. A separating
1317 string can be specified by --om-separator (default empty).
1318
1319 19. Improving the first n character searches.
1320
1321 20. Turn case lists for horizontal and vertical white space into macros so that
1322 they are defined only once.
1323
1324 21. This set of changes together give more compatible Unicode case-folding
1325 behaviour for characters that have more than one other case when UCP
1326 support is available.
1327
1328 (a) The Unicode property table now has offsets into a new table of sets of
1329 three or more characters that are case-equivalent. The MultiStage2.py
1330 script that generates these tables (the pcre_ucd.c file) now scans
1331 CaseFolding.txt instead of UnicodeData.txt for character case
1332 information.
1333
1334 (b) The code for adding characters or ranges of characters to a character
1335 class has been abstracted into a generalized function that also handles
1336 case-independence. In UTF-mode with UCP support, this uses the new data
1337 to handle characters with more than one other case.
1338
1339 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1340 whose other case is greater than 256 are now correctly matched
1341 caselessly. Previously, the high codepoint matched the low one, but not
1342 vice versa.
1343
1344 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1345 of the new class addition function, using character lists defined as
1346 macros alongside the case definitions of 20 above.
1347
1348 (e) Caseless back references now work with characters that have more than
1349 one other case.
1350
1351 (f) General caseless matching of characters with more than one other case
1352 is supported.
1353
1354 22. Unicode character properties were updated from Unicode 6.2.0
1355
1356 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1357
1358 24. Add support for 32-bit character strings, and UTF-32
1359
1360 25. Major JIT compiler update (code refactoring and bugfixing).
1361 Experimental Sparc 32 support is added.
1362
1363 26. Applied a modified version of Daniel Richard G's patch to create
1364 pcre.h.generic and config.h.generic by "make" instead of in the
1365 PrepareRelease script.
1366
1367 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1368 pcre_compile.c when checking for a zero character.
1369
1370 28. Introducing a native interface for JIT. Through this interface, the compiled
1371 machine code can be directly executed. The purpose of this interface is to
1372 provide fast pattern matching, so several sanity checks are not performed.
1373 However, feature tests are still performed. The new interface provides
1374 1.4x speedup compared to the old one.
1375
1376 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1377 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1378 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1379
1380 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1381 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1382 the "old" RFC 2279). Instead, it ended up passing a negative length to
1383 pcre_exec().
1384
1385 31. Add support for GCC's visibility feature to hide internal functions.
1386
1387 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1388 "unknown -C option" after outputting 0 or 1.
1389
1390 33. There is now support for generating a code coverage report for the test
1391 suite in environments where gcc is the compiler and lcov is installed. This
1392 is mainly for the benefit of the developers.
1393
1394 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1395 unaddressable using valgrind annotations, allowing valgrind to detect
1396 invalid memory accesses. This is mainly for the benefit of the developers.
1397
1398 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1399
1400 26. Give configure error if --enable-cpp but no C++ compiler found.
1401
1402
1403 Version 8.31 06-July-2012
1404 -------------------------
1405
1406 1. Fixing a wrong JIT test case and some compiler warnings.
1407
1408 2. Removed a bashism from the RunTest script.
1409
1410 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1411 to unsigned type, result still unsigned" that was given by an MS compiler
1412 on encountering the code "-sizeof(xxx)".
1413
1414 4. Partial matching support is added to the JIT compiler.
1415
1416 5. Fixed several bugs concerned with partial matching of items that consist
1417 of more than one character:
1418
1419 (a) /^(..)\1/ did not partially match "aba" because checking references was
1420 done on an "all or nothing" basis. This also applied to repeated
1421 references.
1422
1423 (b) \R did not give a hard partial match if \r was found at the end of the
1424 subject.
1425
1426 (c) \X did not give a hard partial match after matching one or more
1427 characters at the end of the subject.
1428
1429 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1430 a partial match for the string "\r".
1431
1432 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1433 a partial match for a CR character at the end of the subject string.
1434
1435 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1436 running pcretest, the text "(JIT)" added to the output whenever JIT is
1437 actually used to run the match.
1438
1439 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1440 or /S+[+] with a digit between 1 and 7.
1441
1442 8. OP_NOT now supports any UTF character not just single-byte ones.
1443
1444 9. (*MARK) control verb is now supported by the JIT compiler.
1445
1446 10. The command "./RunTest list" lists the available tests without actually
1447 running any of them. (Because I keep forgetting what they all are.)
1448
1449 11. Add PCRE_INFO_MAXLOOKBEHIND.
1450
1451 12. Applied a (slightly modified) user-supplied patch that improves performance
1452 when the heap is used for recursion (compiled with --disable-stack-for-
1453 recursion). Instead of malloc and free for each heap frame each time a
1454 logical recursion happens, frames are retained on a chain and re-used where
1455 possible. This sometimes gives as much as 30% improvement.
1456
1457 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1458 call.
1459
1460 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1461
1462 15. It is now possible to link pcretest with libedit as an alternative to
1463 libreadline.
1464
1465 16. (*COMMIT) control verb is now supported by the JIT compiler.
1466
1467 17. The Unicode data tables have been updated to Unicode 6.1.0.
1468
1469 18. Added --file-list option to pcregrep.
1470
1471 19. Added binary file support to pcregrep, including the -a, --binary-files,
1472 -I, and --text options.
1473
1474 20. The madvise function is renamed for posix_madvise for QNX compatibility
1475 reasons. Fixed by Giuseppe D'Angelo.
1476
1477 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1478
1479 22. Changed the option for creating symbolic links for 16-bit man pages from
1480 -s to -sf so that re-installing does not cause issues.
1481
1482 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1483
1484 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1485 matches in certain environments (the workspace was not being correctly
1486 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1487 some of the workspace data at the beginning of a restart.
1488
1489 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1490 was not doing so when it should - probably a typo introduced by SVN 528
1491 (change 8.10/14).
1492
1493 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1494 \w+ when the character tables indicated that \x{c4} was a word character.
1495 There were several related cases, all because the tests for doing a table
1496 lookup were testing for characters less than 127 instead of 255.
1497
1498 27. If a pattern contains capturing parentheses that are not used in a match,
1499 their slots in the ovector are set to -1. For those that are higher than
1500 any matched groups, this happens at the end of processing. In the case when
1501 there were back references that the ovector was too small to contain
1502 (causing temporary malloc'd memory to be used during matching), and the
1503 highest capturing number was not used, memory off the end of the ovector
1504 was incorrectly being set to -1. (It was using the size of the temporary
1505 memory instead of the true size.)
1506
1507 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1508 ovector size, it uses memory at the end of the block that it has got.
1509
1510 29. Check for an overlong MARK name and give an error at compile time. The
1511 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1512
1513 30. JIT compiler update.
1514
1515 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1516 Rill for the patch.
1517
1518 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1519
1520 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1521
1522 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1523 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1524 was enabled.
1525
1526 35. Improve JIT code generation for greedy plus quantifier.
1527
1528 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1529 1 to "aa" instead of to an empty string. The bug affected repeated groups
1530 that could potentially match an empty string.
1531
1532 37. Optimizing single character iterators in JIT.
1533
1534 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1535 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1536 codepoints that are too big for the mode are faulted, and in a UTF mode,
1537 disallowed codepoints are also faulted.
1538
1539 39. If PCRE was compiled with UTF support, in three places in the DFA
1540 matcher there was code that should only have been obeyed in UTF mode, but
1541 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1542 processing when bytes with values greater than 127 were present. In 16-bit
1543 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1544 both cases the values are those that cannot be the first data item in a UTF
1545 character. The three items that might have provoked this were recursions,
1546 possessively repeated groups, and atomic groups.
1547
1548 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1549 and pcregrep, because some OS require shared objects to be explicitly
1550 passed to ld, causing the link step to fail if they are not.
1551
1552 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1553 mode, patterns that started with \h* or \R* might be incorrectly matched.
1554
1555
1556 Version 8.30 04-February-2012
1557 -----------------------------
1558
1559 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1560 name is defined in ctype.h.
1561
1562 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1563 only in quite long subpatterns.
1564
1565 3. Removed the function pcre_info(), which has been obsolete and deprecated
1566 since it was replaced by pcre_fullinfo() in February 2000.
1567
1568 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1569 match a (*MARK), and the match failed at the start of the subject, a
1570 reference to memory before the start of the subject could occur. This bug
1571 was introduced by fix 17 of release 8.21.
1572
1573 5. A reference to an unset group with zero minimum repetition was giving
1574 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1575 /(another)?(\1?)test/ matched against "hello world test". This bug was
1576 introduced in release 8.13.
1577
1578 6. Add support for 16-bit character strings (a large amount of work involving
1579 many changes and refactorings).
1580
1581 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1582 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1583 from a file.
1584
1585 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1586 rounding is not applied in this particular case).
1587
1588 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1589 if they appear, or are escaped, in patterns.
1590
1591 10. Get rid of a number of -Wunused-but-set-variable warnings.
1592
1593 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1594 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1595 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1596 also returns the mark "x". This bug applied to capturing parentheses,
1597 non-capturing parentheses, and atomic parentheses. It also applied to some
1598 assertions.
1599
1600 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1601 information out of configure.ac instead of relying on pcre.h.generic, which
1602 is not stored in the repository.
1603
1604 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1605 -lreadline.
1606
1607 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1608
1609 15. Applied Graycode's patch to put the top-level frame on the stack rather
1610 than the heap when not using the stack for recursion. This gives a
1611 performance improvement in many cases when recursion is not deep.
1612
1613 16. Experimental code added to "pcretest -C" to output the stack frame size.
1614
1615
1616 Version 8.21 12-Dec-2011
1617 ------------------------
1618
1619 1. Updating the JIT compiler.
1620
1621 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1622 are added as well.
1623
1624 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1625 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1626 calling _pcre_jit_exec. Some extra comments are added.
1627
1628 4. (*MARK) settings inside atomic groups that do not contain any capturing
1629 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1630 was introduced by change 18 for 8.20.
1631
1632 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1633 ECMA-262 standard.
1634
1635 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1636 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1637 This bug was probably introduced by change 9 of 8.13.
1638
1639 7. While fixing 6 above, I noticed that a number of other items were being
1640 incorrectly rejected as "not fixed length". This arose partly because newer
1641 opcodes had not been added to the fixed-length checking code. I have (a)
1642 corrected the bug and added tests for these items, and (b) arranged for an
1643 error to occur if an unknown opcode is encountered while checking for fixed
1644 length instead of just assuming "not fixed length". The items that were
1645 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1646 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1647 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1648
1649 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1650 being incorrectly compiled and would have given unpredicatble results.
1651
1652 9. A possessively repeated subpattern with minimum repeat count greater than
1653 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1654 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1655 the first (A) could occur when it should not.
1656
1657 10. Add a cast and remove a redundant test from the code.
1658
1659 11. JIT should use pcre_malloc/pcre_free for allocation.
1660
1661 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1662 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1663 is anything other than /usr, -L is still shown).
1664
1665 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1666
1667 14. Perl does not support \N without a following name in a [] class; PCRE now
1668 also gives an error.
1669
1670 15. If a forward reference was repeated with an upper limit of around 2000,
1671 it caused the error "internal error: overran compiling workspace". The
1672 maximum number of forward references (including repeats) was limited by the
1673 internal workspace, and dependent on the LINK_SIZE. The code has been
1674 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1675 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1676 of around 200,000 forward references. While doing this, I also speeded up
1677 the filling in of repeated forward references.
1678
1679 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1680 incorrectly expecting the subject to contain another "a" after the start.
1681
1682 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1683 in the match, the SKIP should be ignored. This was not happening; instead
1684 the SKIP was being treated as NOMATCH. For patterns such as
1685 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1686 tested.
1687
1688 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1689 now much more compatible with Perl, in particular in cases where the result
1690 is a non-match for a non-anchored pattern. For example, if
1691 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1692 "m", where previously it did not return a name. A side effect of this
1693 change is that for partial matches, the last encountered mark name is
1694 returned, as for non matches. A number of tests that were previously not
1695 Perl-compatible have been moved into the Perl-compatible test files. The
1696 refactoring has had the pleasing side effect of removing one argument from
1697 the match() function, thus reducing its stack requirements.
1698
1699 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1700 subsequent uses of /S (without +) incorrectly behaved like /S+.
1701
1702 21. Retrieve executable code size support for the JIT compiler and fixing
1703 some warnings.
1704
1705 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1706 not work when the shorter character appeared right at the end of the
1707 subject string.
1708
1709 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1710 systems.
1711
1712 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1713 output it when the /M option is used in pcretest.
1714
1715 25. The CheckMan script was not being included in the distribution. Also, added
1716 an explicit "perl" to run Perl scripts from the PrepareRelease script
1717 because this is reportedly needed in Windows.
1718
1719 26. If study data was being save in a file and studying had not found a set of
1720 "starts with" bytes for the pattern, the data written to the file (though
1721 never used) was taken from uninitialized memory and so caused valgrind to
1722 complain.
1723
1724 27. Updated RunTest.bat as provided by Sheri Pierce.
1725
1726 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1727
1728 29. Computation of memory usage for the table of capturing group names was
1729 giving an unnecessarily large value.
1730
1731
1732 Version 8.20 21-Oct-2011
1733 ------------------------
1734
1735 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1736 a POSIX class. After further experiments with Perl, which convinced me that
1737 Perl has bugs and confusions, a closing square bracket is no longer allowed
1738 in a POSIX name. This bug also affected patterns with classes that started
1739 with full stops.
1740
1741 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1742 captured substring, but while checking the failing first alternative,
1743 substring 1 is temporarily captured. If the output vector supplied to
1744 pcre_exec() was not big enough for this capture, the yield of the function
1745 was still zero ("insufficient space for captured substrings"). This cannot
1746 be totally fixed without adding another stack variable, which seems a lot
1747 of expense for a edge case. However, I have improved the situation in cases
1748 such as /(a)(b)x|abc/ matched against "abc", where the return code
1749 indicates that fewer than the maximum number of slots in the ovector have
1750 been set.
1751
1752 3. Related to (2) above: when there are more back references in a pattern than
1753 slots in the output vector, pcre_exec() uses temporary memory during
1754 matching, and copies in the captures as far as possible afterwards. It was
1755 using the entire output vector, but this conflicts with the specification
1756 that only 2/3 is used for passing back captured substrings. Now it uses
1757 only the first 2/3, for compatibility. This is, of course, another edge
1758 case.
1759
1760 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1761 main code base, and can be used by building with --enable-jit. When this is
1762 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1763 runtime --no-jit option is given.
1764
1765 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1766 ovector, the return from the function was zero, implying that there were
1767 other matches that did not fit. The correct "exactly full" value is now
1768 returned.
1769
1770 6. If a subpattern that was called recursively or as a subroutine contained
1771 (*PRUNE) or any other control that caused it to give a non-standard return,
1772 invalid errors such as "Error -26 (nested recursion at the same subject
1773 position)" or even infinite loops could occur.
1774
1775 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1776 computing the minimum length on reaching *ACCEPT, and so ended up with the
1777 wrong value of 1 rather than 0. Further investigation indicates that
1778 computing a minimum subject length in the presence of *ACCEPT is difficult
1779 (think back references, subroutine calls), and so I have changed the code
1780 so that no minimum is registered for a pattern that contains *ACCEPT.
1781
1782 8. If (*THEN) was present in the first (true) branch of a conditional group,
1783 it was not handled as intended. [But see 16 below.]
1784
1785 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1786 Sheri Pierce.
1787
1788 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1789 the first byte in a match must be "a".
1790
1791 11. Change 17 for 8.13 increased the recursion depth for patterns like
1792 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1793 pattern contains any instances of (*THEN). If it does not, the old
1794 optimizations are restored. It would be nice to do this on a per-group
1795 basis, but at the moment that is not feasible.
1796
1797 12. In some environments, the output of pcretest -C is CRLF terminated. This
1798 broke RunTest's code that checks for the link size. A single white space
1799 character after the value is now allowed for.
1800
1801 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1802 For "fr", it uses the Windows-specific input and output files.
1803
1804 14. If (*THEN) appeared in a group that was called recursively or as a
1805 subroutine, it did not work as intended. [But see next item.]
1806
1807 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1808 pattern fragments (but not containing any | characters). If A and B are
1809 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1810 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1811 D. In other words, Perl considers parentheses that do not contain any |
1812 characters to be part of a surrounding alternative, whereas PCRE was
1813 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1814 differently. PCRE now behaves in the same way as Perl, except in the case
1815 of subroutine/recursion calls such as (?1) which have in any case always
1816 been different (but PCRE had them first :-).
1817
1818 16. Related to 15 above: Perl does not treat the | in a conditional group as
1819 creating alternatives. Such a group is treated in the same way as an
1820 ordinary group without any | characters when processing (*THEN). PCRE has
1821 been changed to match Perl's behaviour.
1822
1823 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1824 RunGrepTest script failed.
1825
1826 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1827 inevitable for groups that contain captures, but it can lead to a lot of
1828 stack use in large patterns. The old behaviour has been restored for atomic
1829 groups that do not contain any capturing parentheses.
1830
1831 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1832 suppress the check for a minimum subject length at run time. (If it was
1833 given to pcre_exec() or pcre_dfa_exec() it did work.)
1834
1835 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1836 fail to work when decoding hex characters in data strings in EBCDIC
1837 environments.
1838
1839 21. It appears that in at least one Mac OS environment, the isxdigit() function
1840 is implemented as a macro that evaluates to its argument more than once,
1841 contravening the C 90 Standard (I haven't checked a later standard). There
1842 was an instance in pcretest which caused it to go wrong when processing
1843 \x{...} escapes in subject strings. The has been rewritten to avoid using
1844 things like p++ in the argument of isxdigit().
1845
1846
1847 Version 8.13 16-Aug-2011
1848 ------------------------
1849
1850 1. The Unicode data tables have been updated to Unicode 6.0.0.
1851
1852 2. Two minor typos in pcre_internal.h have been fixed.
1853
1854 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1855 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1856 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1857
1858 4. There were a number of related bugs in the code for matching backrefences
1859 caselessly in UTF-8 mode when codes for the characters concerned were
1860 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1861 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1862 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1863 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1864 2-byte code at the end of the subject (it thought there wasn't enough data
1865 left).
1866
1867 5. Comprehensive information about what went wrong is now returned by
1868 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1869 as the output vector has at least 2 elements. The offset of the start of
1870 the failing character and a reason code are placed in the vector.
1871
1872 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1873 now returned is for the first byte of the failing character, instead of the
1874 last byte inspected. This is an incompatible change, but I hope it is small
1875 enough not to be a problem. It makes the returned offset consistent with
1876 pcre_exec() and pcre_dfa_exec().
1877
1878 7. pcretest now gives a text phrase as well as the error number when
1879 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1880 failure, the offset and reason code are output.
1881
1882 8. When \R was used with a maximizing quantifier it failed to skip backwards
1883 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1884 back over a single character (\n). This seems wrong (because it treated the
1885 two characters as a single entity when going forwards), conflicts with the
1886 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1887 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1888 has been changed.
1889
1890 9. Some internal refactoring has changed the processing so that the handling
1891 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1892 time (the PCRE_DOTALL option was changed this way some time ago: version
1893 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1894 which was always a bit of a fudge. It also means that there is one less
1895 argument for the match() function, which reduces its stack requirements
1896 slightly. This change also fixes an incompatibility with Perl: the pattern
1897 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1898
1899 10. More internal refactoring has drastically reduced the number of recursive
1900 calls to match() for possessively repeated groups such as (abc)++ when
1901 using pcre_exec().
1902
1903 11. While implementing 10, a number of bugs in the handling of groups were
1904 discovered and fixed:
1905
1906 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1907 (a|)*(?1) gave a compile-time internal error.
1908 ((a|)+)+ did not notice that the outer group could match an empty string.
1909 (^a|^)+ was not marked as anchored.
1910 (.*a|.*)+ was not marked as matching at start or after a newline.
1911
1912 12. Yet more internal refactoring has removed another argument from the match()
1913 function. Special calls to this function are now indicated by setting a
1914 value in a variable in the "match data" data block.
1915
1916 13. Be more explicit in pcre_study() instead of relying on "default" for
1917 opcodes that mean there is no starting character; this means that when new
1918 ones are added and accidentally left out of pcre_study(), testing should
1919 pick them up.
1920
1921 14. The -s option of pcretest has been documented for ages as being an old
1922 synonym of -m (show memory usage). I have changed it to mean "force study
1923 for every regex", that is, assume /S for every regex. This is similar to -i
1924 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1925 using it. It makes it easier to run collections of tests with and without
1926 study enabled, and thereby test pcre_study() more easily. All the standard
1927 tests are now run with and without -s (but some patterns can be marked as
1928 "never study" - see 20 below).
1929
1930 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1931 restoration of the capturing data to the outer values was not happening
1932 correctly.
1933
1934 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1935 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1936 pattern had matched an empty string, and so incorrectly returned a no
1937 match.
1938
1939 17. There was optimizing code for the last branch of non-capturing parentheses,
1940 and also for the obeyed branch of a conditional subexpression, which used
1941 tail recursion to cut down on stack usage. Unfortunately, now that there is
1942 the possibility of (*THEN) occurring in these branches, tail recursion is
1943 no longer possible because the return has to be checked for (*THEN). These
1944 two optimizations have therefore been removed. [But see 8.20/11 above.]
1945
1946 18. If a pattern containing \R was studied, it was assumed that \R always
1947 matched two bytes, thus causing the minimum subject length to be
1948 incorrectly computed because \R can also match just one byte.
1949
1950 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1951 was incorrectly computed.
1952
1953 20. If /S is present twice on a test pattern in pcretest input, it now
1954 *disables* studying, thereby overriding the use of -s on the command line
1955 (see 14 above). This is necessary for one or two tests to keep the output
1956 identical in both cases.
1957
1958 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1959 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1960
1961 22. When an atomic group that contained a capturing parenthesis was
1962 successfully matched, but the branch in which it appeared failed, the
1963 capturing was not being forgotten if a higher numbered group was later
1964 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1965 group 1 to "a", when in fact it should be unset. This applied to multi-
1966 branched capturing and non-capturing groups, repeated or not, and also to
1967 positive assertions (capturing in negative assertions does not happen
1968 in PCRE) and also to nested atomic groups.
1969
1970 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1971 subject after a captured substring, to make it easier to tell which of a
1972 number of identical substrings has been captured.
1973
1974 24. The way atomic groups are processed by pcre_exec() has been changed so that
1975 if they are repeated, backtracking one repetition now resets captured
1976 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1977 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1978 "aaa". Previously, it would have been "a". As part of this code
1979 refactoring, the way recursive calls are handled has also been changed.
1980
1981 25. If an assertion condition captured any substrings, they were not passed
1982 back unless some other capturing happened later. For example, if
1983 (?(?=(a))a) was matched against "a", no capturing was returned.
1984
1985 26. When studying a pattern that contained subroutine calls or assertions,
1986 the code for finding the minimum length of a possible match was handling
1987 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1988 group 1 called group 2 while simultaneously a separate group 2 called group
1989 1). A stack overflow occurred in this case. I have fixed this by limiting
1990 the recursion depth to 10.
1991
1992 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1993 Fortmann. This supports explicit test numbers on the command line, and has
1994 argument validation and error reporting.
1995
1996 28. An instance of \X with an unlimited repeat could fail if at any point the
1997 first character it looked at was a mark character.
1998
1999 29. Some minor code refactoring concerning Unicode properties and scripts
2000 should reduce the stack requirement of match() slightly.
2001
2002 30. Added the '=' option to pcretest to check the setting of unused capturing
2003 slots at the end of the pattern, which are documented as being -1, but are
2004 not included in the return count.
2005
2006 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
2007 compiled something random. Now it gives a compile-time error (as does
2008 Perl).
2009
2010 32. A *MARK encountered during the processing of a positive assertion is now
2011 recorded and passed back (compatible with Perl).
2012
2013 33. If --only-matching or --colour was set on a pcregrep call whose pattern
2014 had alternative anchored branches, the search for a second match in a line
2015 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
2016 matched the line "0102" twice. The same bug affected patterns that started
2017 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
2018 twice.
2019
2020 34. Previously, PCRE did not allow quantification of assertions. However, Perl
2021 does, and because of capturing effects, quantifying parenthesized
2022 assertions may at times be useful. Quantifiers are now allowed for
2023 parenthesized assertions.
2024
2025 35. A minor code tidy in pcre_compile() when checking options for \R usage.
2026
2027 36. \g was being checked for fancy things in a character class, when it should
2028 just be a literal "g".
2029
2030 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
2031 appearance of a nested POSIX class supersedes an apparent external class.
2032 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
2033 unescaped square brackets may also appear as part of class names. For
2034 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
2035 more like Perl. (But see 8.20/1 above.)
2036
2037 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
2038 was because it thought it was \N{name}, which is not supported).
2039
2040 39. Add minix to OS list not supporting the -S option in pcretest.
2041
2042 40. PCRE tries to detect cases of infinite recursion at compile time, but it
2043 cannot analyze patterns in sufficient detail to catch mutual recursions
2044 such as ((?1))((?2)). There is now a runtime test that gives an error if a
2045 subgroup is called recursively as a subpattern for a second time at the
2046 same position in the subject string. In previous releases this might have
2047 been caught by the recursion limit, or it might have run out of stack.
2048
2049 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
2050 happen only once. PCRE was, however incorrectly giving a compile time error
2051 "recursive call could loop indefinitely" because it cannot analyze the
2052 pattern in sufficient detail. The compile time test no longer happens when
2053 PCRE is compiling a conditional subpattern, but actual runaway loops are
2054 now caught at runtime (see 40 above).
2055
2056 42. It seems that Perl allows any characters other than a closing parenthesis
2057 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
2058 has been changed to be the same.
2059
2060 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
2061 as not to get warnings when autogen.sh is called. Also changed
2062 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
2063
2064 44. To help people who use pcregrep to scan files containing exceedingly long
2065 lines, the following changes have been made:
2066
2067 (a) The default value of the buffer size parameter has been increased from
2068 8K to 20K. (The actual buffer used is three times this size.)
2069
2070 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
2071 PCRE is built.
2072
2073 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
2074 to be set at run time.
2075
2076 (d) Numerical values in pcregrep options can be followed by K or M, for
2077 example --buffer-size=50K.
2078
2079 (e) If a line being scanned overflows pcregrep's buffer, an error is now
2080 given and the return code is set to 2.
2081
2082 45. Add a pointer to the latest mark to the callout data block.
2083
2084 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
2085 partial match of an empty string instead of no match. This was specific to
2086 the use of ".".
2087
2088 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
2089 complete match instead of a partial match. This bug was dependent on both
2090 the PCRE_UTF8 and PCRE_DOTALL options being set.
2091
2092 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
2093 starting byte set, because \b was not being ignored.
2094
2095
2096 Version 8.12 15-Jan-2011
2097 ------------------------
2098
2099 1. Fixed some typos in the markup of the man pages, and wrote a script that
2100 checks for such things as part of the documentation building process.
2101
2102 2. On a big-endian 64-bit system, pcregrep did not correctly process the
2103 --match-limit and --recursion-limit options (added for 8.11). In
2104 particular, this made one of the standard tests fail. (The integer value
2105 went into the wrong half of a long int.)
2106
2107 3. If the --colour option was given to pcregrep with -v (invert match), it
2108 did strange things, either producing crazy output, or crashing. It should,
2109 of course, ignore a request for colour when reporting lines that do not
2110 match.
2111
2112 4. Another pcregrep bug caused similar problems if --colour was specified with
2113 -M (multiline) and the pattern match finished with a line ending.
2114
2115 5. In pcregrep, when a pattern that ended with a literal newline sequence was
2116 matched in multiline mode, the following line was shown as part of the
2117 match. This seems wrong, so I have changed it.
2118
2119 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
2120 the check for further matches in the same line (so they could be coloured)
2121 to overrun the end of the current line. If another match was found, it was
2122 incorrectly shown (and then shown again when found in the next line).
2123
2124 7. If pcregrep was compiled under Windows, there was a reference to the
2125 function pcregrep_exit() before it was defined. I am assuming this was
2126 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
2127 reported by a user. I've moved the definition above the reference.
2128
2129
2130 Version 8.11 10-Dec-2010
2131 ------------------------
2132
2133 1. (*THEN) was not working properly if there were untried alternatives prior
2134 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
2135 backtracked to try for "b" instead of moving to the next alternative branch
2136 at the same level (in this case, to look for "c"). The Perl documentation
2137 is clear that when (*THEN) is backtracked onto, it goes to the "next
2138 alternative in the innermost enclosing group".
2139
2140 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
2141 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
2142 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
2143 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
2144 (*THEN).
2145
2146 3. If \s appeared in a character class, it removed the VT character from
2147 the class, even if it had been included by some previous item, for example
2148 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
2149 of \s, but is part of the POSIX "space" class.)
2150
2151 4. A partial match never returns an empty string (because you can always
2152 match an empty string at the end of the subject); however the checking for
2153 an empty string was starting at the "start of match" point. This has been
2154 changed to the "earliest inspected character" point, because the returned
2155 data for a partial match starts at this character. This means that, for
2156 example, /(?<=abc)def/ gives a partial match for the subject "abc"
2157 (previously it gave "no match").
2158
2159 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2160 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2161 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2162 has an implication that the given string is incomplete (because a partial
2163 match is preferred over a full match). For this reason, these items now
2164 give a partial match in this situation. [Aside: previously, the one case
2165 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2166 match rather than a full match, which was wrong by the old rules, but is
2167 now correct.]
2168
2169 6. There was a bug in the handling of #-introduced comments, recognized when
2170 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2171 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2172 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2173 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2174 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2175 places in pcre_compile().
2176
2177 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2178 comments when looking ahead for named forward references to subpatterns,
2179 the only newline sequence it recognized was NL. It now handles newlines
2180 according to the set newline convention.
2181
2182 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2183 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2184 cater for a lack of strerror(). These oversights have been fixed.
2185
2186 9. Added --match-limit and --recursion-limit to pcregrep.
2187
2188 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2189
2190 11. When the -o option was used, pcregrep was setting a return code of 1, even
2191 when matches were found, and --line-buffered was not being honoured.
2192
2193 12. Added an optional parentheses number to the -o and --only-matching options
2194 of pcregrep.
2195
2196 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2197 can match an empty string. The code to do it in pcretest and pcredemo
2198 needed fixing:
2199
2200 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2201 only one byte after an empty string match just before CRLF (this case
2202 just got forgotten; "any" and "anycrlf" were OK).
2203
2204 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2205 mode when an empty string match preceded an ASCII character followed by
2206 a non-ASCII character. (The code for advancing by one character rather
2207 than one byte was nonsense.)
2208
2209 (c) The pcredemo.c sample program did not have any code at all to handle
2210 the cases when CRLF is a valid newline sequence.
2211
2212 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2213 as a starting offset was within the subject string. There is now a new
2214 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2215 negative or greater than the length of the string. In order to test this,
2216 pcretest is extended to allow the setting of negative starting offsets.
2217
2218 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2219 starting offset points to the beginning of a UTF-8 character was
2220 unnecessarily clumsy. I tidied it up.
2221
2222 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2223 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2224
2225 17. Nobody had reported that the --include_dir option, which was added in
2226 release 7.7 should have been called --include-dir (hyphen, not underscore)
2227 for compatibility with GNU grep. I have changed it to --include-dir, but
2228 left --include_dir as an undocumented synonym, and the same for
2229 --exclude-dir, though that is not available in GNU grep, at least as of
2230 release 2.5.4.
2231
2232 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2233 characters from a string of bytes) have been redefined so as not to use
2234 loops, in order to improve performance in some environments. At the same
2235 time, I abstracted some of the common code into auxiliary macros to save
2236 repetition (this should not affect the compiled code).
2237
2238 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2239 compile-time error is now given if \c is not followed by an ASCII
2240 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2241 different, and any byte value is allowed.)
2242
2243 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2244 START_OPTIMIZE option, which is now allowed at compile time - but just
2245 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2246 to pcregrep and other applications that have no direct access to PCRE
2247 options. The new /Y option in pcretest sets this option when calling
2248 pcre_compile().
2249
2250 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2251 back references. Groups containing recursive back references were forced to
2252 be atomic by that change, but in the case of named groups, the amount of
2253 memory required was incorrectly computed, leading to "Failed: internal
2254 error: code overflow". This has been fixed.
2255
2256 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2257 pcretest.c, to avoid build problems in some Borland environments.
2258
2259
2260 Version 8.10 25-Jun-2010
2261 ------------------------
2262
2263 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2264 THEN.
2265
2266 2. (*ACCEPT) was not working when inside an atomic group.
2267
2268 3. Inside a character class, \B is treated as a literal by default, but
2269 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2270 causes the error). The code is unchanged, but I tidied the documentation.
2271
2272 4. Inside a character class, PCRE always treated \R and \X as literals,
2273 whereas Perl faults them if its -w option is set. I have changed PCRE so
2274 that it faults them when PCRE_EXTRA is set.
2275
2276 5. Added support for \N, which always matches any character other than
2277 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2278
2279 6. When compiling pcregrep with newer versions of gcc which may have
2280 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2281 declared with attribute warn_unused_result" were given. Just casting the
2282 result to (void) does not stop the warnings; a more elaborate fudge is
2283 needed. I've used a macro to implement this.
2284
2285 7. Minor change to pcretest.c to avoid a compiler warning.
2286
2287 8. Added four artifical Unicode properties to help with an option to make
2288 \s etc use properties (see next item). The new properties are: Xan
2289 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2290
2291 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2292 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2293 this option. Modified pcretest to add /W to test this facility. Added
2294 REG_UCP to make it available via the POSIX interface.
2295
2296 10. Added --line-buffered to pcregrep.
2297
2298 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2299 studied, and the match started with a letter with a code point greater than
2300 127 whose first byte was different to the first byte of the other case of
2301 the letter, the other case of this starting letter was not recognized
2302 (#976).
2303
2304 12. If a pattern that was studied started with a repeated Unicode property
2305 test, for example, \p{Nd}+, there was the theoretical possibility of
2306 setting up an incorrect bitmap of starting bytes, but fortunately it could
2307 not have actually happened in practice until change 8 above was made (it
2308 added property types that matched character-matching opcodes).
2309
2310 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2311 possible starting bytes for non-anchored patterns.
2312
2313 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2314 \R, and also a number of cases that involve Unicode properties, both
2315 explicit and implicit when PCRE_UCP is set.
2316
2317 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2318 input, it could crash or give wrong results if characters with values
2319 greater than 0xc0 were present in the subject string. (Detail: it assumed
2320 UTF-8 input when processing these items.)
2321
2322 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2323 size_t is 64-bit (#991).
2324
2325 17. Added a check for running out of memory when PCRE is compiled with
2326 --disable-stack-for-recursion (#990).
2327
2328 18. If the last data line in a file for pcretest does not have a newline on
2329 the end, a newline was missing in the output.
2330
2331 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2332 less than 128) in its various bitmaps. However, there is a facility for
2333 generating tables according to the current locale when PCRE is compiled. It
2334 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2335 characters, are recognized by isspace() and therefore were getting set in
2336 these tables, and indeed these tables seem to approximate to ISO 8859. This
2337 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2338 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2339 which of course cannot start UTF-8 characters. I have changed the code so
2340 that only real ASCII characters (less than 128) and the correct starting
2341 bytes for UTF-8 encodings are set for characters greater than 127 when in
2342 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2343 altogether.)
2344
2345 20. Added the /T option to pcretest so as to be able to run tests with non-
2346 standard character tables, thus making it possible to include the tests
2347 used for 19 above in the standard set of tests.
2348
2349 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2350 reference to a subpattern the other side of a comment that contains an
2351 opening parenthesis caused either an internal compiling error, or a
2352 reference to the wrong subpattern.
2353
2354
2355 Version 8.02 19-Mar-2010
2356 ------------------------
2357
2358 1. The Unicode data tables have been updated to Unicode 5.2.0.
2359
2360 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2361 configured.
2362
2363 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2364 original author of that file, following a query about its status.
2365
2366 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2367 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2368
2369 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2370 quantifier applied to a forward-referencing subroutine call, could compile
2371 incorrect code or give the error "internal error: previously-checked
2372 referenced subpattern not found".
2373
2374 6. Both MS Visual Studio and Symbian OS have problems with initializing
2375 variables to point to external functions. For these systems, therefore,
2376 pcre_malloc etc. are now initialized to local functions that call the
2377 relevant global functions.
2378
2379 7. There were two entries missing in the vectors called coptable and poptable
2380 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2381 I've fixed the data, and added a kludgy way of testing at compile time that
2382 the lengths are correct (equal to the number of opcodes).
2383
2384 8. Following on from 7, I added a similar kludge to check the length of the
2385 eint vector in pcreposix.c.
2386
2387 9. Error texts for pcre_compile() are held as one long string to avoid too
2388 much relocation at load time. To find a text, the string is searched,
2389 counting zeros. There was no check for running off the end of the string,
2390 which could happen if a new error number was added without updating the
2391 string.
2392
2393 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2394
2395 11. \K was not working if it appeared in an atomic group or in a group that
2396 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2397 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2398 the assertion is positive, but not if it is negative.
2399
2400 12. Change 11 fortuitously reduced the size of the stack frame used in the
2401 "match()" function of pcre_exec.c by one pointer. Forthcoming
2402 implementation of support for (*MARK) will need an extra pointer on the
2403 stack; I have reserved it now, so that the stack frame size does not
2404 decrease.
2405
2406 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2407 item in branch that calls a recursion is a subroutine call - as in the
2408 second branch in the above example - was incorrectly given the compile-
2409 time error "recursive call could loop indefinitely" because pcre_compile()
2410 was not correctly checking the subroutine for matching a non-empty string.
2411
2412 14. The checks for overrunning compiling workspace could trigger after an
2413 overrun had occurred. This is a "should never occur" error, but it can be
2414 triggered by pathological patterns such as hundreds of nested parentheses.
2415 The checks now trigger 100 bytes before the end of the workspace.
2416
2417 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2418
2419
2420 Version 8.01 19-Jan-2010
2421 ------------------------
2422
2423 1. If a pattern contained a conditional subpattern with only one branch (in
2424 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2425 computed the wrong minimum data length (which is of course zero for such
2426 subpatterns). This could cause incorrect "no match" results.
2427
2428 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2429 the pattern is reset in the first branch, pcre_compile() failed with
2430 "internal error: code overflow at offset...". This happened only when
2431 the reset was to the original external option setting. (An optimization
2432 abstracts leading options settings into an external setting, which was the
2433 cause of this.)
2434
2435 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2436 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2437 assertion pattern did not match (meaning that the assertion was true), it
2438 was incorrectly treated as false if the SKIP had been reached during the
2439 matching. This also applied to assertions used as conditions.
2440
2441 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2442 assertion subpattern, including such a pattern used as a condition,
2443 unpredictable results occurred, instead of the error return
2444 PCRE_ERROR_DFA_UITEM.
2445
2446 5. The C++ GlobalReplace function was not working like Perl for the special
2447 situation when an empty string is matched. It now does the fancy magic
2448 stuff that is necessary.
2449
2450 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2451 removed. (These were left over from very, very early versions of PCRE.)
2452
2453 7. Some cosmetic changes to the code to make life easier when compiling it
2454 as part of something else:
2455
2456 (a) Change DEBUG to PCRE_DEBUG.
2457
2458 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2459 called "current" as "current_branch", to prevent a collision with the
2460 Linux macro when compiled as a kernel module.
2461
2462 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2463 prevent a collision with the Linux macro when compiled as a kernel
2464 module.
2465
2466 8. In pcre_compile() there are some checks for integer overflows that used to
2467 cast potentially large values to (double). This has been changed to that
2468 when building, a check for int64_t is made, and if it is found, it is used
2469 instead, thus avoiding the use of floating point arithmetic. (There is no
2470 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2471 double.
2472
2473 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2474 2005 (difference between two addresses compared to an unsigned value).
2475
2476 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2477 custom one, because of the following reported problem in Windows:
2478
2479 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2480 under Win32.
2481 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2482 therefore missing the function definition.
2483 - The compiler thus generates a "C" signature for the test function.
2484 - The linker fails to find the "C" function.
2485 - PCRE fails to configure if asked to do so against libbz2.
2486
2487 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2488 messages were output:
2489
2490 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2491 rerunning libtoolize, to keep the correct libtool macros in-tree.
2492 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2493
2494 I have done both of these things.
2495
2496 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2497 most of the time, it *can* run out if it is given a pattern that contains a
2498 runaway infinite recursion. I updated the discussion in the pcrestack man
2499 page.
2500
2501 13. Now that we have gone to the x.xx style of version numbers, the minor
2502 version may start with zero. Using 08 or 09 is a bad idea because users
2503 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2504 interpreted as invalid octal numbers. I've updated the previous comment in
2505 configure.ac, and also added a check that gives an error if 08 or 09 are
2506 used.
2507
2508 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2509 causing partial matching to fail when the end of the subject matched \W
2510 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2511
2512 15. There were some discrepancies between the declarations in pcre_internal.h
2513 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2514 their definitions. The declarations used "const uschar *" and the
2515 definitions used USPTR. Even though USPTR is normally defined as "const
2516 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2517 reported that: "This difference in casting confuses some C++ compilers, for
2518 example, SunCC recognizes above declarations as different functions and
2519 generates broken code for hbpcre." I have changed the declarations to use
2520 USPTR.
2521
2522 16. GNU libtool is named differently on some systems. The autogen.sh script now
2523 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2524 (FreeBSD).
2525
2526 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2527 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2528 comment: "Figure out how to create a longlong from a string: strtoll and
2529 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2530 instance, but it only takes 2 args instead of 3!"
2531
2532 18. A subtle bug concerned with back references has been fixed by a change of
2533 specification, with a corresponding code fix. A pattern such as
2534 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2535 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2536 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2537 same bug. Such groups have to be quantified to be useful, or contained
2538 inside another quantified group. (If there's no repetition, the reference
2539 can never match.) The problem arises because, having left the group and
2540 moved on to the rest of the pattern, a later failure that backtracks into
2541 the group uses the captured value from the final iteration of the group
2542 rather than the correct earlier one. I have fixed this in PCRE by forcing
2543 any group that contains a reference to itself to be an atomic group; that
2544 is, there cannot be any backtracking into it once it has completed. This is
2545 similar to recursive and subroutine calls.
2546
2547
2548 Version 8.00 19-Oct-09
2549 ----------------------
2550
2551 1. The table for translating pcre_compile() error codes into POSIX error codes
2552 was out-of-date, and there was no check on the pcre_compile() error code
2553 being within the table. This could lead to an OK return being given in
2554 error.
2555
2556 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2557 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2558 in a Windows environment.
2559
2560 3. The pcregrep --count option prints the count for each file even when it is
2561 zero, as does GNU grep. However, pcregrep was also printing all files when
2562 --files-with-matches was added. Now, when both options are given, it prints
2563 counts only for those files that have at least one match. (GNU grep just
2564 prints the file name in this circumstance, but including the count seems
2565 more useful - otherwise, why use --count?) Also ensured that the
2566 combination -clh just lists non-zero counts, with no names.
2567
2568 4. The long form of the pcregrep -F option was incorrectly implemented as
2569 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2570 but it seems right to fix it, and I didn't think it was worth preserving
2571 the old behaviour.
2572
2573 5. The command line items --regex=pattern and --regexp=pattern were not
2574 recognized by pcregrep, which required --regex pattern or --regexp pattern
2575 (with a space rather than an '='). The man page documented the '=' forms,
2576 which are compatible with GNU grep; these now work.
2577
2578 6. No libpcreposix.pc file was created for pkg-config; there was just
2579 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2580
2581 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2582 when UCP support is not needed, by modifying the Python script that
2583 generates it from Unicode data files. This should not matter if the module
2584 is correctly used as a library, but I received one complaint about 50K of
2585 unwanted data. My guess is that the person linked everything into his
2586 program rather than using a library. Anyway, it does no harm.
2587
2588 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2589 was a minimum greater than 1 for a wide character in a possessive
2590 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2591 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2592 character. Chaos in the form of incorrect output or a compiling loop could
2593 result.
2594
2595 9. The restrictions on what a pattern can contain when partial matching is
2596 requested for pcre_exec() have been removed. All patterns can now be
2597 partially matched by this function. In addition, if there are at least two
2598 slots in the offset vector, the offset of the earliest inspected character
2599 for the match and the offset of the end of the subject are set in them when
2600 PCRE_ERROR_PARTIAL is returned.
2601
2602 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2603 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2604 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2605 and may be more useful for multi-segment matching.
2606
2607 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2608 used to be given if ever the end of the subject was reached; now it is
2609 given only if matching could not proceed because another character was
2610 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2611 string "Z", which now yields "no match" instead of "partial match". In the
2612 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2613 final character ended with (*FAIL).
2614
2615 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2616 if the pattern had a "must contain" character that was already found in the
2617 earlier partial match, unless partial matching was again requested. For
2618 example, with the pattern /dog.(body)?/, the "must contain" character is
2619 "g". If the first part-match was for the string "dog", restarting with
2620 "sbody" failed. This bug has been fixed.
2621
2622 13. The string returned by pcre_dfa_exec() after a partial match has been
2623 changed so that it starts at the first inspected character rather than the
2624 first character of the match. This makes a difference only if the pattern
2625 starts with a lookbehind assertion or \b or \B (\K is not supported by
2626 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2627 matching functions compatible, and I think it's the right thing to do.
2628
2629 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2630 so that the demonstration program is easily available in environments where
2631 PCRE has not been installed from source.
2632
2633 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2634 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2635 library.
2636
2637 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2638 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2639 is not the first non-POSIX option to be added. Clearly some people find
2640 these options useful.
2641
2642 17. If a caller to the POSIX matching function regexec() passes a non-zero
2643 value for nmatch with a NULL value for pmatch, the value of
2644 nmatch is forced to zero.
2645
2646 18. RunGrepTest did not have a test for the availability of the -u option of
2647 the diff command, as RunTest does. It now checks in the same way as
2648 RunTest, and also checks for the -b option.
2649
2650 19. If an odd number of negated classes containing just a single character
2651 interposed, within parentheses, between a forward reference to a named
2652 subpattern and the definition of the subpattern, compilation crashed with
2653 an internal error, complaining that it could not find the referenced
2654 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2655 [The bug was that it was starting one character too far in when skipping
2656 over the character class, thus treating the ] as data rather than
2657 terminating the class. This meant it could skip too much.]
2658
2659 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2660 /g option in pcretest when the pattern contains \K, which makes it possible
2661 to have an empty string match not at the start, even when the pattern is
2662 anchored. Updated pcretest and pcredemo to use this option.
2663
2664 21. If the maximum number of capturing subpatterns in a recursion was greater
2665 than the maximum at the outer level, the higher number was returned, but
2666 with unset values at the outer level. The correct (outer level) value is
2667 now given.
2668
2669 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2670 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2671 make it do so. The string so far is captured, making this feature
2672 compatible with Perl.
2673
2674 23. The tests have been re-organized, adding tests 11 and 12, to make it
2675 possible to check the Perl 5.10 features against Perl 5.10.
2676
2677 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2678 pattern matches a fixed length string. PCRE did not allow this; now it
2679 does. Neither allows recursion.
2680
2681 25. I finally figured out how to implement a request to provide the minimum
2682 length of subject string that was needed in order to match a given pattern.
2683 (It was back references and recursion that I had previously got hung up
2684 on.) This code has now been added to pcre_study(); it finds a lower bound
2685 to the length of subject needed. It is not necessarily the greatest lower
2686 bound, but using it to avoid searching strings that are too short does give
2687 some useful speed-ups. The value is available to calling programs via
2688 pcre_fullinfo().
2689
2690 26. While implementing 25, I discovered to my embarrassment that pcretest had
2691 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2692 study optimizations had never been tested with that matching function.
2693 Oops. What is worse, even when it was passed study data, there was a bug in
2694 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2695 were also very few tests of studied patterns with pcre_dfa_exec().
2696
2697 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2698 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2699 on the other side of the coin, they are no longer allowed to have different
2700 names, because these cannot be distinguished in PCRE, and this has caused
2701 confusion. (This is a difference from Perl.)
2702
2703 28. When duplicate subpattern names are present (necessarily with different
2704 numbers, as required by 27 above), and a test is made by name in a
2705 conditional pattern, either for a subpattern having been matched, or for
2706 recursion in such a pattern, all the associated numbered subpatterns are
2707 tested, and the overall condition is true if the condition is true for any
2708 one of them. This is the way Perl works, and is also more like the way
2709 testing by number works.
2710
2711
2712 Version 7.9 11-Apr-09
2713 ---------------------
2714
2715 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2716 (pcretest), all targets were linked against these libraries. This included
2717 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2718 libraries. This caused unwanted dependencies to be created. This problem
2719 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2720 pcretest is linked with readline.
2721
2722 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2723 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2724 moved outside it again, because FALSE and TRUE are already defined in AIX,
2725 but BOOL is not.
2726
2727 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2728 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2729
2730 4. The pcregrep documentation said spaces were inserted as well as colons (or
2731 hyphens) following file names and line numbers when outputting matching
2732 lines. This is not true; no spaces are inserted. I have also clarified the
2733 wording for the --colour (or --color) option.
2734
2735 5. In pcregrep, when --colour was used with -o, the list of matching strings
2736 was not coloured; this is different to GNU grep, so I have changed it to be
2737 the same.
2738
2739 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2740 each matching line was coloured. Now it goes on to look for further matches
2741 of any of the test patterns, which is the same behaviour as GNU grep.
2742
2743 7. A pattern that could match an empty string could cause pcregrep to loop; it
2744 doesn't make sense to accept an empty string match in pcregrep, so I have
2745 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2746 seems to be how GNU grep behaves. [But see later change 40 for release
2747 8.33.]
2748
2749 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2750 start or after a newline", because the conditional assertion was not being
2751 correctly handled. The rule now is that both the assertion and what follows
2752 in the first alternative must satisfy the test.
2753
2754 9. If auto-callout was enabled in a pattern with a conditional group whose
2755 condition was an assertion, PCRE could crash during matching, both with
2756 pcre_exec() and pcre_dfa_exec().
2757
2758 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2759 used for matching.
2760
2761 11. Unicode property support in character classes was not working for
2762 characters (bytes) greater than 127 when not in UTF-8 mode.
2763
2764 12. Added the -M command line option to pcretest.
2765
2766 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2767
2768 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2769
2770 16. Added comments and documentation about mis-use of no_arg in the C++
2771 wrapper.
2772
2773 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2774 from Martin Jerabek that uses macro names for all relevant character and
2775 string constants.
2776
2777 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2778 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2779 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2780 these, but not everybody uses configure.
2781
2782 19. A conditional group that had only one branch was not being correctly
2783 recognized as an item that could match an empty string. This meant that an
2784 enclosing group might also not be so recognized, causing infinite looping
2785 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2786 with the subject "ab", where knowledge that the repeated group can match
2787 nothing is needed in order to break the loop.
2788
2789 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2790 exec(), but without supplying a callout function, matching went wrong.
2791
2792 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2793 leak if the size of the offset vector was greater than 30. When the vector
2794 is smaller, the saved offsets during recursion go onto a local stack
2795 vector, but for larger vectors malloc() is used. It was failing to free
2796 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2797 error, in fact).
2798
2799 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2800 heapframe that is used only when UTF-8 support is enabled. This caused no
2801 problem, but was untidy.
2802
2803 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2804 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2805 included within another project.
2806
2807 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2808 slightly modified by me:
2809
2810 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2811 not building pcregrep.
2812
2813 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2814 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2815
2816 25. Forward references, both numeric and by name, in patterns that made use of
2817 duplicate group numbers, could behave incorrectly or give incorrect errors,
2818 because when scanning forward to find the reference group, PCRE was not
2819 taking into account the duplicate group numbers. A pattern such as
2820 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2821
2822 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2823 the feature of a custom pointer more persuasive (as requested by a user).
2824
2825 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2826 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2827 pre-defined.
2828
2829 28. Added support for (*UTF8) at the start of a pattern.
2830
2831 29. Arrange for flags added by the "release type" setting in CMake to be shown
2832 in the configuration summary.
2833
2834
2835 Version 7.8 05-Sep-08
2836 ---------------------
2837
2838 1. Replaced UCP searching code with optimized version as implemented for Ad
2839 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2840 stage table and inline lookup instead of a function, giving speed ups of 2
2841 to 5 times on some simple patterns that I tested. Permission was given to
2842 distribute the MultiStage2.py script that generates the tables (it's not in
2843 the tarball, but is in the Subversion repository).
2844
2845 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2846 scripts.
2847
2848 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2849 a group with a zero qualifier. The result of the study could be incorrect,
2850 or the function might crash, depending on the pattern.
2851
2852 4. Caseless matching was not working for non-ASCII characters in back
2853 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2854 It now works when Unicode Property Support is available.
2855
2856 5. In pcretest, an escape such as \x{de} in the data was always generating
2857 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2858 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2859 truncation.
2860
2861 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2862
2863 7. Added two (int) casts to pcregrep when printing the difference of two
2864 pointers, in case they are 64-bit values.
2865
2866 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2867 test 2 if it fails.
2868
2869 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2870 and a #define of that name to empty if it is not externally set. This is to
2871 allow users of MSVC to set it if necessary.
2872
2873 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2874 the convenience functions in the pcre_get.c source file.
2875
2876 11. An option change at the start of a pattern that had top-level alternatives
2877 could cause overwriting and/or a crash. This command provoked a crash in
2878 some environments:
2879
2880 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2881
2882 This potential security problem was recorded as CVE-2008-2371.
2883
2884 12. For a pattern where the match had to start at the beginning or immediately
2885 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2886 pcre_dfa_exec() could read past the end of the passed subject if there was
2887 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2888 pcretest so that it places the subject at the end of its malloc-ed buffer.
2889
2890 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2891 exec() might read past the end of the data buffer in UTF-8 mode.
2892
2893 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2894 the data contained the byte 0x85 as part of a UTF-8 character within its
2895 first line. This applied both to normal and DFA matching.
2896
2897 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2898 /^[^d]*?$/8 failed to match "abc".
2899
2900 16. Added a missing copyright notice to pcrecpp_internal.h.
2901
2902 17. Make it more clear in the documentation that values returned from
2903 pcre_exec() in ovector are byte offsets, not character counts.
2904
2905 18. Tidied a few places to stop certain compilers from issuing warnings.
2906
2907 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2908 supplied by Stefan Weber. I made a further small update for 7.8 because
2909 there is a change of source arrangements: the pcre_searchfuncs.c module is
2910 replaced by pcre_ucd.c.
2911
2912
2913 Version 7.7 07-May-08
2914 ---------------------
2915
2916 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2917 a string to a long long, pretend we don't even have a long long." This is
2918 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2919
2920 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2921 pre-7.6 versions, which defined a global no_arg variable instead of putting
2922 it in the RE class. (See also #8 below.)
2923
2924 3. Remove a line of dead code, identified by coverity and reported by Nuno
2925 Lopes.
2926
2927 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2928
2929 (1) The include/exclude patterns were being applied to the whole pathnames
2930 of files, instead of just to the final components.
2931
2932 (2) If there was more than one level of directory, the subdirectories were
2933 skipped unless they satisfied the include/exclude conditions. This is
2934 inconsistent with GNU grep (and could even be seen as contrary to the
2935 pcregrep specification - which I improved to make it absolutely clear).
2936 The action now is always to scan all levels of directory, and just
2937 apply the include/exclude patterns to regular files.
2938
2939 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2940 --exclude_dir in the tests to avoid scanning .svn directories.
2941
2942 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2943 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2944 doesn't support NULs in patterns.
2945
2946 7. Added some missing "const"s to declarations of static tables in
2947 pcre_compile.c and pcre_dfa_exec.c.
2948
2949 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2950 caused by fix #2 above. (Subsequently also a second patch to fix the
2951 first patch. And a third patch - this was a messy problem.)
2952
2953 9. Applied Craig's patch to remove the use of push_back().
2954
2955 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2956 matching function regexec().
2957
2958 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2959 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2960 references. PCRE supports relative numbers with this syntax (I don't think
2961 Oniguruma does).
2962
2963 12. Previously, a group with a zero repeat such as (...){0} was completely
2964 omitted from the compiled regex. However, this means that if the group
2965 was called as a subroutine from elsewhere in the pattern, things went wrong
2966 (an internal error was given). Such groups are now left in the compiled
2967 pattern, with a new opcode that causes them to be skipped at execution
2968 time.
2969
2970 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2971 to the way PCRE behaves:
2972
2973 (a) A lone ] character is dis-allowed (Perl treats it as data).
2974
2975 (b) A back reference to an unmatched subpattern matches an empty string
2976 (Perl fails the current match path).
2977
2978 (c) A data ] in a character class must be notated as \] because if the
2979 first data character in a class is ], it defines an empty class. (In
2980 Perl it is not possible to have an empty class.) The empty class []
2981 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2982 The negative empty class [^] matches any one character, independently
2983 of the DOTALL setting.
2984
2985 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2986 non-existent subpattern following a character class starting with ']' and
2987 containing () gave an internal compiling error instead of "reference to
2988 non-existent subpattern". Fortunately, when the pattern did exist, the
2989 compiled code was correct. (When scanning forwards to check for the
2990 existence of the subpattern, it was treating the data ']' as terminating
2991 the class, so got the count wrong. When actually compiling, the reference
2992 was subsequently set up correctly.)
2993
2994 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2995 it was being rejected as not supported by pcre_dfa_exec(), even though
2996 other assertions are supported. I have made pcre_dfa_exec() support
2997 (*FAIL).
2998
2999 16. The implementation of 13c above involved the invention of a new opcode,
3000 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
3001 cannot be changed at match time, I realized I could make a small
3002 improvement to matching performance by compiling OP_ALLANY instead of
3003 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
3004 on the OP_ANY path.
3005
3006 17. Compiling pcretest on Windows with readline support failed without the
3007 following two fixes: (1) Make the unistd.h include conditional on
3008 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
3009
3010 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
3011 ncurses library to be included for pcretest when ReadLine support is
3012 requested, but also to allow for it to be overridden. This patch came from
3013 Daniel Bergström.
3014
3015 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
3016 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
3017 any errors with the current Unicode tables. Thanks to Peter Kankowski for
3018 spotting this.
3019
3020
3021 Version 7.6 28-Jan-08
3022 ---------------------
3023
3024 1. A character class containing a very large number of characters with
3025 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
3026 overflow.
3027
3028 2. Patch to cut out the "long long" test in pcrecpp_unittest when
3029 HAVE_LONG_LONG is not defined.
3030
3031 3. Applied Christian Ehrlicher's patch to update the CMake build files to
3032 bring them up to date and include new features. This patch includes:
3033
3034 - Fixed PH's badly added libz and libbz2 support.
3035 - Fixed a problem with static linking.
3036 - Added pcredemo. [But later removed - see 7 below.]
3037 - Fixed dftables problem and added an option.
3038 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
3039 HAVE_LONG_LONG.
3040 - Added readline support for pcretest.
3041 - Added an listing of the option settings after cmake has run.
3042
3043 4. A user submitted a patch to Makefile that makes it easy to create
3044 "pcre.dll" under mingw when using Configure/Make. I added stuff to
3045 Makefile.am that cause it to include this special target, without
3046 affecting anything else. Note that the same mingw target plus all
3047 the other distribution libraries and programs are now supported
3048 when configuring with CMake (see 6 below) instead of with
3049 Configure/Make.
3050
3051 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
3052 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
3053 exported in the Windows port". It has not yet been confirmed that the patch
3054 solves the problem, but it does no harm.
3055
3056 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
3057 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
3058 with CMake, and also correct the comment about stack recursion.
3059
3060 7. Remove the automatic building of pcredemo from the ./configure system and
3061 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
3062 of a program that users should build themselves after PCRE is installed, so
3063 building it automatically is not really right. What is more, it gave
3064 trouble in some build environments.
3065
3066 8. Further tidies to CMakeLists.txt from Sheri and Christian.
3067
3068
3069 Version 7.5 10-Jan-08
3070 ---------------------
3071
3072 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
3073 values in parens when parsing an RE using the C++ wrapper."
3074
3075 2. Negative specials like \S did not work in character classes in UTF-8 mode.
3076 Characters greater than 255 were excluded from the class instead of being
3077 included.
3078
3079 3. The same bug as (2) above applied to negated POSIX classes such as
3080 [:^space:].
3081
3082 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
3083 defined or documented. It seems to have been a typo for PCRE_STATIC, so
3084 I have changed it.
3085
3086 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
3087 first named subpattern) and a construct such as (?&a) would reference the
3088 first named subpattern whose name started with "a" (in other words, the
3089 length check was missing). Both these problems are fixed. "Subpattern name
3090 expected" is now given for (?&) (a zero-length name), and this patch also
3091 makes it give the same error for \k'' (previously it complained that that
3092 was a reference to a non-existent subpattern).
3093
3094 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
3095 this is right because (?- can be followed by option settings as well as by
3096 digits. I have, however, made the messages clearer.
3097
3098 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
3099 than the number used in the conditional) now cause a compile-time error.
3100 This is actually not compatible with Perl, which accepts such patterns, but
3101 treats the conditional as always being FALSE (as PCRE used to), but it
3102 seems to me that giving a diagnostic is better.
3103
3104 8. Change "alphameric" to the more common word "alphanumeric" in comments
3105 and messages.
3106
3107 9. Fix two occurrences of "backslash" in comments that should have been
3108 "backspace".
3109
3110 10. Remove two redundant lines of code that can never be obeyed (their function
3111 was moved elsewhere).
3112
3113 11. The program that makes PCRE's Unicode character property table had a bug
3114 which caused it to generate incorrect table entries for sequences of
3115 characters that have the same character type, but are in different scripts.
3116 It amalgamated them into a single range, with the script of the first of
3117 them. In other words, some characters were in the wrong script. There were
3118 thirteen such cases, affecting characters in the following ranges:
3119
3120 U+002b0 - U+002c1
3121 U+0060c - U+0060d
3122 U+0061e - U+00612
3123 U+0064b - U+0065e
3124 U+0074d - U+0076d
3125 U+01800 - U+01805
3126 U+01d00 - U+01d77
3127 U+01d9b - U+01dbf
3128 U+0200b - U+0200f
3129 U+030fc - U+030fe
3130 U+03260 - U+0327f
3131 U+0fb46 - U+0fbb1
3132 U+10450 - U+1049d
3133
3134 12. The -o option (show only the matching part of a line) for pcregrep was not
3135 compatible with GNU grep in that, if there was more than one match in a
3136 line, it showed only the first of them. It now behaves in the same way as
3137 GNU grep.
3138
3139 13. If the -o and -v options were combined for pcregrep, it printed a blank
3140 line for every non-matching line. GNU grep prints nothing, and pcregrep now
3141 does the same. The return code can be used to tell if there were any
3142 non-matching lines.
3143
3144 14. Added --file-offsets and --line-offsets to pcregrep.
3145
3146 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
3147 infinitely looping recursion. The bug was that positive lookaheads were not
3148 being skipped when checking for a possible empty match (negative lookaheads
3149 and both kinds of lookbehind were skipped).
3150
3151 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
3152 inclusion of <windows.h> to before rather than after the definition of
3153 INVALID_FILE_ATTRIBUTES (patch from David Byron).
3154
3155 17. Specifying a possessive quantifier with a specific limit for a Unicode
3156 character property caused pcre_compile() to compile bad code, which led at
3157 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
3158 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
3159 caused the error; without that there was no problem.
3160
3161 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3162
3163 19. Added --enable-pcretest-libreadline.
3164
3165 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3166 RE::GlobalReplace(). As a result, the number of replacements returned was
3167 double what it should be. I removed one of the increments, but Craig sent a
3168 later patch that removed the other one (the right fix) and added unit tests
3169 that check the return values (which was not done before).
3170
3171 21. Several CMake things:
3172
3173 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3174 the names libpcre and libpcreposix, not just pcre and pcreposix.
3175
3176 (2) The above change means that pcretest and pcregrep are now correctly
3177 linked with the newly-built libraries, not previously installed ones.
3178
3179 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
3180
3181 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3182 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3183 UTF-8 newline character). The key issue is that the pattern starts .*;
3184 this means that the match must be either at the beginning, or after a
3185 newline. The bug was in the code for advancing after a failed match and
3186 checking that the new position followed a newline. It was not taking
3187 account of UTF-8 characters correctly.
3188
3189 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3190 character classes. PCRE was not treating the sequence [:...:] as a
3191 character class unless the ... were all letters. Perl, however, seems to
3192 allow any characters between [: and :], though of course it rejects as
3193 unknown any "names" that contain non-letters, because all the known class
3194 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3195 for example, whereas PCRE did not - it did not recognize a POSIX character
3196 class. This seemed a bit dangerous, so the code has been changed to be
3197 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3198 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3199 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3200 Perl does, and where it didn't before.
3201
3202 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3203 Windows environments %n is disabled by default.
3204
3205
3206 Version 7.4 21-Sep-07
3207 ---------------------
3208
3209 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3210 means that a class such as [\s] counted as "explicit reference to CR or
3211 LF". That isn't really right - the whole point of the change was to try to
3212 help when there was an actual mention of one of the two characters. So now
3213 the change happens only if \r or \n (or a literal CR or LF) character is
3214 encountered.
3215
3216 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3217 of both had grown to the point where there were only 3 bits left.
3218 Fortunately, there was spare space in the data structure, and so I have
3219 moved the internal flags into a new 16-bit field to free up more option
3220 bits.
3221
3222 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3223 but did not set the internal JCHANGED flag - either of these is enough to
3224 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3225 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3226 start sets both bits.
3227
3228 4. Added options (at build time, compile time, exec time) to change \R from
3229 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3230
3231 5. doc/pcresyntax.html was missing from the distribution.
3232
3233 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3234 compatibility, even though it is no longer used.
3235
3236 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3237 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3238 windows.h file is present (where different names are used). [This was
3239 reversed later after testing - see 16 below.]
3240
3241 8. Changed all #include <config.h> to #include "config.h". There were also
3242 some further <pcre.h> cases that I changed to "pcre.h".
3243
3244 9. When pcregrep was used with the --colour option, it missed the line ending
3245 sequence off the lines that it output.
3246
3247 10. It was pointed out to me that arrays of string pointers cause lots of
3248 relocations when a shared library is dynamically loaded. A technique of
3249 using a single long string with a table of offsets can drastically reduce
3250 these. I have refactored PCRE in four places to do this. The result is
3251 dramatic:
3252
3253 Originally: 290
3254 After changing UCP table: 187
3255 After changing error message table: 43
3256 After changing table of "verbs" 36
3257 After changing table of Posix names 22
3258
3259 Thanks to the folks working on Gregex for glib for this insight.
3260
3261 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3262 unicode-properties was also set.
3263
3264 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3265
3266 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3267 checked only for CRLF.
3268
3269 14. Added casts to pcretest.c to avoid compiler warnings.
3270
3271 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3272
3273 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3274 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3275 entirely. This removes changes made in 7 above.
3276
3277 17. The CMake files have been updated, and there is now more information about
3278 building with CMake in the NON-UNIX-USE document.
3279
3280
3281 Version 7.3 28-Aug-07
3282 ---------------------
3283
3284 1. In the rejigging of the build system that eventually resulted in 7.1, the
3285 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3286 brackets there is not right, since it causes compilers to look for an
3287 installed pcre.h, not the version that is in the source that is being
3288 compiled (which of course may be different). I have changed it back to:
3289
3290 #include "pcre.h"
3291
3292 I have a vague recollection that the change was concerned with compiling in
3293 different directories, but in the new build system, that is taken care of
3294 by the VPATH setting the Makefile.
3295
3296 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3297 when the subject happened to end in the byte 0x85 (e.g. if the last
3298 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3299 characters but of course it shouldn't be taken as a newline when it is part
3300 of another character. The bug was that, for an unlimited repeat of . in
3301 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3302 characters when looking for a newline.
3303
3304 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3305
3306 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3307 in debug output.
3308
3309 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3310 long printing in the pcrecpp unittest when running under MinGW.
3311
3312 6. ESC_K was left out of the EBCDIC table.
3313
3314 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3315 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3316 limit also applies to "virtual nesting" when a pattern is recursive, and in
3317 this case 1000 isn't so big. I have been able to remove this limit at the
3318 expense of backing off one optimization in certain circumstances. Normally,
3319 when pcre_exec() would call its internal match() function recursively and
3320 immediately return the result unconditionally, it uses a "tail recursion"
3321 feature to save stack. However, when a subpattern that can match an empty
3322 string has an unlimited repetition quantifier, it no longer makes this
3323 optimization. That gives it a stack frame in which to save the data for
3324 checking that an empty string has been matched. Previously this was taken
3325 from the 1000-entry workspace that had been reserved. So now there is no
3326 explicit limit, but more stack is used.
3327
3328 8. Applied Daniel's patches to solve problems with the import/export magic
3329 syntax that is required for Windows, and which was going wrong for the
3330 pcreposix and pcrecpp parts of the library. These were overlooked when this
3331 problem was solved for the main library.
3332
3333 9. There were some crude static tests to avoid integer overflow when computing
3334 the size of patterns that contain repeated groups with explicit upper
3335 limits. As the maximum quantifier is 65535, the maximum group length was
3336 set at 30,000 so that the product of these two numbers did not overflow a
3337 32-bit integer. However, it turns out that people want to use groups that
3338 are longer than 30,000 bytes (though not repeat them that many times).
3339 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3340 made it possible to implement the integer overflow checks in a much more
3341 dynamic way, which I have now done. The artificial limitation on group
3342 length has been removed - we now have only the limit on the total length of
3343 the compiled pattern, which depends on the LINK_SIZE setting.
3344
3345 10. Fixed a bug in the documentation for get/copy named substring when
3346 duplicate names are permitted. If none of the named substrings are set, the
3347 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3348 empty string.
3349
3350 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3351 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3352 because the ] is interpreted as the first data character and the
3353 terminating ] is not found. PCRE has been made compatible with Perl in this
3354 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3355 cause memory overwriting.
3356
3357 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3358 string has been matched (to stop an infinite loop). It was not recognizing
3359 a conditional subpattern that could match an empty string if that
3360 subpattern was within another subpattern. For example, it looped when
3361 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3362 condition was not nested. This bug has been fixed.
3363
3364 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3365 past the start of the subject in the presence of bytes with the top bit
3366 set, for example "\x8aBCD".
3367
3368 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3369 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3370
3371 14. Optimized (?!) to (*FAIL).
3372
3373 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3374 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3375 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3376 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3377 does: it's just the validity check that is more restrictive.
3378
3379 16. Inserted checks for integer overflows during escape sequence (backslash)
3380 processing, and also fixed erroneous offset values for syntax errors during
3381 backslash processing.
3382
3383 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3384 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3385
3386 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3387 caused an overrun.
3388
3389 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3390 something other than just ASCII characters) inside a group that had an
3391 unlimited repeat caused a loop at compile time (while checking to see
3392 whether the group could match an empty string).
3393
3394 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3395 [\P{Any}] did so. (Error in the code for printing property names.)
3396
3397 21. An orphan \E inside a character class could cause a crash.
3398
3399 22. A repeated capturing bracket such as (A)? could cause a wild memory
3400 reference during compilation.
3401
3402 23. There are several functions in pcre_compile() that scan along a compiled
3403 expression for various reasons (e.g. to see if it's fixed length for look
3404 behind). There were bugs in these functions when a repeated \p or \P was
3405 present in the pattern. These operators have additional parameters compared
3406 with \d, etc, and these were not being taken into account when moving along
3407 the compiled data. Specifically:
3408
3409 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3410 length.
3411
3412 (b) An item such as \pL+ within a repeated group could cause crashes or
3413 loops.
3414
3415 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3416 "reference to non-existent subpattern" error.
3417
3418 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3419
3420 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3421 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3422
3423 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3424
3425 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3426 character were causing crashes (broken optimization).
3427
3428 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3429 \p or \P) caused a compile-time loop.
3430
3431 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3432 break. For example, the unstudied pattern [\r\n]A does not match the string
3433 "\r\nA" because change 7.0/46 below moves the current point on by two
3434 characters after failing to match at the start. However, the pattern \nA
3435 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3436 the same is true. There doesn't seem any very clean way out of this, but
3437 what I have chosen to do makes the common cases work: PCRE now takes note
3438 of whether there can be an explicit match for \r or \n anywhere in the
3439 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3440 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3441 pattern has explicit CR or LF references.
3442
3443 29. Added (*CR) etc for changing newline setting at start of pattern.
3444
3445
3446 Version 7.2 19-Jun-07
3447 ---------------------
3448
3449 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3450 which is apparently normally available under Windows.
3451
3452 2. Re-jig the pcregrep tests with different newline settings in an attempt
3453 to make them independent of the local environment's newline setting.
3454
3455 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3456
3457 4. Some of the "internals" tests were previously cut out when the link size
3458 was not 2, because the output contained actual offsets. The recent new
3459 "Z" feature of pcretest means that these can be cut out, making the tests
3460 usable with all link sizes.
3461
3462 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3463 stack recursion. This gives a massive performance boost under BSD, but just
3464 a small improvement under Linux. However, it saves one field in the frame
3465 in all cases.
3466
3467 6. Added more features from the forthcoming Perl 5.10:
3468
3469 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3470 recursion call. It refers to the nth most recently opened parentheses.
3471
3472 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3473 to be opened parentheses.
3474
3475 (c) Conditions that refer to capturing parentheses can be specified
3476 relatively, for example, (?(-2)... or (?(+3)...
3477
3478 (d) \K resets the start of the current match so that everything before
3479 is not part of it.
3480
3481 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3482
3483 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3484 reference syntax.
3485
3486 (g) (?| introduces a group in which the numbering of parentheses in each
3487 alternative starts with the same number.
3488
3489 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3490
3491 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3492 PCRE_INFO_JCHANGED.
3493
3494 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3495 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3496 for detecting groups that can match an empty string.
3497
3498 9. A pattern with a very large number of alternatives (more than several
3499 hundred) was running out of internal workspace during the pre-compile
3500 phase, where pcre_compile() figures out how much memory will be needed. A
3501 bit of new cunning has reduced the workspace needed for groups with
3502 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3503 workspace instead of running out of the 4096 that are available.
3504
3505 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3506
3507 11. Applied patch from Google to remove an optimization that didn't quite work.
3508 The report of the bug said:
3509
3510 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3511 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3512 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3513
3514 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3515 it matched the wrong number of bytes.
3516
3517
3518 Version 7.1 24-Apr-07
3519 ---------------------
3520
3521 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3522 that is more "standard", making use of automake and other Autotools. There
3523 is some re-arrangement of the files and adjustment of comments consequent
3524 on this.
3525
3526 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3527 for recursive directory scanning broke on some systems because the files
3528 are not scanned in any specific order and on different systems the order
3529 was different. A call to "sort" has been inserted into RunGrepTest for the
3530 approprate test as a short-term fix. In the longer term there may be an
3531 alternative.
3532
3533 3. I had an email from Eric Raymond about problems translating some of PCRE's
3534 man pages to HTML (despite the fact that I distribute HTML pages, some
3535 people do their own conversions for various reasons). The problems
3536 concerned the use of low-level troff macros .br and .in. I have therefore
3537 removed all such uses from the man pages (some were redundant, some could
3538 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3539 HTML has been updated to handle .nf/.fi and to complain if it encounters
3540 .br or .in.
3541
3542 4. Updated comments in configure.ac that get placed in config.h.in and also
3543 arranged for config.h to be included in the distribution, with the name
3544 config.h.generic, for the benefit of those who have to compile without
3545 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3546
3547 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3548 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3549 makevp.bat for the current PCRE, using the additional files
3550 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3551
3552 6. A Windows user reported a minor discrepancy with test 2, which turned out
3553 to be caused by a trailing space on an input line that had got lost in his
3554 copy. The trailing space was an accident, so I've just removed it.
3555
3556 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3557 that is needed.
3558
3559 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3560 as "const" (a) because they are and (b) because it helps the PHP
3561 maintainers who have recently made a script to detect big data structures
3562 in the php code that should be moved to the .rodata section. I remembered
3563 to update Builducptable as well, so it won't revert if ucptable.h is ever
3564 re-created.
3565
3566 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3567 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3568 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3569 support is not required. This saves 1.5-2K of code, which is important in
3570 some applications.
3571
3572 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3573 so as not to refer to the tables, even though these functions will never be
3574 called when UTF-8 support is disabled. Otherwise there are problems with a
3575 shared library.
3576
3577 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3578
3579 (a) It was defining its arguments as char * instead of void *.
3580
3581 (b) It was assuming that all moves were upwards in memory; this was true
3582 a long time ago when I wrote it, but is no longer the case.
3583
3584 The emulated memove() is provided for those environments that have neither
3585 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3586 is clearly not the case, as these two bugs were recently reported.
3587
3588 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3589 and Detrail to create the HTML documentation, the .txt form of the man
3590 pages, and it removes trailing spaces from listed files. It also creates
3591 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3592 case, it wraps all the #defines with #ifndefs. This script should be run
3593 before "make dist".
3594
3595 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3596 with Unicode property support.
3597
3598 (a) For a maximizing quantifier, if the two different cases of the
3599 character were of different lengths in their UTF-8 codings (there are
3600 some cases like this - I found 11), and the matching function had to
3601 back up over a mixture of the two cases, it incorrectly assumed they
3602 were both the same length.
3603
3604 (b) When PCRE was configured to use the heap rather than the stack for
3605 recursion during matching, it was not correctly preserving the data for
3606 the other case of a UTF-8 character when checking ahead for a match
3607 while processing a minimizing repeat. If the check also involved
3608 matching a wide character, but failed, corruption could cause an
3609 erroneous result when trying to check for a repeat of the original
3610 character.
3611
3612 13. Some tidying changes to the testing mechanism:
3613
3614 (a) The RunTest script now detects the internal link size and whether there
3615 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3616 values substituted by "configure". (The RunGrepTest script already did
3617 this for UTF-8.) The configure.ac script no longer substitutes the
3618 relevant variables.
3619
3620 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3621 with length and offset values. This means that the output is different
3622 for different internal link sizes. Test 2 is skipped for link sizes
3623 other than 2 because of this, bypassing the problem. Unfortunately,
3624 there was also a test in test 3 (the locale tests) that used /B and
3625 failed for link sizes other than 2. Rather than cut the whole test out,
3626 I have added a new /Z option to pcretest that replaces the length and
3627 offset values with spaces. This is now used to make test 3 independent
3628 of link size. (Test 2 will be tidied up later.)
3629
3630 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3631 segmentation fault instead of returning the appropriate error message.
3632
3633 15. In multiline mode when the newline sequence was set to "any", the pattern
3634 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3635 This doesn't seem right; it now treats the CRLF combination as the line
3636 ending, and so does not match in that case. It's only a pattern such as ^$
3637 that would hit this one: something like ^ABC$ would have failed after \r
3638 and then tried again after \r\n.
3639
3640 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3641 in an attempt to make files that differ only in their line terminators
3642 compare equal. This works on Linux.
3643
3644 17. Under certain error circumstances pcregrep might try to free random memory
3645 as it exited. This is now fixed, thanks to valgrind.
3646
3647 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3648 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3649 was because its rules for how to advance for /g after matching an empty
3650 string at the end of a line did not allow for this case. They now check for
3651 it specially.
3652
3653 20. pcretest is supposed to handle patterns and data of any length, by
3654 extending its buffers when necessary. It was getting this wrong when the
3655 buffer for a data line had to be extended.
3656
3657 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3658 CRLF as a newline sequence.
3659
3660 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3661 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3662 I have nevertheless tidied it up.
3663
3664 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3665
3666 24. Added a man page for pcre-config.
3667
3668
3669 Version 7.0 19-Dec-06
3670 ---------------------
3671
3672 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3673 moving to gcc 4.1.1.
3674
3675 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3676 sys/time.h, which is documented as needed for this function. It doesn't
3677 seem to matter on Linux, but it showed up on some releases of OS X.
3678
3679 3. It seems that there are systems where bytes whose values are greater than
3680 127 match isprint() in the "C" locale. The "C" locale should be the
3681 default when a C program starts up. In most systems, only ASCII printing
3682 characters match isprint(). This difference caused the output from pcretest
3683 to vary, making some of the tests fail. I have changed pcretest so that:
3684
3685 (a) When it is outputting text in the compiled version of a pattern, bytes
3686 other than 32-126 are always shown as hex escapes.
3687
3688 (b) When it is outputting text that is a matched part of a subject string,
3689 it does the same, unless a different locale has been set for the match
3690 (using the /L modifier). In this case, it uses isprint() to decide.
3691
3692 4. Fixed a major bug that caused incorrect computation of the amount of memory
3693 required for a compiled pattern when options that changed within the
3694 pattern affected the logic of the preliminary scan that determines the
3695 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3696 that the computed length was too small. The symptoms of this bug were
3697 either the PCRE error "internal error: code overflow" from pcre_compile(),
3698 or a glibc crash with a message such as "pcretest: free(): invalid next
3699 size (fast)". Examples of patterns that provoked this bug (shown in
3700 pcretest format) are:
3701
3702 /(?-x: )/x
3703 /(?x)(?-x: \s*#\s*)/
3704 /((?i)[\x{c0}])/8
3705 /(?i:[\x{c0}])/8
3706
3707 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3708 is now done differently.
3709
3710 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3711 wrapper classes; (b) implement a new function in the C++ scanner that is
3712 more efficient than the old way of doing things because it avoids levels of
3713 recursion in the regex matching; (c) add a paragraph to the documentation
3714 for the FullMatch() function.
3715
3716 6. The escape sequence \n was being treated as whatever was defined as
3717 "newline". Not only was this contrary to the documentation, which states
3718 that \n is character 10 (hex 0A), but it also went horribly wrong when
3719 "newline" was defined as CRLF. This has been fixed.
3720
3721 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3722 was being set to -1 for the "end of line" case (supposedly a value that no
3723 character can have). Though this value is never used (the check for end of
3724 line is "zero bytes in current character"), it caused compiler complaints.
3725 I've changed it to 0xffffffff.
3726
3727 8. In pcre_version.c, the version string was being built by a sequence of
3728 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3729 string (as it is for production releases) called a macro with an empty
3730 argument. The C standard says the result of this is undefined. The gcc
3731 compiler treats it as an empty string (which was what was wanted) but it is
3732 reported that Visual C gives an error. The source has been hacked around to
3733 avoid this problem.
3734
3735 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3736 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3737 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3738 of them did).
3739
3740 10. Originally, pcretest opened its input and output without "b"; then I was
3741 told that "b" was needed in some environments, so it was added for release
3742 5.0 to both the input and output. (It makes no difference on Unix-like
3743 systems.) Later I was told that it is wrong for the input on Windows. I've
3744 now abstracted the modes into two macros, to make it easier to fiddle with
3745 them, and removed "b" from the input mode under Windows.
3746
3747 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3748
3749 12. Added -help and --help to pcretest as an official way of being reminded
3750 of the options.
3751
3752 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3753 and pcrecpp.cc because they annoy compilers at high warning levels.
3754
3755 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3756
3757 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3758 scripts are not C programs :-) and which was not noticed because it works
3759 on Linux.
3760
3761 16. pcretest is supposed to handle any length of pattern and data line (as one
3762 line or as a continued sequence of lines) by extending its input buffer if
3763 necessary. This feature was broken for very long pattern lines, leading to
3764 a string of junk being passed to pcre_compile() if the pattern was longer
3765 than about 50K.
3766
3767 17. I have done a major re-factoring of the way pcre_compile() computes the
3768 amount of memory needed for a compiled pattern. Previously, there was code
3769 that made a preliminary scan of the pattern in order to do this. That was
3770 OK when PCRE was new, but as the facilities have expanded, it has become
3771 harder and harder to keep it in step with the real compile phase, and there
3772 have been a number of bugs (see for example, 4 above). I have now found a
3773 cunning way of running the real compile function in a "fake" mode that
3774 enables it to compute how much memory it would need, while actually only
3775 ever using a few hundred bytes of working memory and without too many
3776 tests of the mode. This should make future maintenance and development
3777 easier. A side effect of this work is that the limit of 200 on the nesting
3778 depth of parentheses has been removed (though this was never a serious
3779 limitation, I suspect). However, there is a downside: pcre_compile() now
3780 runs more slowly than before (30% or more, depending on the pattern). I
3781 hope this isn't a big issue. There is no effect on runtime performance.
3782
3783 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3784 newline (only possible for the last line of a file) and it was a
3785 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3786
3787 19. Added additional timing features to pcretest. (1) The -tm option now times
3788 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3789 separate command line item, by a number that specifies the number of
3790 repeats to use when timing. The default is 50000; this gives better
3791 precision, but takes uncomfortably long for very large patterns.
3792
3793 20. Extended pcre_study() to be more clever in cases where a branch of a
3794 subpattern has no definite first character. For example, (a*|b*)[cd] would
3795 previously give no result from pcre_study(). Now it recognizes that the
3796 first character must be a, b, c, or d.
3797
3798 21. There was an incorrect error "recursive call could loop indefinitely" if
3799 a subpattern (or the entire pattern) that was being tested for matching an
3800 empty string contained only one non-empty item after a nested subpattern.
3801 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3802 incorrectly, because the \d was being skipped in the check.
3803
3804 22. The pcretest program now has a new pattern option /B and a command line
3805 option -b, which is equivalent to adding /B to every pattern. This causes
3806 it to show the compiled bytecode, without the additional information that
3807 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3808 is the same as /B/I).
3809
3810 23. A new optimization is now able automatically to treat some sequences such
3811 as a*b as a*+b. More specifically, if something simple (such as a character
3812 or a simple class like \d) has an unlimited quantifier, and is followed by
3813 something that cannot possibly match the quantified thing, the quantifier
3814 is automatically "possessified".
3815
3816 24. A recursive reference to a subpattern whose number was greater than 39
3817 went wrong under certain circumstances in UTF-8 mode. This bug could also
3818 have affected the operation of pcre_study().
3819
3820 25. Realized that a little bit of performance could be had by replacing
3821 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3822
3823 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3824
3825 27. Possessive quantifiers such as a++ were previously implemented by turning
3826 them into atomic groups such as ($>a+). Now they have their own opcodes,
3827 which improves performance. This includes the automatically created ones
3828 from 23 above.
3829
3830 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3831 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3832 the first matched character to be a colon. This applied both to named and
3833 numbered groups.
3834
3835 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3836
3837 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3838 building PCRE on Windows easier, so I have included it in the distribution.
3839
3840 31. There is now a check in pcretest against a ridiculously large number being
3841 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3842 loop, the loop is abandoned.
3843
3844 32. Forward references to subpatterns in conditions such as (?(2)...) where
3845 subpattern 2 is defined later cause pcre_compile() to search forwards in
3846 the pattern for the relevant set of parentheses. This search went wrong
3847 when there were unescaped parentheses in a character class, parentheses
3848 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3849
3850 33. "Subroutine" calls and backreferences were previously restricted to
3851 referencing subpatterns earlier in the regex. This restriction has now
3852 been removed.
3853
3854 34. Added a number of extra features that are going to be in Perl 5.10. On the
3855 whole, these are just syntactic alternatives for features that PCRE had
3856 previously implemented using the Python syntax or my own invention. The
3857 other formats are all retained for compatibility.
3858
3859 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3860 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3861 also .NET compatible.
3862
3863 (b) A recursion or subroutine call to a named group can now be defined as
3864 (?&name) as well as (?P>name).
3865
3866 (c) A backreference to a named group can now be defined as \k<name> or
3867 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3868 5.10, are also .NET compatible.
3869
3870 (d) A conditional reference to a named group can now use the syntax
3871 (?(<name>) or (?('name') as well as (?(name).
3872
3873 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3874 groups (named and numbered) that are never evaluated inline, but can be
3875 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3876 is always false. There may be only one alternative in such a group.
3877
3878 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3879 as the simple (?(R). The condition is true only if the most recent
3880 recursion is that of the given number or name. It does not search out
3881 through the entire recursion stack.
3882
3883 (g) The escape \gN or \g{N} has been added, where N is a positive or
3884 negative number, specifying an absolute or relative reference.
3885
3886 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3887 some "unreachable code" warnings.
3888
3889 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3890 things, this adds five new scripts.
3891
3892 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3893 There were also incompatibilities regarding the handling of \Q..\E inside
3894 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3895 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3896
3897 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3898 matches an empty string, and forcibly breaks the loop. There were bugs in
3899 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3900 against aaaa the result was just "a" rather than "aaaa", for example. Two
3901 separate and independent bugs (that affected different cases) have been
3902 fixed.
3903
3904 39. Refactored the code to abolish the use of different opcodes for small
3905 capturing bracket numbers. This is a tidy that I avoided doing when I
3906 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3907 The new approach is not only tidier, it makes it possible to reduce the
3908 memory needed to fix the previous bug (38).
3909
3910 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3911 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3912 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3913 mode.
3914
3915 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3916 report.
3917
3918 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3919 copy construction and assignment in the C++ wrapper.
3920
3921 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3922 couple of bugs that could have given wrong results in the "--newline=crlf"
3923 case.
3924
3925 44. Added a number of casts and did some reorganization of signed/unsigned int
3926 variables following suggestions from Dair Grant. Also renamed the variable
3927 "this" as "item" because it is a C++ keyword.
3928
3929 45. Arranged for dftables to add
3930
3931 #include "pcre_internal.h"
3932
3933 to pcre_chartables.c because without it, gcc 4.x may remove the array
3934 definition from the final binary if PCRE is built into a static library and
3935 dead code stripping is activated.
3936
3937 46. For an unanchored pattern, if a match attempt fails at the start of a
3938 newline sequence, and the newline setting is CRLF or ANY, and the next two
3939 characters are CRLF, advance by two characters instead of one.
3940
3941
3942 Version 6.7 04-Jul-06
3943 ---------------------
3944
3945 1. In order to handle tests when input lines are enormously long, pcretest has
3946 been re-factored so that it automatically extends its buffers when
3947 necessary. The code is crude, but this _is_ just a test program. The
3948 default size has been increased from 32K to 50K.
3949
3950 2. The code in pcre_study() was using the value of the re argument before
3951 testing it for NULL. (Of course, in any sensible call of the function, it
3952 won't be NULL.)
3953
3954 3. The memmove() emulation function in pcre_internal.h, which is used on
3955 systems that lack both memmove() and bcopy() - that is, hardly ever -
3956 was missing a "static" storage class specifier.
3957
3958 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3959 containing an extended class (one that cannot be represented by a bitmap
3960 because it contains high-valued characters or Unicode property items, e.g.
3961 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3962 pattern, but PCRE should not loop if you do not (it no longer does).
3963 [Detail: two cases were found: (a) a repeated subpattern containing an
3964 extended class; (b) a recursive reference to a subpattern that followed a
3965 previous extended class. It wasn't skipping over the extended class
3966 correctly when UTF-8 mode was not set.]
3967
3968 5. A negated single-character class was not being recognized as fixed-length
3969 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3970 compile error "lookbehind assertion is not fixed length".
3971
3972 6. The RunPerlTest auxiliary script was showing an unexpected difference
3973 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3974 write a Perl script that can interpret lines of an input file either as
3975 byte characters or as UTF-8, which is what "perltest" was being required to
3976 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3977 can't do is switch easily at run time between having the "use utf8;" pragma
3978 or not. In the end, I fudged it by using the RunPerlTest script to insert
3979 "use utf8;" explicitly for the UTF-8 tests.
3980
3981 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3982 the end of the subject string, contrary to the documentation and to what
3983 Perl does. This was true of both matching functions. Now it matches only at
3984 the start of the subject and immediately after *internal* newlines.
3985
3986 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3987 a pointer to an int instead of a pointer to an unsigned long int. This
3988 caused problems on 64-bit systems.
3989
3990 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3991 instance of the 'standard' template library not being so standard".
3992
3993 10. There was no check on the number of named subpatterns nor the maximum
3994 length of a subpattern name. The product of these values is used to compute
3995 the size of the memory block for a compiled pattern. By supplying a very
3996 long subpattern name and a large number of named subpatterns, the size
3997 computation could be caused to overflow. This is now prevented by limiting
3998 the length of names to 32 characters, and the number of named subpatterns
3999 to 10,000.
4000
4001 11. Subpatterns that are repeated with specific counts have to be replicated in
4002 the compiled pattern. The size of memory for this was computed from the
4003 length of the subpattern and the repeat count. The latter is limited to
4004 65535, but there was no limit on the former, meaning that integer overflow
4005 could in principle occur. The compiled length of a repeated subpattern is
4006 now limited to 30,000 bytes in order to prevent this.
4007
4008 12. Added the optional facility to have named substrings with the same name.
4009
4010 13. Added the ability to use a named substring as a condition, using the
4011 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
4012 are numbers (not recommended). Forward references are permitted.
4013
4014 14. Added forward references in named backreferences (if you see what I mean).
4015
4016 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
4017 pattern could run off the end of the subject. For example, the pattern
4018 "(?s)(.{1,5})"8 did this with the subject "ab".
4019
4020 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
4021 PCRE_CASELESS was set when matching characters that were quantified with ?
4022 or *.
4023
4024 17. A character class other than a single negated character that had a minimum
4025 but no maximum quantifier - for example [ab]{6,} - was not handled
4026 correctly by pce_dfa_exec(). It would match only one character.
4027
4028 18. A valid (though odd) pattern that looked like a POSIX character
4029 class but used an invalid character after [ (for example [[,abc,]]) caused
4030 pcre_compile() to give the error "Failed: internal error: code overflow" or
4031 in some cases to crash with a glibc free() error. This could even happen if
4032 the pattern terminated after [[ but there just happened to be a sequence of
4033 letters, a binary zero, and a closing ] in the memory that followed.
4034
4035 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
4036 over the years. Originally (before any Unicode support), just the bottom 8
4037 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
4038 output from "man perlunicode" includes this:
4039
4040 The regular expression compiler produces polymorphic opcodes. That
4041 is, the pattern adapts to the data and automatically switches to
4042 the Unicode character scheme when presented with Unicode data--or
4043 instead uses a traditional byte scheme when presented with byte
4044 data.
4045
4046 Sadly, a wide octal escape does not cause a switch, and in a string with
4047 no other multibyte characters, these octal escapes are treated as before.
4048 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
4049 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
4050 Unicode string.
4051
4052 I have not perpetrated such confusion in PCRE. Up till now, it took just
4053 the bottom 8 bits, as in old Perl. I have now made octal escapes with
4054 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
4055 translate to the appropriate multibyte character.
4056
4057 29. Applied some refactoring to reduce the number of warnings from Microsoft
4058 and Borland compilers. This has included removing the fudge introduced
4059 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
4060 a warning about an unused variable.
4061
4062 21. PCRE has not included VT (character 0x0b) in the set of whitespace
4063 characters since release 4.0, because Perl (from release 5.004) does not.
4064 [Or at least, is documented not to: some releases seem to be in conflict
4065 with the documentation.] However, when a pattern was studied with
4066 pcre_study() and all its branches started with \s, PCRE still included VT
4067 as a possible starting character. Of course, this did no harm; it just
4068 caused an unnecessary match attempt.
4069
4070 22. Removed a now-redundant internal flag bit that recorded the fact that case
4071 dependency changed within the pattern. This was once needed for "required
4072 byte" processing, but is no longer used. This recovers a now-scarce options
4073 bit. Also moved the least significant internal flag bit to the most-
4074 significant bit of the word, which was not previously used (hangover from
4075 the days when it was an int rather than a uint) to free up another bit for
4076 the future.
4077
4078 23. Added support for CRLF line endings as well as CR and LF. As well as the
4079 default being selectable at build time, it can now be changed at runtime
4080 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
4081 specify that it is scanning data with non-default line endings.
4082
4083 24. Changed the definition of CXXLINK to make it agree with the definition of
4084 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
4085
4086 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
4087 recursions. This makes a big different to stack usage for some patterns.
4088
4089 26. If a subpattern containing a named recursion or subroutine reference such
4090 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
4091 the space required for the compiled pattern went wrong and gave too small a
4092 value. Depending on the environment, this could lead to "Failed: internal
4093 error: code overflow at offset 49" or "glibc detected double free or
4094 corruption" errors.
4095
4096 27. Applied patches from Google (a) to support the new newline modes and (b) to
4097 advance over multibyte UTF-8 characters in GlobalReplace.
4098
4099 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
4100 difference for some implementation of PCRE in some Windows version.
4101
4102 29. Added some extra testing facilities to pcretest:
4103
4104 \q<number> in a data line sets the "match limit" value
4105 \Q<number> in a data line sets the "match recursion limt" value
4106 -S <number> sets the stack size, where <number> is in megabytes
4107
4108 The -S option isn't available for Windows.
4109
4110
4111 Version 6.6 06-Feb-06
4112 ---------------------
4113
4114 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
4115 in pcreposix.h. I have copied the definition from pcre.h.
4116
4117 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
4118 because pcre.h is no longer a built file.
4119
4120 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
4121 not normally included in the compiled code.
4122
4123
4124 Version 6.5 01-Feb-06
4125 ---------------------
4126
4127 1. When using the partial match feature with pcre_dfa_exec(), it was not
4128 anchoring the second and subsequent partial matches at the new starting
4129 point. This could lead to incorrect results. For example, with the pattern
4130 /1234/, partially matching against "123" and then "a4" gave a match.
4131
4132 2. Changes to pcregrep:
4133
4134 (a) All non-match returns from pcre_exec() were being treated as failures
4135 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
4136 error message is output. Some extra information is given for the
4137 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
4138 probably the only errors that are likely to be caused by users (by
4139 specifying a regex that has nested indefinite repeats, for instance).
4140 If there are more than 20 of these errors, pcregrep is abandoned.
4141
4142 (b) A binary zero was treated as data while matching, but terminated the
4143 output line if it was written out. This has been fixed: binary zeroes
4144 are now no different to any other data bytes.
4145
4146 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
4147 used to set a locale for matching. The --locale=xxxx long option has
4148 been added (no short equivalent) to specify a locale explicitly on the
4149 pcregrep command, overriding the environment variables.
4150
4151 (d) When -B was used with -n, some line numbers in the output were one less
4152 than they should have been.
4153
4154 (e) Added the -o (--only-matching) option.
4155
4156 (f) If -A or -C was used with -c (count only), some lines of context were
4157 accidentally printed for the final match.
4158
4159 (g) Added the -H (--with-filename) option.
4160
4161 (h) The combination of options -rh failed to suppress file names for files
4162 that were found from directory arguments.
4163
4164 (i) Added the -D (--devices) and -d (--directories) options.
4165
4166 (j) Added the -F (--fixed-strings) option.
4167
4168 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4169
4170 (l) Added the --colo(u)r option.
4171
4172 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4173 is not present by default.
4174
4175 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4176 items such as (?R) or (?1), when the recursion could match a number of
4177 alternatives. If it matched one of the alternatives, but subsequently,
4178 outside the recursion, there was a failure, the code tried to back up into
4179 the recursion. However, because of the way PCRE is implemented, this is not
4180 possible, and the result was an incorrect result from the match.
4181
4182 In order to prevent this happening, the specification of recursion has
4183 been changed so that all such subpatterns are automatically treated as
4184 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4185
4186 4. I had overlooked the fact that, in some locales, there are characters for
4187 which isalpha() is true but neither isupper() nor islower() are true. In
4188 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4189 and ordfeminine) are like this. This affected the treatment of \w and \W
4190 when they appeared in character classes, but not when they appeared outside
4191 a character class. The bit map for "word" characters is now created
4192 separately from the results of isalnum() instead of just taking it from the
4193 upper, lower, and digit maps. (Plus the underscore character, of course.)
4194
4195 5. The above bug also affected the handling of POSIX character classes such as
4196 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4197 permanent tables. Instead, the bit maps for such a class were previously
4198 created as the appropriate unions of the upper, lower, and digit bitmaps.
4199 Now they are created by subtraction from the [[:word:]] class, which has
4200 its own bitmap.
4201
4202 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4203 It is created by subtracting the vertical space characters (\x09, \x0a,
4204 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4205 subtraction was done in the overall bitmap for a character class, meaning
4206 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4207 be recognized. This bug has been fixed.
4208
4209 7. Patches from the folks at Google:
4210
4211 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4212 real life, but is still worth protecting against".
4213
4214 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4215 regular expressions".
4216
4217 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4218 have it.
4219
4220 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4221 "configure" and the latter not, in order to fix a problem somebody had
4222 with compiling the Arg class on HP-UX.
4223
4224 (e) Improve the error-handling of the C++ wrapper a little bit.
4225
4226 (f) New tests for checking recursion limiting.
4227
4228 8. The pcre_memmove() function, which is used only if the environment does not
4229 have a standard memmove() function (and is therefore rarely compiled),
4230 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4231 returning a result (though PCRE never actually uses the result).
4232
4233 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4234 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4235 returned instead of calling malloc() with an overflowing number that would
4236 most likely cause subsequent chaos.
4237
4238 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4239
4240 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4241 with this option is matched, the nmatch and pmatch options of regexec() are
4242 ignored.
4243
4244 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4245 provided in case anyone wants to the the POSIX interface with UTF-8
4246 strings.
4247
4248 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4249 C++ linking (needed for some HP-UX environments).
4250
4251 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4252 (unused parameter) and in the pcre_printint() function (omitted "default"
4253 switch label when the default is to do nothing).
4254
4255 15. Added some code to make it possible, when PCRE is compiled as a C++
4256 library, to replace subject pointers for pcre_exec() with a smart pointer
4257 class, thus making it possible to process discontinuous strings.
4258
4259 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4260 much the same function. They were added by different people who were trying
4261 to make PCRE easy to compile on non-Unix systems. It has been suggested
4262 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4263 for compiling on Windows systems. I have therefore replaced it with
4264 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4265 defaults to "extern" for C or "extern C" for C++, which works fine on
4266 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4267 SCOPE with something explicit in config.h. In addition:
4268
4269 (a) pcreposix.h still had just "extern" instead of either of these macros;
4270 I have replaced it with PCRE_DATA_SCOPE.
4271
4272 (b) Functions such as _pcre_xclass(), which are internal to the library,
4273 but external in the C sense, all had PCRE_EXPORT in their definitions.
4274 This is apparently wrong for the Windows case, so I have removed it.
4275 (It makes no difference on Unix-like systems.)
4276
4277 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4278 of recursive calls to match(). This is different to MATCH_LIMIT because
4279 that limits the total number of calls to match(), not all of which increase
4280 the depth of recursion. Limiting the recursion depth limits the amount of
4281 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4282 when PCRE is compiled, and changed at run time. A patch from Google adds
4283 this functionality to the C++ interface.
4284
4285 18. Changes to the handling of Unicode character properties:
4286
4287 (a) Updated the table to Unicode 4.1.0.
4288
4289 (b) Recognize characters that are not in the table as "Cn" (undefined).
4290
4291 (c) I revised the way the table is implemented to a much improved format
4292 which includes recognition of ranges. It now supports the ranges that
4293 are defined in UnicodeData.txt, and it also amalgamates other
4294 characters into ranges. This has reduced the number of entries in the
4295 table from around 16,000 to around 3,000, thus reducing its size
4296 considerably. I realized I did not need to use a tree structure after
4297 all - a binary chop search is just as efficient. Having reduced the
4298 number of entries, I extended their size from 6 bytes to 8 bytes to
4299 allow for more data.
4300
4301 (d) Added support for Unicode script names via properties such as \p{Han}.
4302
4303 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4304 matching that character.
4305
4306 20. When matching a repeated Unicode property with a minimum greater than zero,
4307 (for example \pL{2,}), PCRE could look past the end of the subject if it
4308 reached it while seeking the minimum number of characters. This could
4309 happen only if some of the characters were more than one byte long, because
4310 there is a check for at least the minimum number of bytes.
4311
4312 21. Refactored the implementation of \p and \P so as to be more general, to
4313 allow for more different types of property in future. This has changed the
4314 compiled form incompatibly. Anybody with saved compiled patterns that use
4315 \p or \P will have to recompile them.
4316
4317 22. Added "Any" and "L&" to the supported property types.
4318
4319 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4320 but give a compile time error if the value is greater than 0xff.
4321
4322 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4323 accidentally not being installed or uninstalled.
4324
4325 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4326 made were to insert the current release number. This seemed silly, because
4327 it made things harder for people building PCRE on systems that don't run
4328 "configure". I have turned pcre.h into a distributed file, no longer built
4329 by "configure", with the version identification directly included. There is
4330 no longer a pcre.h.in file.
4331
4332 However, this change necessitated a change to the pcre-config script as
4333 well. It is built from pcre-config.in, and one of the substitutions was the
4334 release number. I have updated configure.ac so that ./configure now finds
4335 the release number by grepping pcre.h.
4336
4337 26. Added the ability to run the tests under valgrind.
4338
4339
4340 Version 6.4 05-Sep-05
4341 ---------------------
4342
4343 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4344 "--" to be printed when multiple files were scanned, even when none of the
4345 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4346 consider it to be a bug, and have restored the previous behaviour.
4347
4348 2. A couple of code tidies to get rid of compiler warnings.
4349
4350 3. The pcretest program used to cheat by referring to symbols in the library
4351 whose names begin with _pcre_. These are internal symbols that are not
4352 really supposed to be visible externally, and in some environments it is
4353 possible to suppress them. The cheating is now confined to including
4354 certain files from the library's source, which is a bit cleaner.
4355
4356 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4357 file's purpose clearer.
4358
4359 5. Reorganized pcre_ucp_findchar().
4360
4361
4362 Version 6.3 15-Aug-05
4363 ---------------------
4364
4365 1. The file libpcre.pc.in did not have general read permission in the tarball.
4366
4367 2. There were some problems when building without C++ support:
4368
4369 (a) If C++ support was not built, "make install" and "make test" still
4370 tried to test it.
4371
4372 (b) There were problems when the value of CXX was explicitly set. Some
4373 changes have been made to try to fix these, and ...
4374
4375 (c) --disable-cpp can now be used to explicitly disable C++ support.
4376
4377 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4378 backslash in a target when C++ was disabled. This confuses some
4379 versions of "make", apparently. Using an intermediate variable solves
4380 this. (Same for CPP_LOBJ.)
4381
4382 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4383 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4384 necessary on certain architectures.
4385
4386 4. Added a setting of -export-symbols-regex to the link command to remove
4387 those symbols that are exported in the C sense, but actually are local
4388 within the library, and not documented. Their names all begin with
4389 "_pcre_". This is not a perfect job, because (a) we have to except some
4390 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4391 available (and never for static libraries). I have made a note to try to
4392 find a way round (a) in the future.
4393
4394
4395 Version 6.2 01-Aug-05
4396 ---------------------
4397
4398 1. There was no test for integer overflow of quantifier values. A construction
4399 such as {1111111111111111} would give undefined results. What is worse, if
4400 a minimum quantifier for a parenthesized subpattern overflowed and became
4401 negative, the calculation of the memory size went wrong. This could have
4402 led to memory overwriting.
4403
4404 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4405
4406 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4407 operating environments where this matters.
4408
4409 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4410 PCRE options from within the C++ wrapper.
4411
4412 5. Named capturing subpatterns were not being correctly counted when a pattern
4413 was compiled. This caused two problems: (a) If there were more than 100
4414 such subpatterns, the calculation of the memory needed for the whole
4415 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4416 back references of the form \12, where the number was greater than 9, were
4417 not recognized as back references, even though there were sufficient
4418 previous subpatterns.
4419
4420 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4421 versions of gcc, e.g. 2.95.4.
4422
4423
4424 Version 6.1 21-Jun-05
4425 ---------------------
4426
4427 1. There was one reference to the variable "posix" in pcretest.c that was not
4428 surrounded by "#if !defined NOPOSIX".
4429
4430 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4431 the cross-check on the old pcre_info() function, for the benefit of the
4432 cut-down version of PCRE that is currently imported into Exim.
4433
4434 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4435 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4436 patterns that start (?i) but I don't think that matters. The use of (?i) is
4437 just an example; this all applies to the other options as well.
4438
4439 4. Since libtool seems to echo the compile commands it is issuing, the output
4440 from "make" can be reduced a bit by putting "@" in front of each libtool
4441 compile command.
4442
4443 5. Patch from the folks at Google for configure.in to be a bit more thorough
4444 in checking for a suitable C++ installation before trying to compile the
4445 C++ stuff. This should fix a reported problem when a compiler was present,
4446 but no suitable headers.
4447
4448 6. The man pages all had just "PCRE" as their title. I have changed them to
4449 be the relevant file name. I have also arranged that these names are
4450 retained in the file doc/pcre.txt, which is a concatenation in text format
4451 of all the man pages except the little individual ones for each function.
4452
4453 7. The NON-UNIX-USE file had not been updated for the different set of source
4454 files that come with release 6. I also added a few comments about the C++
4455 wrapper.
4456
4457
4458 Version 6.0 07-Jun-05
4459 ---------------------
4460
4461 1. Some minor internal re-organization to help with my DFA experiments.
4462
4463 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4464 didn't matter for the library itself when fully configured, but did matter
4465 when compiling without UCP support, or within Exim, where the ucp files are
4466 not imported.
4467
4468 3. Refactoring of the library code to split up the various functions into
4469 different source modules. The addition of the new DFA matching code (see
4470 below) to a single monolithic source would have made it really too
4471 unwieldy, quite apart from causing all the code to be include in a
4472 statically linked application, when only some functions are used. This is
4473 relevant even without the DFA addition now that patterns can be compiled in
4474 one application and matched in another.
4475
4476 The downside of splitting up is that there have to be some external
4477 functions and data tables that are used internally in different modules of
4478 the library but which are not part of the API. These have all had their
4479 names changed to start with "_pcre_" so that they are unlikely to clash
4480 with other external names.
4481
4482 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4483 a different (DFA) algorithm. Although it is slower than the original
4484 function, it does have some advantages for certain types of matching
4485 problem.
4486
4487 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4488 including restarting after a partial match.
4489
4490 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4491 defined when compiling for Windows was sent to me. I have put it into the
4492 code, though I have no means of testing or verifying it.
4493
4494 7. Added the pcre_refcount() auxiliary function.
4495
4496 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4497 match before or at the first newline in the subject string. In pcretest,
4498 the /f option on a pattern can be used to set this.
4499
4500 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4501 would behave wrongly. This has been present in PCRE since release 4.0.
4502
4503 10. A number of changes to the pcregrep command:
4504
4505 (a) Refactored how -x works; insert ^(...)$ instead of setting
4506 PCRE_ANCHORED and checking the length, in preparation for adding
4507 something similar for -w.
4508
4509 (b) Added the -w (match as a word) option.
4510
4511 (c) Refactored the way lines are read and buffered so as to have more
4512 than one at a time available.
4513
4514 (d) Implemented a pcregrep test script.
4515
4516 (e) Added the -M (multiline match) option. This allows patterns to match
4517 over several lines of the subject. The buffering ensures that at least
4518 8K, or the rest of the document (whichever is the shorter) is available
4519 for matching (and similarly the previous 8K for lookbehind assertions).
4520
4521 (f) Changed the --help output so that it now says
4522
4523 -w, --word-regex(p)
4524
4525 instead of two lines, one with "regex" and the other with "regexp"
4526 because that confused at least one person since the short forms are the
4527 same. (This required a bit of code, as the output is generated
4528 automatically from a table. It wasn't just a text change.)
4529
4530 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4531 option but starts with a hyphen. Could be a pattern or a path name
4532 starting with a hyphen, for instance.
4533
4534 (h) "-" can be given as a file name to represent stdin.
4535
4536 (i) When file names are being printed, "(standard input)" is used for
4537 the standard input, for compatibility with GNU grep. Previously
4538 "<stdin>" was used.
4539
4540 (j) The option --label=xxx can be used to supply a name to be used for
4541 stdin when file names are being printed. There is no short form.
4542
4543 (k) Re-factored the options decoding logic because we are going to add
4544 two more options that take data. Such options can now be given in four
4545 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4546
4547 (l) Added the -A, -B, and -C options for requesting that lines of context
4548 around matches be printed.
4549
4550 (m) Added the -L option to print the names of files that do not contain
4551 any matching lines, that is, the complement of -l.
4552
4553 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4554 continue to scan other files.
4555
4556 (o) The -s option was incorrectly implemented. For compatibility with other
4557 greps, it now suppresses the error message for a non-existent or non-
4558 accessible file (but not the return code). There is a new option called
4559 -q that suppresses the output of matching lines, which was what -s was
4560 previously doing.
4561
4562 (p) Added --include and --exclude options to specify files for inclusion
4563 and exclusion when recursing.
4564
4565 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4566 Hopefully, it now does.
4567
4568 12. Missing cast in pcre_study().
4569
4570 13. Added an "uninstall" target to the makefile.
4571
4572 14. Replaced "extern" in the function prototypes in Makefile.in with
4573 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4574 world, but is set differently for Windows.
4575
4576 15. Added a second compiling function called pcre_compile2(). The only
4577 difference is that it has an extra argument, which is a pointer to an
4578 integer error code. When there is a compile-time failure, this is set
4579 non-zero, in addition to the error test pointer being set to point to an
4580 error message. The new argument may be NULL if no error number is required
4581 (but then you may as well call pcre_compile(), which is now just a
4582 wrapper). This facility is provided because some applications need a
4583 numeric error indication, but it has also enabled me to tidy up the way
4584 compile-time errors are handled in the POSIX wrapper.
4585
4586 16. Added VPATH=.libs to the makefile; this should help when building with one
4587 prefix path and installing with another. (Or so I'm told by someone who
4588 knows more about this stuff than I do.)
4589
4590 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4591 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4592 match everything, including newlines. This is not POSIX-compatible, but
4593 somebody wanted the feature. From pcretest it can be activated by using
4594 both the P and the s flags.
4595
4596 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4597
4598 19. libpcre.pc was being incorrectly installed as executable.
4599
4600 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4601 it now also looks for '\r' so that it will work unmodified on Windows.
4602
4603 21. Added Google's contributed C++ wrapper to the distribution.
4604
4605 22. Added some untidy missing memory free() calls in pcretest, to keep
4606 Electric Fence happy when testing.
4607
4608
4609
4610 Version 5.0 13-Sep-04
4611 ---------------------
4612
4613 1. Internal change: literal characters are no longer packed up into items
4614 containing multiple characters in a single byte-string. Each character
4615 is now matched using a separate opcode. However, there may be more than one
4616 byte in the character in UTF-8 mode.
4617
4618 2. The pcre_callout_block structure has two new fields: pattern_position and
4619 next_item_length. These contain the offset in the pattern to the next match
4620 item, and its length, respectively.
4621
4622 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4623 insertion of callouts before each pattern item. Added the /C option to
4624 pcretest to make use of this.
4625
4626 4. On the advice of a Windows user, the lines
4627
4628 #if defined(_WIN32) || defined(WIN32)
4629 _setmode( _fileno( stdout ), 0x8000 );
4630 #endif /* defined(_WIN32) || defined(WIN32) */
4631
4632 have been added to the source of pcretest. This apparently does useful
4633 magic in relation to line terminators.
4634
4635 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4636 for the benefit of those environments where the "b" makes a difference.
4637
4638 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4639 to know about it. I have put a hack into configure.in that adds in code
4640 to set GCC=yes if CC=icc. This seems to end up at a point in the
4641 generated configure script that is early enough to affect the setting of
4642 compiler options, which is what is needed, but I have no means of testing
4643 whether it really works. (The user who reported this had patched the
4644 generated configure script, which of course I cannot do.)
4645
4646 LATER: After change 22 below (new libtool files), the configure script
4647 seems to know about icc (and also ecc). Therefore, I have commented out
4648 this hack in configure.in.
4649
4650 7. Added support for pkg-config (2 patches were sent in).
4651
4652 8. Negated POSIX character classes that used a combination of internal tables
4653 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4654 [[:^ascii]]. Typically, they would match almost any characters. The other
4655 POSIX classes were not broken in this way.
4656
4657 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4658 to find the match, as PCRE was deluded into thinking that the match had to
4659 start at the start point or following a newline. The same bug applied to
4660 patterns with negative forward assertions or any backward assertions
4661 preceding ".*" at the start, unless the pattern required a fixed first
4662 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4663
4664 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4665 starting at the last subject character, bytes beyond the end of the subject
4666 string were read.
4667
4668 11. Renamed the variable "class" as "classbits" to make life easier for C++
4669 users. (Previously there was a macro definition, but it apparently wasn't
4670 enough.)
4671
4672 12. Added the new field "tables" to the extra data so that tables can be passed
4673 in at exec time, or the internal tables can be re-selected. This allows
4674 a compiled regex to be saved and re-used at a later time by a different
4675 program that might have everything at different addresses.
4676
4677 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4678 -R library as well as a -L library.
4679
4680 14. The debugging options of pcretest (-d on the command line or D on a
4681 pattern) showed incorrect output for anything following an extended class
4682 that contained multibyte characters and which was followed by a quantifier.
4683
4684 15. Added optional support for general category Unicode character properties
4685 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4686 support. It adds about 90K to the size of the library. The meanings of the
4687 inbuilt class escapes such as \d and \s have NOT been changed.
4688
4689 16. Updated pcredemo.c to include calls to free() to release the memory for the
4690 compiled pattern.
4691
4692 17. The generated file chartables.c was being created in the source directory
4693 instead of in the building directory. This caused the build to fail if the
4694 source directory was different from the building directory, and was
4695 read-only.
4696
4697 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4698 file. No doubt somebody will tell me if they don't make sense... Also added
4699 Dan Mooney's comments about building on OpenVMS.
4700
4701 19. Added support for partial matching via the PCRE_PARTIAL option for
4702 pcre_exec() and the \P data escape in pcretest.
4703
4704 20. Extended pcretest with 3 new pattern features:
4705
4706 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4707 write the compiled pattern to the file whose name is "rest-of-line".
4708 This is a straight binary dump of the data, with the saved pointer to
4709 the character tables forced to be NULL. The study data, if any, is
4710 written too. After writing, pcretest reads a new pattern.
4711
4712 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4713 compiled pattern from the given file. There must not be any
4714 occurrences of "<" in the file name (pretty unlikely); if there are,
4715 pcretest will instead treat the initial "<" as a pattern delimiter.
4716 After reading in the pattern, pcretest goes on to read data lines as
4717 usual.
4718
4719 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4720 and 16-bit fields in a compiled pattern, to simulate a pattern that
4721 was compiled on a host of opposite endianness.
4722
4723 21. The pcre-exec() function can now cope with patterns that were compiled on
4724 hosts of opposite endianness, with this restriction:
4725
4726 As for any compiled expression that is saved and used later, the tables
4727 pointer field cannot be preserved; the extra_data field in the arguments
4728 to pcre_exec() should be used to pass in a tables address if a value
4729 other than the default internal tables were used at compile time.
4730
4731 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4732 now diagnosed as an error. Previously, most of the time, a negative number
4733 would have been treated as zero, but if in addition "ovector" was passed as
4734 NULL, a crash could occur.
4735
4736 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4737 new versions from the libtool 1.5 distribution (the last one is a copy of
4738 a file called libtool.m4). This seems to have fixed the need to patch
4739 "configure" to support Darwin 1.3 (which I used to do). However, I still
4740 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4741 workstation).
4742
4743 24. Changed the PCRE licence to be the more standard "BSD" licence.
4744
4745
4746 Version 4.5 01-Dec-03
4747 ---------------------
4748
4749 1. There has been some re-arrangement of the code for the match() function so
4750 that it can be compiled in a version that does not call itself recursively.
4751 Instead, it keeps those local variables that need separate instances for
4752 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4753 needs to "recurse". Keeping track of where control must go is done by means
4754 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4755 hide most of the details from the main code, and operates only if
4756 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4757 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4758 operating.
4759
4760 To make it easier for callers to provide specially tailored get/free
4761 functions for this usage, two new functions, pcre_stack_malloc, and
4762 pcre_stack_free, are used. They are always called in strict stacking order,
4763 and the size of block requested is always the same.
4764
4765 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4766 PCRE has been compiled to use the stack or the heap for recursion. The
4767 -C option of pcretest uses this to show which version is compiled.
4768
4769 A new data escape \S, is added to pcretest; it causes the amounts of store
4770 obtained and freed by both kinds of malloc/free at match time to be added
4771 to the output.
4772
4773 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4774 what's available on my current Linux desktop machine.
4775
4776 3. When matching a UTF-8 string, the test for a valid string at the start has
4777 been extended. If start_offset is not zero, PCRE now checks that it points
4778 to a byte that is the start of a UTF-8 character. If not, it returns
4779 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4780 this is necessary because there may be backward assertions in the pattern.
4781 When matching the same subject several times, it may save resources to use
4782 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4783
4784 4. The code for checking the validity of UTF-8 strings has been tightened so
4785 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4786 containing "overlong sequences".
4787
4788 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4789 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4790 should have been "&", but it just so happened that all the cases this let
4791 through by mistake were picked up later in the function.
4792
4793 6. I had used a variable called "isblank" - this is a C99 function, causing
4794 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4795
4796 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4797 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4798 so the version is shown; (c) stop "make test" echoing "./RunTest".
4799
4800 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4801
4802 9. The prototype for memmove() for systems that don't have it was using
4803 size_t, but the inclusion of the header that defines size_t was later. I've
4804 moved the #includes for the C headers earlier to avoid this.
4805
4806 10. Added some adjustments to the code to make it easier to compiler on certain
4807 special systems:
4808
4809 (a) Some "const" qualifiers were missing.
4810 (b) Added the macro EXPORT before all exported functions; by default this
4811 is defined to be empty.
4812 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4813 that it reads its output file name as an argument instead of writing
4814 to the standard output and assuming this can be redirected.
4815
4816 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4817 class containing characters with values greater than 255, PCRE compilation
4818 went into a loop.
4819
4820 12. A recursive reference to a subpattern that was within another subpattern
4821 that had a minimum quantifier of zero caused PCRE to crash. For example,
4822 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4823 recursion. If the recursively-called subpattern itself had a zero repeat,
4824 that was OK.
4825
4826 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4827 buffer into which it was copied (for escape processing) was still set at
4828 1024, so long lines caused crashes.
4829
4830 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4831 "internal error: code overflow...". This applied to any character class
4832 that was followed by a possessive quantifier.
4833
4834 15. Modified the Makefile to add libpcre.la as a prerequisite for
4835 libpcreposix.la because I was told this is needed for a parallel build to
4836 work.
4837
4838 16. If a pattern that contained .* following optional items at the start was
4839 studied, the wrong optimizing data was generated, leading to matching
4840 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4841 matching string must start with a or b or c. The correct conclusion for
4842 this pattern is that a match can start with any character.
4843
4844
4845 Version 4.4 13-Aug-03
4846 ---------------------
4847
4848 1. In UTF-8 mode, a character class containing characters with values between
4849 127 and 255 was not handled correctly if the compiled pattern was studied.
4850 In fixing this, I have also improved the studying algorithm for such
4851 classes (slightly).
4852
4853 2. Three internal functions had redundant arguments passed to them. Removal
4854 might give a very teeny performance improvement.
4855
4856 3. Documentation bug: the value of the capture_top field in a callout is *one
4857 more than* the number of the hightest numbered captured substring.
4858
4859 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4860 in incorrectly linking with a previously installed version. They now link
4861 explicitly with libpcre.la.
4862
4863 5. configure.in no longer needs to recognize Cygwin specially.
4864
4865 6. A problem in pcre.in for Windows platforms is fixed.
4866
4867 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4868 pcretest, it used to include the size of the study block as part of its
4869 output. Unfortunately, the structure contains a field that has a different
4870 size on different hardware architectures. This meant that the tests that
4871 showed this size failed. As the block is currently always of a fixed size,
4872 this information isn't actually particularly useful in pcretest output, so
4873 I have just removed it.
4874
4875 8. Three pre-processor statements accidentally did not start in column 1.
4876 Sadly, there are *still* compilers around that complain, even though
4877 standard C has not required this for well over a decade. Sigh.
4878
4879 9. In pcretest, the code for checking callouts passed small integers in the
4880 callout_data field, which is a void * field. However, some picky compilers
4881 complained about the casts involved for this on 64-bit systems. Now
4882 pcretest passes the address of the small integer instead, which should get
4883 rid of the warnings.
4884
4885 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4886 both compile and run time, and gives an error if an invalid UTF-8 sequence
4887 is found. There is a option for disabling this check in cases where the
4888 string is known to be correct and/or the maximum performance is wanted.
4889
4890 11. In response to a bug report, I changed one line in Makefile.in from
4891
4892 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4893 to
4894 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4895
4896 to look similar to other lines, but I have no way of telling whether this
4897 is the right thing to do, as I do not use Windows. No doubt I'll get told
4898 if it's wrong...
4899
4900
4901 Version 4.3 21-May-03
4902 ---------------------
4903
4904 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4905 Makefile.
4906
4907 2. Some refactoring to improve the quality of the code:
4908
4909 (i) The utf8_table... variables are now declared "const".
4910
4911 (ii) The code for \cx, which used the "case flipping" table to upper case
4912 lower case letters, now just substracts 32. This is ASCII-specific,
4913 but the whole concept of \cx is ASCII-specific, so it seems
4914 reasonable.
4915
4916 (iii) PCRE was using its character types table to recognize decimal and
4917 hexadecimal digits in the pattern. This is silly, because it handles
4918 only 0-9, a-f, and A-F, but the character types table is locale-
4919 specific, which means strange things might happen. A private
4920 table is now used for this - though it costs 256 bytes, a table is
4921 much faster than multiple explicit tests. Of course, the standard
4922 character types table is still used for matching digits in subject
4923 strings against \d.
4924
4925 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4926 ending in _t are). So I've renamed it as ESC_tee.
4927
4928 3. The first argument for regexec() in the POSIX wrapper should have been
4929 defined as "const".
4930
4931 4. Changed pcretest to use malloc() for its buffers so that they can be
4932 Electric Fenced for debugging.
4933
4934 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4935 to read one or more bytes before the start of the subject string. Often this
4936 had no effect on PCRE's behaviour, but in some circumstances it could
4937 provoke a segmentation fault.
4938
4939 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4940 to try to read one or more bytes before the start of the subject string.
4941
4942 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4943 UTF-8 support could misbehave in various ways if the subject string
4944 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4945 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4946 back over UTF-8 characters.)
4947
4948
4949 Version 4.2 14-Apr-03
4950 ---------------------
4951
4952 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4953
4954 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4955 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4956 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4957 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4958 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4959 and BUILD_EXEEXT
4960 Note: automatic setting of the BUILD variables is not yet working
4961 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4962 compile-time but not at link-time
4963 [LINK]: use for linking executables only
4964 make different versions for Windows and non-Windows
4965 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4966 libraries
4967 [LINK_FOR_BUILD]: new variable
4968 [OBJEXT]: use throughout
4969 [EXEEXT]: use throughout
4970 <winshared>: new target
4971 <wininstall>: new target
4972 <dftables.o>: use native compiler
4973 <dftables>: use native linker
4974 <install>: handle Windows platform correctly
4975 <clean>: ditto
4976 <check>: ditto
4977 copy DLL to top builddir before testing
4978
4979 As part of these changes, -no-undefined was removed again. This was reported
4980 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4981 in any case.
4982
4983 3. Some tidies to get rid of compiler warnings:
4984
4985 . In the match_data structure, match_limit was an unsigned long int, whereas
4986 match_call_count was an int. I've made them both unsigned long ints.
4987
4988 . In pcretest the fact that a const uschar * doesn't automatically cast to
4989 a void * provoked a warning.
4990
4991 . Turning on some more compiler warnings threw up some "shadow" variables
4992 and a few more missing casts.
4993
4994 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4995 option, a class that contained a single character with a value between 128
4996 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4997
4998 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4999 option, a class that contained several characters, but with at least one
5000 whose value was between 128 and 255 caused PCRE to crash.
5001
5002
5003 Version 4.1 12-Mar-03
5004 ---------------------
5005
5006 1. Compiling with gcc -pedantic found a couple of places where casts were
5007 needed, and a string in dftables.c that was longer than standard compilers are
5008 required to support.
5009
5010 2. Compiling with Sun's compiler found a few more places where the code could
5011 be tidied up in order to avoid warnings.
5012
5013 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
5014 first of these names is deprecated in the latest Autoconf in favour of the name
5015 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
5016 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
5017 analogy I have changed it to CFLAGS_FOR_BUILD.
5018
5019 4. Added -no-undefined to the linking command in the Makefile, because this is
5020 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
5021 linking step for the pcreposix library.
5022
5023 5. PCRE was failing to diagnose the case of two named groups with the same
5024 name.
5025
5026 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
5027 literal character that is needed in the subject for a match, and scans along to
5028 ensure that it is present before embarking on the full matching process. This
5029 saves time in cases of nested unlimited repeats that are never going to match.
5030 Problem: the scan can take a lot of time if the subject is very long (e.g.
5031 megabytes), thus penalizing straightforward matches. It is now done only if the
5032 amount of subject to be scanned is less than 1000 bytes.
5033
5034 7. A lesser problem with the same optimization is that it was recording the
5035 first character of an anchored pattern as "needed", thus provoking a search
5036 right along the subject, even when the first match of the pattern was going to
5037 fail. The "needed" character is now not set for anchored patterns, unless it
5038 follows something in the pattern that is of non-fixed length. Thus, it still
5039 fulfils its original purpose of finding quick non-matches in cases of nested
5040 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
5041
5042
5043 Version 4.0 17-Feb-03
5044 ---------------------
5045
5046 1. If a comment in an extended regex that started immediately after a meta-item
5047 extended to the end of string, PCRE compiled incorrect data. This could lead to
5048 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
5049
5050 2. Moved to autoconf 2.53 and libtool 1.4.2.
5051
5052 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
5053 the special perltest8 script is no longer needed - all the tests can be run
5054 from a single perltest script.
5055
5056 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
5057 by \s. It has now been removed in PCRE. This means it isn't recognized as
5058 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
5059 class [:space:] *does* include VT, thereby creating a mess.
5060
5061 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
5062 space and tab.
5063
5064 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
5065 its new features into the main test script, reducing the number of scripts.
5066
5067 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
5068 were backward compatible, and made the (?i) apply to the whole pattern, as if
5069 /i were given. Now it behaves more logically, and applies the option setting
5070 only to what follows. PCRE has been changed to follow suit. However, if it
5071 finds options settings right at the start of the pattern, it extracts them into
5072 the global options, as before. Thus, they show up in the info data.
5073
5074 8. Added support for the \Q...\E escape sequence. Characters in between are
5075 treated as literals. This is slightly different from Perl in that $ and @ are
5076 also handled as literals inside the quotes. In Perl, they will cause variable
5077 interpolation. Note the following examples:
5078
5079 Pattern PCRE matches Perl matches
5080
5081 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
5082 \Qabc\$xyz\E abc\$xyz abc\$xyz
5083 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
5084
5085 For compatibility with Perl, \Q...\E sequences are recognized inside character
5086 classes as well as outside them.
5087
5088 9. Re-organized 3 code statements in pcretest to avoid "overflow in
5089 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
5090 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
5091 signed/unsigned warnings.
5092
5093 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
5094 option for pcretest, so I've replaced it by a simple function that does just
5095 that job.
5096
5097 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
5098 "pcregrep -".
5099
5100 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
5101 Java package. This provides some syntactic sugar for simple cases of what my
5102 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
5103 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
5104 item, you can use this simplified notation. Note that only makes sense with
5105 greedy quantifiers. Consequently, the use of the possessive quantifier forces
5106 greediness, whatever the setting of the PCRE_UNGREEDY option.
5107
5108 13. A change of greediness default within a pattern was not taking effect at
5109 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
5110 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
5111 was abstracted outside.
5112
5113 14. PCRE now supports the \G assertion. It is true when the current matching
5114 position is at the start point of the match. This differs from \A when the
5115 starting offset is non-zero. Used with the /g option of pcretest (or similar
5116 code), it works in the same way as it does for Perl's /g option. If all
5117 alternatives of a regex begin with \G, the expression is anchored to the start
5118 match position, and the "anchored" flag is set in the compiled expression.
5119
5120 15. Some bugs concerning the handling of certain option changes within patterns
5121 have been fixed. These applied to options other than (?ims). For example,
5122 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
5123 been the other way round. Some of this was related to change 7 above.
5124
5125 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
5126 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
5127 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
5128 POSIX classes only within a class (e.g. /[[:alpha:]]/).
5129
5130 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
5131 mode. Unlike ".", it always matches newline, whatever the setting of
5132 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
5133 assertions. Perl allows it, but it doesn't (in general) work because it can't
5134 calculate the length of the lookbehind. At least, that's the case for Perl
5135 5.8.0 - I've been told they are going to document that it doesn't work in
5136 future.
5137
5138 18. Added an error diagnosis for escapes that PCRE does not support: these are
5139 \L, \l, \N, \P, \p, \U, \u, and \X.
5140
5141 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
5142 reading past the end of the pattern in cases such as /[abcd/.
5143
5144 20. PCRE was getting more memory than necessary for patterns with classes that
5145 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
5146
5147 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
5148 compiling PCRE for use with Virtual Pascal.
5149
5150 22. Small fix to the Makefile to make it work properly if the build is done
5151 outside the source tree.
5152
5153 23. Added a new extension: a condition to go with recursion. If a conditional
5154 subpattern starts with (?(R) the "true" branch is used if recursion has
5155 happened, whereas the "false" branch is used only at the top level.
5156
5157 24. When there was a very long string of literal characters (over 255 bytes
5158 without UTF support, over 250 bytes with UTF support), the computation of how
5159 much memory was required could be incorrect, leading to segfaults or other
5160 strange effects.
5161
5162 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5163 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5164 there was a subsequent back reference to those brackets. This meant that, for
5165 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5166 possible to check for precisely this case. All we can do is abandon the
5167 optimization if .* occurs inside capturing brackets when there are any back
5168 references whatsoever. (See below for a better fix that came later.)
5169
5170 26. The handling of the optimization for finding the first character of a
5171 non-anchored pattern, and for finding a character that is required later in the
5172 match were failing in some cases. This didn't break the matching; it just
5173 failed to optimize when it could. The way this is done has been re-implemented.
5174
5175 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5176
5177 28. Added a new feature that provides some of the functionality that Perl
5178 provides with (?{...}). The facility is termed a "callout". The way it is done
5179 in PCRE is for the caller to provide an optional function, by setting
5180 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5181 global variable. By default it is unset, which disables all calling out. To get
5182 the function called, the regex must include (?C) at appropriate points. This
5183 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5184 This provides a means of identifying different callout points. When PCRE
5185 reaches such a point in the regex, if pcre_callout has been set, the external
5186 function is called. It is provided with data in a structure called
5187 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5188 matching continues; if it returns a non-zero value, the match at the current
5189 point fails. However, backtracking will occur if possible. [This was changed
5190 later and other features added - see item 49 below.]
5191
5192 29. pcretest is upgraded to test the callout functionality. It provides a
5193 callout function that displays information. By default, it shows the start of
5194 the match and the current position in the text. There are some new data escapes
5195 to vary what happens:
5196
5197 \C+ in addition, show current contents of captured substrings
5198 \C- do not supply a callout function
5199 \C!n return 1 when callout number n is reached
5200 \C!n!m return 1 when callout number n is reached for the mth time
5201
5202 30. If pcregrep was called with the -l option and just a single file name, it
5203 output "<stdin>" if a match was found, instead of the file name.
5204
5205 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5206 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5207 pcre_exec(). This saves a malloc/free per call. The default value of
5208 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5209 when configuring.
5210
5211 32. The default maximum size of a compiled pattern is 64K. There have been a
5212 few cases of people hitting this limit. The code now uses macros to handle the
5213 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5214 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5215 configuring. Tests 2 and 5 work only with 2-byte links because they output
5216 debugging information about compiled patterns.
5217
5218 33. Internal code re-arrangements:
5219
5220 (a) Moved the debugging function for printing out a compiled regex into
5221 its own source file (printint.c) and used #include to pull it into
5222 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5223 separate copies.
5224
5225 (b) Defined the list of op-code names for debugging as a macro in
5226 internal.h so that it is next to the definition of the opcodes.
5227
5228 (c) Defined a table of op-code lengths for simpler skipping along compiled
5229 code. This is again a macro in internal.h so that it is next to the
5230 definition of the opcodes.
5231
5232 34. Added support for recursive calls to individual subpatterns, along the
5233 lines of Robin Houston's patch (but implemented somewhat differently).
5234
5235 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5236 allow it to read and process whole directories in Win32. This code was
5237 contributed by Lionel Fourquaux; it has not been tested by me.
5238
5239 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5240 used to name a group. Names consist of alphanumerics and underscores, and must
5241 be unique. Back references use the syntax (?P=name) and recursive calls use
5242 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5243 numbers. The function pcre_fullinfo() can be used after compilation to extract
5244 a name/number map. There are three relevant calls:
5245
5246 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5247 PCRE_INFO_NAMECOUNT yields the number of entries
5248 PCRE_INFO_NAMETABLE yields a pointer to the map.
5249
5250 The map is a vector of fixed-size entries. The size of each entry depends on
5251 the length of the longest name used. The first two bytes of each entry are the
5252 group number, most significant byte first. There follows the corresponding
5253 name, zero terminated. The names are in alphabetical order.
5254
5255 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5256 case instead of 255. Making it the same both with and without UTF-8 support
5257 means that the same test output works with both.
5258
5259 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5260 calling malloc() with a zero argument.
5261
5262 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5263 optimization. I've improved things by keeping a bitmap of backreferences with
5264 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5265 fact referenced, the optimization can be applied. It is unlikely that a
5266 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5267 the match to follow \n) will appear inside brackets with a number greater than
5268 31, but if it does, any back reference > 31 suppresses the optimization.
5269
5270 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5271 of disabling numbered capturing parentheses. Any opening parenthesis that is
5272 not followed by ? behaves as if it were followed by ?: but named parentheses
5273 can still be used for capturing (and they will acquire numbers in the usual
5274 way).
5275
5276 41. Redesigned the return codes from the match() function into yes/no/error so
5277 that errors can be passed back from deep inside the nested calls. A malloc
5278 failure while inside a recursive subpattern call now causes the
5279 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5280
5281 42. It is now possible to set a limit on the number of times the match()
5282 function is called in a call to pcre_exec(). This facility makes it possible to
5283 limit the amount of recursion and backtracking, though not in a directly
5284 obvious way, because the match() function is used in a number of different
5285 circumstances. The count starts from zero for each position in the subject
5286 string (for non-anchored patterns). The default limit is, for compatibility, a
5287 large number, namely 10 000 000. You can change this in two ways:
5288
5289 (a) When configuring PCRE before making, you can use --with-match-limit=n
5290 to set a default value for the compiled library.
5291
5292 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5293 a different value is set. See 45 below.
5294
5295 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5296
5297 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5298 of things that can be changed at compile time. The first argument specifies
5299 what is wanted and the second points to where the information is to be placed.
5300 The current list of available information is:
5301
5302 PCRE_CONFIG_UTF8
5303
5304 The output is an integer that is set to one if UTF-8 support is available;
5305 otherwise it is set to zero.
5306
5307 PCRE_CONFIG_NEWLINE
5308
5309 The output is an integer that it set to the value of the code that is used for
5310 newline. It is either LF (10) or CR (13).
5311
5312 PCRE_CONFIG_LINK_SIZE
5313
5314 The output is an integer that contains the number of bytes used for internal
5315 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5316
5317 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5318
5319 The output is an integer that contains the threshold above which the POSIX
5320 interface uses malloc() for output vectors. See item 31 above.
5321
5322 PCRE_CONFIG_MATCH_LIMIT
5323
5324 The output is an unsigned integer that contains the default limit of the number
5325 of match() calls in a pcre_exec() execution. See 42 above.
5326
5327 44. pcretest has been upgraded by the addition of the -C option. This causes it
5328 to extract all the available output from the new pcre_config() function, and to
5329 output it. The program then exits immediately.
5330
5331 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5332 order to support additional features. One way would have been to define
5333 pcre_exec2() (for example) with extra arguments, but this would not have been
5334 extensible, and would also have required all calls to the original function to
5335 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5336 is used for passing in "extra" data from pcre_study().
5337
5338 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5339 contains the following fields:
5340
5341 flags a bitmap indicating which of the following fields are set
5342 study_data opaque data from pcre_study()
5343 match_limit a way of specifying a limit on match() calls for a specific
5344 call to pcre_exec()
5345 callout_data data for callouts (see 49 below)
5346
5347 The flag bits are also defined in pcre.h, and are
5348
5349 PCRE_EXTRA_STUDY_DATA
5350 PCRE_EXTRA_MATCH_LIMIT
5351 PCRE_EXTRA_CALLOUT_DATA
5352
5353 The pcre_study() function now returns one of these new pcre_extra blocks, with
5354 the actual study data pointed to by the study_data field, and the
5355 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5356 before. That is, this change is entirely upwards-compatible and requires no
5357 change to existing code.
5358
5359 If you want to pass in additional data to pcre_exec(), you can either place it
5360 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5361 block.
5362
5363 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5364 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5365 times with different match limits, until it finds the minimum value needed for
5366 pcre_exec() to complete. The value is then output. This can be instructive; for
5367 most simple matches the number is quite small, but for pathological cases it
5368 gets very large very quickly.
5369
5370 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5371 returns the size of the data block pointed to by the study_data field in a
5372 pcre_extra block, that is, the value that was passed as the argument to
5373 pcre_malloc() when PCRE was getting memory in which to place the information
5374 created by pcre_study(). The fourth argument should point to a size_t variable.
5375 pcretest has been extended so that this information is shown after a successful
5376 pcre_study() call when information about the compiled regex is being displayed.
5377
5378 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5379 because what follows is always an absolute path. (Later: it turns out that this
5380 is more than cosmetic for MinGW, because it doesn't like empty path
5381 components.)
5382
5383 49. Some changes have been made to the callout feature (see 28 above):
5384
5385 (i) A callout function now has three choices for what it returns:
5386
5387 0 => success, carry on matching
5388 > 0 => failure at this point, but backtrack if possible
5389 < 0 => serious error, return this value from pcre_exec()
5390
5391 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5392 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5393 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5394 use by callout functions. It will never be used by PCRE itself.
5395
5396 (ii) The pcre_extra structure (see 45 above) has a void * field called
5397 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5398 pcre_callout_block structure has a field of the same name. The contents of
5399 the field passed in the pcre_extra structure are passed to the callout
5400 function in the corresponding field in the callout block. This makes it
5401 easier to use the same callout-containing regex from multiple threads. For
5402 testing, the pcretest program has a new data escape
5403
5404 \C*n pass the number n (may be negative) as callout_data
5405
5406 If the callout function in pcretest receives a non-zero value as
5407 callout_data, it returns that value.
5408
5409 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5410 there were some redundant $(CFLAGS) in commands that are now specified as
5411 $(LINK), which already includes $(CFLAGS).
5412
5413 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5414 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5415 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5416 one-byte characters throughout. Note that case-insensitive matching applies
5417 only to characters whose values are less than 256. PCRE doesn't support the
5418 notion of cases for higher-valued characters.
5419
5420 (i) A character class whose characters are all within 0-255 is handled as
5421 a bit map, and the map is inverted for negative classes. Previously, a
5422 character > 255 always failed to match such a class; however it should
5423 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5424
5425 (ii) A negated character class with a single character < 255 is coded as
5426 "not this character" (OP_NOT). This wasn't working properly when the test
5427 character was multibyte, either singly or repeated.
5428
5429 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5430 mode, for example: \x{100}{2,3}.
5431
5432 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5433 singly or repeated) now correctly test multibyte characters. However,
5434 PCRE doesn't recognize any characters with values greater than 255 as
5435 digits, spaces, or word characters. Such characters always match \D, \S,
5436 and \W, and never match \d, \s, or \w.
5437
5438 (v) Classes may now contain characters and character ranges with values
5439 greater than 255. For example: [ab\x{100}-\x{400}].
5440
5441 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5442 PCRE in UTF-8 mode.
5443
5444 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5445 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5446 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5447 value.)
5448
5449 53. The single man page has become too large. I have therefore split it up into
5450 a number of separate man pages. These also give rise to individual HTML pages;
5451 these are now put in a separate directory, and there is an index.html page that
5452 lists them all. Some hyperlinking between the pages has been installed.
5453
5454 54. Added convenience functions for handling named capturing parentheses.
5455
5456 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5457 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5458 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5459 are faulted.
5460
5461 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5462 calling configure. These values are used when compiling the dftables.c program
5463 which is run to generate the source of the default character tables. They
5464 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5465 you will need to set these values.
5466
5467 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5468
5469
5470 Version 3.9 02-Jan-02
5471 ---------------------
5472
5473 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5474
5475 2. If --disable-static was given, the building process failed when trying to
5476 build pcretest and pcregrep. (For some reason it was using libtool to compile
5477 them, which is not right, as they aren't part of the library.)
5478
5479
5480 Version 3.8 18-Dec-01
5481 ---------------------
5482
5483 1. The experimental UTF-8 code was completely screwed up. It was packing the
5484 bytes in the wrong order. How dumb can you get?
5485
5486
5487 Version 3.7 29-Oct-01
5488 ---------------------
5489
5490 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5491 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5492 this didn't happen under Solaris 8, where I normally test things.
5493
5494 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5495 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5496 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5497 this fix an hour or so after the initial 3.7 release.)
5498
5499
5500 Version 3.6 23-Oct-01
5501 ---------------------
5502
5503 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5504 offsets passed as NULL with zero offset count.
5505
5506 2. The config.guess and config.sub files had not been updated when I moved to
5507 the latest autoconf.
5508
5509
5510 Version 3.5 15-Aug-01
5511 ---------------------
5512
5513 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5514 had been forgotten.
5515
5516 2. By using declared but undefined structures, we can avoid using "void"
5517 definitions in pcre.h while keeping the internal definitions of the structures
5518 private.
5519
5520 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5521 user point of view, this means that both static and shared libraries are built
5522 by default, but this can be individually controlled. More of the work of
5523 handling this static/shared cases is now inside libtool instead of PCRE's make
5524 file.
5525
5526 4. The pcretest utility is now installed along with pcregrep because it is
5527 useful for users (to test regexs) and by doing this, it automatically gets
5528 relinked by libtool. The documentation has been turned into a man page, so
5529 there are now .1, .txt, and .html versions in /doc.
5530
5531 5. Upgrades to pcregrep:
5532 (i) Added long-form option names like gnu grep.
5533 (ii) Added --help to list all options with an explanatory phrase.
5534 (iii) Added -r, --recursive to recurse into sub-directories.
5535 (iv) Added -f, --file to read patterns from a file.
5536
5537 6. pcre_exec() was referring to its "code" argument before testing that
5538 argument for NULL (and giving an error if it was NULL).
5539
5540 7. Upgraded Makefile.in to allow for compiling in a different directory from
5541 the source directory.
5542
5543 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5544 options bits, the pointer it was passed was to an int instead of to an unsigned
5545 long int. This mattered only on 64-bit systems.
5546
5547 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5548 generated) instead of pcre.in, which it its source. Also made the same change
5549 in several of the .c files.
5550
5551 10. A new release of gcc defines printf() as a macro, which broke pcretest
5552 because it had an ifdef in the middle of a string argument for printf(). Fixed
5553 by using separate calls to printf().
5554
5555 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5556 script, to force use of CR or LF instead of \n in the source. On non-Unix
5557 systems, the value can be set in config.h.
5558
5559 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5560 absolute limit. Changed the text of the error message to make this clear, and
5561 likewise updated the man page.
5562
5563 13. The limit of 99 on the number of capturing subpatterns has been removed.
5564 The new limit is 65535, which I hope will not be a "real" limit.
5565
5566
5567 Version 3.4 22-Aug-00
5568 ---------------------
5569
5570 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5571
5572 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5573
5574
5575 Version 3.3 01-Aug-00
5576 ---------------------
5577
5578 1. If an octal character was given, but the value was greater than \377, it
5579 was not getting masked to the least significant bits, as documented. This could
5580 lead to crashes in some systems.
5581
5582 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5583 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5584
5585 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5586 These just pass their arguments on to (pcre_free)(), but they are provided
5587 because some uses of PCRE bind it to non-C systems that can call its functions,
5588 but cannot call free() or pcre_free() directly.
5589
5590 4. Add "make test" as a synonym for "make check". Corrected some comments in
5591 the Makefile.
5592
5593 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5594 Makefile.
5595
5596 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5597 command called pgrep for grepping around the active processes.
5598
5599 7. Added the beginnings of support for UTF-8 character strings.
5600
5601 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5602 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5603 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5604 out for the ar command.)
5605
5606
5607 Version 3.2 12-May-00
5608 ---------------------
5609
5610 This is purely a bug fixing release.
5611
5612 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5613 of ZA. This was just one example of several cases that could provoke this bug,
5614 which was introduced by change 9 of version 2.00. The code for breaking
5615 infinite loops after an iteration that matches an empty string was't working
5616 correctly.
5617
5618 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5619 when matched against abbab (for example). After matching an empty string, it
5620 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5621 caused it to match further down the string than it should.
5622
5623 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5624 was there because it doesn't seem to be needed, and it causes trouble on some
5625 systems, as it is not a Standard C header. It has been removed.
5626
5627 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5628 were reported on the Macintosh. The changes were from
5629
5630 while ((c = *(++ptr)) != 0 && c != '\n');
5631 to
5632 while ((c = *(++ptr)) != 0 && c != '\n') ;
5633
5634 Totally extraordinary, but if that's what it takes...
5635
5636 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5637 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5638 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5639 assumes the way PCRE uses memmove() (always moving upwards).
5640
5641 6. PCRE is being used in one environment where strchr() is not available. There
5642 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5643 faster code anyway.
5644
5645
5646 Version 3.1 09-Feb-00
5647 ---------------------
5648
5649 The only change in this release is the fixing of some bugs in Makefile.in for
5650 the "install" target:
5651
5652 (1) It was failing to install pcreposix.h.
5653
5654 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5655
5656
5657 Version 3.0 01-Feb-00
5658 ---------------------
5659
5660 1. Add support for the /+ modifier to perltest (to output $` like it does in
5661 pcretest).
5662
5663 2. Add support for the /g modifier to perltest.
5664
5665 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5666 matches null strings.
5667
5668 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5669 pattern. Perl treats empty patterns specially - it reuses the most recent
5670 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5671 effect.
5672
5673 5. The POSIX interface was broken in that it was just handing over the POSIX
5674 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5675 required a bigger vector, with some working space on the end. This means that
5676 the POSIX wrapper now has to get and free some memory, and copy the results.
5677
5678 6. Added some simple autoconf support, placing the test data and the
5679 documentation in separate directories, re-organizing some of the
5680 information files, and making it build pcre-config (a GNU standard). Also added
5681 libtool support for building PCRE as a shared library, which is now the
5682 default.
5683
5684 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5685 09 are not valid octal constants. Single digits will be used for minor values
5686 less than 10.
5687
5688 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5689 existing programs that set these in the POSIX interface can use PCRE without
5690 modification.
5691
5692 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5693 return all that pcre_info() returns, plus additional data. The pcre_info()
5694 function is retained for compatibility, but is considered to be obsolete.
5695
5696 10. Added experimental recursion feature (?R) to handle one common case that
5697 Perl 5.6 will be able to do with (?p{...}).
5698
5699 11. Added support for POSIX character classes like [:alpha:], which Perl is
5700 adopting.
5701
5702
5703 Version 2.08 31-Aug-99
5704 ----------------------
5705
5706 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5707 trying to match at the startoffset position, but instead was moving forward to
5708 the next newline as if a previous match had failed.
5709
5710 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5711 and could get into a loop if a null string was matched other than at the start
5712 of the subject.
5713
5714 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5715 be distinguished at compile time, and for completeness also added PCRE_DATE.
5716
5717 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5718 in GnuWin32 environments.
5719
5720
5721 Version 2.07 29-Jul-99
5722 ----------------------
5723
5724 1. The documentation is now supplied in plain text form and HTML as well as in
5725 the form of man page sources.
5726
5727 2. C++ compilers don't like assigning (void *) values to other pointer types.
5728 In particular this affects malloc(). Although there is no problem in Standard
5729 C, I've put in casts to keep C++ compilers happy.
5730
5731 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5732 should be (const char *).
5733
5734 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5735 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5736 However, I haven't made this a standard facility. The documentation doesn't
5737 mention it, and the Makefile doesn't support it.
5738
5739 5. The Makefile now contains an "install" target, with editable destinations at
5740 the top of the file. The pcretest program is not installed.
5741
5742 6. pgrep -V now gives the PCRE version number and date.
5743
5744 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5745 causing the entire string to be ignored, instead of just the last character.
5746
5747 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5748 non-matching string, it can take a very, very long time, even for strings of
5749 quite modest length, because of the nested recursion. PCRE now does better in
5750 some of these cases. It does this by remembering the last required literal
5751 character in the pattern, and pre-searching the subject to ensure it is present
5752 before running the real match. In other words, it applies a heuristic to detect
5753 some types of certain failure quickly, and in the above example, if presented
5754 with a string that has no trailing " it gives "no match" very quickly.
5755
5756 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5757 other alternatives are tried instead.
5758
5759
5760 Version 2.06 09-Jun-99
5761 ----------------------
5762
5763 1. Change pcretest's output for amount of store used to show just the code
5764 space, because the remainder (the data block) varies in size between 32-bit and
5765 64-bit systems.
5766
5767 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5768 start matching at. This allows lookbehinds to work when searching for multiple
5769 occurrences in a string.
5770
5771 3. Added additional options to pcretest for testing multiple occurrences:
5772
5773 /+ outputs the rest of the string that follows a match
5774 /g loops for multiple occurrences, using the new startoffset argument
5775 /G loops for multiple occurrences by passing an incremented pointer
5776
5777 4. PCRE wasn't doing the "first character" optimization for patterns starting
5778 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5779 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5780 the letter 'x'. On long subject strings, this gives a significant speed-up.
5781
5782
5783 Version 2.05 21-Apr-99
5784 ----------------------
5785
5786 1. Changed the type of magic_number from int to long int so that it works
5787 properly on 16-bit systems.
5788
5789 2. Fixed a bug which caused patterns starting with .* not to work correctly
5790 when the subject string contained newline characters. PCRE was assuming
5791 anchoring for such patterns in all cases, which is not correct because .* will
5792 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5793 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5794 must be retried after every newline in the subject.
5795
5796
5797 Version 2.04 18-Feb-99
5798 ----------------------
5799
5800 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5801 computation of the store needed to hold the pattern was incorrect (too large).
5802 If such patterns were nested a few deep, this could multiply and become a real
5803 problem.
5804
5805 2. Added /M option to pcretest to show the memory requirement of a specific
5806 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5807
5808 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5809 compiled in such a way that the backtracking after subsequent failure was
5810 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5811 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5812
5813
5814 Version 2.03 02-Feb-99
5815 ----------------------
5816
5817 1. Fixed typo and small mistake in man page.
5818
5819 2. Added 4th condition (GPL supersedes if conflict) and created separate
5820 LICENCE file containing the conditions.
5821
5822 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5823 Perl, that is the internal \ allows the delimiter to be included in the
5824 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5825 the final delimiter, add \ to the end of the pattern (to test the error).
5826
5827 4. Added the convenience functions for extracting substrings after a successful
5828 match. Updated pcretest to make it able to test these functions.
5829
5830
5831 Version 2.02 14-Jan-99
5832 ----------------------
5833
5834 1. Initialized the working variables associated with each extraction so that
5835 their saving and restoring doesn't refer to uninitialized store.
5836
5837 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5838 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5839 fix the problem.
5840
5841 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5842 calls, and wasn't printing the correct value for compiling calls. Increased the
5843 default value of LOOPREPEAT, and the number of significant figures in the
5844 times.
5845
5846 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5847
5848 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5849 a building problem on Windows NT with a FAT file system.
5850
5851
5852 Version 2.01 21-Oct-98
5853 ----------------------
5854
5855 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5856 to character tables built by pcre_maketables() in the current locale. If NULL
5857 is passed, the default tables are used.
5858
5859
5860 Version 2.00 24-Sep-98
5861 ----------------------
5862
5863 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5864 it any more.
5865
5866 2. Allow quantification of (?>) groups, and make it work correctly.
5867
5868 3. The first character computation wasn't working for (?>) groups.
5869
5870 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5871 end of the subject) and add 5.005's \z, which really does match only at the
5872 very end of the subject.
5873
5874 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5875
5876 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5877 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5878 localized options. All options to pcre_study() were also removed.
5879
5880 7. Add other new features from 5.005:
5881
5882 $(?<= positive lookbehind
5883 $(?<! negative lookbehind
5884 (?imsx-imsx) added the unsetting capability
5885 such a setting is global if at outer level; local otherwise
5886 (?imsx-imsx:) non-capturing groups with option setting
5887 (?(cond)re|re) conditional pattern matching
5888
5889 A backreference to itself in a repeated group matches the previous
5890 captured string.
5891
5892 8. General tidying up of studying (both automatic and via "study")
5893 consequential on the addition of new assertions.
5894
5895 9. As in 5.005, unlimited repeated groups that could match an empty substring
5896 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5897 runtime if any iteration does actually match an empty substring.
5898
5899 10. Include the RunTest script in the distribution.
5900
5901 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5902 discrepancies, some of which were old and were also with respect to 5.004. They
5903 have now been fixed.
5904
5905
5906 Version 1.09 28-Apr-98
5907 ----------------------
5908
5909 1. A negated single character class followed by a quantifier with a minimum
5910 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5911 program crashes, or just wrong answers. This did not apply to negated classes
5912 containing more than one character, or to minima other than one.
5913
5914
5915 Version 1.08 27-Mar-98
5916 ----------------------
5917
5918 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5919
5920 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5921 latter must appear before anything that relies on it in the pattern.
5922
5923
5924 Version 1.07 16-Feb-98
5925 ----------------------
5926
5927 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5928 repeat of a potentially empty string).
5929
5930
5931 Version 1.06 23-Jan-98
5932 ----------------------
5933
5934 1. Added Markus Oberhumer's little patches for C++.
5935
5936 2. Literal strings longer than 255 characters were broken.
5937
5938
5939 Version 1.05 23-Dec-97
5940 ----------------------
5941
5942 1. Negated character classes containing more than one character were failing if
5943 PCRE_CASELESS was set at run time.
5944
5945
5946 Version 1.04 19-Dec-97
5947 ----------------------
5948
5949 1. Corrected the man page, where some "const" qualifiers had been omitted.
5950
5951 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5952 input syntax.
5953
5954 3. Fixed memory leak which occurred when a regex with back references was
5955 matched with an offsets vector that wasn't big enough. The temporary memory
5956 that is used in this case wasn't being freed if the match failed.
5957
5958 4. Tidied pcretest to ensure it frees memory that it gets.
5959
5960 5. Temporary memory was being obtained in the case where the passed offsets
5961 vector was exactly big enough.
5962
5963 6. Corrected definition of offsetof() from change 5 below.
5964
5965 7. I had screwed up change 6 below and broken the rules for the use of
5966 setjmp(). Now fixed.
5967
5968
5969 Version 1.03 18-Dec-97
5970 ----------------------
5971
5972 1. A erroneous regex with a missing opening parenthesis was correctly
5973 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5974 on some systems.
5975
5976 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5977 it was reported that one broken compiler failed on the former because "code" is
5978 also an independent variable.
5979
5980 3. The erroneous regex a[]b caused an array overrun reference.
5981
5982 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5983 fail on data ending with that character. (It was going on too far, and checking
5984 the next character, typically a binary zero.) This was specific to the
5985 optimized code for single-character negative classes.
5986
5987 5. Added a contributed patch from the TIN world which does the following:
5988
5989 + Add an undef for memmove, in case the the system defines a macro for it.
5990
5991 + Add a definition of offsetof(), in case there isn't one. (I don't know
5992 the reason behind this - offsetof() is part of the ANSI standard - but
5993 it does no harm).
5994
5995 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5996 most of the places where whitespace preceded '#'. I have given up and
5997 allowed the remaining 2 cases to be at the margin.
5998
5999 + Rename some variables in pcre to eliminate shadowing. This seems very
6000 pedantic, but does no harm, of course.
6001
6002 6. Moved the call to setjmp() into its own function, to get rid of warnings
6003 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
6004
6005 7. Constructs such as \d{8,} were compiling into the equivalent of
6006 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
6007 outcome, but in this particular case used more store than had been allocated,
6008 which caused the bug to be discovered because it threw up an internal error.
6009
6010 8. The debugging code in both pcre and pcretest for outputting the compiled
6011 form of a regex was going wrong in the case of back references followed by
6012 curly-bracketed repeats.
6013
6014
6015 Version 1.02 12-Dec-97
6016 ----------------------
6017
6018 1. Typos in pcre.3 and comments in the source fixed.
6019
6020 2. Applied a contributed patch to get rid of places where it used to remove
6021 'const' from variables, and fixed some signed/unsigned and uninitialized
6022 variable warnings.
6023
6024 3. Added the "runtest" target to Makefile.
6025
6026 4. Set default compiler flag to -O2 rather than just -O.
6027
6028
6029 Version 1.01 19-Nov-97
6030 ----------------------
6031
6032 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
6033 like /([ab]*)*/, that is, for classes with more than one character in them.
6034
6035 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
6036 as /((?>a*))*/ (a PCRE_EXTRA facility).
6037
6038
6039 Version 1.00 18-Nov-97
6040 ----------------------
6041
6042 1. Added compile-time macros to support systems such as SunOS4 which don't have
6043 memmove() or strerror() but have other things that can be used instead.
6044
6045 2. Arranged that "make clean" removes the executables.
6046
6047
6048 Version 0.99 27-Oct-97
6049 ----------------------
6050
6051 1. Fixed bug in code for optimizing classes with only one character. It was
6052 initializing a 32-byte map regardless, which could cause it to run off the end
6053 of the memory it had got.
6054
6055 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
6056