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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1699 - (show annotations)
Sun Apr 30 13:42:34 2017 UTC (2 years, 7 months ago) by ph10
File size: 293484 byte(s)
Remove duplication typo.
1 ChangeLog for PCRE
2 ------------------
3
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
6
7 Version 8.41
8 ------------
9
10 1. Fixed typo in CMakeLists.txt (wrong number of arguments for
11 PCRE_STATIC_RUNTIME (affects MSVC only).
12
13 2. Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
14 mode with --only-matching matched several lines, it restarted scanning at the
15 next line instead of moving on to the end of the matched string, which can be
16 several lines after the start.
17
18 3. Fix a missing else in the JIT compiler reported by 'idaifish'.
19
20 4. A (?# style comment is now ignored between a basic quantifier and a
21 following '+' or '?' (example: /X+(?#comment)?Y/.
22
23 5. Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
24 Pisar).
25
26 6. Fuzzers have reported issues in pcretest. These are NOT serious (it is,
27 after all, just a test program). However, to stop the reports, some easy ones
28 are fixed:
29
30 (a) Check for values < 256 when calling isprint() in pcretest.
31 (b) Give an error for too big a number after \O.
32
33 7. In the 32-bit library in non-UTF mode, an attempt to find a Unicode
34 property for a character with a code point greater than 0x10ffff (the Unicode
35 maximum) caused a crash.
36
37 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
38 encountered a character class with a possessive repeat, for example [a-f]{3}+.
39
40 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
41 length incorrectly, which could result in buffer overflow.
42
43 10. Remove redundant line of code (accidentally left in ages ago).
44
45 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
46 with namespace pcrecpp (Bugzilla #2084).
47
48 12. Remove a duplication typo in pcre_tables.c
49
50
51 Version 8.40 11-January-2017
52 ----------------------------
53
54 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
55 the match extended over a line boundary.
56
57 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
58 MSVC static compilation, putting the first patch under a new option.
59
60 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
61
62 4. Ignore "show all captures" (/=) for DFA matching.
63
64 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
65
66 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
67 without PCRE_UCP set, a negative character type such as \D in a positive
68 class should cause all characters greater than 255 to match, whatever else
69 is in the class. There was a bug that caused this not to happen if a
70 Unicode property item was added to such a class, for example [\D\P{Nd}] or
71 [\W\pL].
72
73 7. When pcretest was outputing information from a callout, the caret indicator
74 for the current position in the subject line was incorrect if it was after
75 an escape sequence for a character whose code point was greater than
76 \x{ff}.
77
78 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
79 the conditional was interpreted as a reference to capturing group 1 instead
80 of a test for recursion. Any group whose name began with R was
81 misinterpreted in this way. (The reference interpretation should only
82 happen if the group's name is precisely "R".)
83
84 9. A number of bugs have been mended relating to match start-up optimizations
85 when the first thing in a pattern is a positive lookahead. These all
86 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
87
88 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
89 both an initial 'X' and a following 'X'.
90 (b) Some patterns starting with an assertion that started with .* were
91 incorrectly optimized as having to match at the start of the subject or
92 after a newline. There are cases where this is not true, for example,
93 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
94 start with spaces. Starting .* in an assertion is no longer taken as an
95 indication of matching at the start (or after a newline).
96
97
98 Version 8.39 14-June-2016
99 -------------------------
100
101 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
102 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
103 misbehaved. This bug was found by the LLVM fuzzer.
104
105 2. Similar to the above, if an isolated \E was present between an item and its
106 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
107 bug was found by the LLVM fuzzer.
108
109 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
110 working correctly in UCP mode.
111
112 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
113 was set when the pmatch argument was NULL. It now returns REG_INVARG.
114
115 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
116
117 6. An empty \Q\E sequence between an item and its qualifier caused
118 pcre_compile() to misbehave when auto callouts were enabled. This bug was
119 found by the LLVM fuzzer.
120
121 7. If a pattern that was compiled with PCRE_EXTENDED started with white
122 space or a #-type comment that was followed by (?-x), which turns off
123 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
124 pcre_compile() assumed that (?-x) applied to the whole pattern and
125 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
126
127 8. A call of pcre_copy_named_substring() for a named substring whose number
128 was greater than the space in the ovector could cause a crash.
129
130 9. Yet another buffer overflow bug involved duplicate named groups with a
131 group that reset capture numbers (compare 8.38/7 below). Once again, I have
132 just allowed for more memory, even if not needed. (A proper fix is
133 implemented in PCRE2, but it involves a lot of refactoring.)
134
135 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
136 start of the match to be earlier than the end.
137
138 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
139
140 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
141 assertion, caused pcretest to generate incorrect output, and also to read
142 uninitialized memory (detected by ASAN or valgrind).
143
144 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
145 nested set of parentheses of sufficient size caused an overflow of the
146 compiling workspace (which was diagnosed, but of course is not desirable).
147
148 14. And yet another buffer overflow bug involving duplicate named groups, this
149 time nested, with a nested back reference. Yet again, I have just allowed
150 for more memory, because anything more needs all the refactoring that has
151 been done for PCRE2. An example pattern that provoked this bug is:
152 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
153 registered as CVE-2016-1283.
154
155 15. pcretest went into a loop if global matching was requested with an ovector
156 size less than 2. It now gives an error message. This bug was found by
157 afl-fuzz.
158
159 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
160 ("assertion expected") when (?(?C) was not followed by an opening
161 parenthesis.
162
163 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
164 actually affect anything, by sheer luck.
165
166 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
167 static compilation.
168
169 19. Modified the RunTest script to incorporate a valgrind suppressions file so
170 that certain errors, provoked by the SSE2 instruction set when JIT is used,
171 are ignored.
172
173 20. A racing condition is fixed in JIT reported by Mozilla.
174
175 21. Minor code refactor to avoid "array subscript is below array bounds"
176 compiler warning.
177
178 22. Minor code refactor to avoid "left shift of negative number" warning.
179
180 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
181 UCP support.
182
183 24. Refactor to avoid compiler warnings in pcrecpp.cc.
184
185 25. Refactor to fix a typo in pcre_jit_test.c
186
187 26. Patch to support compiling pcrecpp.cc with Intel compiler.
188
189
190 Version 8.38 23-November-2015
191 -----------------------------
192
193 1. If a group that contained a recursive back reference also contained a
194 forward reference subroutine call followed by a non-forward-reference
195 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
196 compile correct code, leading to undefined behaviour or an internally
197 detected error. This bug was discovered by the LLVM fuzzer.
198
199 2. Quantification of certain items (e.g. atomic back references) could cause
200 incorrect code to be compiled when recursive forward references were
201 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
202 This bug was discovered by the LLVM fuzzer.
203
204 3. A repeated conditional group whose condition was a reference by name caused
205 a buffer overflow if there was more than one group with the given name.
206 This bug was discovered by the LLVM fuzzer.
207
208 4. A recursive back reference by name within a group that had the same name as
209 another group caused a buffer overflow. For example:
210 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
211
212 5. A forward reference by name to a group whose number is the same as the
213 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
214 a buffer overflow at compile time. This bug was discovered by the LLVM
215 fuzzer.
216
217 6. A lookbehind assertion within a set of mutually recursive subpatterns could
218 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
219
220 7. Another buffer overflow bug involved duplicate named groups with a
221 reference between their definition, with a group that reset capture
222 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
223 fixed by always allowing for more memory, even if not needed. (A proper fix
224 is implemented in PCRE2, but it involves more refactoring.)
225
226 8. There was no check for integer overflow in subroutine calls such as (?123).
227
228 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
229 being treated as a literal 'l' instead of causing an error.
230
231 10. There was a buffer overflow if pcre_exec() was called with an ovector of
232 size 1. This bug was found by american fuzzy lop.
233
234 11. If a non-capturing group containing a conditional group that could match
235 an empty string was repeated, it was not identified as matching an empty
236 string itself. For example: /^(?:(?(1)x|)+)+$()/.
237
238 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
239 \a and \e in test subject lines.
240
241 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
242 instead of the EBCDIC value.
243
244 14. The handling of \c in an EBCDIC environment has been revised so that it is
245 now compatible with the specification in Perl's perlebcdic page.
246
247 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
248 ASCII/Unicode. This has now been added to the list of characters that are
249 recognized as white space in EBCDIC.
250
251 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
252 error (correctly) when used outside a class, but did not give an error
253 within a class.
254
255 17. \h within a class was incorrectly compiled in EBCDIC environments.
256
257 18. A pattern with an unmatched closing parenthesis that contained a backward
258 assertion which itself contained a forward reference caused buffer
259 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
260
261 19. JIT should return with error when the compiled pattern requires more stack
262 space than the maximum.
263
264 20. A possessively repeated conditional group that could match an empty string,
265 for example, /(?(R))*+/, was incorrectly compiled.
266
267 21. Fix infinite recursion in the JIT compiler when certain patterns such as
268 /(?:|a|){100}x/ are analysed.
269
270 22. Some patterns with character classes involving [: and \\ were incorrectly
271 compiled and could cause reading from uninitialized memory or an incorrect
272 error diagnosis.
273
274 23. Pathological patterns containing many nested occurrences of [: caused
275 pcre_compile() to run for a very long time.
276
277 24. A conditional group with only one branch has an implicit empty alternative
278 branch and must therefore be treated as potentially matching an empty
279 string.
280
281 25. If (?R was followed by - or + incorrect behaviour happened instead of a
282 diagnostic.
283
284 26. Arrange to give up on finding the minimum matching length for overly
285 complex patterns.
286
287 27. Similar to (4) above: in a pattern with duplicated named groups and an
288 occurrence of (?| it is possible for an apparently non-recursive back
289 reference to become recursive if a later named group with the relevant
290 number is encountered. This could lead to a buffer overflow. Wen Guanxing
291 from Venustech ADLAB discovered this bug.
292
293 28. If pcregrep was given the -q option with -c or -l, or when handling a
294 binary file, it incorrectly wrote output to stdout.
295
296 29. The JIT compiler did not restore the control verb head in case of *THEN
297 control verbs. This issue was found by Karl Skomski with a custom LLVM
298 fuzzer.
299
300 30. Error messages for syntax errors following \g and \k were giving inaccurate
301 offsets in the pattern.
302
303 31. Added a check for integer overflow in conditions (?(<digits>) and
304 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
305 fuzzer.
306
307 32. Handling recursive references such as (?2) when the reference is to a group
308 later in the pattern uses code that is very hacked about and error-prone.
309 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
310 give an internal error if it is obvious that compiling has gone wrong.
311
312 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
313 This issue was found by Karl Skomski with a custom LLVM fuzzer.
314
315 34. The JIT compiler should restore the control chain for empty possessive
316 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
317
318 35. Match limit check added to JIT recursion. This issue was found by Karl
319 Skomski with a custom LLVM fuzzer.
320
321 36. Yet another case similar to 27 above has been circumvented by an
322 unconditional allocation of extra memory. This issue is fixed "properly" in
323 PCRE2 by refactoring the way references are handled. Wen Guanxing
324 from Venustech ADLAB discovered this bug.
325
326 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
327 with a custom LLVM fuzzer.
328
329 38. Fixed a corner case of range optimization in JIT.
330
331 39. An incorrect error "overran compiling workspace" was given if there were
332 exactly enough group forward references such that the last one extended
333 into the workspace safety margin. The next one would have expanded the
334 workspace. The test for overflow was not including the safety margin.
335
336 40. A match limit issue is fixed in JIT which was found by Karl Skomski
337 with a custom LLVM fuzzer.
338
339 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
340 work under Windows. (Why has it taken so long for anyone to notice?)
341
342 42. In a character class such as [\W\p{Any}] where both a negative-type escape
343 ("not a word character") and a property escape were present, the property
344 escape was being ignored.
345
346 43. Fix crash caused by very long (*MARK) or (*THEN) names.
347
348 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
349 by a single ASCII character in a class item, was incorrectly compiled in
350 UCP mode. The POSIX class got lost, but only if the single character
351 followed it.
352
353 45. [:punct:] in UCP mode was matching some characters in the range 128-255
354 that should not have been matched.
355
356 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
357 class, all characters with code points greater than 255 are in the class.
358 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
359 such as \w are turned into Unicode properties), wide characters were not
360 correctly handled, and could fail to match.
361
362
363 Version 8.37 28-April-2015
364 --------------------------
365
366 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
367 for those parentheses to be closed with whatever has been captured so far.
368 However, it was failing to mark any other groups between the hightest
369 capture so far and the currrent group as "unset". Thus, the ovector for
370 those groups contained whatever was previously there. An example is the
371 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
372
373 2. If an assertion condition was quantified with a minimum of zero (an odd
374 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
375
376 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
377 unrecognized modifier, a crash could occur.
378
379 4. An attempt to do global matching in pcretest with a zero-length ovector
380 caused a crash.
381
382 5. Fixed a memory leak during matching that could occur for a subpattern
383 subroutine call (recursive or otherwise) if the number of captured groups
384 that had to be saved was greater than ten.
385
386 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
387 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
388 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
389
390 7. A UTF pattern containing a "not" match of a non-ASCII character and a
391 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
392
393 8. When a pattern is compiled, it remembers the highest back reference so that
394 when matching, if the ovector is too small, extra memory can be obtained to
395 use instead. A conditional subpattern whose condition is a check on a
396 capture having happened, such as, for example in the pattern
397 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
398 setting the highest backreference number. This mattered only if pcre_exec()
399 was called with an ovector that was too small to hold the capture, and there
400 was no other kind of back reference (a situation which is probably quite
401 rare). The effect of the bug was that the condition was always treated as
402 FALSE when the capture could not be consulted, leading to a incorrect
403 behaviour by pcre_exec(). This bug has been fixed.
404
405 9. A reference to a duplicated named group (either a back reference or a test
406 for being set in a conditional) that occurred in a part of the pattern where
407 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
408 to be incorrectly calculated, leading to overwriting.
409
410 10. A mutually recursive set of back references such as (\2)(\1) caused a
411 segfault at study time (while trying to find the minimum matching length).
412 The infinite loop is now broken (with the minimum length unset, that is,
413 zero).
414
415 11. If an assertion that was used as a condition was quantified with a minimum
416 of zero, matching went wrong. In particular, if the whole group had
417 unlimited repetition and could match an empty string, a segfault was
418 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
419 assertions to be quantified, but not if they are being used as conditions,
420 so the above pattern is faulted by Perl. PCRE has now been changed so that
421 it also rejects such patterns.
422
423 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
424 failed to allow the zero-repeat case if pcre2_exec() was called with an
425 ovector too small to capture the group.
426
427 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
428 Red Hat Product Security:
429
430 (a) A crash if /K and /F were both set with the option to save the compiled
431 pattern.
432
433 (b) Another crash if the option to print captured substrings in a callout
434 was combined with setting a null ovector, for example \O\C+ as a subject
435 string.
436
437 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
438 forward reference repeated a large (but limited) number of times within a
439 repeated outer group that has a zero minimum quantifier, caused incorrect
440 code to be compiled, leading to the error "internal error:
441 previously-checked referenced subpattern not found" when an incorrect
442 memory address was read. This bug was reported as "heap overflow",
443 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
444 CVE-2015-2325.
445
446 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
447 call within a group that also contained a recursive back reference caused
448 incorrect code to be compiled. This bug was reported as "heap overflow",
449 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
450 number CVE-2015-2326.
451
452 24. Computing the size of the JIT read-only data in advance has been a source
453 of various issues, and new ones are still appear unfortunately. To fix
454 existing and future issues, size computation is eliminated from the code,
455 and replaced by on-demand memory allocation.
456
457 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
458 adjacent to the end of the range, and the range contained characters with
459 more than one other case, caused incorrect behaviour when compiled in UTF
460 mode. In that example, the range a-j was left out of the class.
461
462 26. Fix JIT compilation of conditional blocks, which assertion
463 is converted to (*FAIL). E.g: /(?(?!))/.
464
465 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
466 discovered by the LLVM fuzzer.
467
468 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
469 when this assertion was used as a condition, for example (?(?!)a|b). In
470 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
471 error about an unsupported item.
472
473 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
474 possessification code could take exponential time to complete. A recursion
475 depth limit of 1000 has been imposed to limit the resources used by this
476 optimization.
477
478 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
479 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
480 because \S ensures they are all in the class. The code for doing this was
481 interacting badly with the code for computing the amount of space needed to
482 compile the pattern, leading to a buffer overflow. This bug was discovered
483 by the LLVM fuzzer.
484
485 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
486 other kinds of group caused stack overflow at compile time. This bug was
487 discovered by the LLVM fuzzer.
488
489 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
490 between a subroutine call and its quantifier was incorrectly compiled,
491 leading to buffer overflow or other errors. This bug was discovered by the
492 LLVM fuzzer.
493
494 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
495 assertion after (?(. The code was failing to check the character after
496 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
497 was discovered by the LLVM fuzzer.
498
499 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
500 a fixed maximum following a group that contains a subroutine reference was
501 incorrectly compiled and could trigger buffer overflow. This bug was
502 discovered by the LLVM fuzzer.
503
504 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
505 caused a stack overflow instead of the diagnosis of a non-fixed length
506 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
507
508 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
509 (e.g. /(?<=\Ka)/) could make pcregrep loop.
510
511 37. There was a similar problem to 36 in pcretest for global matches.
512
513 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
514 and a subsequent item in the pattern caused a non-match, backtracking over
515 the repeated \X did not stop, but carried on past the start of the subject,
516 causing reference to random memory and/or a segfault. There were also some
517 other cases where backtracking after \C could crash. This set of bugs was
518 discovered by the LLVM fuzzer.
519
520 39. The function for finding the minimum length of a matching string could take
521 a very long time if mutual recursion was present many times in a pattern,
522 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
523 method has been implemented. This infelicity was discovered by the LLVM
524 fuzzer.
525
526 40. Static linking against the PCRE library using the pkg-config module was
527 failing on missing pthread symbols.
528
529
530 Version 8.36 26-September-2014
531 ------------------------------
532
533 1. Got rid of some compiler warnings in the C++ modules that were shown up by
534 -Wmissing-field-initializers and -Wunused-parameter.
535
536 2. The tests for quantifiers being too big (greater than 65535) were being
537 applied after reading the number, and stupidly assuming that integer
538 overflow would give a negative number. The tests are now applied as the
539 numbers are read.
540
541 3. Tidy code in pcre_exec.c where two branches that used to be different are
542 now the same.
543
544 4. The JIT compiler did not generate match limit checks for certain
545 bracketed expressions with quantifiers. This may lead to exponential
546 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
547 issue should be resolved now.
548
549 5. Fixed an issue, which occures when nested alternatives are optimized
550 with table jumps.
551
552 6. Inserted two casts and changed some ints to size_t in the light of some
553 reported 64-bit compiler warnings (Bugzilla 1477).
554
555 7. Fixed a bug concerned with zero-minimum possessive groups that could match
556 an empty string, which sometimes were behaving incorrectly in the
557 interpreter (though correctly in the JIT matcher). This pcretest input is
558 an example:
559
560 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
561 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
562
563 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
564 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
565 for an empty string was breaking the inner loop and carrying on at a lower
566 level, when possessive repeated groups should always return to a higher
567 level as they have no backtrack points in them. The empty string test now
568 occurs at the outer level.
569
570 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
571 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
572
573 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
574 doesn't).
575
576 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
577 to hold a pattern to be incorrectly computed (too small) when there were
578 named back references to duplicated names. This could cause "internal
579 error: code overflow" or "double free or corruption" or other memory
580 handling errors.
581
582 11. When named subpatterns had the same prefixes, back references could be
583 confused. For example, in this pattern:
584
585 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
586
587 the reference to 'Name' was incorrectly treated as a reference to a
588 duplicate name.
589
590 12. A pattern such as /^s?c/mi8 where the optional character has more than
591 one "other case" was incorrectly compiled such that it would only try to
592 match starting at "c".
593
594 13. When a pattern starting with \s was studied, VT was not included in the
595 list of possible starting characters; this should have been part of the
596 8.34/18 patch.
597
598 14. If a character class started [\Qx]... where x is any character, the class
599 was incorrectly terminated at the ].
600
601 15. If a pattern that started with a caseless match for a character with more
602 than one "other case" was studied, PCRE did not set up the starting code
603 unit bit map for the list of possible characters. Now it does. This is an
604 optimization improvement, not a bug fix.
605
606 16. The Unicode data tables have been updated to Unicode 7.0.0.
607
608 17. Fixed a number of memory leaks in pcregrep.
609
610 18. Avoid a compiler warning (from some compilers) for a function call with
611 a cast that removes "const" from an lvalue by using an intermediate
612 variable (to which the compiler does not object).
613
614 19. Incorrect code was compiled if a group that contained an internal recursive
615 back reference was optional (had quantifier with a minimum of zero). This
616 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
617 caused segmentation faults because of stack overflows at compile time.
618
619 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
620 group that is quantified with an indefinite repeat, caused a compile-time
621 loop which used up all the system stack and provoked a segmentation fault.
622 This was not the same bug as 19 above.
623
624 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
625 Patch by Mike Frysinger.
626
627
628 Version 8.35 04-April-2014
629 --------------------------
630
631 1. A new flag is set, when property checks are present in an XCLASS.
632 When this flag is not set, PCRE can perform certain optimizations
633 such as studying these XCLASS-es.
634
635 2. The auto-possessification of character sets were improved: a normal
636 and an extended character set can be compared now. Furthermore
637 the JIT compiler optimizes more character set checks.
638
639 3. Got rid of some compiler warnings for potentially uninitialized variables
640 that show up only when compiled with -O2.
641
642 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
643 of a match later then the end of the match. The pcretest program was not
644 handling the case sensibly - it was outputting from the start to the next
645 binary zero. It now reports this situation in a message, and outputs the
646 text from the end to the start.
647
648 5. Fast forward search is improved in JIT. Instead of the first three
649 characters, any three characters with fixed position can be searched.
650 Search order: first, last, middle.
651
652 6. Improve character range checks in JIT. Characters are read by an inprecise
653 function now, which returns with an unknown value if the character code is
654 above a certain threshold (e.g: 256). The only limitation is that the value
655 must be bigger than the threshold as well. This function is useful when
656 the characters above the threshold are handled in the same way.
657
658 7. The macros whose names start with RAWUCHAR are placeholders for a future
659 mode in which only the bottom 21 bits of 32-bit data items are used. To
660 make this more memorable for those maintaining the code, the names have
661 been changed to start with UCHAR21, and an extensive comment has been added
662 to their definition.
663
664 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
665 to the export list in Makefile.am (they were accidentally omitted from the
666 8.34 tarball).
667
668 9. The informational output from pcretest used the phrase "starting byte set"
669 which is inappropriate for the 16-bit and 32-bit libraries. As the output
670 for "first char" and "need char" really means "non-UTF-char", I've changed
671 "byte" to "char", and slightly reworded the output. The documentation about
672 these values has also been (I hope) clarified.
673
674 10. Another JIT related optimization: use table jumps for selecting the correct
675 backtracking path, when more than four alternatives are present inside a
676 bracket.
677
678 11. Empty match is not possible, when the minimum length is greater than zero,
679 and there is no \K in the pattern. JIT should avoid empty match checks in
680 such cases.
681
682 12. In a caseless character class with UCP support, when a character with more
683 than one alternative case was not the first character of a range, not all
684 the alternative cases were added to the class. For example, s and \x{17f}
685 are both alternative cases for S: the class [RST] was handled correctly,
686 but [R-T] was not.
687
688 13. The configure.ac file always checked for pthread support when JIT was
689 enabled. This is not used in Windows, so I have put this test inside a
690 check for the presence of windows.h (which was already tested for).
691
692 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
693 The algorithm provides a way to skip certain starting offsets, and usually
694 faster than linear prefix searches.
695
696 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
697 as for 'fr_FR' and 'french'. For some reason, however, it then used the
698 Windows-specific input and output files, which have 'french' screwed in.
699 So this could never have worked. One of the problems with locales is that
700 they aren't always the same. I have now updated RunTest so that it checks
701 the output of the locale test (test 3) against three different output
702 files, and it allows the test to pass if any one of them matches. With luck
703 this should make the test pass on some versions of Solaris where it was
704 failing. Because of the uncertainty, the script did not used to stop if
705 test 3 failed; it now does. If further versions of a French locale ever
706 come to light, they can now easily be added.
707
708 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
709 there was a message during ./configure, but it did not stop. This now
710 provokes an error. The invalid example in README has been corrected.
711 If a value less than the minimum is given, the minimum value has always
712 been used, but now a warning is given.
713
714 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
715 was a bug in the test system, which is now fixed. Also, the list of various
716 configurations that are tested for each release did not have one with both
717 16/32 bits and --enable-bar-anycrlf. It now does.
718
719 18. pcretest was missing "-C bsr" for displaying the \R default setting.
720
721 19. Little endian PowerPC systems are supported now by the JIT compiler.
722
723 20. The fast forward newline mechanism could enter to an infinite loop on
724 certain invalid UTF-8 input. Although we don't support these cases
725 this issue can be fixed by a performance optimization.
726
727 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
728 not take account if existing stack usage. There is now a new global
729 variable called pcre_stack_guard that can be set to point to an external
730 function to check stack availability. It is called at the start of
731 processing every parenthesized group.
732
733 22. A typo in the code meant that in ungreedy mode the max/min qualifier
734 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
735 not match "ab".
736
737 23. When UTF was disabled, the JIT program reported some incorrect compile
738 errors. These messages are silenced now.
739
740 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
741 compiler.
742
743 25. Change all the temporary files used in RunGrepTest to be different to those
744 used by RunTest so that the tests can be run simultaneously, for example by
745 "make -j check".
746
747
748 Version 8.34 15-December-2013
749 -----------------------------
750
751 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
752 executable memory. Patch inspired by Carsten Klein.
753
754 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
755 this macro is never tested and has no effect, because the work to support
756 coverage involves only compiling and linking options and special targets in
757 the Makefile. The comment in config.h implied that defining the macro would
758 enable coverage support, which is totally false. There was also support for
759 setting this macro in the CMake files (my fault, I just copied it from
760 configure). SUPPORT_GCOV has now been removed.
761
762 3. Make a small performance improvement in strlen16() and strlen32() in
763 pcretest.
764
765 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
766 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
767 these cases). There was also one in pcretest.c.
768
769 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
770
771 6. In UTF mode, the code for checking whether a group could match an empty
772 string (which is used for indefinitely repeated groups to allow for
773 breaking an infinite loop) was broken when the group contained a repeated
774 negated single-character class with a character that occupied more than one
775 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
776 UTF-8 mode). The effect was undefined: the group might or might not be
777 deemed as matching an empty string, or the program might have crashed.
778
779 7. The code for checking whether a group could match an empty string was not
780 recognizing that \h, \H, \v, \V, and \R must match a character.
781
782 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
783 an empty string. If it can, pcretest shows this in its information output.
784
785 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
786 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
787 matched by pcre_exec() without using JIT:
788
789 (a) If the rest of the pattern did not match after a maximal run of
790 grapheme clusters, the code for backing up to try with fewer of them
791 did not always back up over a full grapheme when characters that do not
792 have the modifier quality were involved, e.g. Hangul syllables.
793
794 (b) If the match point in a subject started with modifier character, and
795 there was no match, the code could incorrectly back up beyond the match
796 point, and potentially beyond the first character in the subject,
797 leading to a segfault or an incorrect match result.
798
799 10. A conditional group with an assertion condition could lead to PCRE
800 recording an incorrect first data item for a match if no other first data
801 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
802 first data item, and therefore matched "ca" after "c" instead of at the
803 start.
804
805 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
806 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
807
808 12. The source of pcregrep now includes z/OS-specific code so that it can be
809 compiled for z/OS as part of the special z/OS distribution.
810
811 13. Added the -T and -TM options to pcretest.
812
813 14. The code in pcre_compile.c for creating the table of named capturing groups
814 has been refactored. Instead of creating the table dynamically during the
815 actual compiling pass, the information is remembered during the pre-compile
816 pass (on the stack unless there are more than 20 named groups, in which
817 case malloc() is used) and the whole table is created before the actual
818 compile happens. This has simplified the code (it is now nearly 150 lines
819 shorter) and prepared the way for better handling of references to groups
820 with duplicate names.
821
822 15. A back reference to a named subpattern when there is more than one of the
823 same name now checks them in the order in which they appear in the pattern.
824 The first one that is set is used for the reference. Previously only the
825 first one was inspected. This change makes PCRE more compatible with Perl.
826
827 16. Unicode character properties were updated from Unicode 6.3.0.
828
829 17. The compile-time code for auto-possessification has been refactored, based
830 on a patch by Zoltan Herczeg. It now happens after instead of during
831 compilation. The code is cleaner, and more cases are handled. The option
832 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
833 options in pcretest are provided to set it. It can also be set by
834 (*NO_AUTO_POSSESS) at the start of a pattern.
835
836 18. The character VT has been added to the default ("C" locale) set of
837 characters that match \s and are generally treated as white space,
838 following this same change in Perl 5.18. There is now no difference between
839 "Perl space" and "POSIX space". Whether VT is treated as white space in
840 other locales depends on the locale.
841
842 19. The code for checking named groups as conditions, either for being set or
843 for being recursed, has been refactored (this is related to 14 and 15
844 above). Processing unduplicated named groups should now be as fast at
845 numerical groups, and processing duplicated groups should be faster than
846 before.
847
848 20. Two patches to the CMake build system, by Alexander Barkov:
849
850 (1) Replace the "source" command by "." in CMakeLists.txt because
851 "source" is a bash-ism.
852
853 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
854 without these the CMake build does not work on Solaris.
855
856 21. Perl has changed its handling of \8 and \9. If there is no previously
857 encountered capturing group of those numbers, they are treated as the
858 literal characters 8 and 9 instead of a binary zero followed by the
859 literals. PCRE now does the same.
860
861 22. Following Perl, added \o{} to specify codepoints in octal, making it
862 possible to specify values greater than 0777 and also making them
863 unambiguous.
864
865 23. Perl now gives an error for missing closing braces after \x{... instead of
866 treating the string as literal. PCRE now does the same.
867
868 24. RunTest used to grumble if an inappropriate test was selected explicitly,
869 but just skip it when running all tests. This make it awkward to run ranges
870 of tests when one of them was inappropriate. Now it just skips any
871 inappropriate tests, as it always did when running all tests.
872
873 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
874 character types such as \d or \w, too many callouts were inserted, and the
875 data that they returned was rubbish.
876
877 26. In UCP mode, \s was not matching two of the characters that Perl matches,
878 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
879 were matched by \h. The code has now been refactored so that the lists of
880 the horizontal and vertical whitespace characters used for \h and \v (which
881 are defined only in one place) are now also used for \s.
882
883 27. Add JIT support for the 64 bit TileGX architecture.
884 Patch by Jiong Wang (Tilera Corporation).
885
886 28. Possessive quantifiers for classes (both explicit and automatically
887 generated) now use special opcodes instead of wrapping in ONCE brackets.
888
889 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
890 (because it's meaningless), this was not happening when PCRE_CASELESS was
891 set. Not wrong, but inefficient.
892
893 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
894 properties for \w, \d, etc) is present in a test regex. Otherwise if the
895 test contains no characters greater than 255, Perl doesn't realise it
896 should be using Unicode semantics.
897
898 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
899 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
900 does in Unicode mode.
901
902 32. Added the "forbid" facility to pcretest so that putting tests into the
903 wrong test files can sometimes be quickly detected.
904
905 33. There is now a limit (default 250) on the depth of nesting of parentheses.
906 This limit is imposed to control the amount of system stack used at compile
907 time. It can be changed at build time by --with-parens-nest-limit=xxx or
908 the equivalent in CMake.
909
910 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
911 errors. Perl warns for these when in warning mode, but PCRE has no facility
912 for giving warnings.
913
914 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
915 However, this was not working for (?!) because it is optimized to (*FAIL),
916 for which PCRE does not allow quantifiers. The optimization is now disabled
917 when a quantifier follows (?!). I can't see any use for this, but it makes
918 things uniform.
919
920 36. Perl no longer allows group names to start with digits, so I have made this
921 change also in PCRE. It simplifies the code a bit.
922
923 37. In extended mode, Perl ignores spaces before a + that indicates a
924 possessive quantifier. PCRE allowed a space before the quantifier, but not
925 before the possessive +. It now does.
926
927 38. The use of \K (reset reported match start) within a repeated possessive
928 group such as (a\Kb)*+ was not working.
929
930 40. Document that the same character tables must be used at compile time and
931 run time, and that the facility to pass tables to pcre_exec() and
932 pcre_dfa_exec() is for use only with saved/restored patterns.
933
934 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
935 features for Builds with MSVC:
936
937 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
938 builds). The .rc files can be used to set FileDescription and many other
939 attributes.
940
941 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
942 This allows higher-level build scripts which want .pdb files to avoid
943 hard-coding the exact files needed."
944
945 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
946 mean "start of word" and "end of word", respectively, as a transition aid.
947
948 43. A minimizing repeat of a class containing codepoints greater than 255 in
949 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
950 compiled to use the heap for recursion.
951
952 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
953 is configured.
954
955
956 Version 8.33 28-May-2013
957 ------------------------
958
959 1. Added 'U' to some constants that are compared to unsigned integers, to
960 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
961 variables that are added to signed variables, to ensure the result is
962 signed and can be negated.
963
964 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
965 CMake config files.
966
967 3. Revise the creation of config.h.generic so that all boolean macros are
968 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
969 overriding via -D on the command line possible.
970
971 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
972 to unsigned int is reported to make a quite noticeable speed difference in
973 a specific Windows environment. Testing on Linux did also appear to show
974 some benefit (and it is clearly not harmful). Also fixed the definition of
975 Xop which should be unsigned.
976
977 5. Related to (4), changing the definition of the intermediate variable cc
978 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
979 improvements.
980
981 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
982 unnecessary spaces.
983
984 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
985 and later.
986
987 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
988
989 9. Optimizing fast_forward_start_bits in JIT.
990
991 10. Adding support for callouts in JIT, and fixing some issues revealed
992 during this work. Namely:
993
994 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
995
996 (b) Minimum length was not checked before the matching is started.
997
998 11. The value of capture_last that is passed to callouts was incorrect in some
999 cases when there was a capture on one path that was subsequently abandoned
1000 after a backtrack. Also, the capture_last value is now reset after a
1001 recursion, since all captures are also reset in this case.
1002
1003 12. The interpreter no longer returns the "too many substrings" error in the
1004 case when an overflowing capture is in a branch that is subsequently
1005 abandoned after a backtrack.
1006
1007 13. In the pathological case when an offset vector of size 2 is used, pcretest
1008 now prints out the matched string after a yield of 0 or 1.
1009
1010 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
1011 Only supported by the JIT compiler at the moment.
1012
1013 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
1014
1015 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
1016 offset of the starting point of the matching process, provided the offsets
1017 vector is large enough.
1018
1019 17. The \A escape now records a lookbehind value of 1, though its execution
1020 does not actually inspect the previous character. This is to ensure that,
1021 in partial multi-segment matching, at least one character from the old
1022 segment is retained when a new segment is processed. Otherwise, if there
1023 are no lookbehinds in the pattern, \A might match incorrectly at the start
1024 of a new segment.
1025
1026 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
1027
1028 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
1029 gives some modest performance improvement in 8-bit mode.
1030
1031 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
1032 be expressed in certain programming languages using Universal Character
1033 Names.
1034
1035 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
1036 which points out that "non characters" are not "characters that may not
1037 appear in Unicode strings" but rather "characters that are reserved for
1038 internal use and have only local meaning".
1039
1040 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
1041 there was a conditional group that depended on an assertion, if the
1042 assertion was false, the callout that immediately followed the alternation
1043 in the condition was skipped when pcre_exec() was used for matching.
1044
1045 23. Allow an explicit callout to be inserted before an assertion that is the
1046 condition for a conditional group, for compatibility with automatic
1047 callouts, which always insert a callout at this point.
1048
1049 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1050 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1051
1052 25. (*PRUNE) is now supported by the JIT compiler.
1053
1054 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1055
1056 27. Fix the case where there are two or more SKIPs with arguments that may be
1057 ignored.
1058
1059 28. (*SKIP) is now supported by the JIT compiler.
1060
1061 29. (*THEN) is now supported by the JIT compiler.
1062
1063 30. Update RunTest with additional test selector options.
1064
1065 31. The way PCRE handles backtracking verbs has been changed in two ways.
1066
1067 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1068 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1069 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1070 rather obscure rules do not always do the same thing.
1071
1072 (2) Previously, backtracking verbs were confined within assertions. This is
1073 no longer the case for positive assertions, except for (*ACCEPT). Again,
1074 this sometimes improves Perl compatibility, and sometimes does not.
1075
1076 32. A number of tests that were in test 2 because Perl did things differently
1077 have been moved to test 1, because either Perl or PCRE has changed, and
1078 these tests are now compatible.
1079
1080 32. Backtracking control verbs are now handled in the same way in JIT and
1081 interpreter.
1082
1083 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1084 contained a forward subroutine reference caused a compile error.
1085
1086 34. Auto-detect and optimize limited repetitions in JIT.
1087
1088 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1089 blocking (*UTF) etc.
1090
1091 36. In the interpreter, maximizing pattern repetitions for characters and
1092 character types now use tail recursion, which reduces stack usage.
1093
1094 37. The value of the max lookbehind was not correctly preserved if a compiled
1095 and saved regex was reloaded on a host of different endianness.
1096
1097 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1098 of the compiled pattern block, expand the flags field from 16 to 32 bits
1099 because it was almost full.
1100
1101 39. Try madvise first before posix_madvise.
1102
1103 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1104 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1105 original change locked out all matches of empty strings. This has been
1106 changed so that one match of an empty string per line is recognized.
1107 Subsequent searches on the same line (for colouring or for --only-matching,
1108 for example) do not recognize empty strings.
1109
1110 41. Applied a user patch to fix a number of spelling mistakes in comments.
1111
1112 42. Data lines longer than 65536 caused pcretest to crash.
1113
1114 43. Clarified the data type for length and startoffset arguments for pcre_exec
1115 and pcre_dfa_exec in the function-specific man pages, where they were
1116 explicitly stated to be in bytes, never having been updated. I also added
1117 some clarification to the pcreapi man page.
1118
1119 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1120 a segmentation fault.
1121
1122
1123 Version 8.32 30-November-2012
1124 -----------------------------
1125
1126 1. Improved JIT compiler optimizations for first character search and single
1127 character iterators.
1128
1129 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1130 Patch by Daniel Richard G.
1131
1132 3. Single character iterator optimizations in the JIT compiler.
1133
1134 4. Improved JIT compiler optimizations for character ranges.
1135
1136 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1137 Reported by Giuseppe D'Angelo.
1138
1139 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1140 of a line, was being set incorrectly in cases where .* appeared inside
1141 atomic brackets at the start of a pattern, or where there was a subsequent
1142 *PRUNE or *SKIP.
1143
1144 7. Improved instruction cache flush for POWER/PowerPC.
1145 Patch by Daniel Richard G.
1146
1147 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1148 grep:
1149
1150 (a) There is now no limit to the number of patterns to be matched.
1151
1152 (b) An error is given if a pattern is too long.
1153
1154 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1155 are now supported.
1156
1157 (d) --exclude-from and --include-from (multiple use) have been added.
1158
1159 (e) Exclusions and inclusions now apply to all files and directories, not
1160 just to those obtained from scanning a directory recursively.
1161
1162 (f) Multiple uses of -f and --file-list are now supported.
1163
1164 (g) In a Windows environment, the default for -d has been changed from
1165 "read" (the GNU grep default) to "skip", because otherwise the presence
1166 of a directory in the file list provokes an error.
1167
1168 (h) The documentation has been revised and clarified in places.
1169
1170 9. Improve the matching speed of capturing brackets.
1171
1172 10. Changed the meaning of \X so that it now matches a Unicode extended
1173 grapheme cluster.
1174
1175 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1176 out POSIX threads when JIT support is configured.
1177
1178 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1179
1180 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1181 structure could go wrong in environments where size_t is not the same size
1182 as int.
1183
1184 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1185 set.
1186
1187 15. The EBCDIC support had decayed; later updates to the code had included
1188 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1189 general tidy up of EBCDIC-related issues, and the documentation was also
1190 not quite right. There is now a test that can be run on ASCII systems to
1191 check some of the EBCDIC-related things (but is it not a full test).
1192
1193 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1194 in a small tidy to the code.
1195
1196 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1197
1198 18. If the --only-matching (-o) option in pcregrep is specified multiple
1199 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1200 substrings matched by the 1st and 2nd capturing parentheses. A separating
1201 string can be specified by --om-separator (default empty).
1202
1203 19. Improving the first n character searches.
1204
1205 20. Turn case lists for horizontal and vertical white space into macros so that
1206 they are defined only once.
1207
1208 21. This set of changes together give more compatible Unicode case-folding
1209 behaviour for characters that have more than one other case when UCP
1210 support is available.
1211
1212 (a) The Unicode property table now has offsets into a new table of sets of
1213 three or more characters that are case-equivalent. The MultiStage2.py
1214 script that generates these tables (the pcre_ucd.c file) now scans
1215 CaseFolding.txt instead of UnicodeData.txt for character case
1216 information.
1217
1218 (b) The code for adding characters or ranges of characters to a character
1219 class has been abstracted into a generalized function that also handles
1220 case-independence. In UTF-mode with UCP support, this uses the new data
1221 to handle characters with more than one other case.
1222
1223 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1224 whose other case is greater than 256 are now correctly matched
1225 caselessly. Previously, the high codepoint matched the low one, but not
1226 vice versa.
1227
1228 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1229 of the new class addition function, using character lists defined as
1230 macros alongside the case definitions of 20 above.
1231
1232 (e) Caseless back references now work with characters that have more than
1233 one other case.
1234
1235 (f) General caseless matching of characters with more than one other case
1236 is supported.
1237
1238 22. Unicode character properties were updated from Unicode 6.2.0
1239
1240 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1241
1242 24. Add support for 32-bit character strings, and UTF-32
1243
1244 25. Major JIT compiler update (code refactoring and bugfixing).
1245 Experimental Sparc 32 support is added.
1246
1247 26. Applied a modified version of Daniel Richard G's patch to create
1248 pcre.h.generic and config.h.generic by "make" instead of in the
1249 PrepareRelease script.
1250
1251 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1252 pcre_compile.c when checking for a zero character.
1253
1254 28. Introducing a native interface for JIT. Through this interface, the compiled
1255 machine code can be directly executed. The purpose of this interface is to
1256 provide fast pattern matching, so several sanity checks are not performed.
1257 However, feature tests are still performed. The new interface provides
1258 1.4x speedup compared to the old one.
1259
1260 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1261 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1262 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1263
1264 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1265 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1266 the "old" RFC 2279). Instead, it ended up passing a negative length to
1267 pcre_exec().
1268
1269 31. Add support for GCC's visibility feature to hide internal functions.
1270
1271 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1272 "unknown -C option" after outputting 0 or 1.
1273
1274 33. There is now support for generating a code coverage report for the test
1275 suite in environments where gcc is the compiler and lcov is installed. This
1276 is mainly for the benefit of the developers.
1277
1278 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1279 unaddressable using valgrind annotations, allowing valgrind to detect
1280 invalid memory accesses. This is mainly for the benefit of the developers.
1281
1282 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1283
1284 26. Give configure error if --enable-cpp but no C++ compiler found.
1285
1286
1287 Version 8.31 06-July-2012
1288 -------------------------
1289
1290 1. Fixing a wrong JIT test case and some compiler warnings.
1291
1292 2. Removed a bashism from the RunTest script.
1293
1294 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1295 to unsigned type, result still unsigned" that was given by an MS compiler
1296 on encountering the code "-sizeof(xxx)".
1297
1298 4. Partial matching support is added to the JIT compiler.
1299
1300 5. Fixed several bugs concerned with partial matching of items that consist
1301 of more than one character:
1302
1303 (a) /^(..)\1/ did not partially match "aba" because checking references was
1304 done on an "all or nothing" basis. This also applied to repeated
1305 references.
1306
1307 (b) \R did not give a hard partial match if \r was found at the end of the
1308 subject.
1309
1310 (c) \X did not give a hard partial match after matching one or more
1311 characters at the end of the subject.
1312
1313 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1314 a partial match for the string "\r".
1315
1316 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1317 a partial match for a CR character at the end of the subject string.
1318
1319 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1320 running pcretest, the text "(JIT)" added to the output whenever JIT is
1321 actually used to run the match.
1322
1323 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1324 or /S+[+] with a digit between 1 and 7.
1325
1326 8. OP_NOT now supports any UTF character not just single-byte ones.
1327
1328 9. (*MARK) control verb is now supported by the JIT compiler.
1329
1330 10. The command "./RunTest list" lists the available tests without actually
1331 running any of them. (Because I keep forgetting what they all are.)
1332
1333 11. Add PCRE_INFO_MAXLOOKBEHIND.
1334
1335 12. Applied a (slightly modified) user-supplied patch that improves performance
1336 when the heap is used for recursion (compiled with --disable-stack-for-
1337 recursion). Instead of malloc and free for each heap frame each time a
1338 logical recursion happens, frames are retained on a chain and re-used where
1339 possible. This sometimes gives as much as 30% improvement.
1340
1341 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1342 call.
1343
1344 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1345
1346 15. It is now possible to link pcretest with libedit as an alternative to
1347 libreadline.
1348
1349 16. (*COMMIT) control verb is now supported by the JIT compiler.
1350
1351 17. The Unicode data tables have been updated to Unicode 6.1.0.
1352
1353 18. Added --file-list option to pcregrep.
1354
1355 19. Added binary file support to pcregrep, including the -a, --binary-files,
1356 -I, and --text options.
1357
1358 20. The madvise function is renamed for posix_madvise for QNX compatibility
1359 reasons. Fixed by Giuseppe D'Angelo.
1360
1361 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1362
1363 22. Changed the option for creating symbolic links for 16-bit man pages from
1364 -s to -sf so that re-installing does not cause issues.
1365
1366 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1367
1368 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1369 matches in certain environments (the workspace was not being correctly
1370 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1371 some of the workspace data at the beginning of a restart.
1372
1373 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1374 was not doing so when it should - probably a typo introduced by SVN 528
1375 (change 8.10/14).
1376
1377 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1378 \w+ when the character tables indicated that \x{c4} was a word character.
1379 There were several related cases, all because the tests for doing a table
1380 lookup were testing for characters less than 127 instead of 255.
1381
1382 27. If a pattern contains capturing parentheses that are not used in a match,
1383 their slots in the ovector are set to -1. For those that are higher than
1384 any matched groups, this happens at the end of processing. In the case when
1385 there were back references that the ovector was too small to contain
1386 (causing temporary malloc'd memory to be used during matching), and the
1387 highest capturing number was not used, memory off the end of the ovector
1388 was incorrectly being set to -1. (It was using the size of the temporary
1389 memory instead of the true size.)
1390
1391 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1392 ovector size, it uses memory at the end of the block that it has got.
1393
1394 29. Check for an overlong MARK name and give an error at compile time. The
1395 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1396
1397 30. JIT compiler update.
1398
1399 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1400 Rill for the patch.
1401
1402 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1403
1404 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1405
1406 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1407 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1408 was enabled.
1409
1410 35. Improve JIT code generation for greedy plus quantifier.
1411
1412 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1413 1 to "aa" instead of to an empty string. The bug affected repeated groups
1414 that could potentially match an empty string.
1415
1416 37. Optimizing single character iterators in JIT.
1417
1418 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1419 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1420 codepoints that are too big for the mode are faulted, and in a UTF mode,
1421 disallowed codepoints are also faulted.
1422
1423 39. If PCRE was compiled with UTF support, in three places in the DFA
1424 matcher there was code that should only have been obeyed in UTF mode, but
1425 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1426 processing when bytes with values greater than 127 were present. In 16-bit
1427 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1428 both cases the values are those that cannot be the first data item in a UTF
1429 character. The three items that might have provoked this were recursions,
1430 possessively repeated groups, and atomic groups.
1431
1432 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1433 and pcregrep, because some OS require shared objects to be explicitly
1434 passed to ld, causing the link step to fail if they are not.
1435
1436 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1437 mode, patterns that started with \h* or \R* might be incorrectly matched.
1438
1439
1440 Version 8.30 04-February-2012
1441 -----------------------------
1442
1443 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1444 name is defined in ctype.h.
1445
1446 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1447 only in quite long subpatterns.
1448
1449 3. Removed the function pcre_info(), which has been obsolete and deprecated
1450 since it was replaced by pcre_fullinfo() in February 2000.
1451
1452 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1453 match a (*MARK), and the match failed at the start of the subject, a
1454 reference to memory before the start of the subject could occur. This bug
1455 was introduced by fix 17 of release 8.21.
1456
1457 5. A reference to an unset group with zero minimum repetition was giving
1458 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1459 /(another)?(\1?)test/ matched against "hello world test". This bug was
1460 introduced in release 8.13.
1461
1462 6. Add support for 16-bit character strings (a large amount of work involving
1463 many changes and refactorings).
1464
1465 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1466 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1467 from a file.
1468
1469 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1470 rounding is not applied in this particular case).
1471
1472 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1473 if they appear, or are escaped, in patterns.
1474
1475 10. Get rid of a number of -Wunused-but-set-variable warnings.
1476
1477 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1478 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1479 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1480 also returns the mark "x". This bug applied to capturing parentheses,
1481 non-capturing parentheses, and atomic parentheses. It also applied to some
1482 assertions.
1483
1484 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1485 information out of configure.ac instead of relying on pcre.h.generic, which
1486 is not stored in the repository.
1487
1488 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1489 -lreadline.
1490
1491 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1492
1493 15. Applied Graycode's patch to put the top-level frame on the stack rather
1494 than the heap when not using the stack for recursion. This gives a
1495 performance improvement in many cases when recursion is not deep.
1496
1497 16. Experimental code added to "pcretest -C" to output the stack frame size.
1498
1499
1500 Version 8.21 12-Dec-2011
1501 ------------------------
1502
1503 1. Updating the JIT compiler.
1504
1505 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1506 are added as well.
1507
1508 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1509 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1510 calling _pcre_jit_exec. Some extra comments are added.
1511
1512 4. (*MARK) settings inside atomic groups that do not contain any capturing
1513 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1514 was introduced by change 18 for 8.20.
1515
1516 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1517 ECMA-262 standard.
1518
1519 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1520 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1521 This bug was probably introduced by change 9 of 8.13.
1522
1523 7. While fixing 6 above, I noticed that a number of other items were being
1524 incorrectly rejected as "not fixed length". This arose partly because newer
1525 opcodes had not been added to the fixed-length checking code. I have (a)
1526 corrected the bug and added tests for these items, and (b) arranged for an
1527 error to occur if an unknown opcode is encountered while checking for fixed
1528 length instead of just assuming "not fixed length". The items that were
1529 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1530 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1531 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1532
1533 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1534 being incorrectly compiled and would have given unpredicatble results.
1535
1536 9. A possessively repeated subpattern with minimum repeat count greater than
1537 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1538 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1539 the first (A) could occur when it should not.
1540
1541 10. Add a cast and remove a redundant test from the code.
1542
1543 11. JIT should use pcre_malloc/pcre_free for allocation.
1544
1545 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1546 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1547 is anything other than /usr, -L is still shown).
1548
1549 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1550
1551 14. Perl does not support \N without a following name in a [] class; PCRE now
1552 also gives an error.
1553
1554 15. If a forward reference was repeated with an upper limit of around 2000,
1555 it caused the error "internal error: overran compiling workspace". The
1556 maximum number of forward references (including repeats) was limited by the
1557 internal workspace, and dependent on the LINK_SIZE. The code has been
1558 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1559 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1560 of around 200,000 forward references. While doing this, I also speeded up
1561 the filling in of repeated forward references.
1562
1563 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1564 incorrectly expecting the subject to contain another "a" after the start.
1565
1566 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1567 in the match, the SKIP should be ignored. This was not happening; instead
1568 the SKIP was being treated as NOMATCH. For patterns such as
1569 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1570 tested.
1571
1572 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1573 now much more compatible with Perl, in particular in cases where the result
1574 is a non-match for a non-anchored pattern. For example, if
1575 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1576 "m", where previously it did not return a name. A side effect of this
1577 change is that for partial matches, the last encountered mark name is
1578 returned, as for non matches. A number of tests that were previously not
1579 Perl-compatible have been moved into the Perl-compatible test files. The
1580 refactoring has had the pleasing side effect of removing one argument from
1581 the match() function, thus reducing its stack requirements.
1582
1583 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1584 subsequent uses of /S (without +) incorrectly behaved like /S+.
1585
1586 21. Retrieve executable code size support for the JIT compiler and fixing
1587 some warnings.
1588
1589 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1590 not work when the shorter character appeared right at the end of the
1591 subject string.
1592
1593 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1594 systems.
1595
1596 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1597 output it when the /M option is used in pcretest.
1598
1599 25. The CheckMan script was not being included in the distribution. Also, added
1600 an explicit "perl" to run Perl scripts from the PrepareRelease script
1601 because this is reportedly needed in Windows.
1602
1603 26. If study data was being save in a file and studying had not found a set of
1604 "starts with" bytes for the pattern, the data written to the file (though
1605 never used) was taken from uninitialized memory and so caused valgrind to
1606 complain.
1607
1608 27. Updated RunTest.bat as provided by Sheri Pierce.
1609
1610 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1611
1612 29. Computation of memory usage for the table of capturing group names was
1613 giving an unnecessarily large value.
1614
1615
1616 Version 8.20 21-Oct-2011
1617 ------------------------
1618
1619 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1620 a POSIX class. After further experiments with Perl, which convinced me that
1621 Perl has bugs and confusions, a closing square bracket is no longer allowed
1622 in a POSIX name. This bug also affected patterns with classes that started
1623 with full stops.
1624
1625 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1626 captured substring, but while checking the failing first alternative,
1627 substring 1 is temporarily captured. If the output vector supplied to
1628 pcre_exec() was not big enough for this capture, the yield of the function
1629 was still zero ("insufficient space for captured substrings"). This cannot
1630 be totally fixed without adding another stack variable, which seems a lot
1631 of expense for a edge case. However, I have improved the situation in cases
1632 such as /(a)(b)x|abc/ matched against "abc", where the return code
1633 indicates that fewer than the maximum number of slots in the ovector have
1634 been set.
1635
1636 3. Related to (2) above: when there are more back references in a pattern than
1637 slots in the output vector, pcre_exec() uses temporary memory during
1638 matching, and copies in the captures as far as possible afterwards. It was
1639 using the entire output vector, but this conflicts with the specification
1640 that only 2/3 is used for passing back captured substrings. Now it uses
1641 only the first 2/3, for compatibility. This is, of course, another edge
1642 case.
1643
1644 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1645 main code base, and can be used by building with --enable-jit. When this is
1646 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1647 runtime --no-jit option is given.
1648
1649 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1650 ovector, the return from the function was zero, implying that there were
1651 other matches that did not fit. The correct "exactly full" value is now
1652 returned.
1653
1654 6. If a subpattern that was called recursively or as a subroutine contained
1655 (*PRUNE) or any other control that caused it to give a non-standard return,
1656 invalid errors such as "Error -26 (nested recursion at the same subject
1657 position)" or even infinite loops could occur.
1658
1659 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1660 computing the minimum length on reaching *ACCEPT, and so ended up with the
1661 wrong value of 1 rather than 0. Further investigation indicates that
1662 computing a minimum subject length in the presence of *ACCEPT is difficult
1663 (think back references, subroutine calls), and so I have changed the code
1664 so that no minimum is registered for a pattern that contains *ACCEPT.
1665
1666 8. If (*THEN) was present in the first (true) branch of a conditional group,
1667 it was not handled as intended. [But see 16 below.]
1668
1669 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1670 Sheri Pierce.
1671
1672 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1673 the first byte in a match must be "a".
1674
1675 11. Change 17 for 8.13 increased the recursion depth for patterns like
1676 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1677 pattern contains any instances of (*THEN). If it does not, the old
1678 optimizations are restored. It would be nice to do this on a per-group
1679 basis, but at the moment that is not feasible.
1680
1681 12. In some environments, the output of pcretest -C is CRLF terminated. This
1682 broke RunTest's code that checks for the link size. A single white space
1683 character after the value is now allowed for.
1684
1685 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1686 For "fr", it uses the Windows-specific input and output files.
1687
1688 14. If (*THEN) appeared in a group that was called recursively or as a
1689 subroutine, it did not work as intended. [But see next item.]
1690
1691 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1692 pattern fragments (but not containing any | characters). If A and B are
1693 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1694 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1695 D. In other words, Perl considers parentheses that do not contain any |
1696 characters to be part of a surrounding alternative, whereas PCRE was
1697 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1698 differently. PCRE now behaves in the same way as Perl, except in the case
1699 of subroutine/recursion calls such as (?1) which have in any case always
1700 been different (but PCRE had them first :-).
1701
1702 16. Related to 15 above: Perl does not treat the | in a conditional group as
1703 creating alternatives. Such a group is treated in the same way as an
1704 ordinary group without any | characters when processing (*THEN). PCRE has
1705 been changed to match Perl's behaviour.
1706
1707 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1708 RunGrepTest script failed.
1709
1710 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1711 inevitable for groups that contain captures, but it can lead to a lot of
1712 stack use in large patterns. The old behaviour has been restored for atomic
1713 groups that do not contain any capturing parentheses.
1714
1715 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1716 suppress the check for a minimum subject length at run time. (If it was
1717 given to pcre_exec() or pcre_dfa_exec() it did work.)
1718
1719 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1720 fail to work when decoding hex characters in data strings in EBCDIC
1721 environments.
1722
1723 21. It appears that in at least one Mac OS environment, the isxdigit() function
1724 is implemented as a macro that evaluates to its argument more than once,
1725 contravening the C 90 Standard (I haven't checked a later standard). There
1726 was an instance in pcretest which caused it to go wrong when processing
1727 \x{...} escapes in subject strings. The has been rewritten to avoid using
1728 things like p++ in the argument of isxdigit().
1729
1730
1731 Version 8.13 16-Aug-2011
1732 ------------------------
1733
1734 1. The Unicode data tables have been updated to Unicode 6.0.0.
1735
1736 2. Two minor typos in pcre_internal.h have been fixed.
1737
1738 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1739 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1740 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1741
1742 4. There were a number of related bugs in the code for matching backrefences
1743 caselessly in UTF-8 mode when codes for the characters concerned were
1744 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1745 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1746 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1747 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1748 2-byte code at the end of the subject (it thought there wasn't enough data
1749 left).
1750
1751 5. Comprehensive information about what went wrong is now returned by
1752 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1753 as the output vector has at least 2 elements. The offset of the start of
1754 the failing character and a reason code are placed in the vector.
1755
1756 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1757 now returned is for the first byte of the failing character, instead of the
1758 last byte inspected. This is an incompatible change, but I hope it is small
1759 enough not to be a problem. It makes the returned offset consistent with
1760 pcre_exec() and pcre_dfa_exec().
1761
1762 7. pcretest now gives a text phrase as well as the error number when
1763 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1764 failure, the offset and reason code are output.
1765
1766 8. When \R was used with a maximizing quantifier it failed to skip backwards
1767 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1768 back over a single character (\n). This seems wrong (because it treated the
1769 two characters as a single entity when going forwards), conflicts with the
1770 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1771 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1772 has been changed.
1773
1774 9. Some internal refactoring has changed the processing so that the handling
1775 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1776 time (the PCRE_DOTALL option was changed this way some time ago: version
1777 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1778 which was always a bit of a fudge. It also means that there is one less
1779 argument for the match() function, which reduces its stack requirements
1780 slightly. This change also fixes an incompatibility with Perl: the pattern
1781 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1782
1783 10. More internal refactoring has drastically reduced the number of recursive
1784 calls to match() for possessively repeated groups such as (abc)++ when
1785 using pcre_exec().
1786
1787 11. While implementing 10, a number of bugs in the handling of groups were
1788 discovered and fixed:
1789
1790 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1791 (a|)*(?1) gave a compile-time internal error.
1792 ((a|)+)+ did not notice that the outer group could match an empty string.
1793 (^a|^)+ was not marked as anchored.
1794 (.*a|.*)+ was not marked as matching at start or after a newline.
1795
1796 12. Yet more internal refactoring has removed another argument from the match()
1797 function. Special calls to this function are now indicated by setting a
1798 value in a variable in the "match data" data block.
1799
1800 13. Be more explicit in pcre_study() instead of relying on "default" for
1801 opcodes that mean there is no starting character; this means that when new
1802 ones are added and accidentally left out of pcre_study(), testing should
1803 pick them up.
1804
1805 14. The -s option of pcretest has been documented for ages as being an old
1806 synonym of -m (show memory usage). I have changed it to mean "force study
1807 for every regex", that is, assume /S for every regex. This is similar to -i
1808 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1809 using it. It makes it easier to run collections of tests with and without
1810 study enabled, and thereby test pcre_study() more easily. All the standard
1811 tests are now run with and without -s (but some patterns can be marked as
1812 "never study" - see 20 below).
1813
1814 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1815 restoration of the capturing data to the outer values was not happening
1816 correctly.
1817
1818 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1819 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1820 pattern had matched an empty string, and so incorrectly returned a no
1821 match.
1822
1823 17. There was optimizing code for the last branch of non-capturing parentheses,
1824 and also for the obeyed branch of a conditional subexpression, which used
1825 tail recursion to cut down on stack usage. Unfortunately, now that there is
1826 the possibility of (*THEN) occurring in these branches, tail recursion is
1827 no longer possible because the return has to be checked for (*THEN). These
1828 two optimizations have therefore been removed. [But see 8.20/11 above.]
1829
1830 18. If a pattern containing \R was studied, it was assumed that \R always
1831 matched two bytes, thus causing the minimum subject length to be
1832 incorrectly computed because \R can also match just one byte.
1833
1834 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1835 was incorrectly computed.
1836
1837 20. If /S is present twice on a test pattern in pcretest input, it now
1838 *disables* studying, thereby overriding the use of -s on the command line
1839 (see 14 above). This is necessary for one or two tests to keep the output
1840 identical in both cases.
1841
1842 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1843 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1844
1845 22. When an atomic group that contained a capturing parenthesis was
1846 successfully matched, but the branch in which it appeared failed, the
1847 capturing was not being forgotten if a higher numbered group was later
1848 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1849 group 1 to "a", when in fact it should be unset. This applied to multi-
1850 branched capturing and non-capturing groups, repeated or not, and also to
1851 positive assertions (capturing in negative assertions does not happen
1852 in PCRE) and also to nested atomic groups.
1853
1854 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1855 subject after a captured substring, to make it easier to tell which of a
1856 number of identical substrings has been captured.
1857
1858 24. The way atomic groups are processed by pcre_exec() has been changed so that
1859 if they are repeated, backtracking one repetition now resets captured
1860 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1861 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1862 "aaa". Previously, it would have been "a". As part of this code
1863 refactoring, the way recursive calls are handled has also been changed.
1864
1865 25. If an assertion condition captured any substrings, they were not passed
1866 back unless some other capturing happened later. For example, if
1867 (?(?=(a))a) was matched against "a", no capturing was returned.
1868
1869 26. When studying a pattern that contained subroutine calls or assertions,
1870 the code for finding the minimum length of a possible match was handling
1871 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1872 group 1 called group 2 while simultaneously a separate group 2 called group
1873 1). A stack overflow occurred in this case. I have fixed this by limiting
1874 the recursion depth to 10.
1875
1876 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1877 Fortmann. This supports explicit test numbers on the command line, and has
1878 argument validation and error reporting.
1879
1880 28. An instance of \X with an unlimited repeat could fail if at any point the
1881 first character it looked at was a mark character.
1882
1883 29. Some minor code refactoring concerning Unicode properties and scripts
1884 should reduce the stack requirement of match() slightly.
1885
1886 30. Added the '=' option to pcretest to check the setting of unused capturing
1887 slots at the end of the pattern, which are documented as being -1, but are
1888 not included in the return count.
1889
1890 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1891 compiled something random. Now it gives a compile-time error (as does
1892 Perl).
1893
1894 32. A *MARK encountered during the processing of a positive assertion is now
1895 recorded and passed back (compatible with Perl).
1896
1897 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1898 had alternative anchored branches, the search for a second match in a line
1899 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1900 matched the line "0102" twice. The same bug affected patterns that started
1901 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1902 twice.
1903
1904 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1905 does, and because of capturing effects, quantifying parenthesized
1906 assertions may at times be useful. Quantifiers are now allowed for
1907 parenthesized assertions.
1908
1909 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1910
1911 36. \g was being checked for fancy things in a character class, when it should
1912 just be a literal "g".
1913
1914 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1915 appearance of a nested POSIX class supersedes an apparent external class.
1916 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1917 unescaped square brackets may also appear as part of class names. For
1918 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1919 more like Perl. (But see 8.20/1 above.)
1920
1921 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1922 was because it thought it was \N{name}, which is not supported).
1923
1924 39. Add minix to OS list not supporting the -S option in pcretest.
1925
1926 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1927 cannot analyze patterns in sufficient detail to catch mutual recursions
1928 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1929 subgroup is called recursively as a subpattern for a second time at the
1930 same position in the subject string. In previous releases this might have
1931 been caught by the recursion limit, or it might have run out of stack.
1932
1933 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1934 happen only once. PCRE was, however incorrectly giving a compile time error
1935 "recursive call could loop indefinitely" because it cannot analyze the
1936 pattern in sufficient detail. The compile time test no longer happens when
1937 PCRE is compiling a conditional subpattern, but actual runaway loops are
1938 now caught at runtime (see 40 above).
1939
1940 42. It seems that Perl allows any characters other than a closing parenthesis
1941 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1942 has been changed to be the same.
1943
1944 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1945 as not to get warnings when autogen.sh is called. Also changed
1946 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1947
1948 44. To help people who use pcregrep to scan files containing exceedingly long
1949 lines, the following changes have been made:
1950
1951 (a) The default value of the buffer size parameter has been increased from
1952 8K to 20K. (The actual buffer used is three times this size.)
1953
1954 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1955 PCRE is built.
1956
1957 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1958 to be set at run time.
1959
1960 (d) Numerical values in pcregrep options can be followed by K or M, for
1961 example --buffer-size=50K.
1962
1963 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1964 given and the return code is set to 2.
1965
1966 45. Add a pointer to the latest mark to the callout data block.
1967
1968 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1969 partial match of an empty string instead of no match. This was specific to
1970 the use of ".".
1971
1972 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1973 complete match instead of a partial match. This bug was dependent on both
1974 the PCRE_UTF8 and PCRE_DOTALL options being set.
1975
1976 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1977 starting byte set, because \b was not being ignored.
1978
1979
1980 Version 8.12 15-Jan-2011
1981 ------------------------
1982
1983 1. Fixed some typos in the markup of the man pages, and wrote a script that
1984 checks for such things as part of the documentation building process.
1985
1986 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1987 --match-limit and --recursion-limit options (added for 8.11). In
1988 particular, this made one of the standard tests fail. (The integer value
1989 went into the wrong half of a long int.)
1990
1991 3. If the --colour option was given to pcregrep with -v (invert match), it
1992 did strange things, either producing crazy output, or crashing. It should,
1993 of course, ignore a request for colour when reporting lines that do not
1994 match.
1995
1996 4. Another pcregrep bug caused similar problems if --colour was specified with
1997 -M (multiline) and the pattern match finished with a line ending.
1998
1999 5. In pcregrep, when a pattern that ended with a literal newline sequence was
2000 matched in multiline mode, the following line was shown as part of the
2001 match. This seems wrong, so I have changed it.
2002
2003 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
2004 the check for further matches in the same line (so they could be coloured)
2005 to overrun the end of the current line. If another match was found, it was
2006 incorrectly shown (and then shown again when found in the next line).
2007
2008 7. If pcregrep was compiled under Windows, there was a reference to the
2009 function pcregrep_exit() before it was defined. I am assuming this was
2010 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
2011 reported by a user. I've moved the definition above the reference.
2012
2013
2014 Version 8.11 10-Dec-2010
2015 ------------------------
2016
2017 1. (*THEN) was not working properly if there were untried alternatives prior
2018 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
2019 backtracked to try for "b" instead of moving to the next alternative branch
2020 at the same level (in this case, to look for "c"). The Perl documentation
2021 is clear that when (*THEN) is backtracked onto, it goes to the "next
2022 alternative in the innermost enclosing group".
2023
2024 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
2025 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
2026 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
2027 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
2028 (*THEN).
2029
2030 3. If \s appeared in a character class, it removed the VT character from
2031 the class, even if it had been included by some previous item, for example
2032 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
2033 of \s, but is part of the POSIX "space" class.)
2034
2035 4. A partial match never returns an empty string (because you can always
2036 match an empty string at the end of the subject); however the checking for
2037 an empty string was starting at the "start of match" point. This has been
2038 changed to the "earliest inspected character" point, because the returned
2039 data for a partial match starts at this character. This means that, for
2040 example, /(?<=abc)def/ gives a partial match for the subject "abc"
2041 (previously it gave "no match").
2042
2043 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2044 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2045 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2046 has an implication that the given string is incomplete (because a partial
2047 match is preferred over a full match). For this reason, these items now
2048 give a partial match in this situation. [Aside: previously, the one case
2049 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2050 match rather than a full match, which was wrong by the old rules, but is
2051 now correct.]
2052
2053 6. There was a bug in the handling of #-introduced comments, recognized when
2054 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2055 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2056 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2057 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2058 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2059 places in pcre_compile().
2060
2061 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2062 comments when looking ahead for named forward references to subpatterns,
2063 the only newline sequence it recognized was NL. It now handles newlines
2064 according to the set newline convention.
2065
2066 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2067 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2068 cater for a lack of strerror(). These oversights have been fixed.
2069
2070 9. Added --match-limit and --recursion-limit to pcregrep.
2071
2072 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2073
2074 11. When the -o option was used, pcregrep was setting a return code of 1, even
2075 when matches were found, and --line-buffered was not being honoured.
2076
2077 12. Added an optional parentheses number to the -o and --only-matching options
2078 of pcregrep.
2079
2080 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2081 can match an empty string. The code to do it in pcretest and pcredemo
2082 needed fixing:
2083
2084 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2085 only one byte after an empty string match just before CRLF (this case
2086 just got forgotten; "any" and "anycrlf" were OK).
2087
2088 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2089 mode when an empty string match preceded an ASCII character followed by
2090 a non-ASCII character. (The code for advancing by one character rather
2091 than one byte was nonsense.)
2092
2093 (c) The pcredemo.c sample program did not have any code at all to handle
2094 the cases when CRLF is a valid newline sequence.
2095
2096 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2097 as a starting offset was within the subject string. There is now a new
2098 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2099 negative or greater than the length of the string. In order to test this,
2100 pcretest is extended to allow the setting of negative starting offsets.
2101
2102 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2103 starting offset points to the beginning of a UTF-8 character was
2104 unnecessarily clumsy. I tidied it up.
2105
2106 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2107 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2108
2109 17. Nobody had reported that the --include_dir option, which was added in
2110 release 7.7 should have been called --include-dir (hyphen, not underscore)
2111 for compatibility with GNU grep. I have changed it to --include-dir, but
2112 left --include_dir as an undocumented synonym, and the same for
2113 --exclude-dir, though that is not available in GNU grep, at least as of
2114 release 2.5.4.
2115
2116 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2117 characters from a string of bytes) have been redefined so as not to use
2118 loops, in order to improve performance in some environments. At the same
2119 time, I abstracted some of the common code into auxiliary macros to save
2120 repetition (this should not affect the compiled code).
2121
2122 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2123 compile-time error is now given if \c is not followed by an ASCII
2124 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2125 different, and any byte value is allowed.)
2126
2127 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2128 START_OPTIMIZE option, which is now allowed at compile time - but just
2129 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2130 to pcregrep and other applications that have no direct access to PCRE
2131 options. The new /Y option in pcretest sets this option when calling
2132 pcre_compile().
2133
2134 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2135 back references. Groups containing recursive back references were forced to
2136 be atomic by that change, but in the case of named groups, the amount of
2137 memory required was incorrectly computed, leading to "Failed: internal
2138 error: code overflow". This has been fixed.
2139
2140 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2141 pcretest.c, to avoid build problems in some Borland environments.
2142
2143
2144 Version 8.10 25-Jun-2010
2145 ------------------------
2146
2147 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2148 THEN.
2149
2150 2. (*ACCEPT) was not working when inside an atomic group.
2151
2152 3. Inside a character class, \B is treated as a literal by default, but
2153 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2154 causes the error). The code is unchanged, but I tidied the documentation.
2155
2156 4. Inside a character class, PCRE always treated \R and \X as literals,
2157 whereas Perl faults them if its -w option is set. I have changed PCRE so
2158 that it faults them when PCRE_EXTRA is set.
2159
2160 5. Added support for \N, which always matches any character other than
2161 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2162
2163 6. When compiling pcregrep with newer versions of gcc which may have
2164 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2165 declared with attribute warn_unused_result" were given. Just casting the
2166 result to (void) does not stop the warnings; a more elaborate fudge is
2167 needed. I've used a macro to implement this.
2168
2169 7. Minor change to pcretest.c to avoid a compiler warning.
2170
2171 8. Added four artifical Unicode properties to help with an option to make
2172 \s etc use properties (see next item). The new properties are: Xan
2173 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2174
2175 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2176 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2177 this option. Modified pcretest to add /W to test this facility. Added
2178 REG_UCP to make it available via the POSIX interface.
2179
2180 10. Added --line-buffered to pcregrep.
2181
2182 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2183 studied, and the match started with a letter with a code point greater than
2184 127 whose first byte was different to the first byte of the other case of
2185 the letter, the other case of this starting letter was not recognized
2186 (#976).
2187
2188 12. If a pattern that was studied started with a repeated Unicode property
2189 test, for example, \p{Nd}+, there was the theoretical possibility of
2190 setting up an incorrect bitmap of starting bytes, but fortunately it could
2191 not have actually happened in practice until change 8 above was made (it
2192 added property types that matched character-matching opcodes).
2193
2194 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2195 possible starting bytes for non-anchored patterns.
2196
2197 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2198 \R, and also a number of cases that involve Unicode properties, both
2199 explicit and implicit when PCRE_UCP is set.
2200
2201 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2202 input, it could crash or give wrong results if characters with values
2203 greater than 0xc0 were present in the subject string. (Detail: it assumed
2204 UTF-8 input when processing these items.)
2205
2206 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2207 size_t is 64-bit (#991).
2208
2209 17. Added a check for running out of memory when PCRE is compiled with
2210 --disable-stack-for-recursion (#990).
2211
2212 18. If the last data line in a file for pcretest does not have a newline on
2213 the end, a newline was missing in the output.
2214
2215 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2216 less than 128) in its various bitmaps. However, there is a facility for
2217 generating tables according to the current locale when PCRE is compiled. It
2218 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2219 characters, are recognized by isspace() and therefore were getting set in
2220 these tables, and indeed these tables seem to approximate to ISO 8859. This
2221 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2222 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2223 which of course cannot start UTF-8 characters. I have changed the code so
2224 that only real ASCII characters (less than 128) and the correct starting
2225 bytes for UTF-8 encodings are set for characters greater than 127 when in
2226 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2227 altogether.)
2228
2229 20. Added the /T option to pcretest so as to be able to run tests with non-
2230 standard character tables, thus making it possible to include the tests
2231 used for 19 above in the standard set of tests.
2232
2233 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2234 reference to a subpattern the other side of a comment that contains an
2235 opening parenthesis caused either an internal compiling error, or a
2236 reference to the wrong subpattern.
2237
2238
2239 Version 8.02 19-Mar-2010
2240 ------------------------
2241
2242 1. The Unicode data tables have been updated to Unicode 5.2.0.
2243
2244 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2245 configured.
2246
2247 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2248 original author of that file, following a query about its status.
2249
2250 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2251 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2252
2253 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2254 quantifier applied to a forward-referencing subroutine call, could compile
2255 incorrect code or give the error "internal error: previously-checked
2256 referenced subpattern not found".
2257
2258 6. Both MS Visual Studio and Symbian OS have problems with initializing
2259 variables to point to external functions. For these systems, therefore,
2260 pcre_malloc etc. are now initialized to local functions that call the
2261 relevant global functions.
2262
2263 7. There were two entries missing in the vectors called coptable and poptable
2264 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2265 I've fixed the data, and added a kludgy way of testing at compile time that
2266 the lengths are correct (equal to the number of opcodes).
2267
2268 8. Following on from 7, I added a similar kludge to check the length of the
2269 eint vector in pcreposix.c.
2270
2271 9. Error texts for pcre_compile() are held as one long string to avoid too
2272 much relocation at load time. To find a text, the string is searched,
2273 counting zeros. There was no check for running off the end of the string,
2274 which could happen if a new error number was added without updating the
2275 string.
2276
2277 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2278
2279 11. \K was not working if it appeared in an atomic group or in a group that
2280 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2281 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2282 the assertion is positive, but not if it is negative.
2283
2284 12. Change 11 fortuitously reduced the size of the stack frame used in the
2285 "match()" function of pcre_exec.c by one pointer. Forthcoming
2286 implementation of support for (*MARK) will need an extra pointer on the
2287 stack; I have reserved it now, so that the stack frame size does not
2288 decrease.
2289
2290 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2291 item in branch that calls a recursion is a subroutine call - as in the
2292 second branch in the above example - was incorrectly given the compile-
2293 time error "recursive call could loop indefinitely" because pcre_compile()
2294 was not correctly checking the subroutine for matching a non-empty string.
2295
2296 14. The checks for overrunning compiling workspace could trigger after an
2297 overrun had occurred. This is a "should never occur" error, but it can be
2298 triggered by pathological patterns such as hundreds of nested parentheses.
2299 The checks now trigger 100 bytes before the end of the workspace.
2300
2301 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2302
2303
2304 Version 8.01 19-Jan-2010
2305 ------------------------
2306
2307 1. If a pattern contained a conditional subpattern with only one branch (in
2308 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2309 computed the wrong minimum data length (which is of course zero for such
2310 subpatterns). This could cause incorrect "no match" results.
2311
2312 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2313 the pattern is reset in the first branch, pcre_compile() failed with
2314 "internal error: code overflow at offset...". This happened only when
2315 the reset was to the original external option setting. (An optimization
2316 abstracts leading options settings into an external setting, which was the
2317 cause of this.)
2318
2319 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2320 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2321 assertion pattern did not match (meaning that the assertion was true), it
2322 was incorrectly treated as false if the SKIP had been reached during the
2323 matching. This also applied to assertions used as conditions.
2324
2325 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2326 assertion subpattern, including such a pattern used as a condition,
2327 unpredictable results occurred, instead of the error return
2328 PCRE_ERROR_DFA_UITEM.
2329
2330 5. The C++ GlobalReplace function was not working like Perl for the special
2331 situation when an empty string is matched. It now does the fancy magic
2332 stuff that is necessary.
2333
2334 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2335 removed. (These were left over from very, very early versions of PCRE.)
2336
2337 7. Some cosmetic changes to the code to make life easier when compiling it
2338 as part of something else:
2339
2340 (a) Change DEBUG to PCRE_DEBUG.
2341
2342 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2343 called "current" as "current_branch", to prevent a collision with the
2344 Linux macro when compiled as a kernel module.
2345
2346 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2347 prevent a collision with the Linux macro when compiled as a kernel
2348 module.
2349
2350 8. In pcre_compile() there are some checks for integer overflows that used to
2351 cast potentially large values to (double). This has been changed to that
2352 when building, a check for int64_t is made, and if it is found, it is used
2353 instead, thus avoiding the use of floating point arithmetic. (There is no
2354 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2355 double.
2356
2357 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2358 2005 (difference between two addresses compared to an unsigned value).
2359
2360 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2361 custom one, because of the following reported problem in Windows:
2362
2363 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2364 under Win32.
2365 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2366 therefore missing the function definition.
2367 - The compiler thus generates a "C" signature for the test function.
2368 - The linker fails to find the "C" function.
2369 - PCRE fails to configure if asked to do so against libbz2.
2370
2371 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2372 messages were output:
2373
2374 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2375 rerunning libtoolize, to keep the correct libtool macros in-tree.
2376 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2377
2378 I have done both of these things.
2379
2380 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2381 most of the time, it *can* run out if it is given a pattern that contains a
2382 runaway infinite recursion. I updated the discussion in the pcrestack man
2383 page.
2384
2385 13. Now that we have gone to the x.xx style of version numbers, the minor
2386 version may start with zero. Using 08 or 09 is a bad idea because users
2387 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2388 interpreted as invalid octal numbers. I've updated the previous comment in
2389 configure.ac, and also added a check that gives an error if 08 or 09 are
2390 used.
2391
2392 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2393 causing partial matching to fail when the end of the subject matched \W
2394 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2395
2396 15. There were some discrepancies between the declarations in pcre_internal.h
2397 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2398 their definitions. The declarations used "const uschar *" and the
2399 definitions used USPTR. Even though USPTR is normally defined as "const
2400 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2401 reported that: "This difference in casting confuses some C++ compilers, for
2402 example, SunCC recognizes above declarations as different functions and
2403 generates broken code for hbpcre." I have changed the declarations to use
2404 USPTR.
2405
2406 16. GNU libtool is named differently on some systems. The autogen.sh script now
2407 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2408 (FreeBSD).
2409
2410 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2411 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2412 comment: "Figure out how to create a longlong from a string: strtoll and
2413 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2414 instance, but it only takes 2 args instead of 3!"
2415
2416 18. A subtle bug concerned with back references has been fixed by a change of
2417 specification, with a corresponding code fix. A pattern such as
2418 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2419 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2420 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2421 same bug. Such groups have to be quantified to be useful, or contained
2422 inside another quantified group. (If there's no repetition, the reference
2423 can never match.) The problem arises because, having left the group and
2424 moved on to the rest of the pattern, a later failure that backtracks into
2425 the group uses the captured value from the final iteration of the group
2426 rather than the correct earlier one. I have fixed this in PCRE by forcing
2427 any group that contains a reference to itself to be an atomic group; that
2428 is, there cannot be any backtracking into it once it has completed. This is
2429 similar to recursive and subroutine calls.
2430
2431
2432 Version 8.00 19-Oct-09
2433 ----------------------
2434
2435 1. The table for translating pcre_compile() error codes into POSIX error codes
2436 was out-of-date, and there was no check on the pcre_compile() error code
2437 being within the table. This could lead to an OK return being given in
2438 error.
2439
2440 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2441 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2442 in a Windows environment.
2443
2444 3. The pcregrep --count option prints the count for each file even when it is
2445 zero, as does GNU grep. However, pcregrep was also printing all files when
2446 --files-with-matches was added. Now, when both options are given, it prints
2447 counts only for those files that have at least one match. (GNU grep just
2448 prints the file name in this circumstance, but including the count seems
2449 more useful - otherwise, why use --count?) Also ensured that the
2450 combination -clh just lists non-zero counts, with no names.
2451
2452 4. The long form of the pcregrep -F option was incorrectly implemented as
2453 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2454 but it seems right to fix it, and I didn't think it was worth preserving
2455 the old behaviour.
2456
2457 5. The command line items --regex=pattern and --regexp=pattern were not
2458 recognized by pcregrep, which required --regex pattern or --regexp pattern
2459 (with a space rather than an '='). The man page documented the '=' forms,
2460 which are compatible with GNU grep; these now work.
2461
2462 6. No libpcreposix.pc file was created for pkg-config; there was just
2463 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2464
2465 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2466 when UCP support is not needed, by modifying the Python script that
2467 generates it from Unicode data files. This should not matter if the module
2468 is correctly used as a library, but I received one complaint about 50K of
2469 unwanted data. My guess is that the person linked everything into his
2470 program rather than using a library. Anyway, it does no harm.
2471
2472 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2473 was a minimum greater than 1 for a wide character in a possessive
2474 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2475 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2476 character. Chaos in the form of incorrect output or a compiling loop could
2477 result.
2478
2479 9. The restrictions on what a pattern can contain when partial matching is
2480 requested for pcre_exec() have been removed. All patterns can now be
2481 partially matched by this function. In addition, if there are at least two
2482 slots in the offset vector, the offset of the earliest inspected character
2483 for the match and the offset of the end of the subject are set in them when
2484 PCRE_ERROR_PARTIAL is returned.
2485
2486 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2487 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2488 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2489 and may be more useful for multi-segment matching.
2490
2491 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2492 used to be given if ever the end of the subject was reached; now it is
2493 given only if matching could not proceed because another character was
2494 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2495 string "Z", which now yields "no match" instead of "partial match". In the
2496 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2497 final character ended with (*FAIL).
2498
2499 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2500 if the pattern had a "must contain" character that was already found in the
2501 earlier partial match, unless partial matching was again requested. For
2502 example, with the pattern /dog.(body)?/, the "must contain" character is
2503 "g". If the first part-match was for the string "dog", restarting with
2504 "sbody" failed. This bug has been fixed.
2505
2506 13. The string returned by pcre_dfa_exec() after a partial match has been
2507 changed so that it starts at the first inspected character rather than the
2508 first character of the match. This makes a difference only if the pattern
2509 starts with a lookbehind assertion or \b or \B (\K is not supported by
2510 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2511 matching functions compatible, and I think it's the right thing to do.
2512
2513 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2514 so that the demonstration program is easily available in environments where
2515 PCRE has not been installed from source.
2516
2517 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2518 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2519 library.
2520
2521 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2522 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2523 is not the first non-POSIX option to be added. Clearly some people find
2524 these options useful.
2525
2526 17. If a caller to the POSIX matching function regexec() passes a non-zero
2527 value for nmatch with a NULL value for pmatch, the value of
2528 nmatch is forced to zero.
2529
2530 18. RunGrepTest did not have a test for the availability of the -u option of
2531 the diff command, as RunTest does. It now checks in the same way as
2532 RunTest, and also checks for the -b option.
2533
2534 19. If an odd number of negated classes containing just a single character
2535 interposed, within parentheses, between a forward reference to a named
2536 subpattern and the definition of the subpattern, compilation crashed with
2537 an internal error, complaining that it could not find the referenced
2538 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2539 [The bug was that it was starting one character too far in when skipping
2540 over the character class, thus treating the ] as data rather than
2541 terminating the class. This meant it could skip too much.]
2542
2543 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2544 /g option in pcretest when the pattern contains \K, which makes it possible
2545 to have an empty string match not at the start, even when the pattern is
2546 anchored. Updated pcretest and pcredemo to use this option.
2547
2548 21. If the maximum number of capturing subpatterns in a recursion was greater
2549 than the maximum at the outer level, the higher number was returned, but
2550 with unset values at the outer level. The correct (outer level) value is
2551 now given.
2552
2553 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2554 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2555 make it do so. The string so far is captured, making this feature
2556 compatible with Perl.
2557
2558 23. The tests have been re-organized, adding tests 11 and 12, to make it
2559 possible to check the Perl 5.10 features against Perl 5.10.
2560
2561 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2562 pattern matches a fixed length string. PCRE did not allow this; now it
2563 does. Neither allows recursion.
2564
2565 25. I finally figured out how to implement a request to provide the minimum
2566 length of subject string that was needed in order to match a given pattern.
2567 (It was back references and recursion that I had previously got hung up
2568 on.) This code has now been added to pcre_study(); it finds a lower bound
2569 to the length of subject needed. It is not necessarily the greatest lower
2570 bound, but using it to avoid searching strings that are too short does give
2571 some useful speed-ups. The value is available to calling programs via
2572 pcre_fullinfo().
2573
2574 26. While implementing 25, I discovered to my embarrassment that pcretest had
2575 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2576 study optimizations had never been tested with that matching function.
2577 Oops. What is worse, even when it was passed study data, there was a bug in
2578 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2579 were also very few tests of studied patterns with pcre_dfa_exec().
2580
2581 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2582 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2583 on the other side of the coin, they are no longer allowed to have different
2584 names, because these cannot be distinguished in PCRE, and this has caused
2585 confusion. (This is a difference from Perl.)
2586
2587 28. When duplicate subpattern names are present (necessarily with different
2588 numbers, as required by 27 above), and a test is made by name in a
2589 conditional pattern, either for a subpattern having been matched, or for
2590 recursion in such a pattern, all the associated numbered subpatterns are
2591 tested, and the overall condition is true if the condition is true for any
2592 one of them. This is the way Perl works, and is also more like the way
2593 testing by number works.
2594
2595
2596 Version 7.9 11-Apr-09
2597 ---------------------
2598
2599 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2600 (pcretest), all targets were linked against these libraries. This included
2601 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2602 libraries. This caused unwanted dependencies to be created. This problem
2603 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2604 pcretest is linked with readline.
2605
2606 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2607 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2608 moved outside it again, because FALSE and TRUE are already defined in AIX,
2609 but BOOL is not.
2610
2611 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2612 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2613
2614 4. The pcregrep documentation said spaces were inserted as well as colons (or
2615 hyphens) following file names and line numbers when outputting matching
2616 lines. This is not true; no spaces are inserted. I have also clarified the
2617 wording for the --colour (or --color) option.
2618
2619 5. In pcregrep, when --colour was used with -o, the list of matching strings
2620 was not coloured; this is different to GNU grep, so I have changed it to be
2621 the same.
2622
2623 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2624 each matching line was coloured. Now it goes on to look for further matches
2625 of any of the test patterns, which is the same behaviour as GNU grep.
2626
2627 7. A pattern that could match an empty string could cause pcregrep to loop; it
2628 doesn't make sense to accept an empty string match in pcregrep, so I have
2629 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2630 seems to be how GNU grep behaves. [But see later change 40 for release
2631 8.33.]
2632
2633 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2634 start or after a newline", because the conditional assertion was not being
2635 correctly handled. The rule now is that both the assertion and what follows
2636 in the first alternative must satisfy the test.
2637
2638 9. If auto-callout was enabled in a pattern with a conditional group whose
2639 condition was an assertion, PCRE could crash during matching, both with
2640 pcre_exec() and pcre_dfa_exec().
2641
2642 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2643 used for matching.
2644
2645 11. Unicode property support in character classes was not working for
2646 characters (bytes) greater than 127 when not in UTF-8 mode.
2647
2648 12. Added the -M command line option to pcretest.
2649
2650 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2651
2652 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2653
2654 16. Added comments and documentation about mis-use of no_arg in the C++
2655 wrapper.
2656
2657 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2658 from Martin Jerabek that uses macro names for all relevant character and
2659 string constants.
2660
2661 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2662 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2663 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2664 these, but not everybody uses configure.
2665
2666 19. A conditional group that had only one branch was not being correctly
2667 recognized as an item that could match an empty string. This meant that an
2668 enclosing group might also not be so recognized, causing infinite looping
2669 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2670 with the subject "ab", where knowledge that the repeated group can match
2671 nothing is needed in order to break the loop.
2672
2673 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2674 exec(), but without supplying a callout function, matching went wrong.
2675
2676 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2677 leak if the size of the offset vector was greater than 30. When the vector
2678 is smaller, the saved offsets during recursion go onto a local stack
2679 vector, but for larger vectors malloc() is used. It was failing to free
2680 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2681 error, in fact).
2682
2683 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2684 heapframe that is used only when UTF-8 support is enabled. This caused no
2685 problem, but was untidy.
2686
2687 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2688 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2689 included within another project.
2690
2691 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2692 slightly modified by me:
2693
2694 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2695 not building pcregrep.
2696
2697 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2698 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2699
2700 25. Forward references, both numeric and by name, in patterns that made use of
2701 duplicate group numbers, could behave incorrectly or give incorrect errors,
2702 because when scanning forward to find the reference group, PCRE was not
2703 taking into account the duplicate group numbers. A pattern such as
2704 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2705
2706 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2707 the feature of a custom pointer more persuasive (as requested by a user).
2708
2709 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2710 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2711 pre-defined.
2712
2713 28. Added support for (*UTF8) at the start of a pattern.
2714
2715 29. Arrange for flags added by the "release type" setting in CMake to be shown
2716 in the configuration summary.
2717
2718
2719 Version 7.8 05-Sep-08
2720 ---------------------
2721
2722 1. Replaced UCP searching code with optimized version as implemented for Ad
2723 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2724 stage table and inline lookup instead of a function, giving speed ups of 2
2725 to 5 times on some simple patterns that I tested. Permission was given to
2726 distribute the MultiStage2.py script that generates the tables (it's not in
2727 the tarball, but is in the Subversion repository).
2728
2729 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2730 scripts.
2731
2732 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2733 a group with a zero qualifier. The result of the study could be incorrect,
2734 or the function might crash, depending on the pattern.
2735
2736 4. Caseless matching was not working for non-ASCII characters in back
2737 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2738 It now works when Unicode Property Support is available.
2739
2740 5. In pcretest, an escape such as \x{de} in the data was always generating
2741 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2742 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2743 truncation.
2744
2745 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2746
2747 7. Added two (int) casts to pcregrep when printing the difference of two
2748 pointers, in case they are 64-bit values.
2749
2750 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2751 test 2 if it fails.
2752
2753 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2754 and a #define of that name to empty if it is not externally set. This is to
2755 allow users of MSVC to set it if necessary.
2756
2757 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2758 the convenience functions in the pcre_get.c source file.
2759
2760 11. An option change at the start of a pattern that had top-level alternatives
2761 could cause overwriting and/or a crash. This command provoked a crash in
2762 some environments:
2763
2764 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2765
2766 This potential security problem was recorded as CVE-2008-2371.
2767
2768 12. For a pattern where the match had to start at the beginning or immediately
2769 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2770 pcre_dfa_exec() could read past the end of the passed subject if there was
2771 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2772 pcretest so that it places the subject at the end of its malloc-ed buffer.
2773
2774 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2775 exec() might read past the end of the data buffer in UTF-8 mode.
2776
2777 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2778 the data contained the byte 0x85 as part of a UTF-8 character within its
2779 first line. This applied both to normal and DFA matching.
2780
2781 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2782 /^[^d]*?$/8 failed to match "abc".
2783
2784 16. Added a missing copyright notice to pcrecpp_internal.h.
2785
2786 17. Make it more clear in the documentation that values returned from
2787 pcre_exec() in ovector are byte offsets, not character counts.
2788
2789 18. Tidied a few places to stop certain compilers from issuing warnings.
2790
2791 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2792 supplied by Stefan Weber. I made a further small update for 7.8 because
2793 there is a change of source arrangements: the pcre_searchfuncs.c module is
2794 replaced by pcre_ucd.c.
2795
2796
2797 Version 7.7 07-May-08
2798 ---------------------
2799
2800 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2801 a string to a long long, pretend we don't even have a long long." This is
2802 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2803
2804 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2805 pre-7.6 versions, which defined a global no_arg variable instead of putting
2806 it in the RE class. (See also #8 below.)
2807
2808 3. Remove a line of dead code, identified by coverity and reported by Nuno
2809 Lopes.
2810
2811 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2812
2813 (1) The include/exclude patterns were being applied to the whole pathnames
2814 of files, instead of just to the final components.
2815
2816 (2) If there was more than one level of directory, the subdirectories were
2817 skipped unless they satisfied the include/exclude conditions. This is
2818 inconsistent with GNU grep (and could even be seen as contrary to the
2819 pcregrep specification - which I improved to make it absolutely clear).
2820 The action now is always to scan all levels of directory, and just
2821 apply the include/exclude patterns to regular files.
2822
2823 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2824 --exclude_dir in the tests to avoid scanning .svn directories.
2825
2826 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2827 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2828 doesn't support NULs in patterns.
2829
2830 7. Added some missing "const"s to declarations of static tables in
2831 pcre_compile.c and pcre_dfa_exec.c.
2832
2833 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2834 caused by fix #2 above. (Subsequently also a second patch to fix the
2835 first patch. And a third patch - this was a messy problem.)
2836
2837 9. Applied Craig's patch to remove the use of push_back().
2838
2839 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2840 matching function regexec().
2841
2842 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2843 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2844 references. PCRE supports relative numbers with this syntax (I don't think
2845 Oniguruma does).
2846
2847 12. Previously, a group with a zero repeat such as (...){0} was completely
2848 omitted from the compiled regex. However, this means that if the group
2849 was called as a subroutine from elsewhere in the pattern, things went wrong
2850 (an internal error was given). Such groups are now left in the compiled
2851 pattern, with a new opcode that causes them to be skipped at execution
2852 time.
2853
2854 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2855 to the way PCRE behaves:
2856
2857 (a) A lone ] character is dis-allowed (Perl treats it as data).
2858
2859 (b) A back reference to an unmatched subpattern matches an empty string
2860 (Perl fails the current match path).
2861
2862 (c) A data ] in a character class must be notated as \] because if the
2863 first data character in a class is ], it defines an empty class. (In
2864 Perl it is not possible to have an empty class.) The empty class []
2865 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2866 The negative empty class [^] matches any one character, independently
2867 of the DOTALL setting.
2868
2869 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2870 non-existent subpattern following a character class starting with ']' and
2871 containing () gave an internal compiling error instead of "reference to
2872 non-existent subpattern". Fortunately, when the pattern did exist, the
2873 compiled code was correct. (When scanning forwards to check for the
2874 existence of the subpattern, it was treating the data ']' as terminating
2875 the class, so got the count wrong. When actually compiling, the reference
2876 was subsequently set up correctly.)
2877
2878 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2879 it was being rejected as not supported by pcre_dfa_exec(), even though
2880 other assertions are supported. I have made pcre_dfa_exec() support
2881 (*FAIL).
2882
2883 16. The implementation of 13c above involved the invention of a new opcode,
2884 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2885 cannot be changed at match time, I realized I could make a small
2886 improvement to matching performance by compiling OP_ALLANY instead of
2887 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2888 on the OP_ANY path.
2889
2890 17. Compiling pcretest on Windows with readline support failed without the
2891 following two fixes: (1) Make the unistd.h include conditional on
2892 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2893
2894 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2895 ncurses library to be included for pcretest when ReadLine support is
2896 requested, but also to allow for it to be overridden. This patch came from
2897 Daniel Bergström.
2898
2899 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2900 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2901 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2902 spotting this.
2903
2904
2905 Version 7.6 28-Jan-08
2906 ---------------------
2907
2908 1. A character class containing a very large number of characters with
2909 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2910 overflow.
2911
2912 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2913 HAVE_LONG_LONG is not defined.
2914
2915 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2916 bring them up to date and include new features. This patch includes:
2917
2918 - Fixed PH's badly added libz and libbz2 support.
2919 - Fixed a problem with static linking.
2920 - Added pcredemo. [But later removed - see 7 below.]
2921 - Fixed dftables problem and added an option.
2922 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2923 HAVE_LONG_LONG.
2924 - Added readline support for pcretest.
2925 - Added an listing of the option settings after cmake has run.
2926
2927 4. A user submitted a patch to Makefile that makes it easy to create
2928 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2929 Makefile.am that cause it to include this special target, without
2930 affecting anything else. Note that the same mingw target plus all
2931 the other distribution libraries and programs are now supported
2932 when configuring with CMake (see 6 below) instead of with
2933 Configure/Make.
2934
2935 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2936 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2937 exported in the Windows port". It has not yet been confirmed that the patch
2938 solves the problem, but it does no harm.
2939
2940 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2941 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2942 with CMake, and also correct the comment about stack recursion.
2943
2944 7. Remove the automatic building of pcredemo from the ./configure system and
2945 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2946 of a program that users should build themselves after PCRE is installed, so
2947 building it automatically is not really right. What is more, it gave
2948 trouble in some build environments.
2949
2950 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2951
2952
2953 Version 7.5 10-Jan-08
2954 ---------------------
2955
2956 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2957 values in parens when parsing an RE using the C++ wrapper."
2958
2959 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2960 Characters greater than 255 were excluded from the class instead of being
2961 included.
2962
2963 3. The same bug as (2) above applied to negated POSIX classes such as
2964 [:^space:].
2965
2966 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2967 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2968 I have changed it.
2969
2970 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2971 first named subpattern) and a construct such as (?&a) would reference the
2972 first named subpattern whose name started with "a" (in other words, the
2973 length check was missing). Both these problems are fixed. "Subpattern name
2974 expected" is now given for (?&) (a zero-length name), and this patch also
2975 makes it give the same error for \k'' (previously it complained that that
2976 was a reference to a non-existent subpattern).
2977
2978 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2979 this is right because (?- can be followed by option settings as well as by
2980 digits. I have, however, made the messages clearer.
2981
2982 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2983 than the number used in the conditional) now cause a compile-time error.
2984 This is actually not compatible with Perl, which accepts such patterns, but
2985 treats the conditional as always being FALSE (as PCRE used to), but it
2986 seems to me that giving a diagnostic is better.
2987
2988 8. Change "alphameric" to the more common word "alphanumeric" in comments
2989 and messages.
2990
2991 9. Fix two occurrences of "backslash" in comments that should have been
2992 "backspace".
2993
2994 10. Remove two redundant lines of code that can never be obeyed (their function
2995 was moved elsewhere).
2996
2997 11. The program that makes PCRE's Unicode character property table had a bug
2998 which caused it to generate incorrect table entries for sequences of
2999 characters that have the same character type, but are in different scripts.
3000 It amalgamated them into a single range, with the script of the first of
3001 them. In other words, some characters were in the wrong script. There were
3002 thirteen such cases, affecting characters in the following ranges:
3003
3004 U+002b0 - U+002c1
3005 U+0060c - U+0060d
3006 U+0061e - U+00612
3007 U+0064b - U+0065e
3008 U+0074d - U+0076d
3009 U+01800 - U+01805
3010 U+01d00 - U+01d77
3011 U+01d9b - U+01dbf
3012 U+0200b - U+0200f
3013 U+030fc - U+030fe
3014 U+03260 - U+0327f
3015 U+0fb46 - U+0fbb1
3016 U+10450 - U+1049d
3017
3018 12. The -o option (show only the matching part of a line) for pcregrep was not
3019 compatible with GNU grep in that, if there was more than one match in a
3020 line, it showed only the first of them. It now behaves in the same way as
3021 GNU grep.
3022
3023 13. If the -o and -v options were combined for pcregrep, it printed a blank
3024 line for every non-matching line. GNU grep prints nothing, and pcregrep now
3025 does the same. The return code can be used to tell if there were any
3026 non-matching lines.
3027
3028 14. Added --file-offsets and --line-offsets to pcregrep.
3029
3030 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
3031 infinitely looping recursion. The bug was that positive lookaheads were not
3032 being skipped when checking for a possible empty match (negative lookaheads
3033 and both kinds of lookbehind were skipped).
3034
3035 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
3036 inclusion of <windows.h> to before rather than after the definition of
3037 INVALID_FILE_ATTRIBUTES (patch from David Byron).
3038
3039 17. Specifying a possessive quantifier with a specific limit for a Unicode
3040 character property caused pcre_compile() to compile bad code, which led at
3041 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
3042 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
3043 caused the error; without that there was no problem.
3044
3045 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3046
3047 19. Added --enable-pcretest-libreadline.
3048
3049 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3050 RE::GlobalReplace(). As a result, the number of replacements returned was
3051 double what it should be. I removed one of the increments, but Craig sent a
3052 later patch that removed the other one (the right fix) and added unit tests
3053 that check the return values (which was not done before).
3054
3055 21. Several CMake things:
3056
3057 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3058 the names libpcre and libpcreposix, not just pcre and pcreposix.
3059
3060 (2) The above change means that pcretest and pcregrep are now correctly
3061 linked with the newly-built libraries, not previously installed ones.
3062
3063 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
3064
3065 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3066 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3067 UTF-8 newline character). The key issue is that the pattern starts .*;
3068 this means that the match must be either at the beginning, or after a
3069 newline. The bug was in the code for advancing after a failed match and
3070 checking that the new position followed a newline. It was not taking
3071 account of UTF-8 characters correctly.
3072
3073 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3074 character classes. PCRE was not treating the sequence [:...:] as a
3075 character class unless the ... were all letters. Perl, however, seems to
3076 allow any characters between [: and :], though of course it rejects as
3077 unknown any "names" that contain non-letters, because all the known class
3078 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3079 for example, whereas PCRE did not - it did not recognize a POSIX character
3080 class. This seemed a bit dangerous, so the code has been changed to be
3081 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3082 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3083 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3084 Perl does, and where it didn't before.
3085
3086 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3087 Windows environments %n is disabled by default.
3088
3089
3090 Version 7.4 21-Sep-07
3091 ---------------------
3092
3093 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3094 means that a class such as [\s] counted as "explicit reference to CR or
3095 LF". That isn't really right - the whole point of the change was to try to
3096 help when there was an actual mention of one of the two characters. So now
3097 the change happens only if \r or \n (or a literal CR or LF) character is
3098 encountered.
3099
3100 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3101 of both had grown to the point where there were only 3 bits left.
3102 Fortunately, there was spare space in the data structure, and so I have
3103 moved the internal flags into a new 16-bit field to free up more option
3104 bits.
3105
3106 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3107 but did not set the internal JCHANGED flag - either of these is enough to
3108 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3109 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3110 start sets both bits.
3111
3112 4. Added options (at build time, compile time, exec time) to change \R from
3113 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3114
3115 5. doc/pcresyntax.html was missing from the distribution.
3116
3117 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3118 compatibility, even though it is no longer used.
3119
3120 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3121 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3122 windows.h file is present (where different names are used). [This was
3123 reversed later after testing - see 16 below.]
3124
3125 8. Changed all #include <config.h> to #include "config.h". There were also
3126 some further <pcre.h> cases that I changed to "pcre.h".
3127
3128 9. When pcregrep was used with the --colour option, it missed the line ending
3129 sequence off the lines that it output.
3130
3131 10. It was pointed out to me that arrays of string pointers cause lots of
3132 relocations when a shared library is dynamically loaded. A technique of
3133 using a single long string with a table of offsets can drastically reduce
3134 these. I have refactored PCRE in four places to do this. The result is
3135 dramatic:
3136
3137 Originally: 290
3138 After changing UCP table: 187
3139 After changing error message table: 43
3140 After changing table of "verbs" 36
3141 After changing table of Posix names 22
3142
3143 Thanks to the folks working on Gregex for glib for this insight.
3144
3145 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3146 unicode-properties was also set.
3147
3148 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3149
3150 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3151 checked only for CRLF.
3152
3153 14. Added casts to pcretest.c to avoid compiler warnings.
3154
3155 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3156
3157 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3158 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3159 entirely. This removes changes made in 7 above.
3160
3161 17. The CMake files have been updated, and there is now more information about
3162 building with CMake in the NON-UNIX-USE document.
3163
3164
3165 Version 7.3 28-Aug-07
3166 ---------------------
3167
3168 1. In the rejigging of the build system that eventually resulted in 7.1, the
3169 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3170 brackets there is not right, since it causes compilers to look for an
3171 installed pcre.h, not the version that is in the source that is being
3172 compiled (which of course may be different). I have changed it back to:
3173
3174 #include "pcre.h"
3175
3176 I have a vague recollection that the change was concerned with compiling in
3177 different directories, but in the new build system, that is taken care of
3178 by the VPATH setting the Makefile.
3179
3180 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3181 when the subject happened to end in the byte 0x85 (e.g. if the last
3182 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3183 characters but of course it shouldn't be taken as a newline when it is part
3184 of another character. The bug was that, for an unlimited repeat of . in
3185 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3186 characters when looking for a newline.
3187
3188 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3189
3190 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3191 in debug output.
3192
3193 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3194 long printing in the pcrecpp unittest when running under MinGW.
3195
3196 6. ESC_K was left out of the EBCDIC table.
3197
3198 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3199 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3200 limit also applies to "virtual nesting" when a pattern is recursive, and in
3201 this case 1000 isn't so big. I have been able to remove this limit at the
3202 expense of backing off one optimization in certain circumstances. Normally,
3203 when pcre_exec() would call its internal match() function recursively and
3204 immediately return the result unconditionally, it uses a "tail recursion"
3205 feature to save stack. However, when a subpattern that can match an empty
3206 string has an unlimited repetition quantifier, it no longer makes this
3207 optimization. That gives it a stack frame in which to save the data for
3208 checking that an empty string has been matched. Previously this was taken
3209 from the 1000-entry workspace that had been reserved. So now there is no
3210 explicit limit, but more stack is used.
3211
3212 8. Applied Daniel's patches to solve problems with the import/export magic
3213 syntax that is required for Windows, and which was going wrong for the
3214 pcreposix and pcrecpp parts of the library. These were overlooked when this
3215 problem was solved for the main library.
3216
3217 9. There were some crude static tests to avoid integer overflow when computing
3218 the size of patterns that contain repeated groups with explicit upper
3219 limits. As the maximum quantifier is 65535, the maximum group length was
3220 set at 30,000 so that the product of these two numbers did not overflow a
3221 32-bit integer. However, it turns out that people want to use groups that
3222 are longer than 30,000 bytes (though not repeat them that many times).
3223 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3224 made it possible to implement the integer overflow checks in a much more
3225 dynamic way, which I have now done. The artificial limitation on group
3226 length has been removed - we now have only the limit on the total length of
3227 the compiled pattern, which depends on the LINK_SIZE setting.
3228
3229 10. Fixed a bug in the documentation for get/copy named substring when
3230 duplicate names are permitted. If none of the named substrings are set, the
3231 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3232 empty string.
3233
3234 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3235 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3236 because the ] is interpreted as the first data character and the
3237 terminating ] is not found. PCRE has been made compatible with Perl in this
3238 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3239 cause memory overwriting.
3240
3241 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3242 string has been matched (to stop an infinite loop). It was not recognizing
3243 a conditional subpattern that could match an empty string if that
3244 subpattern was within another subpattern. For example, it looped when
3245 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3246 condition was not nested. This bug has been fixed.
3247
3248 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3249 past the start of the subject in the presence of bytes with the top bit
3250 set, for example "\x8aBCD".
3251
3252 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3253 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3254
3255 14. Optimized (?!) to (*FAIL).
3256
3257 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3258 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3259 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3260 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3261 does: it's just the validity check that is more restrictive.
3262
3263 16. Inserted checks for integer overflows during escape sequence (backslash)
3264 processing, and also fixed erroneous offset values for syntax errors during
3265 backslash processing.
3266
3267 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3268 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3269
3270 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3271 caused an overrun.
3272
3273 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3274 something other than just ASCII characters) inside a group that had an
3275 unlimited repeat caused a loop at compile time (while checking to see
3276 whether the group could match an empty string).
3277
3278 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3279 [\P{Any}] did so. (Error in the code for printing property names.)
3280
3281 21. An orphan \E inside a character class could cause a crash.
3282
3283 22. A repeated capturing bracket such as (A)? could cause a wild memory
3284 reference during compilation.
3285
3286 23. There are several functions in pcre_compile() that scan along a compiled
3287 expression for various reasons (e.g. to see if it's fixed length for look
3288 behind). There were bugs in these functions when a repeated \p or \P was
3289 present in the pattern. These operators have additional parameters compared
3290 with \d, etc, and these were not being taken into account when moving along
3291 the compiled data. Specifically:
3292
3293 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3294 length.
3295
3296 (b) An item such as \pL+ within a repeated group could cause crashes or
3297 loops.
3298
3299 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3300 "reference to non-existent subpattern" error.
3301
3302 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3303
3304 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3305 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3306
3307 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3308
3309 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3310 character were causing crashes (broken optimization).
3311
3312 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3313 \p or \P) caused a compile-time loop.
3314
3315 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3316 break. For example, the unstudied pattern [\r\n]A does not match the string
3317 "\r\nA" because change 7.0/46 below moves the current point on by two
3318 characters after failing to match at the start. However, the pattern \nA
3319 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3320 the same is true. There doesn't seem any very clean way out of this, but
3321 what I have chosen to do makes the common cases work: PCRE now takes note
3322 of whether there can be an explicit match for \r or \n anywhere in the
3323 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3324 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3325 pattern has explicit CR or LF references.
3326
3327 29. Added (*CR) etc for changing newline setting at start of pattern.
3328
3329
3330 Version 7.2 19-Jun-07
3331 ---------------------
3332
3333 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3334 which is apparently normally available under Windows.
3335
3336 2. Re-jig the pcregrep tests with different newline settings in an attempt
3337 to make them independent of the local environment's newline setting.
3338
3339 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3340
3341 4. Some of the "internals" tests were previously cut out when the link size
3342 was not 2, because the output contained actual offsets. The recent new
3343 "Z" feature of pcretest means that these can be cut out, making the tests
3344 usable with all link sizes.
3345
3346 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3347 stack recursion. This gives a massive performance boost under BSD, but just
3348 a small improvement under Linux. However, it saves one field in the frame
3349 in all cases.
3350
3351 6. Added more features from the forthcoming Perl 5.10:
3352
3353 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3354 recursion call. It refers to the nth most recently opened parentheses.
3355
3356 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3357 to be opened parentheses.
3358
3359 (c) Conditions that refer to capturing parentheses can be specified
3360 relatively, for example, (?(-2)... or (?(+3)...
3361
3362 (d) \K resets the start of the current match so that everything before
3363 is not part of it.
3364
3365 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3366
3367 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3368 reference syntax.
3369
3370 (g) (?| introduces a group in which the numbering of parentheses in each
3371 alternative starts with the same number.
3372
3373 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3374
3375 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3376 PCRE_INFO_JCHANGED.
3377
3378 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3379 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3380 for detecting groups that can match an empty string.
3381
3382 9. A pattern with a very large number of alternatives (more than several
3383 hundred) was running out of internal workspace during the pre-compile
3384 phase, where pcre_compile() figures out how much memory will be needed. A
3385 bit of new cunning has reduced the workspace needed for groups with
3386 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3387 workspace instead of running out of the 4096 that are available.
3388
3389 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3390
3391 11. Applied patch from Google to remove an optimization that didn't quite work.
3392 The report of the bug said:
3393
3394 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3395 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3396 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3397
3398 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3399 it matched the wrong number of bytes.
3400
3401
3402 Version 7.1 24-Apr-07
3403 ---------------------
3404
3405 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3406 that is more "standard", making use of automake and other Autotools. There
3407 is some re-arrangement of the files and adjustment of comments consequent
3408 on this.
3409
3410 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3411 for recursive directory scanning broke on some systems because the files
3412 are not scanned in any specific order and on different systems the order
3413 was different. A call to "sort" has been inserted into RunGrepTest for the
3414 approprate test as a short-term fix. In the longer term there may be an
3415 alternative.
3416
3417 3. I had an email from Eric Raymond about problems translating some of PCRE's
3418 man pages to HTML (despite the fact that I distribute HTML pages, some
3419 people do their own conversions for various reasons). The problems
3420 concerned the use of low-level troff macros .br and .in. I have therefore
3421 removed all such uses from the man pages (some were redundant, some could
3422 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3423 HTML has been updated to handle .nf/.fi and to complain if it encounters
3424 .br or .in.
3425
3426 4. Updated comments in configure.ac that get placed in config.h.in and also
3427 arranged for config.h to be included in the distribution, with the name
3428 config.h.generic, for the benefit of those who have to compile without
3429 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3430
3431 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3432 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3433 makevp.bat for the current PCRE, using the additional files
3434 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3435
3436 6. A Windows user reported a minor discrepancy with test 2, which turned out
3437 to be caused by a trailing space on an input line that had got lost in his
3438 copy. The trailing space was an accident, so I've just removed it.
3439
3440 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3441 that is needed.
3442
3443 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3444 as "const" (a) because they are and (b) because it helps the PHP
3445 maintainers who have recently made a script to detect big data structures
3446 in the php code that should be moved to the .rodata section. I remembered
3447 to update Builducptable as well, so it won't revert if ucptable.h is ever
3448 re-created.
3449
3450 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3451 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3452 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3453 support is not required. This saves 1.5-2K of code, which is important in
3454 some applications.
3455
3456 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3457 so as not to refer to the tables, even though these functions will never be
3458 called when UTF-8 support is disabled. Otherwise there are problems with a
3459 shared library.
3460
3461 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3462
3463 (a) It was defining its arguments as char * instead of void *.
3464
3465 (b) It was assuming that all moves were upwards in memory; this was true
3466 a long time ago when I wrote it, but is no longer the case.
3467
3468 The emulated memove() is provided for those environments that have neither
3469 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3470 is clearly not the case, as these two bugs were recently reported.
3471
3472 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3473 and Detrail to create the HTML documentation, the .txt form of the man
3474 pages, and it removes trailing spaces from listed files. It also creates
3475 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3476 case, it wraps all the #defines with #ifndefs. This script should be run
3477 before "make dist".
3478
3479 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3480 with Unicode property support.
3481
3482 (a) For a maximizing quantifier, if the two different cases of the
3483 character were of different lengths in their UTF-8 codings (there are
3484 some cases like this - I found 11), and the matching function had to
3485 back up over a mixture of the two cases, it incorrectly assumed they
3486 were both the same length.
3487
3488 (b) When PCRE was configured to use the heap rather than the stack for
3489 recursion during matching, it was not correctly preserving the data for
3490 the other case of a UTF-8 character when checking ahead for a match
3491 while processing a minimizing repeat. If the check also involved
3492 matching a wide character, but failed, corruption could cause an
3493 erroneous result when trying to check for a repeat of the original
3494 character.
3495
3496 13. Some tidying changes to the testing mechanism:
3497
3498 (a) The RunTest script now detects the internal link size and whether there
3499 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3500 values substituted by "configure". (The RunGrepTest script already did
3501 this for UTF-8.) The configure.ac script no longer substitutes the
3502 relevant variables.
3503
3504 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3505 with length and offset values. This means that the output is different
3506 for different internal link sizes. Test 2 is skipped for link sizes
3507 other than 2 because of this, bypassing the problem. Unfortunately,
3508 there was also a test in test 3 (the locale tests) that used /B and
3509 failed for link sizes other than 2. Rather than cut the whole test out,
3510 I have added a new /Z option to pcretest that replaces the length and
3511 offset values with spaces. This is now used to make test 3 independent
3512 of link size. (Test 2 will be tidied up later.)
3513
3514 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3515 segmentation fault instead of returning the appropriate error message.
3516
3517 15. In multiline mode when the newline sequence was set to "any", the pattern
3518 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3519 This doesn't seem right; it now treats the CRLF combination as the line
3520 ending, and so does not match in that case. It's only a pattern such as ^$
3521 that would hit this one: something like ^ABC$ would have failed after \r
3522 and then tried again after \r\n.
3523
3524 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3525 in an attempt to make files that differ only in their line terminators
3526 compare equal. This works on Linux.
3527
3528 17. Under certain error circumstances pcregrep might try to free random memory
3529 as it exited. This is now fixed, thanks to valgrind.
3530
3531 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3532 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3533 was because its rules for how to advance for /g after matching an empty
3534 string at the end of a line did not allow for this case. They now check for
3535 it specially.
3536
3537 20. pcretest is supposed to handle patterns and data of any length, by
3538 extending its buffers when necessary. It was getting this wrong when the
3539 buffer for a data line had to be extended.
3540
3541 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3542 CRLF as a newline sequence.
3543
3544 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3545 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3546 I have nevertheless tidied it up.
3547
3548 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3549
3550 24. Added a man page for pcre-config.
3551
3552
3553 Version 7.0 19-Dec-06
3554 ---------------------
3555
3556 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3557 moving to gcc 4.1.1.
3558
3559 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3560 sys/time.h, which is documented as needed for this function. It doesn't
3561 seem to matter on Linux, but it showed up on some releases of OS X.
3562
3563 3. It seems that there are systems where bytes whose values are greater than
3564 127 match isprint() in the "C" locale. The "C" locale should be the
3565 default when a C program starts up. In most systems, only ASCII printing
3566 characters match isprint(). This difference caused the output from pcretest
3567 to vary, making some of the tests fail. I have changed pcretest so that:
3568
3569 (a) When it is outputting text in the compiled version of a pattern, bytes
3570 other than 32-126 are always shown as hex escapes.
3571
3572 (b) When it is outputting text that is a matched part of a subject string,
3573 it does the same, unless a different locale has been set for the match
3574 (using the /L modifier). In this case, it uses isprint() to decide.
3575
3576 4. Fixed a major bug that caused incorrect computation of the amount of memory
3577 required for a compiled pattern when options that changed within the
3578 pattern affected the logic of the preliminary scan that determines the
3579 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3580 that the computed length was too small. The symptoms of this bug were
3581 either the PCRE error "internal error: code overflow" from pcre_compile(),
3582 or a glibc crash with a message such as "pcretest: free(): invalid next
3583 size (fast)". Examples of patterns that provoked this bug (shown in
3584 pcretest format) are:
3585
3586 /(?-x: )/x
3587 /(?x)(?-x: \s*#\s*)/
3588 /((?i)[\x{c0}])/8
3589 /(?i:[\x{c0}])/8
3590
3591 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3592 is now done differently.
3593
3594 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3595 wrapper classes; (b) implement a new function in the C++ scanner that is
3596 more efficient than the old way of doing things because it avoids levels of
3597 recursion in the regex matching; (c) add a paragraph to the documentation
3598 for the FullMatch() function.
3599
3600 6. The escape sequence \n was being treated as whatever was defined as
3601 "newline". Not only was this contrary to the documentation, which states
3602 that \n is character 10 (hex 0A), but it also went horribly wrong when
3603 "newline" was defined as CRLF. This has been fixed.
3604
3605 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3606 was being set to -1 for the "end of line" case (supposedly a value that no
3607 character can have). Though this value is never used (the check for end of
3608 line is "zero bytes in current character"), it caused compiler complaints.
3609 I've changed it to 0xffffffff.
3610
3611 8. In pcre_version.c, the version string was being built by a sequence of
3612 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3613 string (as it is for production releases) called a macro with an empty
3614 argument. The C standard says the result of this is undefined. The gcc
3615 compiler treats it as an empty string (which was what was wanted) but it is
3616 reported that Visual C gives an error. The source has been hacked around to
3617 avoid this problem.
3618
3619 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3620 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3621 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3622 of them did).
3623
3624 10. Originally, pcretest opened its input and output without "b"; then I was
3625 told that "b" was needed in some environments, so it was added for release
3626 5.0 to both the input and output. (It makes no difference on Unix-like
3627 systems.) Later I was told that it is wrong for the input on Windows. I've
3628 now abstracted the modes into two macros, to make it easier to fiddle with
3629 them, and removed "b" from the input mode under Windows.
3630
3631 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3632
3633 12. Added -help and --help to pcretest as an official way of being reminded
3634 of the options.
3635
3636 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3637 and pcrecpp.cc because they annoy compilers at high warning levels.
3638
3639 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3640
3641 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3642 scripts are not C programs :-) and which was not noticed because it works
3643 on Linux.
3644
3645 16. pcretest is supposed to handle any length of pattern and data line (as one
3646 line or as a continued sequence of lines) by extending its input buffer if
3647 necessary. This feature was broken for very long pattern lines, leading to
3648 a string of junk being passed to pcre_compile() if the pattern was longer
3649 than about 50K.
3650
3651 17. I have done a major re-factoring of the way pcre_compile() computes the
3652 amount of memory needed for a compiled pattern. Previously, there was code
3653 that made a preliminary scan of the pattern in order to do this. That was
3654 OK when PCRE was new, but as the facilities have expanded, it has become
3655 harder and harder to keep it in step with the real compile phase, and there
3656 have been a number of bugs (see for example, 4 above). I have now found a
3657 cunning way of running the real compile function in a "fake" mode that
3658 enables it to compute how much memory it would need, while actually only
3659 ever using a few hundred bytes of working memory and without too many
3660 tests of the mode. This should make future maintenance and development
3661 easier. A side effect of this work is that the limit of 200 on the nesting
3662 depth of parentheses has been removed (though this was never a serious
3663 limitation, I suspect). However, there is a downside: pcre_compile() now
3664 runs more slowly than before (30% or more, depending on the pattern). I
3665 hope this isn't a big issue. There is no effect on runtime performance.
3666
3667 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3668 newline (only possible for the last line of a file) and it was a
3669 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3670
3671 19. Added additional timing features to pcretest. (1) The -tm option now times
3672 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3673 separate command line item, by a number that specifies the number of
3674 repeats to use when timing. The default is 50000; this gives better
3675 precision, but takes uncomfortably long for very large patterns.
3676
3677 20. Extended pcre_study() to be more clever in cases where a branch of a
3678 subpattern has no definite first character. For example, (a*|b*)[cd] would
3679 previously give no result from pcre_study(). Now it recognizes that the
3680 first character must be a, b, c, or d.
3681
3682 21. There was an incorrect error "recursive call could loop indefinitely" if
3683 a subpattern (or the entire pattern) that was being tested for matching an
3684 empty string contained only one non-empty item after a nested subpattern.
3685 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3686 incorrectly, because the \d was being skipped in the check.
3687
3688 22. The pcretest program now has a new pattern option /B and a command line
3689 option -b, which is equivalent to adding /B to every pattern. This causes
3690 it to show the compiled bytecode, without the additional information that
3691 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3692 is the same as /B/I).
3693
3694 23. A new optimization is now able automatically to treat some sequences such
3695 as a*b as a*+b. More specifically, if something simple (such as a character
3696 or a simple class like \d) has an unlimited quantifier, and is followed by
3697 something that cannot possibly match the quantified thing, the quantifier
3698 is automatically "possessified".
3699
3700 24. A recursive reference to a subpattern whose number was greater than 39
3701 went wrong under certain circumstances in UTF-8 mode. This bug could also
3702 have affected the operation of pcre_study().
3703
3704 25. Realized that a little bit of performance could be had by replacing
3705 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3706
3707 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3708
3709 27. Possessive quantifiers such as a++ were previously implemented by turning
3710 them into atomic groups such as ($>a+). Now they have their own opcodes,
3711 which improves performance. This includes the automatically created ones
3712 from 23 above.
3713
3714 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3715 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3716 the first matched character to be a colon. This applied both to named and
3717 numbered groups.
3718
3719 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3720
3721 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3722 building PCRE on Windows easier, so I have included it in the distribution.
3723
3724 31. There is now a check in pcretest against a ridiculously large number being
3725 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3726 loop, the loop is abandoned.
3727
3728 32. Forward references to subpatterns in conditions such as (?(2)...) where
3729 subpattern 2 is defined later cause pcre_compile() to search forwards in
3730 the pattern for the relevant set of parentheses. This search went wrong
3731 when there were unescaped parentheses in a character class, parentheses
3732 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3733
3734 33. "Subroutine" calls and backreferences were previously restricted to
3735 referencing subpatterns earlier in the regex. This restriction has now
3736 been removed.
3737
3738 34. Added a number of extra features that are going to be in Perl 5.10. On the
3739 whole, these are just syntactic alternatives for features that PCRE had
3740 previously implemented using the Python syntax or my own invention. The
3741 other formats are all retained for compatibility.
3742
3743 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3744 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3745 also .NET compatible.
3746
3747 (b) A recursion or subroutine call to a named group can now be defined as
3748 (?&name) as well as (?P>name).
3749
3750 (c) A backreference to a named group can now be defined as \k<name> or
3751 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3752 5.10, are also .NET compatible.
3753
3754 (d) A conditional reference to a named group can now use the syntax
3755 (?(<name>) or (?('name') as well as (?(name).
3756
3757 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3758 groups (named and numbered) that are never evaluated inline, but can be
3759 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3760 is always false. There may be only one alternative in such a group.
3761
3762 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3763 as the simple (?(R). The condition is true only if the most recent
3764 recursion is that of the given number or name. It does not search out
3765 through the entire recursion stack.
3766
3767 (g) The escape \gN or \g{N} has been added, where N is a positive or
3768 negative number, specifying an absolute or relative reference.
3769
3770 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3771 some "unreachable code" warnings.
3772
3773 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3774 things, this adds five new scripts.
3775
3776 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3777 There were also incompatibilities regarding the handling of \Q..\E inside
3778 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3779 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3780
3781 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3782 matches an empty string, and forcibly breaks the loop. There were bugs in
3783 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3784 against aaaa the result was just "a" rather than "aaaa", for example. Two
3785 separate and independent bugs (that affected different cases) have been
3786 fixed.
3787
3788 39. Refactored the code to abolish the use of different opcodes for small
3789 capturing bracket numbers. This is a tidy that I avoided doing when I
3790 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3791 The new approach is not only tidier, it makes it possible to reduce the
3792 memory needed to fix the previous bug (38).
3793
3794 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3795 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3796 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3797 mode.
3798
3799 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3800 report.
3801
3802 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3803 copy construction and assignment in the C++ wrapper.
3804
3805 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3806 couple of bugs that could have given wrong results in the "--newline=crlf"
3807 case.
3808
3809 44. Added a number of casts and did some reorganization of signed/unsigned int
3810 variables following suggestions from Dair Grant. Also renamed the variable
3811 "this" as "item" because it is a C++ keyword.
3812
3813 45. Arranged for dftables to add
3814
3815 #include "pcre_internal.h"
3816
3817 to pcre_chartables.c because without it, gcc 4.x may remove the array
3818 definition from the final binary if PCRE is built into a static library and
3819 dead code stripping is activated.
3820
3821 46. For an unanchored pattern, if a match attempt fails at the start of a
3822 newline sequence, and the newline setting is CRLF or ANY, and the next two
3823 characters are CRLF, advance by two characters instead of one.
3824
3825
3826 Version 6.7 04-Jul-06
3827 ---------------------
3828
3829 1. In order to handle tests when input lines are enormously long, pcretest has
3830 been re-factored so that it automatically extends its buffers when
3831 necessary. The code is crude, but this _is_ just a test program. The
3832 default size has been increased from 32K to 50K.
3833
3834 2. The code in pcre_study() was using the value of the re argument before
3835 testing it for NULL. (Of course, in any sensible call of the function, it
3836 won't be NULL.)
3837
3838 3. The memmove() emulation function in pcre_internal.h, which is used on
3839 systems that lack both memmove() and bcopy() - that is, hardly ever -
3840 was missing a "static" storage class specifier.
3841
3842 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3843 containing an extended class (one that cannot be represented by a bitmap
3844 because it contains high-valued characters or Unicode property items, e.g.
3845 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3846 pattern, but PCRE should not loop if you do not (it no longer does).
3847 [Detail: two cases were found: (a) a repeated subpattern containing an
3848 extended class; (b) a recursive reference to a subpattern that followed a
3849 previous extended class. It wasn't skipping over the extended class
3850 correctly when UTF-8 mode was not set.]
3851
3852 5. A negated single-character class was not being recognized as fixed-length
3853 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3854 compile error "lookbehind assertion is not fixed length".
3855
3856 6. The RunPerlTest auxiliary script was showing an unexpected difference
3857 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3858 write a Perl script that can interpret lines of an input file either as
3859 byte characters or as UTF-8, which is what "perltest" was being required to
3860 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3861 can't do is switch easily at run time between having the "use utf8;" pragma
3862 or not. In the end, I fudged it by using the RunPerlTest script to insert
3863 "use utf8;" explicitly for the UTF-8 tests.
3864
3865 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3866 the end of the subject string, contrary to the documentation and to what
3867 Perl does. This was true of both matching functions. Now it matches only at
3868 the start of the subject and immediately after *internal* newlines.
3869
3870 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3871 a pointer to an int instead of a pointer to an unsigned long int. This
3872 caused problems on 64-bit systems.
3873
3874 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3875 instance of the 'standard' template library not being so standard".
3876
3877 10. There was no check on the number of named subpatterns nor the maximum
3878 length of a subpattern name. The product of these values is used to compute
3879 the size of the memory block for a compiled pattern. By supplying a very
3880 long subpattern name and a large number of named subpatterns, the size
3881 computation could be caused to overflow. This is now prevented by limiting
3882 the length of names to 32 characters, and the number of named subpatterns
3883 to 10,000.
3884
3885 11. Subpatterns that are repeated with specific counts have to be replicated in
3886 the compiled pattern. The size of memory for this was computed from the
3887 length of the subpattern and the repeat count. The latter is limited to
3888 65535, but there was no limit on the former, meaning that integer overflow
3889 could in principle occur. The compiled length of a repeated subpattern is
3890 now limited to 30,000 bytes in order to prevent this.
3891
3892 12. Added the optional facility to have named substrings with the same name.
3893
3894 13. Added the ability to use a named substring as a condition, using the
3895 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3896 are numbers (not recommended). Forward references are permitted.
3897
3898 14. Added forward references in named backreferences (if you see what I mean).
3899
3900 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3901 pattern could run off the end of the subject. For example, the pattern
3902 "(?s)(.{1,5})"8 did this with the subject "ab".
3903
3904 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3905 PCRE_CASELESS was set when matching characters that were quantified with ?
3906 or *.
3907
3908 17. A character class other than a single negated character that had a minimum
3909 but no maximum quantifier - for example [ab]{6,} - was not handled
3910 correctly by pce_dfa_exec(). It would match only one character.
3911
3912 18. A valid (though odd) pattern that looked like a POSIX character
3913 class but used an invalid character after [ (for example [[,abc,]]) caused
3914 pcre_compile() to give the error "Failed: internal error: code overflow" or
3915 in some cases to crash with a glibc free() error. This could even happen if
3916 the pattern terminated after [[ but there just happened to be a sequence of
3917 letters, a binary zero, and a closing ] in the memory that followed.
3918
3919 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3920 over the years. Originally (before any Unicode support), just the bottom 8
3921 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3922 output from "man perlunicode" includes this:
3923
3924 The regular expression compiler produces polymorphic opcodes. That
3925 is, the pattern adapts to the data and automatically switches to
3926 the Unicode character scheme when presented with Unicode data--or
3927 instead uses a traditional byte scheme when presented with byte
3928 data.
3929
3930 Sadly, a wide octal escape does not cause a switch, and in a string with
3931 no other multibyte characters, these octal escapes are treated as before.
3932 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3933 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3934 Unicode string.
3935
3936 I have not perpetrated such confusion in PCRE. Up till now, it took just
3937 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3938 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3939 translate to the appropriate multibyte character.
3940
3941 29. Applied some refactoring to reduce the number of warnings from Microsoft
3942 and Borland compilers. This has included removing the fudge introduced
3943 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3944 a warning about an unused variable.
3945
3946 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3947 characters since release 4.0, because Perl (from release 5.004) does not.
3948 [Or at least, is documented not to: some releases seem to be in conflict
3949 with the documentation.] However, when a pattern was studied with
3950 pcre_study() and all its branches started with \s, PCRE still included VT
3951 as a possible starting character. Of course, this did no harm; it just
3952 caused an unnecessary match attempt.
3953
3954 22. Removed a now-redundant internal flag bit that recorded the fact that case
3955 dependency changed within the pattern. This was once needed for "required
3956 byte" processing, but is no longer used. This recovers a now-scarce options
3957 bit. Also moved the least significant internal flag bit to the most-
3958 significant bit of the word, which was not previously used (hangover from
3959 the days when it was an int rather than a uint) to free up another bit for
3960 the future.
3961
3962 23. Added support for CRLF line endings as well as CR and LF. As well as the
3963 default being selectable at build time, it can now be changed at runtime
3964 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3965 specify that it is scanning data with non-default line endings.
3966
3967 24. Changed the definition of CXXLINK to make it agree with the definition of
3968 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3969
3970 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3971 recursions. This makes a big different to stack usage for some patterns.
3972
3973 26. If a subpattern containing a named recursion or subroutine reference such
3974 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3975 the space required for the compiled pattern went wrong and gave too small a
3976 value. Depending on the environment, this could lead to "Failed: internal
3977 error: code overflow at offset 49" or "glibc detected double free or
3978 corruption" errors.
3979
3980 27. Applied patches from Google (a) to support the new newline modes and (b) to
3981 advance over multibyte UTF-8 characters in GlobalReplace.
3982
3983 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3984 difference for some implementation of PCRE in some Windows version.
3985
3986 29. Added some extra testing facilities to pcretest:
3987
3988 \q<number> in a data line sets the "match limit" value
3989 \Q<number> in a data line sets the "match recursion limt" value
3990 -S <number> sets the stack size, where <number> is in megabytes
3991
3992 The -S option isn't available for Windows.
3993
3994
3995 Version 6.6 06-Feb-06
3996 ---------------------
3997
3998 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3999 in pcreposix.h. I have copied the definition from pcre.h.
4000
4001 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
4002 because pcre.h is no longer a built file.
4003
4004 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
4005 not normally included in the compiled code.
4006
4007
4008 Version 6.5 01-Feb-06
4009 ---------------------
4010
4011 1. When using the partial match feature with pcre_dfa_exec(), it was not
4012 anchoring the second and subsequent partial matches at the new starting
4013 point. This could lead to incorrect results. For example, with the pattern
4014 /1234/, partially matching against "123" and then "a4" gave a match.
4015
4016 2. Changes to pcregrep:
4017
4018 (a) All non-match returns from pcre_exec() were being treated as failures
4019 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
4020 error message is output. Some extra information is given for the
4021 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
4022 probably the only errors that are likely to be caused by users (by
4023 specifying a regex that has nested indefinite repeats, for instance).
4024 If there are more than 20 of these errors, pcregrep is abandoned.
4025
4026 (b) A binary zero was treated as data while matching, but terminated the
4027 output line if it was written out. This has been fixed: binary zeroes
4028 are now no different to any other data bytes.
4029
4030 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
4031 used to set a locale for matching. The --locale=xxxx long option has
4032 been added (no short equivalent) to specify a locale explicitly on the
4033 pcregrep command, overriding the environment variables.
4034
4035 (d) When -B was used with -n, some line numbers in the output were one less
4036 than they should have been.
4037
4038 (e) Added the -o (--only-matching) option.
4039
4040 (f) If -A or -C was used with -c (count only), some lines of context were
4041 accidentally printed for the final match.
4042
4043 (g) Added the -H (--with-filename) option.
4044
4045 (h) The combination of options -rh failed to suppress file names for files
4046 that were found from directory arguments.
4047
4048 (i) Added the -D (--devices) and -d (--directories) options.
4049
4050 (j) Added the -F (--fixed-strings) option.
4051
4052 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4053
4054 (l) Added the --colo(u)r option.
4055
4056 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4057 is not present by default.
4058
4059 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4060 items such as (?R) or (?1), when the recursion could match a number of
4061 alternatives. If it matched one of the alternatives, but subsequently,
4062 outside the recursion, there was a failure, the code tried to back up into
4063 the recursion. However, because of the way PCRE is implemented, this is not
4064 possible, and the result was an incorrect result from the match.
4065
4066 In order to prevent this happening, the specification of recursion has
4067 been changed so that all such subpatterns are automatically treated as
4068 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4069
4070 4. I had overlooked the fact that, in some locales, there are characters for
4071 which isalpha() is true but neither isupper() nor islower() are true. In
4072 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4073 and ordfeminine) are like this. This affected the treatment of \w and \W
4074 when they appeared in character classes, but not when they appeared outside
4075 a character class. The bit map for "word" characters is now created
4076 separately from the results of isalnum() instead of just taking it from the
4077 upper, lower, and digit maps. (Plus the underscore character, of course.)
4078
4079 5. The above bug also affected the handling of POSIX character classes such as
4080 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4081 permanent tables. Instead, the bit maps for such a class were previously
4082 created as the appropriate unions of the upper, lower, and digit bitmaps.
4083 Now they are created by subtraction from the [[:word:]] class, which has
4084 its own bitmap.
4085
4086 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4087 It is created by subtracting the vertical space characters (\x09, \x0a,
4088 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4089 subtraction was done in the overall bitmap for a character class, meaning
4090 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4091 be recognized. This bug has been fixed.
4092
4093 7. Patches from the folks at Google:
4094
4095 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4096 real life, but is still worth protecting against".
4097
4098 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4099 regular expressions".
4100
4101 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4102 have it.
4103
4104 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4105 "configure" and the latter not, in order to fix a problem somebody had
4106 with compiling the Arg class on HP-UX.
4107
4108 (e) Improve the error-handling of the C++ wrapper a little bit.
4109
4110 (f) New tests for checking recursion limiting.
4111
4112 8. The pcre_memmove() function, which is used only if the environment does not
4113 have a standard memmove() function (and is therefore rarely compiled),
4114 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4115 returning a result (though PCRE never actually uses the result).
4116
4117 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4118 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4119 returned instead of calling malloc() with an overflowing number that would
4120 most likely cause subsequent chaos.
4121
4122 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4123
4124 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4125 with this option is matched, the nmatch and pmatch options of regexec() are
4126 ignored.
4127
4128 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4129 provided in case anyone wants to the the POSIX interface with UTF-8
4130 strings.
4131
4132 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4133 C++ linking (needed for some HP-UX environments).
4134
4135 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4136 (unused parameter) and in the pcre_printint() function (omitted "default"
4137 switch label when the default is to do nothing).
4138
4139 15. Added some code to make it possible, when PCRE is compiled as a C++
4140 library, to replace subject pointers for pcre_exec() with a smart pointer
4141 class, thus making it possible to process discontinuous strings.
4142
4143 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4144 much the same function. They were added by different people who were trying
4145 to make PCRE easy to compile on non-Unix systems. It has been suggested
4146 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4147 for compiling on Windows systems. I have therefore replaced it with
4148 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4149 defaults to "extern" for C or "extern C" for C++, which works fine on
4150 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4151 SCOPE with something explicit in config.h. In addition:
4152
4153 (a) pcreposix.h still had just "extern" instead of either of these macros;
4154 I have replaced it with PCRE_DATA_SCOPE.
4155
4156 (b) Functions such as _pcre_xclass(), which are internal to the library,
4157 but external in the C sense, all had PCRE_EXPORT in their definitions.
4158 This is apparently wrong for the Windows case, so I have removed it.
4159 (It makes no difference on Unix-like systems.)
4160
4161 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4162 of recursive calls to match(). This is different to MATCH_LIMIT because
4163 that limits the total number of calls to match(), not all of which increase
4164 the depth of recursion. Limiting the recursion depth limits the amount of
4165 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4166 when PCRE is compiled, and changed at run time. A patch from Google adds
4167 this functionality to the C++ interface.
4168
4169 18. Changes to the handling of Unicode character properties:
4170
4171 (a) Updated the table to Unicode 4.1.0.
4172
4173 (b) Recognize characters that are not in the table as "Cn" (undefined).
4174
4175 (c) I revised the way the table is implemented to a much improved format
4176 which includes recognition of ranges. It now supports the ranges that
4177 are defined in UnicodeData.txt, and it also amalgamates other
4178 characters into ranges. This has reduced the number of entries in the
4179 table from around 16,000 to around 3,000, thus reducing its size
4180 considerably. I realized I did not need to use a tree structure after
4181 all - a binary chop search is just as efficient. Having reduced the
4182 number of entries, I extended their size from 6 bytes to 8 bytes to
4183 allow for more data.
4184
4185 (d) Added support for Unicode script names via properties such as \p{Han}.
4186
4187 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4188 matching that character.
4189
4190 20. When matching a repeated Unicode property with a minimum greater than zero,
4191 (for example \pL{2,}), PCRE could look past the end of the subject if it
4192 reached it while seeking the minimum number of characters. This could
4193 happen only if some of the characters were more than one byte long, because
4194 there is a check for at least the minimum number of bytes.
4195
4196 21. Refactored the implementation of \p and \P so as to be more general, to
4197 allow for more different types of property in future. This has changed the
4198 compiled form incompatibly. Anybody with saved compiled patterns that use
4199 \p or \P will have to recompile them.
4200
4201 22. Added "Any" and "L&" to the supported property types.
4202
4203 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4204 but give a compile time error if the value is greater than 0xff.
4205
4206 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4207 accidentally not being installed or uninstalled.
4208
4209 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4210 made were to insert the current release number. This seemed silly, because
4211 it made things harder for people building PCRE on systems that don't run
4212 "configure". I have turned pcre.h into a distributed file, no longer built
4213 by "configure", with the version identification directly included. There is
4214 no longer a pcre.h.in file.
4215
4216 However, this change necessitated a change to the pcre-config script as
4217 well. It is built from pcre-config.in, and one of the substitutions was the
4218 release number. I have updated configure.ac so that ./configure now finds
4219 the release number by grepping pcre.h.
4220
4221 26. Added the ability to run the tests under valgrind.
4222
4223
4224 Version 6.4 05-Sep-05
4225 ---------------------
4226
4227 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4228 "--" to be printed when multiple files were scanned, even when none of the
4229 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4230 consider it to be a bug, and have restored the previous behaviour.
4231
4232 2. A couple of code tidies to get rid of compiler warnings.
4233
4234 3. The pcretest program used to cheat by referring to symbols in the library
4235 whose names begin with _pcre_. These are internal symbols that are not
4236 really supposed to be visible externally, and in some environments it is
4237 possible to suppress them. The cheating is now confined to including
4238 certain files from the library's source, which is a bit cleaner.
4239
4240 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4241 file's purpose clearer.
4242
4243 5. Reorganized pcre_ucp_findchar().
4244
4245
4246 Version 6.3 15-Aug-05
4247 ---------------------
4248
4249 1. The file libpcre.pc.in did not have general read permission in the tarball.
4250
4251 2. There were some problems when building without C++ support:
4252
4253 (a) If C++ support was not built, "make install" and "make test" still
4254 tried to test it.
4255
4256 (b) There were problems when the value of CXX was explicitly set. Some
4257 changes have been made to try to fix these, and ...
4258
4259 (c) --disable-cpp can now be used to explicitly disable C++ support.
4260
4261 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4262 backslash in a target when C++ was disabled. This confuses some
4263 versions of "make", apparently. Using an intermediate variable solves
4264 this. (Same for CPP_LOBJ.)
4265
4266 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4267 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4268 necessary on certain architectures.
4269
4270 4. Added a setting of -export-symbols-regex to the link command to remove
4271 those symbols that are exported in the C sense, but actually are local
4272 within the library, and not documented. Their names all begin with
4273 "_pcre_". This is not a perfect job, because (a) we have to except some
4274 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4275 available (and never for static libraries). I have made a note to try to
4276 find a way round (a) in the future.
4277
4278
4279 Version 6.2 01-Aug-05
4280 ---------------------
4281
4282 1. There was no test for integer overflow of quantifier values. A construction
4283 such as {1111111111111111} would give undefined results. What is worse, if
4284 a minimum quantifier for a parenthesized subpattern overflowed and became
4285 negative, the calculation of the memory size went wrong. This could have
4286 led to memory overwriting.
4287
4288 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4289
4290 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4291 operating environments where this matters.
4292
4293 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4294 PCRE options from within the C++ wrapper.
4295
4296 5. Named capturing subpatterns were not being correctly counted when a pattern
4297 was compiled. This caused two problems: (a) If there were more than 100
4298 such subpatterns, the calculation of the memory needed for the whole
4299 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4300 back references of the form \12, where the number was greater than 9, were
4301 not recognized as back references, even though there were sufficient
4302 previous subpatterns.
4303
4304 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4305 versions of gcc, e.g. 2.95.4.
4306
4307
4308 Version 6.1 21-Jun-05
4309 ---------------------
4310
4311 1. There was one reference to the variable "posix" in pcretest.c that was not
4312 surrounded by "#if !defined NOPOSIX".
4313
4314 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4315 the cross-check on the old pcre_info() function, for the benefit of the
4316 cut-down version of PCRE that is currently imported into Exim.
4317
4318 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4319 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4320 patterns that start (?i) but I don't think that matters. The use of (?i) is
4321 just an example; this all applies to the other options as well.
4322
4323 4. Since libtool seems to echo the compile commands it is issuing, the output
4324 from "make" can be reduced a bit by putting "@" in front of each libtool
4325 compile command.
4326
4327 5. Patch from the folks at Google for configure.in to be a bit more thorough
4328 in checking for a suitable C++ installation before trying to compile the
4329 C++ stuff. This should fix a reported problem when a compiler was present,
4330 but no suitable headers.
4331
4332 6. The man pages all had just "PCRE" as their title. I have changed them to
4333 be the relevant file name. I have also arranged that these names are
4334 retained in the file doc/pcre.txt, which is a concatenation in text format
4335 of all the man pages except the little individual ones for each function.
4336
4337 7. The NON-UNIX-USE file had not been updated for the different set of source
4338 files that come with release 6. I also added a few comments about the C++
4339 wrapper.
4340
4341
4342 Version 6.0 07-Jun-05
4343 ---------------------
4344
4345 1. Some minor internal re-organization to help with my DFA experiments.
4346
4347 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4348 didn't matter for the library itself when fully configured, but did matter
4349 when compiling without UCP support, or within Exim, where the ucp files are
4350 not imported.
4351
4352 3. Refactoring of the library code to split up the various functions into
4353 different source modules. The addition of the new DFA matching code (see
4354 below) to a single monolithic source would have made it really too
4355 unwieldy, quite apart from causing all the code to be include in a
4356 statically linked application, when only some functions are used. This is
4357 relevant even without the DFA addition now that patterns can be compiled in
4358 one application and matched in another.
4359
4360 The downside of splitting up is that there have to be some external
4361 functions and data tables that are used internally in different modules of
4362 the library but which are not part of the API. These have all had their
4363 names changed to start with "_pcre_" so that they are unlikely to clash
4364 with other external names.
4365
4366 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4367 a different (DFA) algorithm. Although it is slower than the original
4368 function, it does have some advantages for certain types of matching
4369 problem.
4370
4371 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4372 including restarting after a partial match.
4373
4374 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4375 defined when compiling for Windows was sent to me. I have put it into the
4376 code, though I have no means of testing or verifying it.
4377
4378 7. Added the pcre_refcount() auxiliary function.
4379
4380 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4381 match before or at the first newline in the subject string. In pcretest,
4382 the /f option on a pattern can be used to set this.
4383
4384 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4385 would behave wrongly. This has been present in PCRE since release 4.0.
4386
4387 10. A number of changes to the pcregrep command:
4388
4389 (a) Refactored how -x works; insert ^(...)$ instead of setting
4390 PCRE_ANCHORED and checking the length, in preparation for adding
4391 something similar for -w.
4392
4393 (b) Added the -w (match as a word) option.
4394
4395 (c) Refactored the way lines are read and buffered so as to have more
4396 than one at a time available.
4397
4398 (d) Implemented a pcregrep test script.
4399
4400 (e) Added the -M (multiline match) option. This allows patterns to match
4401 over several lines of the subject. The buffering ensures that at least
4402 8K, or the rest of the document (whichever is the shorter) is available
4403 for matching (and similarly the previous 8K for lookbehind assertions).
4404
4405 (f) Changed the --help output so that it now says
4406
4407 -w, --word-regex(p)
4408
4409 instead of two lines, one with "regex" and the other with "regexp"
4410 because that confused at least one person since the short forms are the
4411 same. (This required a bit of code, as the output is generated
4412 automatically from a table. It wasn't just a text change.)
4413
4414 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4415 option but starts with a hyphen. Could be a pattern or a path name
4416 starting with a hyphen, for instance.
4417
4418 (h) "-" can be given as a file name to represent stdin.
4419
4420 (i) When file names are being printed, "(standard input)" is used for
4421 the standard input, for compatibility with GNU grep. Previously
4422 "<stdin>" was used.
4423
4424 (j) The option --label=xxx can be used to supply a name to be used for
4425 stdin when file names are being printed. There is no short form.
4426
4427 (k) Re-factored the options decoding logic because we are going to add
4428 two more options that take data. Such options can now be given in four
4429 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4430
4431 (l) Added the -A, -B, and -C options for requesting that lines of context
4432 around matches be printed.
4433
4434 (m) Added the -L option to print the names of files that do not contain
4435 any matching lines, that is, the complement of -l.
4436
4437 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4438 continue to scan other files.
4439
4440 (o) The -s option was incorrectly implemented. For compatibility with other
4441 greps, it now suppresses the error message for a non-existent or non-
4442 accessible file (but not the return code). There is a new option called
4443 -q that suppresses the output of matching lines, which was what -s was
4444 previously doing.
4445
4446 (p) Added --include and --exclude options to specify files for inclusion
4447 and exclusion when recursing.
4448
4449 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4450 Hopefully, it now does.
4451
4452 12. Missing cast in pcre_study().
4453
4454 13. Added an "uninstall" target to the makefile.
4455
4456 14. Replaced "extern" in the function prototypes in Makefile.in with
4457 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4458 world, but is set differently for Windows.
4459
4460 15. Added a second compiling function called pcre_compile2(). The only
4461 difference is that it has an extra argument, which is a pointer to an
4462 integer error code. When there is a compile-time failure, this is set
4463 non-zero, in addition to the error test pointer being set to point to an
4464 error message. The new argument may be NULL if no error number is required
4465 (but then you may as well call pcre_compile(), which is now just a
4466 wrapper). This facility is provided because some applications need a
4467 numeric error indication, but it has also enabled me to tidy up the way
4468 compile-time errors are handled in the POSIX wrapper.
4469
4470 16. Added VPATH=.libs to the makefile; this should help when building with one
4471 prefix path and installing with another. (Or so I'm told by someone who
4472 knows more about this stuff than I do.)
4473
4474 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4475 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4476 match everything, including newlines. This is not POSIX-compatible, but
4477 somebody wanted the feature. From pcretest it can be activated by using
4478 both the P and the s flags.
4479
4480 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4481
4482 19. libpcre.pc was being incorrectly installed as executable.
4483
4484 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4485 it now also looks for '\r' so that it will work unmodified on Windows.
4486
4487 21. Added Google's contributed C++ wrapper to the distribution.
4488
4489 22. Added some untidy missing memory free() calls in pcretest, to keep
4490 Electric Fence happy when testing.
4491
4492
4493
4494 Version 5.0 13-Sep-04
4495 ---------------------
4496
4497 1. Internal change: literal characters are no longer packed up into items
4498 containing multiple characters in a single byte-string. Each character
4499 is now matched using a separate opcode. However, there may be more than one
4500 byte in the character in UTF-8 mode.
4501
4502 2. The pcre_callout_block structure has two new fields: pattern_position and
4503 next_item_length. These contain the offset in the pattern to the next match
4504 item, and its length, respectively.
4505
4506 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4507 insertion of callouts before each pattern item. Added the /C option to
4508 pcretest to make use of this.
4509
4510 4. On the advice of a Windows user, the lines
4511
4512 #if defined(_WIN32) || defined(WIN32)
4513 _setmode( _fileno( stdout ), 0x8000 );
4514 #endif /* defined(_WIN32) || defined(WIN32) */
4515
4516 have been added to the source of pcretest. This apparently does useful
4517 magic in relation to line terminators.
4518
4519 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4520 for the benefit of those environments where the "b" makes a difference.
4521
4522 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4523 to know about it. I have put a hack into configure.in that adds in code
4524 to set GCC=yes if CC=icc. This seems to end up at a point in the
4525 generated configure script that is early enough to affect the setting of
4526 compiler options, which is what is needed, but I have no means of testing
4527 whether it really works. (The user who reported this had patched the
4528 generated configure script, which of course I cannot do.)
4529
4530 LATER: After change 22 below (new libtool files), the configure script
4531 seems to know about icc (and also ecc). Therefore, I have commented out
4532 this hack in configure.in.
4533
4534 7. Added support for pkg-config (2 patches were sent in).
4535
4536 8. Negated POSIX character classes that used a combination of internal tables
4537 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4538 [[:^ascii]]. Typically, they would match almost any characters. The other
4539 POSIX classes were not broken in this way.
4540
4541 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4542 to find the match, as PCRE was deluded into thinking that the match had to
4543 start at the start point or following a newline. The same bug applied to
4544 patterns with negative forward assertions or any backward assertions
4545 preceding ".*" at the start, unless the pattern required a fixed first
4546 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4547
4548 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4549 starting at the last subject character, bytes beyond the end of the subject
4550 string were read.
4551
4552 11. Renamed the variable "class" as "classbits" to make life easier for C++
4553 users. (Previously there was a macro definition, but it apparently wasn't
4554 enough.)
4555
4556 12. Added the new field "tables" to the extra data so that tables can be passed
4557 in at exec time, or the internal tables can be re-selected. This allows
4558 a compiled regex to be saved and re-used at a later time by a different
4559 program that might have everything at different addresses.
4560
4561 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4562 -R library as well as a -L library.
4563
4564 14. The debugging options of pcretest (-d on the command line or D on a
4565 pattern) showed incorrect output for anything following an extended class
4566 that contained multibyte characters and which was followed by a quantifier.
4567
4568 15. Added optional support for general category Unicode character properties
4569 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4570 support. It adds about 90K to the size of the library. The meanings of the
4571 inbuilt class escapes such as \d and \s have NOT been changed.
4572
4573 16. Updated pcredemo.c to include calls to free() to release the memory for the
4574 compiled pattern.
4575
4576 17. The generated file chartables.c was being created in the source directory
4577 instead of in the building directory. This caused the build to fail if the
4578 source directory was different from the building directory, and was
4579 read-only.
4580
4581 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4582 file. No doubt somebody will tell me if they don't make sense... Also added
4583 Dan Mooney's comments about building on OpenVMS.
4584
4585 19. Added support for partial matching via the PCRE_PARTIAL option for
4586 pcre_exec() and the \P data escape in pcretest.
4587
4588 20. Extended pcretest with 3 new pattern features:
4589
4590 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4591 write the compiled pattern to the file whose name is "rest-of-line".
4592 This is a straight binary dump of the data, with the saved pointer to
4593 the character tables forced to be NULL. The study data, if any, is
4594 written too. After writing, pcretest reads a new pattern.
4595
4596 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4597 compiled pattern from the given file. There must not be any
4598 occurrences of "<" in the file name (pretty unlikely); if there are,
4599 pcretest will instead treat the initial "<" as a pattern delimiter.
4600 After reading in the pattern, pcretest goes on to read data lines as
4601 usual.
4602
4603 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4604 and 16-bit fields in a compiled pattern, to simulate a pattern that
4605 was compiled on a host of opposite endianness.
4606
4607 21. The pcre-exec() function can now cope with patterns that were compiled on
4608 hosts of opposite endianness, with this restriction:
4609
4610 As for any compiled expression that is saved and used later, the tables
4611 pointer field cannot be preserved; the extra_data field in the arguments
4612 to pcre_exec() should be used to pass in a tables address if a value
4613 other than the default internal tables were used at compile time.
4614
4615 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4616 now diagnosed as an error. Previously, most of the time, a negative number
4617 would have been treated as zero, but if in addition "ovector" was passed as
4618 NULL, a crash could occur.
4619
4620 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4621 new versions from the libtool 1.5 distribution (the last one is a copy of
4622 a file called libtool.m4). This seems to have fixed the need to patch
4623 "configure" to support Darwin 1.3 (which I used to do). However, I still
4624 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4625 workstation).
4626
4627 24. Changed the PCRE licence to be the more standard "BSD" licence.
4628
4629
4630 Version 4.5 01-Dec-03
4631 ---------------------
4632
4633 1. There has been some re-arrangement of the code for the match() function so
4634 that it can be compiled in a version that does not call itself recursively.
4635 Instead, it keeps those local variables that need separate instances for
4636 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4637 needs to "recurse". Keeping track of where control must go is done by means
4638 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4639 hide most of the details from the main code, and operates only if
4640 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4641 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4642 operating.
4643
4644 To make it easier for callers to provide specially tailored get/free
4645 functions for this usage, two new functions, pcre_stack_malloc, and
4646 pcre_stack_free, are used. They are always called in strict stacking order,
4647 and the size of block requested is always the same.
4648
4649 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4650 PCRE has been compiled to use the stack or the heap for recursion. The
4651 -C option of pcretest uses this to show which version is compiled.
4652
4653 A new data escape \S, is added to pcretest; it causes the amounts of store
4654 obtained and freed by both kinds of malloc/free at match time to be added
4655 to the output.
4656
4657 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4658 what's available on my current Linux desktop machine.
4659
4660 3. When matching a UTF-8 string, the test for a valid string at the start has
4661 been extended. If start_offset is not zero, PCRE now checks that it points
4662 to a byte that is the start of a UTF-8 character. If not, it returns
4663 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4664 this is necessary because there may be backward assertions in the pattern.
4665 When matching the same subject several times, it may save resources to use
4666 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4667
4668 4. The code for checking the validity of UTF-8 strings has been tightened so
4669 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4670 containing "overlong sequences".
4671
4672 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4673 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4674 should have been "&", but it just so happened that all the cases this let
4675 through by mistake were picked up later in the function.
4676
4677 6. I had used a variable called "isblank" - this is a C99 function, causing
4678 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4679
4680 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4681 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4682 so the version is shown; (c) stop "make test" echoing "./RunTest".
4683
4684 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4685
4686 9. The prototype for memmove() for systems that don't have it was using
4687 size_t, but the inclusion of the header that defines size_t was later. I've
4688 moved the #includes for the C headers earlier to avoid this.
4689
4690 10. Added some adjustments to the code to make it easier to compiler on certain
4691 special systems:
4692
4693 (a) Some "const" qualifiers were missing.
4694 (b) Added the macro EXPORT before all exported functions; by default this
4695 is defined to be empty.
4696 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4697 that it reads its output file name as an argument instead of writing
4698 to the standard output and assuming this can be redirected.
4699
4700 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4701 class containing characters with values greater than 255, PCRE compilation
4702 went into a loop.
4703
4704 12. A recursive reference to a subpattern that was within another subpattern
4705 that had a minimum quantifier of zero caused PCRE to crash. For example,
4706 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4707 recursion. If the recursively-called subpattern itself had a zero repeat,
4708 that was OK.
4709
4710 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4711 buffer into which it was copied (for escape processing) was still set at
4712 1024, so long lines caused crashes.
4713
4714 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4715 "internal error: code overflow...". This applied to any character class
4716 that was followed by a possessive quantifier.
4717
4718 15. Modified the Makefile to add libpcre.la as a prerequisite for
4719 libpcreposix.la because I was told this is needed for a parallel build to
4720 work.
4721
4722 16. If a pattern that contained .* following optional items at the start was
4723 studied, the wrong optimizing data was generated, leading to matching
4724 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4725 matching string must start with a or b or c. The correct conclusion for
4726 this pattern is that a match can start with any character.
4727
4728
4729 Version 4.4 13-Aug-03
4730 ---------------------
4731
4732 1. In UTF-8 mode, a character class containing characters with values between
4733 127 and 255 was not handled correctly if the compiled pattern was studied.
4734 In fixing this, I have also improved the studying algorithm for such
4735 classes (slightly).
4736
4737 2. Three internal functions had redundant arguments passed to them. Removal
4738 might give a very teeny performance improvement.
4739
4740 3. Documentation bug: the value of the capture_top field in a callout is *one
4741 more than* the number of the hightest numbered captured substring.
4742
4743 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4744 in incorrectly linking with a previously installed version. They now link
4745 explicitly with libpcre.la.
4746
4747 5. configure.in no longer needs to recognize Cygwin specially.
4748
4749 6. A problem in pcre.in for Windows platforms is fixed.
4750
4751 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4752 pcretest, it used to include the size of the study block as part of its
4753 output. Unfortunately, the structure contains a field that has a different
4754 size on different hardware architectures. This meant that the tests that
4755 showed this size failed. As the block is currently always of a fixed size,
4756 this information isn't actually particularly useful in pcretest output, so
4757 I have just removed it.
4758
4759 8. Three pre-processor statements accidentally did not start in column 1.
4760 Sadly, there are *still* compilers around that complain, even though
4761 standard C has not required this for well over a decade. Sigh.
4762
4763 9. In pcretest, the code for checking callouts passed small integers in the
4764 callout_data field, which is a void * field. However, some picky compilers
4765 complained about the casts involved for this on 64-bit systems. Now
4766 pcretest passes the address of the small integer instead, which should get
4767 rid of the warnings.
4768
4769 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4770 both compile and run time, and gives an error if an invalid UTF-8 sequence
4771 is found. There is a option for disabling this check in cases where the
4772 string is known to be correct and/or the maximum performance is wanted.
4773
4774 11. In response to a bug report, I changed one line in Makefile.in from
4775
4776 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4777 to
4778 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4779
4780 to look similar to other lines, but I have no way of telling whether this
4781 is the right thing to do, as I do not use Windows. No doubt I'll get told
4782 if it's wrong...
4783
4784
4785 Version 4.3 21-May-03
4786 ---------------------
4787
4788 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4789 Makefile.
4790
4791 2. Some refactoring to improve the quality of the code:
4792
4793 (i) The utf8_table... variables are now declared "const".
4794
4795 (ii) The code for \cx, which used the "case flipping" table to upper case
4796 lower case letters, now just substracts 32. This is ASCII-specific,
4797 but the whole concept of \cx is ASCII-specific, so it seems
4798 reasonable.
4799
4800 (iii) PCRE was using its character types table to recognize decimal and
4801 hexadecimal digits in the pattern. This is silly, because it handles
4802 only 0-9, a-f, and A-F, but the character types table is locale-
4803 specific, which means strange things might happen. A private
4804 table is now used for this - though it costs 256 bytes, a table is
4805 much faster than multiple explicit tests. Of course, the standard
4806 character types table is still used for matching digits in subject
4807 strings against \d.
4808
4809 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4810 ending in _t are). So I've renamed it as ESC_tee.
4811
4812 3. The first argument for regexec() in the POSIX wrapper should have been
4813 defined as "const".
4814
4815 4. Changed pcretest to use malloc() for its buffers so that they can be
4816 Electric Fenced for debugging.
4817
4818 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4819 to read one or more bytes before the start of the subject string. Often this
4820 had no effect on PCRE's behaviour, but in some circumstances it could
4821 provoke a segmentation fault.
4822
4823 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4824 to try to read one or more bytes before the start of the subject string.
4825
4826 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4827 UTF-8 support could misbehave in various ways if the subject string
4828 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4829 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4830 back over UTF-8 characters.)
4831
4832
4833 Version 4.2 14-Apr-03
4834 ---------------------
4835
4836 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4837
4838 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4839 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4840 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4841 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4842 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4843 and BUILD_EXEEXT
4844 Note: automatic setting of the BUILD variables is not yet working
4845 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4846 compile-time but not at link-time
4847 [LINK]: use for linking executables only
4848 make different versions for Windows and non-Windows
4849 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4850 libraries
4851 [LINK_FOR_BUILD]: new variable
4852 [OBJEXT]: use throughout
4853 [EXEEXT]: use throughout
4854 <winshared>: new target
4855 <wininstall>: new target
4856 <dftables.o>: use native compiler
4857 <dftables>: use native linker
4858 <install>: handle Windows platform correctly
4859 <clean>: ditto
4860 <check>: ditto
4861 copy DLL to top builddir before testing
4862
4863 As part of these changes, -no-undefined was removed again. This was reported
4864 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4865 in any case.
4866
4867 3. Some tidies to get rid of compiler warnings:
4868
4869 . In the match_data structure, match_limit was an unsigned long int, whereas
4870 match_call_count was an int. I've made them both unsigned long ints.
4871
4872 . In pcretest the fact that a const uschar * doesn't automatically cast to
4873 a void * provoked a warning.
4874
4875 . Turning on some more compiler warnings threw up some "shadow" variables
4876 and a few more missing casts.
4877
4878 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4879 option, a class that contained a single character with a value between 128
4880 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4881
4882 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4883 option, a class that contained several characters, but with at least one
4884 whose value was between 128 and 255 caused PCRE to crash.
4885
4886
4887 Version 4.1 12-Mar-03
4888 ---------------------
4889
4890 1. Compiling with gcc -pedantic found a couple of places where casts were
4891 needed, and a string in dftables.c that was longer than standard compilers are
4892 required to support.
4893
4894 2. Compiling with Sun's compiler found a few more places where the code could
4895 be tidied up in order to avoid warnings.
4896
4897 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4898 first of these names is deprecated in the latest Autoconf in favour of the name
4899 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4900 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4901 analogy I have changed it to CFLAGS_FOR_BUILD.
4902
4903 4. Added -no-undefined to the linking command in the Makefile, because this is
4904 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4905 linking step for the pcreposix library.
4906
4907 5. PCRE was failing to diagnose the case of two named groups with the same
4908 name.
4909
4910 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4911 literal character that is needed in the subject for a match, and scans along to
4912 ensure that it is present before embarking on the full matching process. This
4913 saves time in cases of nested unlimited repeats that are never going to match.
4914 Problem: the scan can take a lot of time if the subject is very long (e.g.
4915 megabytes), thus penalizing straightforward matches. It is now done only if the
4916 amount of subject to be scanned is less than 1000 bytes.
4917
4918 7. A lesser problem with the same optimization is that it was recording the
4919 first character of an anchored pattern as "needed", thus provoking a search
4920 right along the subject, even when the first match of the pattern was going to
4921 fail. The "needed" character is now not set for anchored patterns, unless it
4922 follows something in the pattern that is of non-fixed length. Thus, it still
4923 fulfils its original purpose of finding quick non-matches in cases of nested
4924 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4925
4926
4927 Version 4.0 17-Feb-03
4928 ---------------------
4929
4930 1. If a comment in an extended regex that started immediately after a meta-item
4931 extended to the end of string, PCRE compiled incorrect data. This could lead to
4932 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4933
4934 2. Moved to autoconf 2.53 and libtool 1.4.2.
4935
4936 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4937 the special perltest8 script is no longer needed - all the tests can be run
4938 from a single perltest script.
4939
4940 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4941 by \s. It has now been removed in PCRE. This means it isn't recognized as
4942 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4943 class [:space:] *does* include VT, thereby creating a mess.
4944
4945 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4946 space and tab.
4947
4948 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4949 its new features into the main test script, reducing the number of scripts.
4950
4951 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4952 were backward compatible, and made the (?i) apply to the whole pattern, as if
4953 /i were given. Now it behaves more logically, and applies the option setting
4954 only to what follows. PCRE has been changed to follow suit. However, if it
4955 finds options settings right at the start of the pattern, it extracts them into
4956 the global options, as before. Thus, they show up in the info data.
4957
4958 8. Added support for the \Q...\E escape sequence. Characters in between are
4959 treated as literals. This is slightly different from Perl in that $ and @ are
4960 also handled as literals inside the quotes. In Perl, they will cause variable
4961 interpolation. Note the following examples:
4962
4963 Pattern PCRE matches Perl matches
4964
4965 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4966 \Qabc\$xyz\E abc\$xyz abc\$xyz
4967 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4968
4969 For compatibility with Perl, \Q...\E sequences are recognized inside character
4970 classes as well as outside them.
4971
4972 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4973 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4974 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4975 signed/unsigned warnings.
4976
4977 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4978 option for pcretest, so I've replaced it by a simple function that does just
4979 that job.
4980
4981 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4982 "pcregrep -".
4983
4984 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4985 Java package. This provides some syntactic sugar for simple cases of what my
4986 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4987 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4988 item, you can use this simplified notation. Note that only makes sense with
4989 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4990 greediness, whatever the setting of the PCRE_UNGREEDY option.
4991
4992 13. A change of greediness default within a pattern was not taking effect at
4993 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4994 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4995 was abstracted outside.
4996
4997 14. PCRE now supports the \G assertion. It is true when the current matching
4998 position is at the start point of the match. This differs from \A when the
4999 starting offset is non-zero. Used with the /g option of pcretest (or similar
5000 code), it works in the same way as it does for Perl's /g option. If all
5001 alternatives of a regex begin with \G, the expression is anchored to the start
5002 match position, and the "anchored" flag is set in the compiled expression.
5003
5004 15. Some bugs concerning the handling of certain option changes within patterns
5005 have been fixed. These applied to options other than (?ims). For example,
5006 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
5007 been the other way round. Some of this was related to change 7 above.
5008
5009 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
5010 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
5011 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
5012 POSIX classes only within a class (e.g. /[[:alpha:]]/).
5013
5014 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
5015 mode. Unlike ".", it always matches newline, whatever the setting of
5016 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
5017 assertions. Perl allows it, but it doesn't (in general) work because it can't
5018 calculate the length of the lookbehind. At least, that's the case for Perl
5019 5.8.0 - I've been told they are going to document that it doesn't work in
5020 future.
5021
5022 18. Added an error diagnosis for escapes that PCRE does not support: these are
5023 \L, \l, \N, \P, \p, \U, \u, and \X.
5024
5025 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
5026 reading past the end of the pattern in cases such as /[abcd/.
5027
5028 20. PCRE was getting more memory than necessary for patterns with classes that
5029 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
5030
5031 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
5032 compiling PCRE for use with Virtual Pascal.
5033
5034 22. Small fix to the Makefile to make it work properly if the build is done
5035 outside the source tree.
5036
5037 23. Added a new extension: a condition to go with recursion. If a conditional
5038 subpattern starts with (?(R) the "true" branch is used if recursion has
5039 happened, whereas the "false" branch is used only at the top level.
5040
5041 24. When there was a very long string of literal characters (over 255 bytes
5042 without UTF support, over 250 bytes with UTF support), the computation of how
5043 much memory was required could be incorrect, leading to segfaults or other
5044 strange effects.
5045
5046 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5047 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5048 there was a subsequent back reference to those brackets. This meant that, for
5049 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5050 possible to check for precisely this case. All we can do is abandon the
5051 optimization if .* occurs inside capturing brackets when there are any back
5052 references whatsoever. (See below for a better fix that came later.)
5053
5054 26. The handling of the optimization for finding the first character of a
5055 non-anchored pattern, and for finding a character that is required later in the
5056 match were failing in some cases. This didn't break the matching; it just
5057 failed to optimize when it could. The way this is done has been re-implemented.
5058
5059 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5060
5061 28. Added a new feature that provides some of the functionality that Perl
5062 provides with (?{...}). The facility is termed a "callout". The way it is done
5063 in PCRE is for the caller to provide an optional function, by setting
5064 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5065 global variable. By default it is unset, which disables all calling out. To get
5066 the function called, the regex must include (?C) at appropriate points. This
5067 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5068 This provides a means of identifying different callout points. When PCRE
5069 reaches such a point in the regex, if pcre_callout has been set, the external
5070 function is called. It is provided with data in a structure called
5071 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5072 matching continues; if it returns a non-zero value, the match at the current
5073 point fails. However, backtracking will occur if possible. [This was changed
5074 later and other features added - see item 49 below.]
5075
5076 29. pcretest is upgraded to test the callout functionality. It provides a
5077 callout function that displays information. By default, it shows the start of
5078 the match and the current position in the text. There are some new data escapes
5079 to vary what happens:
5080
5081 \C+ in addition, show current contents of captured substrings
5082 \C- do not supply a callout function
5083 \C!n return 1 when callout number n is reached
5084 \C!n!m return 1 when callout number n is reached for the mth time
5085
5086 30. If pcregrep was called with the -l option and just a single file name, it
5087 output "<stdin>" if a match was found, instead of the file name.
5088
5089 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5090 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5091 pcre_exec(). This saves a malloc/free per call. The default value of
5092 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5093 when configuring.
5094
5095 32. The default maximum size of a compiled pattern is 64K. There have been a
5096 few cases of people hitting this limit. The code now uses macros to handle the
5097 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5098 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5099 configuring. Tests 2 and 5 work only with 2-byte links because they output
5100 debugging information about compiled patterns.
5101
5102 33. Internal code re-arrangements:
5103
5104 (a) Moved the debugging function for printing out a compiled regex into
5105 its own source file (printint.c) and used #include to pull it into
5106 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5107 separate copies.
5108
5109 (b) Defined the list of op-code names for debugging as a macro in
5110 internal.h so that it is next to the definition of the opcodes.
5111
5112 (c) Defined a table of op-code lengths for simpler skipping along compiled
5113 code. This is again a macro in internal.h so that it is next to the
5114 definition of the opcodes.
5115
5116 34. Added support for recursive calls to individual subpatterns, along the
5117 lines of Robin Houston's patch (but implemented somewhat differently).
5118
5119 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5120 allow it to read and process whole directories in Win32. This code was
5121 contributed by Lionel Fourquaux; it has not been tested by me.
5122
5123 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5124 used to name a group. Names consist of alphanumerics and underscores, and must
5125 be unique. Back references use the syntax (?P=name) and recursive calls use
5126 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5127 numbers. The function pcre_fullinfo() can be used after compilation to extract
5128 a name/number map. There are three relevant calls:
5129
5130 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5131 PCRE_INFO_NAMECOUNT yields the number of entries
5132 PCRE_INFO_NAMETABLE yields a pointer to the map.
5133
5134 The map is a vector of fixed-size entries. The size of each entry depends on
5135 the length of the longest name used. The first two bytes of each entry are the
5136 group number, most significant byte first. There follows the corresponding
5137 name, zero terminated. The names are in alphabetical order.
5138
5139 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5140 case instead of 255. Making it the same both with and without UTF-8 support
5141 means that the same test output works with both.
5142
5143 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5144 calling malloc() with a zero argument.
5145
5146 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5147 optimization. I've improved things by keeping a bitmap of backreferences with
5148 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5149 fact referenced, the optimization can be applied. It is unlikely that a
5150 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5151 the match to follow \n) will appear inside brackets with a number greater than
5152 31, but if it does, any back reference > 31 suppresses the optimization.
5153
5154 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5155 of disabling numbered capturing parentheses. Any opening parenthesis that is
5156 not followed by ? behaves as if it were followed by ?: but named parentheses
5157 can still be used for capturing (and they will acquire numbers in the usual
5158 way).
5159
5160 41. Redesigned the return codes from the match() function into yes/no/error so
5161 that errors can be passed back from deep inside the nested calls. A malloc
5162 failure while inside a recursive subpattern call now causes the
5163 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5164
5165 42. It is now possible to set a limit on the number of times the match()
5166 function is called in a call to pcre_exec(). This facility makes it possible to
5167 limit the amount of recursion and backtracking, though not in a directly
5168 obvious way, because the match() function is used in a number of different
5169 circumstances. The count starts from zero for each position in the subject
5170 string (for non-anchored patterns). The default limit is, for compatibility, a
5171 large number, namely 10 000 000. You can change this in two ways:
5172
5173 (a) When configuring PCRE before making, you can use --with-match-limit=n
5174 to set a default value for the compiled library.
5175
5176 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5177 a different value is set. See 45 below.
5178
5179 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5180
5181 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5182 of things that can be changed at compile time. The first argument specifies
5183 what is wanted and the second points to where the information is to be placed.
5184 The current list of available information is:
5185
5186 PCRE_CONFIG_UTF8
5187
5188 The output is an integer that is set to one if UTF-8 support is available;
5189 otherwise it is set to zero.
5190
5191 PCRE_CONFIG_NEWLINE
5192
5193 The output is an integer that it set to the value of the code that is used for
5194 newline. It is either LF (10) or CR (13).
5195
5196 PCRE_CONFIG_LINK_SIZE
5197
5198 The output is an integer that contains the number of bytes used for internal
5199 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5200
5201 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5202
5203 The output is an integer that contains the threshold above which the POSIX
5204 interface uses malloc() for output vectors. See item 31 above.
5205
5206 PCRE_CONFIG_MATCH_LIMIT
5207
5208 The output is an unsigned integer that contains the default limit of the number
5209 of match() calls in a pcre_exec() execution. See 42 above.
5210
5211 44. pcretest has been upgraded by the addition of the -C option. This causes it
5212 to extract all the available output from the new pcre_config() function, and to
5213 output it. The program then exits immediately.
5214
5215 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5216 order to support additional features. One way would have been to define
5217 pcre_exec2() (for example) with extra arguments, but this would not have been
5218 extensible, and would also have required all calls to the original function to
5219 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5220 is used for passing in "extra" data from pcre_study().
5221
5222 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5223 contains the following fields:
5224
5225 flags a bitmap indicating which of the following fields are set
5226 study_data opaque data from pcre_study()
5227 match_limit a way of specifying a limit on match() calls for a specific
5228 call to pcre_exec()
5229 callout_data data for callouts (see 49 below)
5230
5231 The flag bits are also defined in pcre.h, and are
5232
5233 PCRE_EXTRA_STUDY_DATA
5234 PCRE_EXTRA_MATCH_LIMIT
5235 PCRE_EXTRA_CALLOUT_DATA
5236
5237 The pcre_study() function now returns one of these new pcre_extra blocks, with
5238 the actual study data pointed to by the study_data field, and the
5239 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5240 before. That is, this change is entirely upwards-compatible and requires no
5241 change to existing code.
5242
5243 If you want to pass in additional data to pcre_exec(), you can either place it
5244 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5245 block.
5246
5247 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5248 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5249 times with different match limits, until it finds the minimum value needed for
5250 pcre_exec() to complete. The value is then output. This can be instructive; for
5251 most simple matches the number is quite small, but for pathological cases it
5252 gets very large very quickly.
5253
5254 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5255 returns the size of the data block pointed to by the study_data field in a
5256 pcre_extra block, that is, the value that was passed as the argument to
5257 pcre_malloc() when PCRE was getting memory in which to place the information
5258 created by pcre_study(). The fourth argument should point to a size_t variable.
5259 pcretest has been extended so that this information is shown after a successful
5260 pcre_study() call when information about the compiled regex is being displayed.
5261
5262 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5263 because what follows is always an absolute path. (Later: it turns out that this
5264 is more than cosmetic for MinGW, because it doesn't like empty path
5265 components.)
5266
5267 49. Some changes have been made to the callout feature (see 28 above):
5268
5269 (i) A callout function now has three choices for what it returns:
5270
5271 0 => success, carry on matching
5272 > 0 => failure at this point, but backtrack if possible
5273 < 0 => serious error, return this value from pcre_exec()
5274
5275 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5276 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5277 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5278 use by callout functions. It will never be used by PCRE itself.
5279
5280 (ii) The pcre_extra structure (see 45 above) has a void * field called
5281 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5282 pcre_callout_block structure has a field of the same name. The contents of
5283 the field passed in the pcre_extra structure are passed to the callout
5284 function in the corresponding field in the callout block. This makes it
5285 easier to use the same callout-containing regex from multiple threads. For
5286 testing, the pcretest program has a new data escape
5287
5288 \C*n pass the number n (may be negative) as callout_data
5289
5290 If the callout function in pcretest receives a non-zero value as
5291 callout_data, it returns that value.
5292
5293 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5294 there were some redundant $(CFLAGS) in commands that are now specified as
5295 $(LINK), which already includes $(CFLAGS).
5296
5297 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5298 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5299 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5300 one-byte characters throughout. Note that case-insensitive matching applies
5301 only to characters whose values are less than 256. PCRE doesn't support the
5302 notion of cases for higher-valued characters.
5303
5304 (i) A character class whose characters are all within 0-255 is handled as
5305 a bit map, and the map is inverted for negative classes. Previously, a
5306 character > 255 always failed to match such a class; however it should
5307 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5308
5309 (ii) A negated character class with a single character < 255 is coded as
5310 "not this character" (OP_NOT). This wasn't working properly when the test
5311 character was multibyte, either singly or repeated.
5312
5313 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5314 mode, for example: \x{100}{2,3}.
5315
5316 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5317 singly or repeated) now correctly test multibyte characters. However,
5318 PCRE doesn't recognize any characters with values greater than 255 as
5319 digits, spaces, or word characters. Such characters always match \D, \S,
5320 and \W, and never match \d, \s, or \w.
5321
5322 (v) Classes may now contain characters and character ranges with values
5323 greater than 255. For example: [ab\x{100}-\x{400}].
5324
5325 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5326 PCRE in UTF-8 mode.
5327
5328 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5329 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5330 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5331 value.)
5332
5333 53. The single man page has become too large. I have therefore split it up into
5334 a number of separate man pages. These also give rise to individual HTML pages;
5335 these are now put in a separate directory, and there is an index.html page that
5336 lists them all. Some hyperlinking between the pages has been installed.
5337
5338 54. Added convenience functions for handling named capturing parentheses.
5339
5340 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5341 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5342 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5343 are faulted.
5344
5345 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5346 calling configure. These values are used when compiling the dftables.c program
5347 which is run to generate the source of the default character tables. They
5348 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5349 you will need to set these values.
5350
5351 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5352
5353
5354 Version 3.9 02-Jan-02
5355 ---------------------
5356
5357 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5358
5359 2. If --disable-static was given, the building process failed when trying to
5360 build pcretest and pcregrep. (For some reason it was using libtool to compile
5361 them, which is not right, as they aren't part of the library.)
5362
5363
5364 Version 3.8 18-Dec-01
5365 ---------------------
5366
5367 1. The experimental UTF-8 code was completely screwed up. It was packing the
5368 bytes in the wrong order. How dumb can you get?
5369
5370
5371 Version 3.7 29-Oct-01
5372 ---------------------
5373
5374 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5375 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5376 this didn't happen under Solaris 8, where I normally test things.
5377
5378 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5379 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5380 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5381 this fix an hour or so after the initial 3.7 release.)
5382
5383
5384 Version 3.6 23-Oct-01
5385 ---------------------
5386
5387 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5388 offsets passed as NULL with zero offset count.
5389
5390 2. The config.guess and config.sub files had not been updated when I moved to
5391 the latest autoconf.
5392
5393
5394 Version 3.5 15-Aug-01
5395 ---------------------
5396
5397 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5398 had been forgotten.
5399
5400 2. By using declared but undefined structures, we can avoid using "void"
5401 definitions in pcre.h while keeping the internal definitions of the structures
5402 private.
5403
5404 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5405 user point of view, this means that both static and shared libraries are built
5406 by default, but this can be individually controlled. More of the work of
5407 handling this static/shared cases is now inside libtool instead of PCRE's make
5408 file.
5409
5410 4. The pcretest utility is now installed along with pcregrep because it is
5411 useful for users (to test regexs) and by doing this, it automatically gets
5412 relinked by libtool. The documentation has been turned into a man page, so
5413 there are now .1, .txt, and .html versions in /doc.
5414
5415 5. Upgrades to pcregrep:
5416 (i) Added long-form option names like gnu grep.
5417 (ii) Added --help to list all options with an explanatory phrase.
5418 (iii) Added -r, --recursive to recurse into sub-directories.
5419 (iv) Added -f, --file to read patterns from a file.
5420
5421 6. pcre_exec() was referring to its "code" argument before testing that
5422 argument for NULL (and giving an error if it was NULL).
5423
5424 7. Upgraded Makefile.in to allow for compiling in a different directory from
5425 the source directory.
5426
5427 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5428 options bits, the pointer it was passed was to an int instead of to an unsigned
5429 long int. This mattered only on 64-bit systems.
5430
5431 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5432 generated) instead of pcre.in, which it its source. Also made the same change
5433 in several of the .c files.
5434
5435 10. A new release of gcc defines printf() as a macro, which broke pcretest
5436 because it had an ifdef in the middle of a string argument for printf(). Fixed
5437 by using separate calls to printf().
5438
5439 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5440 script, to force use of CR or LF instead of \n in the source. On non-Unix
5441 systems, the value can be set in config.h.
5442
5443 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5444 absolute limit. Changed the text of the error message to make this clear, and
5445 likewise updated the man page.
5446
5447 13. The limit of 99 on the number of capturing subpatterns has been removed.
5448 The new limit is 65535, which I hope will not be a "real" limit.
5449
5450
5451 Version 3.4 22-Aug-00
5452 ---------------------
5453
5454 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5455
5456 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5457
5458
5459 Version 3.3 01-Aug-00
5460 ---------------------
5461
5462 1. If an octal character was given, but the value was greater than \377, it
5463 was not getting masked to the least significant bits, as documented. This could
5464 lead to crashes in some systems.
5465
5466 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5467 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5468
5469 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5470 These just pass their arguments on to (pcre_free)(), but they are provided
5471 because some uses of PCRE bind it to non-C systems that can call its functions,
5472 but cannot call free() or pcre_free() directly.
5473
5474 4. Add "make test" as a synonym for "make check". Corrected some comments in
5475 the Makefile.
5476
5477 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5478 Makefile.
5479
5480 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5481 command called pgrep for grepping around the active processes.
5482
5483 7. Added the beginnings of support for UTF-8 character strings.
5484
5485 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5486 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5487 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5488 out for the ar command.)
5489
5490
5491 Version 3.2 12-May-00
5492 ---------------------
5493
5494 This is purely a bug fixing release.
5495
5496 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5497 of ZA. This was just one example of several cases that could provoke this bug,
5498 which was introduced by change 9 of version 2.00. The code for breaking
5499 infinite loops after an iteration that matches an empty string was't working
5500 correctly.
5501
5502 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5503 when matched against abbab (for example). After matching an empty string, it
5504 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5505 caused it to match further down the string than it should.
5506
5507 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5508 was there because it doesn't seem to be needed, and it causes trouble on some
5509 systems, as it is not a Standard C header. It has been removed.
5510
5511 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5512 were reported on the Macintosh. The changes were from
5513
5514 while ((c = *(++ptr)) != 0 && c != '\n');
5515 to
5516 while ((c = *(++ptr)) != 0 && c != '\n') ;
5517
5518 Totally extraordinary, but if that's what it takes...
5519
5520 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5521 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5522 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5523 assumes the way PCRE uses memmove() (always moving upwards).
5524
5525 6. PCRE is being used in one environment where strchr() is not available. There
5526 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5527 faster code anyway.
5528
5529
5530 Version 3.1 09-Feb-00
5531 ---------------------
5532
5533 The only change in this release is the fixing of some bugs in Makefile.in for
5534 the "install" target:
5535
5536 (1) It was failing to install pcreposix.h.
5537
5538 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5539
5540
5541 Version 3.0 01-Feb-00
5542 ---------------------
5543
5544 1. Add support for the /+ modifier to perltest (to output $` like it does in
5545 pcretest).
5546
5547 2. Add support for the /g modifier to perltest.
5548
5549 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5550 matches null strings.
5551
5552 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5553 pattern. Perl treats empty patterns specially - it reuses the most recent
5554 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5555 effect.
5556
5557 5. The POSIX interface was broken in that it was just handing over the POSIX
5558 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5559 required a bigger vector, with some working space on the end. This means that
5560 the POSIX wrapper now has to get and free some memory, and copy the results.
5561
5562 6. Added some simple autoconf support, placing the test data and the
5563 documentation in separate directories, re-organizing some of the
5564 information files, and making it build pcre-config (a GNU standard). Also added
5565 libtool support for building PCRE as a shared library, which is now the
5566 default.
5567
5568 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5569 09 are not valid octal constants. Single digits will be used for minor values
5570 less than 10.
5571
5572 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5573 existing programs that set these in the POSIX interface can use PCRE without
5574 modification.
5575
5576 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5577 return all that pcre_info() returns, plus additional data. The pcre_info()
5578 function is retained for compatibility, but is considered to be obsolete.
5579
5580 10. Added experimental recursion feature (?R) to handle one common case that
5581 Perl 5.6 will be able to do with (?p{...}).
5582
5583 11. Added support for POSIX character classes like [:alpha:], which Perl is
5584 adopting.
5585
5586
5587 Version 2.08 31-Aug-99
5588 ----------------------
5589
5590 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5591 trying to match at the startoffset position, but instead was moving forward to
5592 the next newline as if a previous match had failed.
5593
5594 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5595 and could get into a loop if a null string was matched other than at the start
5596 of the subject.
5597
5598 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5599 be distinguished at compile time, and for completeness also added PCRE_DATE.
5600
5601 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5602 in GnuWin32 environments.
5603
5604
5605 Version 2.07 29-Jul-99
5606 ----------------------
5607
5608 1. The documentation is now supplied in plain text form and HTML as well as in
5609 the form of man page sources.
5610
5611 2. C++ compilers don't like assigning (void *) values to other pointer types.
5612 In particular this affects malloc(). Although there is no problem in Standard
5613 C, I've put in casts to keep C++ compilers happy.
5614
5615 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5616 should be (const char *).
5617
5618 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5619 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5620 However, I haven't made this a standard facility. The documentation doesn't
5621 mention it, and the Makefile doesn't support it.
5622
5623 5. The Makefile now contains an "install" target, with editable destinations at
5624 the top of the file. The pcretest program is not installed.
5625
5626 6. pgrep -V now gives the PCRE version number and date.
5627
5628 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5629 causing the entire string to be ignored, instead of just the last character.
5630
5631 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5632 non-matching string, it can take a very, very long time, even for strings of
5633 quite modest length, because of the nested recursion. PCRE now does better in
5634 some of these cases. It does this by remembering the last required literal
5635 character in the pattern, and pre-searching the subject to ensure it is present
5636 before running the real match. In other words, it applies a heuristic to detect
5637 some types of certain failure quickly, and in the above example, if presented
5638 with a string that has no trailing " it gives "no match" very quickly.
5639
5640 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5641 other alternatives are tried instead.
5642
5643
5644 Version 2.06 09-Jun-99
5645 ----------------------
5646
5647 1. Change pcretest's output for amount of store used to show just the code
5648 space, because the remainder (the data block) varies in size between 32-bit and
5649 64-bit systems.
5650
5651 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5652 start matching at. This allows lookbehinds to work when searching for multiple
5653 occurrences in a string.
5654
5655 3. Added additional options to pcretest for testing multiple occurrences:
5656
5657 /+ outputs the rest of the string that follows a match
5658 /g loops for multiple occurrences, using the new startoffset argument
5659 /G loops for multiple occurrences by passing an incremented pointer
5660
5661 4. PCRE wasn't doing the "first character" optimization for patterns starting
5662 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5663 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5664 the letter 'x'. On long subject strings, this gives a significant speed-up.
5665
5666
5667 Version 2.05 21-Apr-99
5668 ----------------------
5669
5670 1. Changed the type of magic_number from int to long int so that it works
5671 properly on 16-bit systems.
5672
5673 2. Fixed a bug which caused patterns starting with .* not to work correctly
5674 when the subject string contained newline characters. PCRE was assuming
5675 anchoring for such patterns in all cases, which is not correct because .* will
5676 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5677 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5678 must be retried after every newline in the subject.
5679
5680
5681 Version 2.04 18-Feb-99
5682 ----------------------
5683
5684 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5685 computation of the store needed to hold the pattern was incorrect (too large).
5686 If such patterns were nested a few deep, this could multiply and become a real
5687 problem.
5688
5689 2. Added /M option to pcretest to show the memory requirement of a specific
5690 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5691
5692 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5693 compiled in such a way that the backtracking after subsequent failure was
5694 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5695 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5696
5697
5698 Version 2.03 02-Feb-99
5699 ----------------------
5700
5701 1. Fixed typo and small mistake in man page.
5702
5703 2. Added 4th condition (GPL supersedes if conflict) and created separate
5704 LICENCE file containing the conditions.
5705
5706 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5707 Perl, that is the internal \ allows the delimiter to be included in the
5708 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5709 the final delimiter, add \ to the end of the pattern (to test the error).
5710
5711 4. Added the convenience functions for extracting substrings after a successful
5712 match. Updated pcretest to make it able to test these functions.
5713
5714
5715 Version 2.02 14-Jan-99
5716 ----------------------
5717
5718 1. Initialized the working variables associated with each extraction so that
5719 their saving and restoring doesn't refer to uninitialized store.
5720
5721 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5722 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5723 fix the problem.
5724
5725 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5726 calls, and wasn't printing the correct value for compiling calls. Increased the
5727 default value of LOOPREPEAT, and the number of significant figures in the
5728 times.
5729
5730 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5731
5732 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5733 a building problem on Windows NT with a FAT file system.
5734
5735
5736 Version 2.01 21-Oct-98
5737 ----------------------
5738
5739 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5740 to character tables built by pcre_maketables() in the current locale. If NULL
5741 is passed, the default tables are used.
5742
5743
5744 Version 2.00 24-Sep-98
5745 ----------------------
5746
5747 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5748 it any more.
5749
5750 2. Allow quantification of (?>) groups, and make it work correctly.
5751
5752 3. The first character computation wasn't working for (?>) groups.
5753
5754 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5755 end of the subject) and add 5.005's \z, which really does match only at the
5756 very end of the subject.
5757
5758 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5759
5760 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5761 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5762 localized options. All options to pcre_study() were also removed.
5763
5764 7. Add other new features from 5.005:
5765
5766 $(?<= positive lookbehind
5767 $(?<! negative lookbehind
5768 (?imsx-imsx) added the unsetting capability
5769 such a setting is global if at outer level; local otherwise
5770 (?imsx-imsx:) non-capturing groups with option setting
5771 (?(cond)re|re) conditional pattern matching
5772
5773 A backreference to itself in a repeated group matches the previous
5774 captured string.
5775
5776 8. General tidying up of studying (both automatic and via "study")
5777 consequential on the addition of new assertions.
5778
5779 9. As in 5.005, unlimited repeated groups that could match an empty substring
5780 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5781 runtime if any iteration does actually match an empty substring.
5782
5783 10. Include the RunTest script in the distribution.
5784
5785 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5786 discrepancies, some of which were old and were also with respect to 5.004. They
5787 have now been fixed.
5788
5789
5790 Version 1.09 28-Apr-98
5791 ----------------------
5792
5793 1. A negated single character class followed by a quantifier with a minimum
5794 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5795 program crashes, or just wrong answers. This did not apply to negated classes
5796 containing more than one character, or to minima other than one.
5797
5798
5799 Version 1.08 27-Mar-98
5800 ----------------------
5801
5802 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5803
5804 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5805 latter must appear before anything that relies on it in the pattern.
5806
5807
5808 Version 1.07 16-Feb-98
5809 ----------------------
5810
5811 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5812 repeat of a potentially empty string).
5813
5814
5815 Version 1.06 23-Jan-98
5816 ----------------------
5817
5818 1. Added Markus Oberhumer's little patches for C++.
5819
5820 2. Literal strings longer than 255 characters were broken.
5821
5822
5823 Version 1.05 23-Dec-97
5824 ----------------------
5825
5826 1. Negated character classes containing more than one character were failing if
5827 PCRE_CASELESS was set at run time.
5828
5829
5830 Version 1.04 19-Dec-97
5831 ----------------------
5832
5833 1. Corrected the man page, where some "const" qualifiers had been omitted.
5834
5835 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5836 input syntax.
5837
5838 3. Fixed memory leak which occurred when a regex with back references was
5839 matched with an offsets vector that wasn't big enough. The temporary memory
5840 that is used in this case wasn't being freed if the match failed.
5841
5842 4. Tidied pcretest to ensure it frees memory that it gets.
5843
5844 5. Temporary memory was being obtained in the case where the passed offsets
5845 vector was exactly big enough.
5846
5847 6. Corrected definition of offsetof() from change 5 below.
5848
5849 7. I had screwed up change 6 below and broken the rules for the use of
5850 setjmp(). Now fixed.
5851
5852
5853 Version 1.03 18-Dec-97
5854 ----------------------
5855
5856 1. A erroneous regex with a missing opening parenthesis was correctly
5857 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5858 on some systems.
5859
5860 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5861 it was reported that one broken compiler failed on the former because "code" is
5862 also an independent variable.
5863
5864 3. The erroneous regex a[]b caused an array overrun reference.
5865
5866 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5867 fail on data ending with that character. (It was going on too far, and checking
5868 the next character, typically a binary zero.) This was specific to the
5869 optimized code for single-character negative classes.
5870
5871 5. Added a contributed patch from the TIN world which does the following:
5872
5873 + Add an undef for memmove, in case the the system defines a macro for it.
5874
5875 + Add a definition of offsetof(), in case there isn't one. (I don't know
5876 the reason behind this - offsetof() is part of the ANSI standard - but
5877 it does no harm).
5878
5879 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5880 most of the places where whitespace preceded '#'. I have given up and
5881 allowed the remaining 2 cases to be at the margin.
5882
5883 + Rename some variables in pcre to eliminate shadowing. This seems very
5884 pedantic, but does no harm, of course.
5885
5886 6. Moved the call to setjmp() into its own function, to get rid of warnings
5887 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5888
5889 7. Constructs such as \d{8,} were compiling into the equivalent of
5890 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5891 outcome, but in this particular case used more store than had been allocated,
5892 which caused the bug to be discovered because it threw up an internal error.
5893
5894 8. The debugging code in both pcre and pcretest for outputting the compiled
5895 form of a regex was going wrong in the case of back references followed by
5896 curly-bracketed repeats.
5897
5898
5899 Version 1.02 12-Dec-97
5900 ----------------------
5901
5902 1. Typos in pcre.3 and comments in the source fixed.
5903
5904 2. Applied a contributed patch to get rid of places where it used to remove
5905 'const' from variables, and fixed some signed/unsigned and uninitialized
5906 variable warnings.
5907
5908 3. Added the "runtest" target to Makefile.
5909
5910 4. Set default compiler flag to -O2 rather than just -O.
5911
5912
5913 Version 1.01 19-Nov-97
5914 ----------------------
5915
5916 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5917 like /([ab]*)*/, that is, for classes with more than one character in them.
5918
5919 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5920 as /((?>a*))*/ (a PCRE_EXTRA facility).
5921
5922
5923 Version 1.00 18-Nov-97
5924 ----------------------
5925
5926 1. Added compile-time macros to support systems such as SunOS4 which don't have
5927 memmove() or strerror() but have other things that can be used instead.
5928
5929 2. Arranged that "make clean" removes the executables.
5930
5931
5932 Version 0.99 27-Oct-97
5933 ----------------------
5934
5935 1. Fixed bug in code for optimizing classes with only one character. It was
5936 initializing a 32-byte map regardless, which could cause it to run off the end
5937 of the memory it had got.
5938
5939 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5940
5941
5942 Version 0.98 22-Oct-97
5943 ----------------------
5944
5945 1. Fixed bug in code for handling temporary memory usage when there are more
5946 back references than supplied space in the ovector. This could cause segfaults.
5947
5948
5949 Version 0.97 21-Oct-97
5950 ----------------------
5951
5952 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5953
5954 2. Optimized negated single characters not to use a bit map.
5955
5956 3. Brought error texts together as macro definitions; clarified some of them;
5957 fixed one that was wrong - it said "range out of order" when it meant "invalid
5958 escape sequence".
5959
5960 4. Changed some char * arguments to const char *.
5961
5962 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5963
5964 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5965 pcretest.
5966
5967
5968 Version 0.96 16-Oct-97
5969 ----------------------
5970
5971 1. Added a simple "pgrep" utility to the distribution.
5972
5973 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5974 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5975 where "ddd" means "one or more decimal digits".
5976
5977 3. Fixed serious bug. If a pattern had a back reference, but the call to
5978 pcre_exec() didn't supply a large enough ovector to record the related
5979 identifying subpattern, the match always failed. PCRE now remembers the number
5980 of the largest back reference, and gets some temporary memory in which to save
5981 the offsets during matching if necessary, in order to ensure that
5982 backreferences always work.
5983
5984 4. Increased the compatibility with Perl in a number of ways:
5985
5986 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5987 to request this handling. The option can be set at compile or exec time.
5988
5989 (b) $ matches before a terminating newline by default; an option
5990 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5991 mode). The option can be set at compile or exec time.
5992
5993 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5994 the same as Perl's. If the decimal number it represents is less than 10
5995 or there aren't that many previous left capturing parentheses, an octal
5996 escape is read. Inside a character class, it's always an octal escape,
5997 even if it is a single digit.
5998
5999 (d) An escaped but undefined alphabetic character is taken as a literal,
6000 unless PCRE_EXTRA is set. Currently this just reserves the remaining
6001 escapes.
6002
6003 (e) {0} is now permitted. (The previous item is removed from the compiled
6004 pattern).
6005
6006 5. Changed all the names of code files so that the basic parts are no longer
6007 than 10 characters, and abolished the teeny "globals.c" file.
6008
6009 6. Changed the handling of character classes; they are now done with a 32-byte
6010 bit map always.
6011
6012 7. Added the -d and /D options to pcretest to make it possible to look at the
6013 internals of compilation without having to recompile pcre.
6014
6015
6016 Version 0.95 23-Sep-97
6017 ----------------------
6018
6019 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
6020 \x20 at the start of a run of normal characters. These were being treated as
6021 real characters, instead of the source characters being re-checked.
6022
6023
6024 Version 0.94 18-Sep-97
6025 ----------------------
6026
6027 1. The functions are now thread-safe, with the caveat that the global variables
6028 containing pointers to malloc() and free() or alternative functions are the
6029 same for all threads.
6030
6031 2. Get pcre_study() to generate a bitmap of initial characters for non-
6032 anchored patterns when this is possible, and use it if passed to pcre_exec().
6033
6034
6035 Version 0.93 15-Sep-97
6036 ----------------------
6037
6038 1. /(b)|(:+)/ was computing an incorrect first character.
6039
6040 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
6041 but not actually doing anything yet.
6042
6043 3. Treat "-" characters in classes that cannot be part of ranges as literals,
6044 as Perl does (e.g. [-az] or [az-]).
6045
6046 4. Set the anchored flag if a branch starts with .* or .*? because that tests
6047 all possible positions.
6048
6049 5. Split up into different modules to avoid including unneeded functions in a
6050 compiled binary. However, compile and exec are still in one module. The "study"
6051 function is split off.
6052
6053 6. The character tables are now in a separate module whose source is generated
6054 by an auxiliary program - but can then be edited by hand if required. There are
6055 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
6056 toupper() in the code.
6057
6058 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
6059 make them global. Abolish the function for setting them, as the caller can now
6060 set them directly.