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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1619 - (show annotations)
Sat Dec 5 16:58:46 2015 UTC (3 years, 10 months ago) by ph10
File size: 287252 byte(s)
Fix (by hacking) another length computation issue.
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 8.39 xx-xxxxxx-201x
8 ---------------------------
9
10 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
11 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
12 misbehaved. This bug was found by the LLVM fuzzer.
13
14 2. Similar to the above, if an isolated \E was present between an item and its
15 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
16 bug was found by the LLVM fuzzer.
17
18 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
19 working correctly in UCP mode.
20
21 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
22 was set when the pmatch argument was NULL. It now returns REG_INVARG.
23
24 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
25
26 6. An empty \Q\E sequence between an item and its qualifier caused
27 pcre_compile() to misbehave when auto callouts were enabled. This bug was
28 found by the LLVM fuzzer.
29
30 7. If a pattern that was compiled with PCRE_EXTENDED started with white
31 space or a #-type comment that was followed by (?-x), which turns off
32 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
33 pcre_compile() assumed that (?-x) applied to the whole pattern and
34 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
35
36 8. An call of pcre_copy_named_substring() for a named substring whose number
37 was greater than the space in the ovector could cause a crash.
38
39 9. Yet another buffer overflow bug involved duplicate named groups with a
40 group that reset capture numbers (compare 8.38/7 below). Once again, I have
41 just allowed for more memory, even if not needed. (A proper fix is
42 implemented in PCRE2, but it involves a lot of refactoring.)
43
44
45 Version 8.38 23-November-2015
46 -----------------------------
47
48 1. If a group that contained a recursive back reference also contained a
49 forward reference subroutine call followed by a non-forward-reference
50 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
51 compile correct code, leading to undefined behaviour or an internally
52 detected error. This bug was discovered by the LLVM fuzzer.
53
54 2. Quantification of certain items (e.g. atomic back references) could cause
55 incorrect code to be compiled when recursive forward references were
56 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
57 This bug was discovered by the LLVM fuzzer.
58
59 3. A repeated conditional group whose condition was a reference by name caused
60 a buffer overflow if there was more than one group with the given name.
61 This bug was discovered by the LLVM fuzzer.
62
63 4. A recursive back reference by name within a group that had the same name as
64 another group caused a buffer overflow. For example:
65 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
66
67 5. A forward reference by name to a group whose number is the same as the
68 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
69 a buffer overflow at compile time. This bug was discovered by the LLVM
70 fuzzer.
71
72 6. A lookbehind assertion within a set of mutually recursive subpatterns could
73 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
74
75 7. Another buffer overflow bug involved duplicate named groups with a
76 reference between their definition, with a group that reset capture
77 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
78 fixed by always allowing for more memory, even if not needed. (A proper fix
79 is implemented in PCRE2, but it involves more refactoring.)
80
81 8. There was no check for integer overflow in subroutine calls such as (?123).
82
83 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
84 being treated as a literal 'l' instead of causing an error.
85
86 10. There was a buffer overflow if pcre_exec() was called with an ovector of
87 size 1. This bug was found by american fuzzy lop.
88
89 11. If a non-capturing group containing a conditional group that could match
90 an empty string was repeated, it was not identified as matching an empty
91 string itself. For example: /^(?:(?(1)x|)+)+$()/.
92
93 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
94 \a and \e in test subject lines.
95
96 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
97 instead of the EBCDIC value.
98
99 14. The handling of \c in an EBCDIC environment has been revised so that it is
100 now compatible with the specification in Perl's perlebcdic page.
101
102 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
103 ASCII/Unicode. This has now been added to the list of characters that are
104 recognized as white space in EBCDIC.
105
106 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
107 error (correctly) when used outside a class, but did not give an error
108 within a class.
109
110 17. \h within a class was incorrectly compiled in EBCDIC environments.
111
112 18. A pattern with an unmatched closing parenthesis that contained a backward
113 assertion which itself contained a forward reference caused buffer
114 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
115
116 19. JIT should return with error when the compiled pattern requires more stack
117 space than the maximum.
118
119 20. A possessively repeated conditional group that could match an empty string,
120 for example, /(?(R))*+/, was incorrectly compiled.
121
122 21. Fix infinite recursion in the JIT compiler when certain patterns such as
123 /(?:|a|){100}x/ are analysed.
124
125 22. Some patterns with character classes involving [: and \\ were incorrectly
126 compiled and could cause reading from uninitialized memory or an incorrect
127 error diagnosis.
128
129 23. Pathological patterns containing many nested occurrences of [: caused
130 pcre_compile() to run for a very long time.
131
132 24. A conditional group with only one branch has an implicit empty alternative
133 branch and must therefore be treated as potentially matching an empty
134 string.
135
136 25. If (?R was followed by - or + incorrect behaviour happened instead of a
137 diagnostic.
138
139 26. Arrange to give up on finding the minimum matching length for overly
140 complex patterns.
141
142 27. Similar to (4) above: in a pattern with duplicated named groups and an
143 occurrence of (?| it is possible for an apparently non-recursive back
144 reference to become recursive if a later named group with the relevant
145 number is encountered. This could lead to a buffer overflow. Wen Guanxing
146 from Venustech ADLAB discovered this bug.
147
148 28. If pcregrep was given the -q option with -c or -l, or when handling a
149 binary file, it incorrectly wrote output to stdout.
150
151 29. The JIT compiler did not restore the control verb head in case of *THEN
152 control verbs. This issue was found by Karl Skomski with a custom LLVM
153 fuzzer.
154
155 30. Error messages for syntax errors following \g and \k were giving inaccurate
156 offsets in the pattern.
157
158 31. Added a check for integer overflow in conditions (?(<digits>) and
159 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
160 fuzzer.
161
162 32. Handling recursive references such as (?2) when the reference is to a group
163 later in the pattern uses code that is very hacked about and error-prone.
164 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
165 give an internal error if it is obvious that compiling has gone wrong.
166
167 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
168 This issue was found by Karl Skomski with a custom LLVM fuzzer.
169
170 34. The JIT compiler should restore the control chain for empty possessive
171 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
172
173 35. Match limit check added to JIT recursion. This issue was found by Karl
174 Skomski with a custom LLVM fuzzer.
175
176 36. Yet another case similar to 27 above has been circumvented by an
177 unconditional allocation of extra memory. This issue is fixed "properly" in
178 PCRE2 by refactoring the way references are handled. Wen Guanxing
179 from Venustech ADLAB discovered this bug.
180
181 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
182 with a custom LLVM fuzzer.
183
184 38. Fixed a corner case of range optimization in JIT.
185
186 39. An incorrect error "overran compiling workspace" was given if there were
187 exactly enough group forward references such that the last one extended
188 into the workspace safety margin. The next one would have expanded the
189 workspace. The test for overflow was not including the safety margin.
190
191 40. A match limit issue is fixed in JIT which was found by Karl Skomski
192 with a custom LLVM fuzzer.
193
194 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
195 work under Windows. (Why has it taken so long for anyone to notice?)
196
197 42. In a character class such as [\W\p{Any}] where both a negative-type escape
198 ("not a word character") and a property escape were present, the property
199 escape was being ignored.
200
201 43. Fix crash caused by very long (*MARK) or (*THEN) names.
202
203 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
204 by a single ASCII character in a class item, was incorrectly compiled in
205 UCP mode. The POSIX class got lost, but only if the single character
206 followed it.
207
208 45. [:punct:] in UCP mode was matching some characters in the range 128-255
209 that should not have been matched.
210
211 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
212 class, all characters with code points greater than 255 are in the class.
213 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
214 such as \w are turned into Unicode properties), wide characters were not
215 correctly handled, and could fail to match.
216
217
218 Version 8.37 28-April-2015
219 --------------------------
220
221 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
222 for those parentheses to be closed with whatever has been captured so far.
223 However, it was failing to mark any other groups between the hightest
224 capture so far and the currrent group as "unset". Thus, the ovector for
225 those groups contained whatever was previously there. An example is the
226 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
227
228 2. If an assertion condition was quantified with a minimum of zero (an odd
229 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
230
231 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
232 unrecognized modifier, a crash could occur.
233
234 4. An attempt to do global matching in pcretest with a zero-length ovector
235 caused a crash.
236
237 5. Fixed a memory leak during matching that could occur for a subpattern
238 subroutine call (recursive or otherwise) if the number of captured groups
239 that had to be saved was greater than ten.
240
241 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
242 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
243 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
244
245 7. A UTF pattern containing a "not" match of a non-ASCII character and a
246 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
247
248 8. When a pattern is compiled, it remembers the highest back reference so that
249 when matching, if the ovector is too small, extra memory can be obtained to
250 use instead. A conditional subpattern whose condition is a check on a
251 capture having happened, such as, for example in the pattern
252 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
253 setting the highest backreference number. This mattered only if pcre_exec()
254 was called with an ovector that was too small to hold the capture, and there
255 was no other kind of back reference (a situation which is probably quite
256 rare). The effect of the bug was that the condition was always treated as
257 FALSE when the capture could not be consulted, leading to a incorrect
258 behaviour by pcre_exec(). This bug has been fixed.
259
260 9. A reference to a duplicated named group (either a back reference or a test
261 for being set in a conditional) that occurred in a part of the pattern where
262 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
263 to be incorrectly calculated, leading to overwriting.
264
265 10. A mutually recursive set of back references such as (\2)(\1) caused a
266 segfault at study time (while trying to find the minimum matching length).
267 The infinite loop is now broken (with the minimum length unset, that is,
268 zero).
269
270 11. If an assertion that was used as a condition was quantified with a minimum
271 of zero, matching went wrong. In particular, if the whole group had
272 unlimited repetition and could match an empty string, a segfault was
273 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
274 assertions to be quantified, but not if they are being used as conditions,
275 so the above pattern is faulted by Perl. PCRE has now been changed so that
276 it also rejects such patterns.
277
278 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
279 failed to allow the zero-repeat case if pcre2_exec() was called with an
280 ovector too small to capture the group.
281
282 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
283 Red Hat Product Security:
284
285 (a) A crash if /K and /F were both set with the option to save the compiled
286 pattern.
287
288 (b) Another crash if the option to print captured substrings in a callout
289 was combined with setting a null ovector, for example \O\C+ as a subject
290 string.
291
292 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
293 forward reference repeated a large (but limited) number of times within a
294 repeated outer group that has a zero minimum quantifier, caused incorrect
295 code to be compiled, leading to the error "internal error:
296 previously-checked referenced subpattern not found" when an incorrect
297 memory address was read. This bug was reported as "heap overflow",
298 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
299 CVE-2015-2325.
300
301 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
302 call within a group that also contained a recursive back reference caused
303 incorrect code to be compiled. This bug was reported as "heap overflow",
304 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
305 number CVE-2015-2326.
306
307 24. Computing the size of the JIT read-only data in advance has been a source
308 of various issues, and new ones are still appear unfortunately. To fix
309 existing and future issues, size computation is eliminated from the code,
310 and replaced by on-demand memory allocation.
311
312 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
313 adjacent to the end of the range, and the range contained characters with
314 more than one other case, caused incorrect behaviour when compiled in UTF
315 mode. In that example, the range a-j was left out of the class.
316
317 26. Fix JIT compilation of conditional blocks, which assertion
318 is converted to (*FAIL). E.g: /(?(?!))/.
319
320 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
321 discovered by the LLVM fuzzer.
322
323 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
324 when this assertion was used as a condition, for example (?(?!)a|b). In
325 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
326 error about an unsupported item.
327
328 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
329 possessification code could take exponential time to complete. A recursion
330 depth limit of 1000 has been imposed to limit the resources used by this
331 optimization.
332
333 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
334 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
335 because \S ensures they are all in the class. The code for doing this was
336 interacting badly with the code for computing the amount of space needed to
337 compile the pattern, leading to a buffer overflow. This bug was discovered
338 by the LLVM fuzzer.
339
340 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
341 other kinds of group caused stack overflow at compile time. This bug was
342 discovered by the LLVM fuzzer.
343
344 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
345 between a subroutine call and its quantifier was incorrectly compiled,
346 leading to buffer overflow or other errors. This bug was discovered by the
347 LLVM fuzzer.
348
349 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
350 assertion after (?(. The code was failing to check the character after
351 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
352 was discovered by the LLVM fuzzer.
353
354 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
355 a fixed maximum following a group that contains a subroutine reference was
356 incorrectly compiled and could trigger buffer overflow. This bug was
357 discovered by the LLVM fuzzer.
358
359 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
360 caused a stack overflow instead of the diagnosis of a non-fixed length
361 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
362
363 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
364 (e.g. /(?<=\Ka)/) could make pcregrep loop.
365
366 37. There was a similar problem to 36 in pcretest for global matches.
367
368 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
369 and a subsequent item in the pattern caused a non-match, backtracking over
370 the repeated \X did not stop, but carried on past the start of the subject,
371 causing reference to random memory and/or a segfault. There were also some
372 other cases where backtracking after \C could crash. This set of bugs was
373 discovered by the LLVM fuzzer.
374
375 39. The function for finding the minimum length of a matching string could take
376 a very long time if mutual recursion was present many times in a pattern,
377 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
378 method has been implemented. This infelicity was discovered by the LLVM
379 fuzzer.
380
381 40. Static linking against the PCRE library using the pkg-config module was
382 failing on missing pthread symbols.
383
384
385 Version 8.36 26-September-2014
386 ------------------------------
387
388 1. Got rid of some compiler warnings in the C++ modules that were shown up by
389 -Wmissing-field-initializers and -Wunused-parameter.
390
391 2. The tests for quantifiers being too big (greater than 65535) were being
392 applied after reading the number, and stupidly assuming that integer
393 overflow would give a negative number. The tests are now applied as the
394 numbers are read.
395
396 3. Tidy code in pcre_exec.c where two branches that used to be different are
397 now the same.
398
399 4. The JIT compiler did not generate match limit checks for certain
400 bracketed expressions with quantifiers. This may lead to exponential
401 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
402 issue should be resolved now.
403
404 5. Fixed an issue, which occures when nested alternatives are optimized
405 with table jumps.
406
407 6. Inserted two casts and changed some ints to size_t in the light of some
408 reported 64-bit compiler warnings (Bugzilla 1477).
409
410 7. Fixed a bug concerned with zero-minimum possessive groups that could match
411 an empty string, which sometimes were behaving incorrectly in the
412 interpreter (though correctly in the JIT matcher). This pcretest input is
413 an example:
414
415 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
416 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
417
418 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
419 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
420 for an empty string was breaking the inner loop and carrying on at a lower
421 level, when possessive repeated groups should always return to a higher
422 level as they have no backtrack points in them. The empty string test now
423 occurs at the outer level.
424
425 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
426 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
427
428 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
429 doesn't).
430
431 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
432 to hold a pattern to be incorrectly computed (too small) when there were
433 named back references to duplicated names. This could cause "internal
434 error: code overflow" or "double free or corruption" or other memory
435 handling errors.
436
437 11. When named subpatterns had the same prefixes, back references could be
438 confused. For example, in this pattern:
439
440 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
441
442 the reference to 'Name' was incorrectly treated as a reference to a
443 duplicate name.
444
445 12. A pattern such as /^s?c/mi8 where the optional character has more than
446 one "other case" was incorrectly compiled such that it would only try to
447 match starting at "c".
448
449 13. When a pattern starting with \s was studied, VT was not included in the
450 list of possible starting characters; this should have been part of the
451 8.34/18 patch.
452
453 14. If a character class started [\Qx]... where x is any character, the class
454 was incorrectly terminated at the ].
455
456 15. If a pattern that started with a caseless match for a character with more
457 than one "other case" was studied, PCRE did not set up the starting code
458 unit bit map for the list of possible characters. Now it does. This is an
459 optimization improvement, not a bug fix.
460
461 16. The Unicode data tables have been updated to Unicode 7.0.0.
462
463 17. Fixed a number of memory leaks in pcregrep.
464
465 18. Avoid a compiler warning (from some compilers) for a function call with
466 a cast that removes "const" from an lvalue by using an intermediate
467 variable (to which the compiler does not object).
468
469 19. Incorrect code was compiled if a group that contained an internal recursive
470 back reference was optional (had quantifier with a minimum of zero). This
471 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
472 caused segmentation faults because of stack overflows at compile time.
473
474 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
475 group that is quantified with an indefinite repeat, caused a compile-time
476 loop which used up all the system stack and provoked a segmentation fault.
477 This was not the same bug as 19 above.
478
479 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
480 Patch by Mike Frysinger.
481
482
483 Version 8.35 04-April-2014
484 --------------------------
485
486 1. A new flag is set, when property checks are present in an XCLASS.
487 When this flag is not set, PCRE can perform certain optimizations
488 such as studying these XCLASS-es.
489
490 2. The auto-possessification of character sets were improved: a normal
491 and an extended character set can be compared now. Furthermore
492 the JIT compiler optimizes more character set checks.
493
494 3. Got rid of some compiler warnings for potentially uninitialized variables
495 that show up only when compiled with -O2.
496
497 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
498 of a match later then the end of the match. The pcretest program was not
499 handling the case sensibly - it was outputting from the start to the next
500 binary zero. It now reports this situation in a message, and outputs the
501 text from the end to the start.
502
503 5. Fast forward search is improved in JIT. Instead of the first three
504 characters, any three characters with fixed position can be searched.
505 Search order: first, last, middle.
506
507 6. Improve character range checks in JIT. Characters are read by an inprecise
508 function now, which returns with an unknown value if the character code is
509 above a certain threshold (e.g: 256). The only limitation is that the value
510 must be bigger than the threshold as well. This function is useful when
511 the characters above the threshold are handled in the same way.
512
513 7. The macros whose names start with RAWUCHAR are placeholders for a future
514 mode in which only the bottom 21 bits of 32-bit data items are used. To
515 make this more memorable for those maintaining the code, the names have
516 been changed to start with UCHAR21, and an extensive comment has been added
517 to their definition.
518
519 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
520 to the export list in Makefile.am (they were accidentally omitted from the
521 8.34 tarball).
522
523 9. The informational output from pcretest used the phrase "starting byte set"
524 which is inappropriate for the 16-bit and 32-bit libraries. As the output
525 for "first char" and "need char" really means "non-UTF-char", I've changed
526 "byte" to "char", and slightly reworded the output. The documentation about
527 these values has also been (I hope) clarified.
528
529 10. Another JIT related optimization: use table jumps for selecting the correct
530 backtracking path, when more than four alternatives are present inside a
531 bracket.
532
533 11. Empty match is not possible, when the minimum length is greater than zero,
534 and there is no \K in the pattern. JIT should avoid empty match checks in
535 such cases.
536
537 12. In a caseless character class with UCP support, when a character with more
538 than one alternative case was not the first character of a range, not all
539 the alternative cases were added to the class. For example, s and \x{17f}
540 are both alternative cases for S: the class [RST] was handled correctly,
541 but [R-T] was not.
542
543 13. The configure.ac file always checked for pthread support when JIT was
544 enabled. This is not used in Windows, so I have put this test inside a
545 check for the presence of windows.h (which was already tested for).
546
547 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
548 The algorithm provides a way to skip certain starting offsets, and usually
549 faster than linear prefix searches.
550
551 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
552 as for 'fr_FR' and 'french'. For some reason, however, it then used the
553 Windows-specific input and output files, which have 'french' screwed in.
554 So this could never have worked. One of the problems with locales is that
555 they aren't always the same. I have now updated RunTest so that it checks
556 the output of the locale test (test 3) against three different output
557 files, and it allows the test to pass if any one of them matches. With luck
558 this should make the test pass on some versions of Solaris where it was
559 failing. Because of the uncertainty, the script did not used to stop if
560 test 3 failed; it now does. If further versions of a French locale ever
561 come to light, they can now easily be added.
562
563 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
564 there was a message during ./configure, but it did not stop. This now
565 provokes an error. The invalid example in README has been corrected.
566 If a value less than the minimum is given, the minimum value has always
567 been used, but now a warning is given.
568
569 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
570 was a bug in the test system, which is now fixed. Also, the list of various
571 configurations that are tested for each release did not have one with both
572 16/32 bits and --enable-bar-anycrlf. It now does.
573
574 18. pcretest was missing "-C bsr" for displaying the \R default setting.
575
576 19. Little endian PowerPC systems are supported now by the JIT compiler.
577
578 20. The fast forward newline mechanism could enter to an infinite loop on
579 certain invalid UTF-8 input. Although we don't support these cases
580 this issue can be fixed by a performance optimization.
581
582 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
583 not take account if existing stack usage. There is now a new global
584 variable called pcre_stack_guard that can be set to point to an external
585 function to check stack availability. It is called at the start of
586 processing every parenthesized group.
587
588 22. A typo in the code meant that in ungreedy mode the max/min qualifier
589 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
590 not match "ab".
591
592 23. When UTF was disabled, the JIT program reported some incorrect compile
593 errors. These messages are silenced now.
594
595 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
596 compiler.
597
598 25. Change all the temporary files used in RunGrepTest to be different to those
599 used by RunTest so that the tests can be run simultaneously, for example by
600 "make -j check".
601
602
603 Version 8.34 15-December-2013
604 -----------------------------
605
606 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
607 executable memory. Patch inspired by Carsten Klein.
608
609 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
610 this macro is never tested and has no effect, because the work to support
611 coverage involves only compiling and linking options and special targets in
612 the Makefile. The comment in config.h implied that defining the macro would
613 enable coverage support, which is totally false. There was also support for
614 setting this macro in the CMake files (my fault, I just copied it from
615 configure). SUPPORT_GCOV has now been removed.
616
617 3. Make a small performance improvement in strlen16() and strlen32() in
618 pcretest.
619
620 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
621 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
622 these cases). There was also one in pcretest.c.
623
624 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
625
626 6. In UTF mode, the code for checking whether a group could match an empty
627 string (which is used for indefinitely repeated groups to allow for
628 breaking an infinite loop) was broken when the group contained a repeated
629 negated single-character class with a character that occupied more than one
630 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
631 UTF-8 mode). The effect was undefined: the group might or might not be
632 deemed as matching an empty string, or the program might have crashed.
633
634 7. The code for checking whether a group could match an empty string was not
635 recognizing that \h, \H, \v, \V, and \R must match a character.
636
637 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
638 an empty string. If it can, pcretest shows this in its information output.
639
640 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
641 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
642 matched by pcre_exec() without using JIT:
643
644 (a) If the rest of the pattern did not match after a maximal run of
645 grapheme clusters, the code for backing up to try with fewer of them
646 did not always back up over a full grapheme when characters that do not
647 have the modifier quality were involved, e.g. Hangul syllables.
648
649 (b) If the match point in a subject started with modifier character, and
650 there was no match, the code could incorrectly back up beyond the match
651 point, and potentially beyond the first character in the subject,
652 leading to a segfault or an incorrect match result.
653
654 10. A conditional group with an assertion condition could lead to PCRE
655 recording an incorrect first data item for a match if no other first data
656 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
657 first data item, and therefore matched "ca" after "c" instead of at the
658 start.
659
660 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
661 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
662
663 12. The source of pcregrep now includes z/OS-specific code so that it can be
664 compiled for z/OS as part of the special z/OS distribution.
665
666 13. Added the -T and -TM options to pcretest.
667
668 14. The code in pcre_compile.c for creating the table of named capturing groups
669 has been refactored. Instead of creating the table dynamically during the
670 actual compiling pass, the information is remembered during the pre-compile
671 pass (on the stack unless there are more than 20 named groups, in which
672 case malloc() is used) and the whole table is created before the actual
673 compile happens. This has simplified the code (it is now nearly 150 lines
674 shorter) and prepared the way for better handling of references to groups
675 with duplicate names.
676
677 15. A back reference to a named subpattern when there is more than one of the
678 same name now checks them in the order in which they appear in the pattern.
679 The first one that is set is used for the reference. Previously only the
680 first one was inspected. This change makes PCRE more compatible with Perl.
681
682 16. Unicode character properties were updated from Unicode 6.3.0.
683
684 17. The compile-time code for auto-possessification has been refactored, based
685 on a patch by Zoltan Herczeg. It now happens after instead of during
686 compilation. The code is cleaner, and more cases are handled. The option
687 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
688 options in pcretest are provided to set it. It can also be set by
689 (*NO_AUTO_POSSESS) at the start of a pattern.
690
691 18. The character VT has been added to the default ("C" locale) set of
692 characters that match \s and are generally treated as white space,
693 following this same change in Perl 5.18. There is now no difference between
694 "Perl space" and "POSIX space". Whether VT is treated as white space in
695 other locales depends on the locale.
696
697 19. The code for checking named groups as conditions, either for being set or
698 for being recursed, has been refactored (this is related to 14 and 15
699 above). Processing unduplicated named groups should now be as fast at
700 numerical groups, and processing duplicated groups should be faster than
701 before.
702
703 20. Two patches to the CMake build system, by Alexander Barkov:
704
705 (1) Replace the "source" command by "." in CMakeLists.txt because
706 "source" is a bash-ism.
707
708 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
709 without these the CMake build does not work on Solaris.
710
711 21. Perl has changed its handling of \8 and \9. If there is no previously
712 encountered capturing group of those numbers, they are treated as the
713 literal characters 8 and 9 instead of a binary zero followed by the
714 literals. PCRE now does the same.
715
716 22. Following Perl, added \o{} to specify codepoints in octal, making it
717 possible to specify values greater than 0777 and also making them
718 unambiguous.
719
720 23. Perl now gives an error for missing closing braces after \x{... instead of
721 treating the string as literal. PCRE now does the same.
722
723 24. RunTest used to grumble if an inappropriate test was selected explicitly,
724 but just skip it when running all tests. This make it awkward to run ranges
725 of tests when one of them was inappropriate. Now it just skips any
726 inappropriate tests, as it always did when running all tests.
727
728 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
729 character types such as \d or \w, too many callouts were inserted, and the
730 data that they returned was rubbish.
731
732 26. In UCP mode, \s was not matching two of the characters that Perl matches,
733 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
734 were matched by \h. The code has now been refactored so that the lists of
735 the horizontal and vertical whitespace characters used for \h and \v (which
736 are defined only in one place) are now also used for \s.
737
738 27. Add JIT support for the 64 bit TileGX architecture.
739 Patch by Jiong Wang (Tilera Corporation).
740
741 28. Possessive quantifiers for classes (both explicit and automatically
742 generated) now use special opcodes instead of wrapping in ONCE brackets.
743
744 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
745 (because it's meaningless), this was not happening when PCRE_CASELESS was
746 set. Not wrong, but inefficient.
747
748 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
749 properties for \w, \d, etc) is present in a test regex. Otherwise if the
750 test contains no characters greater than 255, Perl doesn't realise it
751 should be using Unicode semantics.
752
753 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
754 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
755 does in Unicode mode.
756
757 32. Added the "forbid" facility to pcretest so that putting tests into the
758 wrong test files can sometimes be quickly detected.
759
760 33. There is now a limit (default 250) on the depth of nesting of parentheses.
761 This limit is imposed to control the amount of system stack used at compile
762 time. It can be changed at build time by --with-parens-nest-limit=xxx or
763 the equivalent in CMake.
764
765 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
766 errors. Perl warns for these when in warning mode, but PCRE has no facility
767 for giving warnings.
768
769 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
770 However, this was not working for (?!) because it is optimized to (*FAIL),
771 for which PCRE does not allow quantifiers. The optimization is now disabled
772 when a quantifier follows (?!). I can't see any use for this, but it makes
773 things uniform.
774
775 36. Perl no longer allows group names to start with digits, so I have made this
776 change also in PCRE. It simplifies the code a bit.
777
778 37. In extended mode, Perl ignores spaces before a + that indicates a
779 possessive quantifier. PCRE allowed a space before the quantifier, but not
780 before the possessive +. It now does.
781
782 38. The use of \K (reset reported match start) within a repeated possessive
783 group such as (a\Kb)*+ was not working.
784
785 40. Document that the same character tables must be used at compile time and
786 run time, and that the facility to pass tables to pcre_exec() and
787 pcre_dfa_exec() is for use only with saved/restored patterns.
788
789 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
790 features for Builds with MSVC:
791
792 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
793 builds). The .rc files can be used to set FileDescription and many other
794 attributes.
795
796 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
797 This allows higher-level build scripts which want .pdb files to avoid
798 hard-coding the exact files needed."
799
800 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
801 mean "start of word" and "end of word", respectively, as a transition aid.
802
803 43. A minimizing repeat of a class containing codepoints greater than 255 in
804 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
805 compiled to use the heap for recursion.
806
807 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
808 is configured.
809
810
811 Version 8.33 28-May-2013
812 ------------------------
813
814 1. Added 'U' to some constants that are compared to unsigned integers, to
815 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
816 variables that are added to signed variables, to ensure the result is
817 signed and can be negated.
818
819 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
820 CMake config files.
821
822 3. Revise the creation of config.h.generic so that all boolean macros are
823 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
824 overriding via -D on the command line possible.
825
826 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
827 to unsigned int is reported to make a quite noticeable speed difference in
828 a specific Windows environment. Testing on Linux did also appear to show
829 some benefit (and it is clearly not harmful). Also fixed the definition of
830 Xop which should be unsigned.
831
832 5. Related to (4), changing the definition of the intermediate variable cc
833 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
834 improvements.
835
836 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
837 unnecessary spaces.
838
839 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
840 and later.
841
842 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
843
844 9. Optimizing fast_forward_start_bits in JIT.
845
846 10. Adding support for callouts in JIT, and fixing some issues revealed
847 during this work. Namely:
848
849 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
850
851 (b) Minimum length was not checked before the matching is started.
852
853 11. The value of capture_last that is passed to callouts was incorrect in some
854 cases when there was a capture on one path that was subsequently abandoned
855 after a backtrack. Also, the capture_last value is now reset after a
856 recursion, since all captures are also reset in this case.
857
858 12. The interpreter no longer returns the "too many substrings" error in the
859 case when an overflowing capture is in a branch that is subsequently
860 abandoned after a backtrack.
861
862 13. In the pathological case when an offset vector of size 2 is used, pcretest
863 now prints out the matched string after a yield of 0 or 1.
864
865 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
866 Only supported by the JIT compiler at the moment.
867
868 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
869
870 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
871 offset of the starting point of the matching process, provided the offsets
872 vector is large enough.
873
874 17. The \A escape now records a lookbehind value of 1, though its execution
875 does not actually inspect the previous character. This is to ensure that,
876 in partial multi-segment matching, at least one character from the old
877 segment is retained when a new segment is processed. Otherwise, if there
878 are no lookbehinds in the pattern, \A might match incorrectly at the start
879 of a new segment.
880
881 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
882
883 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
884 gives some modest performance improvement in 8-bit mode.
885
886 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
887 be expressed in certain programming languages using Universal Character
888 Names.
889
890 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
891 which points out that "non characters" are not "characters that may not
892 appear in Unicode strings" but rather "characters that are reserved for
893 internal use and have only local meaning".
894
895 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
896 there was a conditional group that depended on an assertion, if the
897 assertion was false, the callout that immediately followed the alternation
898 in the condition was skipped when pcre_exec() was used for matching.
899
900 23. Allow an explicit callout to be inserted before an assertion that is the
901 condition for a conditional group, for compatibility with automatic
902 callouts, which always insert a callout at this point.
903
904 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
905 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
906
907 25. (*PRUNE) is now supported by the JIT compiler.
908
909 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
910
911 27. Fix the case where there are two or more SKIPs with arguments that may be
912 ignored.
913
914 28. (*SKIP) is now supported by the JIT compiler.
915
916 29. (*THEN) is now supported by the JIT compiler.
917
918 30. Update RunTest with additional test selector options.
919
920 31. The way PCRE handles backtracking verbs has been changed in two ways.
921
922 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
923 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
924 backtracking. In some cases this makes it more Perl-compatible, but Perl's
925 rather obscure rules do not always do the same thing.
926
927 (2) Previously, backtracking verbs were confined within assertions. This is
928 no longer the case for positive assertions, except for (*ACCEPT). Again,
929 this sometimes improves Perl compatibility, and sometimes does not.
930
931 32. A number of tests that were in test 2 because Perl did things differently
932 have been moved to test 1, because either Perl or PCRE has changed, and
933 these tests are now compatible.
934
935 32. Backtracking control verbs are now handled in the same way in JIT and
936 interpreter.
937
938 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
939 contained a forward subroutine reference caused a compile error.
940
941 34. Auto-detect and optimize limited repetitions in JIT.
942
943 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
944 blocking (*UTF) etc.
945
946 36. In the interpreter, maximizing pattern repetitions for characters and
947 character types now use tail recursion, which reduces stack usage.
948
949 37. The value of the max lookbehind was not correctly preserved if a compiled
950 and saved regex was reloaded on a host of different endianness.
951
952 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
953 of the compiled pattern block, expand the flags field from 16 to 32 bits
954 because it was almost full.
955
956 39. Try madvise first before posix_madvise.
957
958 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
959 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
960 original change locked out all matches of empty strings. This has been
961 changed so that one match of an empty string per line is recognized.
962 Subsequent searches on the same line (for colouring or for --only-matching,
963 for example) do not recognize empty strings.
964
965 41. Applied a user patch to fix a number of spelling mistakes in comments.
966
967 42. Data lines longer than 65536 caused pcretest to crash.
968
969 43. Clarified the data type for length and startoffset arguments for pcre_exec
970 and pcre_dfa_exec in the function-specific man pages, where they were
971 explicitly stated to be in bytes, never having been updated. I also added
972 some clarification to the pcreapi man page.
973
974 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
975 a segmentation fault.
976
977
978 Version 8.32 30-November-2012
979 -----------------------------
980
981 1. Improved JIT compiler optimizations for first character search and single
982 character iterators.
983
984 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
985 Patch by Daniel Richard G.
986
987 3. Single character iterator optimizations in the JIT compiler.
988
989 4. Improved JIT compiler optimizations for character ranges.
990
991 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
992 Reported by Giuseppe D'Angelo.
993
994 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
995 of a line, was being set incorrectly in cases where .* appeared inside
996 atomic brackets at the start of a pattern, or where there was a subsequent
997 *PRUNE or *SKIP.
998
999 7. Improved instruction cache flush for POWER/PowerPC.
1000 Patch by Daniel Richard G.
1001
1002 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1003 grep:
1004
1005 (a) There is now no limit to the number of patterns to be matched.
1006
1007 (b) An error is given if a pattern is too long.
1008
1009 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1010 are now supported.
1011
1012 (d) --exclude-from and --include-from (multiple use) have been added.
1013
1014 (e) Exclusions and inclusions now apply to all files and directories, not
1015 just to those obtained from scanning a directory recursively.
1016
1017 (f) Multiple uses of -f and --file-list are now supported.
1018
1019 (g) In a Windows environment, the default for -d has been changed from
1020 "read" (the GNU grep default) to "skip", because otherwise the presence
1021 of a directory in the file list provokes an error.
1022
1023 (h) The documentation has been revised and clarified in places.
1024
1025 9. Improve the matching speed of capturing brackets.
1026
1027 10. Changed the meaning of \X so that it now matches a Unicode extended
1028 grapheme cluster.
1029
1030 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1031 out POSIX threads when JIT support is configured.
1032
1033 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1034
1035 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1036 structure could go wrong in environments where size_t is not the same size
1037 as int.
1038
1039 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1040 set.
1041
1042 15. The EBCDIC support had decayed; later updates to the code had included
1043 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1044 general tidy up of EBCDIC-related issues, and the documentation was also
1045 not quite right. There is now a test that can be run on ASCII systems to
1046 check some of the EBCDIC-related things (but is it not a full test).
1047
1048 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1049 in a small tidy to the code.
1050
1051 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1052
1053 18. If the --only-matching (-o) option in pcregrep is specified multiple
1054 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1055 substrings matched by the 1st and 2nd capturing parentheses. A separating
1056 string can be specified by --om-separator (default empty).
1057
1058 19. Improving the first n character searches.
1059
1060 20. Turn case lists for horizontal and vertical white space into macros so that
1061 they are defined only once.
1062
1063 21. This set of changes together give more compatible Unicode case-folding
1064 behaviour for characters that have more than one other case when UCP
1065 support is available.
1066
1067 (a) The Unicode property table now has offsets into a new table of sets of
1068 three or more characters that are case-equivalent. The MultiStage2.py
1069 script that generates these tables (the pcre_ucd.c file) now scans
1070 CaseFolding.txt instead of UnicodeData.txt for character case
1071 information.
1072
1073 (b) The code for adding characters or ranges of characters to a character
1074 class has been abstracted into a generalized function that also handles
1075 case-independence. In UTF-mode with UCP support, this uses the new data
1076 to handle characters with more than one other case.
1077
1078 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1079 whose other case is greater than 256 are now correctly matched
1080 caselessly. Previously, the high codepoint matched the low one, but not
1081 vice versa.
1082
1083 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1084 of the new class addition function, using character lists defined as
1085 macros alongside the case definitions of 20 above.
1086
1087 (e) Caseless back references now work with characters that have more than
1088 one other case.
1089
1090 (f) General caseless matching of characters with more than one other case
1091 is supported.
1092
1093 22. Unicode character properties were updated from Unicode 6.2.0
1094
1095 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1096
1097 24. Add support for 32-bit character strings, and UTF-32
1098
1099 25. Major JIT compiler update (code refactoring and bugfixing).
1100 Experimental Sparc 32 support is added.
1101
1102 26. Applied a modified version of Daniel Richard G's patch to create
1103 pcre.h.generic and config.h.generic by "make" instead of in the
1104 PrepareRelease script.
1105
1106 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1107 pcre_compile.c when checking for a zero character.
1108
1109 28. Introducing a native interface for JIT. Through this interface, the compiled
1110 machine code can be directly executed. The purpose of this interface is to
1111 provide fast pattern matching, so several sanity checks are not performed.
1112 However, feature tests are still performed. The new interface provides
1113 1.4x speedup compared to the old one.
1114
1115 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1116 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1117 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1118
1119 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1120 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1121 the "old" RFC 2279). Instead, it ended up passing a negative length to
1122 pcre_exec().
1123
1124 31. Add support for GCC's visibility feature to hide internal functions.
1125
1126 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1127 "unknown -C option" after outputting 0 or 1.
1128
1129 33. There is now support for generating a code coverage report for the test
1130 suite in environments where gcc is the compiler and lcov is installed. This
1131 is mainly for the benefit of the developers.
1132
1133 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1134 unaddressable using valgrind annotations, allowing valgrind to detect
1135 invalid memory accesses. This is mainly for the benefit of the developers.
1136
1137 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1138
1139 26. Give configure error if --enable-cpp but no C++ compiler found.
1140
1141
1142 Version 8.31 06-July-2012
1143 -------------------------
1144
1145 1. Fixing a wrong JIT test case and some compiler warnings.
1146
1147 2. Removed a bashism from the RunTest script.
1148
1149 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1150 to unsigned type, result still unsigned" that was given by an MS compiler
1151 on encountering the code "-sizeof(xxx)".
1152
1153 4. Partial matching support is added to the JIT compiler.
1154
1155 5. Fixed several bugs concerned with partial matching of items that consist
1156 of more than one character:
1157
1158 (a) /^(..)\1/ did not partially match "aba" because checking references was
1159 done on an "all or nothing" basis. This also applied to repeated
1160 references.
1161
1162 (b) \R did not give a hard partial match if \r was found at the end of the
1163 subject.
1164
1165 (c) \X did not give a hard partial match after matching one or more
1166 characters at the end of the subject.
1167
1168 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1169 a partial match for the string "\r".
1170
1171 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1172 a partial match for a CR character at the end of the subject string.
1173
1174 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1175 running pcretest, the text "(JIT)" added to the output whenever JIT is
1176 actually used to run the match.
1177
1178 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1179 or /S+[+] with a digit between 1 and 7.
1180
1181 8. OP_NOT now supports any UTF character not just single-byte ones.
1182
1183 9. (*MARK) control verb is now supported by the JIT compiler.
1184
1185 10. The command "./RunTest list" lists the available tests without actually
1186 running any of them. (Because I keep forgetting what they all are.)
1187
1188 11. Add PCRE_INFO_MAXLOOKBEHIND.
1189
1190 12. Applied a (slightly modified) user-supplied patch that improves performance
1191 when the heap is used for recursion (compiled with --disable-stack-for-
1192 recursion). Instead of malloc and free for each heap frame each time a
1193 logical recursion happens, frames are retained on a chain and re-used where
1194 possible. This sometimes gives as much as 30% improvement.
1195
1196 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1197 call.
1198
1199 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1200
1201 15. It is now possible to link pcretest with libedit as an alternative to
1202 libreadline.
1203
1204 16. (*COMMIT) control verb is now supported by the JIT compiler.
1205
1206 17. The Unicode data tables have been updated to Unicode 6.1.0.
1207
1208 18. Added --file-list option to pcregrep.
1209
1210 19. Added binary file support to pcregrep, including the -a, --binary-files,
1211 -I, and --text options.
1212
1213 20. The madvise function is renamed for posix_madvise for QNX compatibility
1214 reasons. Fixed by Giuseppe D'Angelo.
1215
1216 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1217
1218 22. Changed the option for creating symbolic links for 16-bit man pages from
1219 -s to -sf so that re-installing does not cause issues.
1220
1221 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1222
1223 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1224 matches in certain environments (the workspace was not being correctly
1225 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1226 some of the workspace data at the beginning of a restart.
1227
1228 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1229 was not doing so when it should - probably a typo introduced by SVN 528
1230 (change 8.10/14).
1231
1232 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1233 \w+ when the character tables indicated that \x{c4} was a word character.
1234 There were several related cases, all because the tests for doing a table
1235 lookup were testing for characters less than 127 instead of 255.
1236
1237 27. If a pattern contains capturing parentheses that are not used in a match,
1238 their slots in the ovector are set to -1. For those that are higher than
1239 any matched groups, this happens at the end of processing. In the case when
1240 there were back references that the ovector was too small to contain
1241 (causing temporary malloc'd memory to be used during matching), and the
1242 highest capturing number was not used, memory off the end of the ovector
1243 was incorrectly being set to -1. (It was using the size of the temporary
1244 memory instead of the true size.)
1245
1246 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1247 ovector size, it uses memory at the end of the block that it has got.
1248
1249 29. Check for an overlong MARK name and give an error at compile time. The
1250 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1251
1252 30. JIT compiler update.
1253
1254 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1255 Rill for the patch.
1256
1257 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1258
1259 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1260
1261 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1262 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1263 was enabled.
1264
1265 35. Improve JIT code generation for greedy plus quantifier.
1266
1267 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1268 1 to "aa" instead of to an empty string. The bug affected repeated groups
1269 that could potentially match an empty string.
1270
1271 37. Optimizing single character iterators in JIT.
1272
1273 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1274 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1275 codepoints that are too big for the mode are faulted, and in a UTF mode,
1276 disallowed codepoints are also faulted.
1277
1278 39. If PCRE was compiled with UTF support, in three places in the DFA
1279 matcher there was code that should only have been obeyed in UTF mode, but
1280 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1281 processing when bytes with values greater than 127 were present. In 16-bit
1282 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1283 both cases the values are those that cannot be the first data item in a UTF
1284 character. The three items that might have provoked this were recursions,
1285 possessively repeated groups, and atomic groups.
1286
1287 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1288 and pcregrep, because some OS require shared objects to be explicitly
1289 passed to ld, causing the link step to fail if they are not.
1290
1291 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1292 mode, patterns that started with \h* or \R* might be incorrectly matched.
1293
1294
1295 Version 8.30 04-February-2012
1296 -----------------------------
1297
1298 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1299 name is defined in ctype.h.
1300
1301 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1302 only in quite long subpatterns.
1303
1304 3. Removed the function pcre_info(), which has been obsolete and deprecated
1305 since it was replaced by pcre_fullinfo() in February 2000.
1306
1307 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1308 match a (*MARK), and the match failed at the start of the subject, a
1309 reference to memory before the start of the subject could occur. This bug
1310 was introduced by fix 17 of release 8.21.
1311
1312 5. A reference to an unset group with zero minimum repetition was giving
1313 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1314 /(another)?(\1?)test/ matched against "hello world test". This bug was
1315 introduced in release 8.13.
1316
1317 6. Add support for 16-bit character strings (a large amount of work involving
1318 many changes and refactorings).
1319
1320 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1321 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1322 from a file.
1323
1324 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1325 rounding is not applied in this particular case).
1326
1327 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1328 if they appear, or are escaped, in patterns.
1329
1330 10. Get rid of a number of -Wunused-but-set-variable warnings.
1331
1332 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1333 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1334 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1335 also returns the mark "x". This bug applied to capturing parentheses,
1336 non-capturing parentheses, and atomic parentheses. It also applied to some
1337 assertions.
1338
1339 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1340 information out of configure.ac instead of relying on pcre.h.generic, which
1341 is not stored in the repository.
1342
1343 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1344 -lreadline.
1345
1346 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1347
1348 15. Applied Graycode's patch to put the top-level frame on the stack rather
1349 than the heap when not using the stack for recursion. This gives a
1350 performance improvement in many cases when recursion is not deep.
1351
1352 16. Experimental code added to "pcretest -C" to output the stack frame size.
1353
1354
1355 Version 8.21 12-Dec-2011
1356 ------------------------
1357
1358 1. Updating the JIT compiler.
1359
1360 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1361 are added as well.
1362
1363 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1364 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1365 calling _pcre_jit_exec. Some extra comments are added.
1366
1367 4. (*MARK) settings inside atomic groups that do not contain any capturing
1368 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1369 was introduced by change 18 for 8.20.
1370
1371 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1372 ECMA-262 standard.
1373
1374 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1375 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1376 This bug was probably introduced by change 9 of 8.13.
1377
1378 7. While fixing 6 above, I noticed that a number of other items were being
1379 incorrectly rejected as "not fixed length". This arose partly because newer
1380 opcodes had not been added to the fixed-length checking code. I have (a)
1381 corrected the bug and added tests for these items, and (b) arranged for an
1382 error to occur if an unknown opcode is encountered while checking for fixed
1383 length instead of just assuming "not fixed length". The items that were
1384 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1385 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1386 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1387
1388 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1389 being incorrectly compiled and would have given unpredicatble results.
1390
1391 9. A possessively repeated subpattern with minimum repeat count greater than
1392 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1393 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1394 the first (A) could occur when it should not.
1395
1396 10. Add a cast and remove a redundant test from the code.
1397
1398 11. JIT should use pcre_malloc/pcre_free for allocation.
1399
1400 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1401 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1402 is anything other than /usr, -L is still shown).
1403
1404 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1405
1406 14. Perl does not support \N without a following name in a [] class; PCRE now
1407 also gives an error.
1408
1409 15. If a forward reference was repeated with an upper limit of around 2000,
1410 it caused the error "internal error: overran compiling workspace". The
1411 maximum number of forward references (including repeats) was limited by the
1412 internal workspace, and dependent on the LINK_SIZE. The code has been
1413 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1414 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1415 of around 200,000 forward references. While doing this, I also speeded up
1416 the filling in of repeated forward references.
1417
1418 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1419 incorrectly expecting the subject to contain another "a" after the start.
1420
1421 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1422 in the match, the SKIP should be ignored. This was not happening; instead
1423 the SKIP was being treated as NOMATCH. For patterns such as
1424 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1425 tested.
1426
1427 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1428 now much more compatible with Perl, in particular in cases where the result
1429 is a non-match for a non-anchored pattern. For example, if
1430 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1431 "m", where previously it did not return a name. A side effect of this
1432 change is that for partial matches, the last encountered mark name is
1433 returned, as for non matches. A number of tests that were previously not
1434 Perl-compatible have been moved into the Perl-compatible test files. The
1435 refactoring has had the pleasing side effect of removing one argument from
1436 the match() function, thus reducing its stack requirements.
1437
1438 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1439 subsequent uses of /S (without +) incorrectly behaved like /S+.
1440
1441 21. Retrieve executable code size support for the JIT compiler and fixing
1442 some warnings.
1443
1444 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1445 not work when the shorter character appeared right at the end of the
1446 subject string.
1447
1448 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1449 systems.
1450
1451 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1452 output it when the /M option is used in pcretest.
1453
1454 25. The CheckMan script was not being included in the distribution. Also, added
1455 an explicit "perl" to run Perl scripts from the PrepareRelease script
1456 because this is reportedly needed in Windows.
1457
1458 26. If study data was being save in a file and studying had not found a set of
1459 "starts with" bytes for the pattern, the data written to the file (though
1460 never used) was taken from uninitialized memory and so caused valgrind to
1461 complain.
1462
1463 27. Updated RunTest.bat as provided by Sheri Pierce.
1464
1465 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1466
1467 29. Computation of memory usage for the table of capturing group names was
1468 giving an unnecessarily large value.
1469
1470
1471 Version 8.20 21-Oct-2011
1472 ------------------------
1473
1474 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1475 a POSIX class. After further experiments with Perl, which convinced me that
1476 Perl has bugs and confusions, a closing square bracket is no longer allowed
1477 in a POSIX name. This bug also affected patterns with classes that started
1478 with full stops.
1479
1480 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1481 captured substring, but while checking the failing first alternative,
1482 substring 1 is temporarily captured. If the output vector supplied to
1483 pcre_exec() was not big enough for this capture, the yield of the function
1484 was still zero ("insufficient space for captured substrings"). This cannot
1485 be totally fixed without adding another stack variable, which seems a lot
1486 of expense for a edge case. However, I have improved the situation in cases
1487 such as /(a)(b)x|abc/ matched against "abc", where the return code
1488 indicates that fewer than the maximum number of slots in the ovector have
1489 been set.
1490
1491 3. Related to (2) above: when there are more back references in a pattern than
1492 slots in the output vector, pcre_exec() uses temporary memory during
1493 matching, and copies in the captures as far as possible afterwards. It was
1494 using the entire output vector, but this conflicts with the specification
1495 that only 2/3 is used for passing back captured substrings. Now it uses
1496 only the first 2/3, for compatibility. This is, of course, another edge
1497 case.
1498
1499 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1500 main code base, and can be used by building with --enable-jit. When this is
1501 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1502 runtime --no-jit option is given.
1503
1504 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1505 ovector, the return from the function was zero, implying that there were
1506 other matches that did not fit. The correct "exactly full" value is now
1507 returned.
1508
1509 6. If a subpattern that was called recursively or as a subroutine contained
1510 (*PRUNE) or any other control that caused it to give a non-standard return,
1511 invalid errors such as "Error -26 (nested recursion at the same subject
1512 position)" or even infinite loops could occur.
1513
1514 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1515 computing the minimum length on reaching *ACCEPT, and so ended up with the
1516 wrong value of 1 rather than 0. Further investigation indicates that
1517 computing a minimum subject length in the presence of *ACCEPT is difficult
1518 (think back references, subroutine calls), and so I have changed the code
1519 so that no minimum is registered for a pattern that contains *ACCEPT.
1520
1521 8. If (*THEN) was present in the first (true) branch of a conditional group,
1522 it was not handled as intended. [But see 16 below.]
1523
1524 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1525 Sheri Pierce.
1526
1527 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1528 the first byte in a match must be "a".
1529
1530 11. Change 17 for 8.13 increased the recursion depth for patterns like
1531 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1532 pattern contains any instances of (*THEN). If it does not, the old
1533 optimizations are restored. It would be nice to do this on a per-group
1534 basis, but at the moment that is not feasible.
1535
1536 12. In some environments, the output of pcretest -C is CRLF terminated. This
1537 broke RunTest's code that checks for the link size. A single white space
1538 character after the value is now allowed for.
1539
1540 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1541 For "fr", it uses the Windows-specific input and output files.
1542
1543 14. If (*THEN) appeared in a group that was called recursively or as a
1544 subroutine, it did not work as intended. [But see next item.]
1545
1546 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1547 pattern fragments (but not containing any | characters). If A and B are
1548 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1549 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1550 D. In other words, Perl considers parentheses that do not contain any |
1551 characters to be part of a surrounding alternative, whereas PCRE was
1552 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1553 differently. PCRE now behaves in the same way as Perl, except in the case
1554 of subroutine/recursion calls such as (?1) which have in any case always
1555 been different (but PCRE had them first :-).
1556
1557 16. Related to 15 above: Perl does not treat the | in a conditional group as
1558 creating alternatives. Such a group is treated in the same way as an
1559 ordinary group without any | characters when processing (*THEN). PCRE has
1560 been changed to match Perl's behaviour.
1561
1562 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1563 RunGrepTest script failed.
1564
1565 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1566 inevitable for groups that contain captures, but it can lead to a lot of
1567 stack use in large patterns. The old behaviour has been restored for atomic
1568 groups that do not contain any capturing parentheses.
1569
1570 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1571 suppress the check for a minimum subject length at run time. (If it was
1572 given to pcre_exec() or pcre_dfa_exec() it did work.)
1573
1574 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1575 fail to work when decoding hex characters in data strings in EBCDIC
1576 environments.
1577
1578 21. It appears that in at least one Mac OS environment, the isxdigit() function
1579 is implemented as a macro that evaluates to its argument more than once,
1580 contravening the C 90 Standard (I haven't checked a later standard). There
1581 was an instance in pcretest which caused it to go wrong when processing
1582 \x{...} escapes in subject strings. The has been rewritten to avoid using
1583 things like p++ in the argument of isxdigit().
1584
1585
1586 Version 8.13 16-Aug-2011
1587 ------------------------
1588
1589 1. The Unicode data tables have been updated to Unicode 6.0.0.
1590
1591 2. Two minor typos in pcre_internal.h have been fixed.
1592
1593 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1594 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1595 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1596
1597 4. There were a number of related bugs in the code for matching backrefences
1598 caselessly in UTF-8 mode when codes for the characters concerned were
1599 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1600 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1601 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1602 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1603 2-byte code at the end of the subject (it thought there wasn't enough data
1604 left).
1605
1606 5. Comprehensive information about what went wrong is now returned by
1607 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1608 as the output vector has at least 2 elements. The offset of the start of
1609 the failing character and a reason code are placed in the vector.
1610
1611 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1612 now returned is for the first byte of the failing character, instead of the
1613 last byte inspected. This is an incompatible change, but I hope it is small
1614 enough not to be a problem. It makes the returned offset consistent with
1615 pcre_exec() and pcre_dfa_exec().
1616
1617 7. pcretest now gives a text phrase as well as the error number when
1618 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1619 failure, the offset and reason code are output.
1620
1621 8. When \R was used with a maximizing quantifier it failed to skip backwards
1622 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1623 back over a single character (\n). This seems wrong (because it treated the
1624 two characters as a single entity when going forwards), conflicts with the
1625 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1626 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1627 has been changed.
1628
1629 9. Some internal refactoring has changed the processing so that the handling
1630 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1631 time (the PCRE_DOTALL option was changed this way some time ago: version
1632 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1633 which was always a bit of a fudge. It also means that there is one less
1634 argument for the match() function, which reduces its stack requirements
1635 slightly. This change also fixes an incompatibility with Perl: the pattern
1636 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1637
1638 10. More internal refactoring has drastically reduced the number of recursive
1639 calls to match() for possessively repeated groups such as (abc)++ when
1640 using pcre_exec().
1641
1642 11. While implementing 10, a number of bugs in the handling of groups were
1643 discovered and fixed:
1644
1645 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1646 (a|)*(?1) gave a compile-time internal error.
1647 ((a|)+)+ did not notice that the outer group could match an empty string.
1648 (^a|^)+ was not marked as anchored.
1649 (.*a|.*)+ was not marked as matching at start or after a newline.
1650
1651 12. Yet more internal refactoring has removed another argument from the match()
1652 function. Special calls to this function are now indicated by setting a
1653 value in a variable in the "match data" data block.
1654
1655 13. Be more explicit in pcre_study() instead of relying on "default" for
1656 opcodes that mean there is no starting character; this means that when new
1657 ones are added and accidentally left out of pcre_study(), testing should
1658 pick them up.
1659
1660 14. The -s option of pcretest has been documented for ages as being an old
1661 synonym of -m (show memory usage). I have changed it to mean "force study
1662 for every regex", that is, assume /S for every regex. This is similar to -i
1663 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1664 using it. It makes it easier to run collections of tests with and without
1665 study enabled, and thereby test pcre_study() more easily. All the standard
1666 tests are now run with and without -s (but some patterns can be marked as
1667 "never study" - see 20 below).
1668
1669 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1670 restoration of the capturing data to the outer values was not happening
1671 correctly.
1672
1673 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1674 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1675 pattern had matched an empty string, and so incorrectly returned a no
1676 match.
1677
1678 17. There was optimizing code for the last branch of non-capturing parentheses,
1679 and also for the obeyed branch of a conditional subexpression, which used
1680 tail recursion to cut down on stack usage. Unfortunately, now that there is
1681 the possibility of (*THEN) occurring in these branches, tail recursion is
1682 no longer possible because the return has to be checked for (*THEN). These
1683 two optimizations have therefore been removed. [But see 8.20/11 above.]
1684
1685 18. If a pattern containing \R was studied, it was assumed that \R always
1686 matched two bytes, thus causing the minimum subject length to be
1687 incorrectly computed because \R can also match just one byte.
1688
1689 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1690 was incorrectly computed.
1691
1692 20. If /S is present twice on a test pattern in pcretest input, it now
1693 *disables* studying, thereby overriding the use of -s on the command line
1694 (see 14 above). This is necessary for one or two tests to keep the output
1695 identical in both cases.
1696
1697 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1698 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1699
1700 22. When an atomic group that contained a capturing parenthesis was
1701 successfully matched, but the branch in which it appeared failed, the
1702 capturing was not being forgotten if a higher numbered group was later
1703 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1704 group 1 to "a", when in fact it should be unset. This applied to multi-
1705 branched capturing and non-capturing groups, repeated or not, and also to
1706 positive assertions (capturing in negative assertions does not happen
1707 in PCRE) and also to nested atomic groups.
1708
1709 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1710 subject after a captured substring, to make it easier to tell which of a
1711 number of identical substrings has been captured.
1712
1713 24. The way atomic groups are processed by pcre_exec() has been changed so that
1714 if they are repeated, backtracking one repetition now resets captured
1715 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1716 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1717 "aaa". Previously, it would have been "a". As part of this code
1718 refactoring, the way recursive calls are handled has also been changed.
1719
1720 25. If an assertion condition captured any substrings, they were not passed
1721 back unless some other capturing happened later. For example, if
1722 (?(?=(a))a) was matched against "a", no capturing was returned.
1723
1724 26. When studying a pattern that contained subroutine calls or assertions,
1725 the code for finding the minimum length of a possible match was handling
1726 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1727 group 1 called group 2 while simultaneously a separate group 2 called group
1728 1). A stack overflow occurred in this case. I have fixed this by limiting
1729 the recursion depth to 10.
1730
1731 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1732 Fortmann. This supports explicit test numbers on the command line, and has
1733 argument validation and error reporting.
1734
1735 28. An instance of \X with an unlimited repeat could fail if at any point the
1736 first character it looked at was a mark character.
1737
1738 29. Some minor code refactoring concerning Unicode properties and scripts
1739 should reduce the stack requirement of match() slightly.
1740
1741 30. Added the '=' option to pcretest to check the setting of unused capturing
1742 slots at the end of the pattern, which are documented as being -1, but are
1743 not included in the return count.
1744
1745 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1746 compiled something random. Now it gives a compile-time error (as does
1747 Perl).
1748
1749 32. A *MARK encountered during the processing of a positive assertion is now
1750 recorded and passed back (compatible with Perl).
1751
1752 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1753 had alternative anchored branches, the search for a second match in a line
1754 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1755 matched the line "0102" twice. The same bug affected patterns that started
1756 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1757 twice.
1758
1759 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1760 does, and because of capturing effects, quantifying parenthesized
1761 assertions may at times be useful. Quantifiers are now allowed for
1762 parenthesized assertions.
1763
1764 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1765
1766 36. \g was being checked for fancy things in a character class, when it should
1767 just be a literal "g".
1768
1769 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1770 appearance of a nested POSIX class supersedes an apparent external class.
1771 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1772 unescaped square brackets may also appear as part of class names. For
1773 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1774 more like Perl. (But see 8.20/1 above.)
1775
1776 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1777 was because it thought it was \N{name}, which is not supported).
1778
1779 39. Add minix to OS list not supporting the -S option in pcretest.
1780
1781 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1782 cannot analyze patterns in sufficient detail to catch mutual recursions
1783 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1784 subgroup is called recursively as a subpattern for a second time at the
1785 same position in the subject string. In previous releases this might have
1786 been caught by the recursion limit, or it might have run out of stack.
1787
1788 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1789 happen only once. PCRE was, however incorrectly giving a compile time error
1790 "recursive call could loop indefinitely" because it cannot analyze the
1791 pattern in sufficient detail. The compile time test no longer happens when
1792 PCRE is compiling a conditional subpattern, but actual runaway loops are
1793 now caught at runtime (see 40 above).
1794
1795 42. It seems that Perl allows any characters other than a closing parenthesis
1796 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1797 has been changed to be the same.
1798
1799 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1800 as not to get warnings when autogen.sh is called. Also changed
1801 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1802
1803 44. To help people who use pcregrep to scan files containing exceedingly long
1804 lines, the following changes have been made:
1805
1806 (a) The default value of the buffer size parameter has been increased from
1807 8K to 20K. (The actual buffer used is three times this size.)
1808
1809 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1810 PCRE is built.
1811
1812 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1813 to be set at run time.
1814
1815 (d) Numerical values in pcregrep options can be followed by K or M, for
1816 example --buffer-size=50K.
1817
1818 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1819 given and the return code is set to 2.
1820
1821 45. Add a pointer to the latest mark to the callout data block.
1822
1823 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1824 partial match of an empty string instead of no match. This was specific to
1825 the use of ".".
1826
1827 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1828 complete match instead of a partial match. This bug was dependent on both
1829 the PCRE_UTF8 and PCRE_DOTALL options being set.
1830
1831 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1832 starting byte set, because \b was not being ignored.
1833
1834
1835 Version 8.12 15-Jan-2011
1836 ------------------------
1837
1838 1. Fixed some typos in the markup of the man pages, and wrote a script that
1839 checks for such things as part of the documentation building process.
1840
1841 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1842 --match-limit and --recursion-limit options (added for 8.11). In
1843 particular, this made one of the standard tests fail. (The integer value
1844 went into the wrong half of a long int.)
1845
1846 3. If the --colour option was given to pcregrep with -v (invert match), it
1847 did strange things, either producing crazy output, or crashing. It should,
1848 of course, ignore a request for colour when reporting lines that do not
1849 match.
1850
1851 4. Another pcregrep bug caused similar problems if --colour was specified with
1852 -M (multiline) and the pattern match finished with a line ending.
1853
1854 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1855 matched in multiline mode, the following line was shown as part of the
1856 match. This seems wrong, so I have changed it.
1857
1858 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1859 the check for further matches in the same line (so they could be coloured)
1860 to overrun the end of the current line. If another match was found, it was
1861 incorrectly shown (and then shown again when found in the next line).
1862
1863 7. If pcregrep was compiled under Windows, there was a reference to the
1864 function pcregrep_exit() before it was defined. I am assuming this was
1865 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1866 reported by a user. I've moved the definition above the reference.
1867
1868
1869 Version 8.11 10-Dec-2010
1870 ------------------------
1871
1872 1. (*THEN) was not working properly if there were untried alternatives prior
1873 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1874 backtracked to try for "b" instead of moving to the next alternative branch
1875 at the same level (in this case, to look for "c"). The Perl documentation
1876 is clear that when (*THEN) is backtracked onto, it goes to the "next
1877 alternative in the innermost enclosing group".
1878
1879 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1880 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1881 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1882 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1883 (*THEN).
1884
1885 3. If \s appeared in a character class, it removed the VT character from
1886 the class, even if it had been included by some previous item, for example
1887 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1888 of \s, but is part of the POSIX "space" class.)
1889
1890 4. A partial match never returns an empty string (because you can always
1891 match an empty string at the end of the subject); however the checking for
1892 an empty string was starting at the "start of match" point. This has been
1893 changed to the "earliest inspected character" point, because the returned
1894 data for a partial match starts at this character. This means that, for
1895 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1896 (previously it gave "no match").
1897
1898 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1899 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1900 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1901 has an implication that the given string is incomplete (because a partial
1902 match is preferred over a full match). For this reason, these items now
1903 give a partial match in this situation. [Aside: previously, the one case
1904 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1905 match rather than a full match, which was wrong by the old rules, but is
1906 now correct.]
1907
1908 6. There was a bug in the handling of #-introduced comments, recognized when
1909 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1910 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1911 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1912 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1913 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1914 places in pcre_compile().
1915
1916 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1917 comments when looking ahead for named forward references to subpatterns,
1918 the only newline sequence it recognized was NL. It now handles newlines
1919 according to the set newline convention.
1920
1921 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1922 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1923 cater for a lack of strerror(). These oversights have been fixed.
1924
1925 9. Added --match-limit and --recursion-limit to pcregrep.
1926
1927 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1928
1929 11. When the -o option was used, pcregrep was setting a return code of 1, even
1930 when matches were found, and --line-buffered was not being honoured.
1931
1932 12. Added an optional parentheses number to the -o and --only-matching options
1933 of pcregrep.
1934
1935 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1936 can match an empty string. The code to do it in pcretest and pcredemo
1937 needed fixing:
1938
1939 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1940 only one byte after an empty string match just before CRLF (this case
1941 just got forgotten; "any" and "anycrlf" were OK).
1942
1943 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1944 mode when an empty string match preceded an ASCII character followed by
1945 a non-ASCII character. (The code for advancing by one character rather
1946 than one byte was nonsense.)
1947
1948 (c) The pcredemo.c sample program did not have any code at all to handle
1949 the cases when CRLF is a valid newline sequence.
1950
1951 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1952 as a starting offset was within the subject string. There is now a new
1953 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1954 negative or greater than the length of the string. In order to test this,
1955 pcretest is extended to allow the setting of negative starting offsets.
1956
1957 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1958 starting offset points to the beginning of a UTF-8 character was
1959 unnecessarily clumsy. I tidied it up.
1960
1961 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1962 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1963
1964 17. Nobody had reported that the --include_dir option, which was added in
1965 release 7.7 should have been called --include-dir (hyphen, not underscore)
1966 for compatibility with GNU grep. I have changed it to --include-dir, but
1967 left --include_dir as an undocumented synonym, and the same for
1968 --exclude-dir, though that is not available in GNU grep, at least as of
1969 release 2.5.4.
1970
1971 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1972 characters from a string of bytes) have been redefined so as not to use
1973 loops, in order to improve performance in some environments. At the same
1974 time, I abstracted some of the common code into auxiliary macros to save
1975 repetition (this should not affect the compiled code).
1976
1977 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1978 compile-time error is now given if \c is not followed by an ASCII
1979 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1980 different, and any byte value is allowed.)
1981
1982 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1983 START_OPTIMIZE option, which is now allowed at compile time - but just
1984 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1985 to pcregrep and other applications that have no direct access to PCRE
1986 options. The new /Y option in pcretest sets this option when calling
1987 pcre_compile().
1988
1989 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1990 back references. Groups containing recursive back references were forced to
1991 be atomic by that change, but in the case of named groups, the amount of
1992 memory required was incorrectly computed, leading to "Failed: internal
1993 error: code overflow". This has been fixed.
1994
1995 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1996 pcretest.c, to avoid build problems in some Borland environments.
1997
1998
1999 Version 8.10 25-Jun-2010
2000 ------------------------
2001
2002 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2003 THEN.
2004
2005 2. (*ACCEPT) was not working when inside an atomic group.
2006
2007 3. Inside a character class, \B is treated as a literal by default, but
2008 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2009 causes the error). The code is unchanged, but I tidied the documentation.
2010
2011 4. Inside a character class, PCRE always treated \R and \X as literals,
2012 whereas Perl faults them if its -w option is set. I have changed PCRE so
2013 that it faults them when PCRE_EXTRA is set.
2014
2015 5. Added support for \N, which always matches any character other than
2016 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2017
2018 6. When compiling pcregrep with newer versions of gcc which may have
2019 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2020 declared with attribute warn_unused_result" were given. Just casting the
2021 result to (void) does not stop the warnings; a more elaborate fudge is
2022 needed. I've used a macro to implement this.
2023
2024 7. Minor change to pcretest.c to avoid a compiler warning.
2025
2026 8. Added four artifical Unicode properties to help with an option to make
2027 \s etc use properties (see next item). The new properties are: Xan
2028 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2029
2030 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2031 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2032 this option. Modified pcretest to add /W to test this facility. Added
2033 REG_UCP to make it available via the POSIX interface.
2034
2035 10. Added --line-buffered to pcregrep.
2036
2037 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2038 studied, and the match started with a letter with a code point greater than
2039 127 whose first byte was different to the first byte of the other case of
2040 the letter, the other case of this starting letter was not recognized
2041 (#976).
2042
2043 12. If a pattern that was studied started with a repeated Unicode property
2044 test, for example, \p{Nd}+, there was the theoretical possibility of
2045 setting up an incorrect bitmap of starting bytes, but fortunately it could
2046 not have actually happened in practice until change 8 above was made (it
2047 added property types that matched character-matching opcodes).
2048
2049 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2050 possible starting bytes for non-anchored patterns.
2051
2052 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2053 \R, and also a number of cases that involve Unicode properties, both
2054 explicit and implicit when PCRE_UCP is set.
2055
2056 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2057 input, it could crash or give wrong results if characters with values
2058 greater than 0xc0 were present in the subject string. (Detail: it assumed
2059 UTF-8 input when processing these items.)
2060
2061 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2062 size_t is 64-bit (#991).
2063
2064 17. Added a check for running out of memory when PCRE is compiled with
2065 --disable-stack-for-recursion (#990).
2066
2067 18. If the last data line in a file for pcretest does not have a newline on
2068 the end, a newline was missing in the output.
2069
2070 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2071 less than 128) in its various bitmaps. However, there is a facility for
2072 generating tables according to the current locale when PCRE is compiled. It
2073 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2074 characters, are recognized by isspace() and therefore were getting set in
2075 these tables, and indeed these tables seem to approximate to ISO 8859. This
2076 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2077 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2078 which of course cannot start UTF-8 characters. I have changed the code so
2079 that only real ASCII characters (less than 128) and the correct starting
2080 bytes for UTF-8 encodings are set for characters greater than 127 when in
2081 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2082 altogether.)
2083
2084 20. Added the /T option to pcretest so as to be able to run tests with non-
2085 standard character tables, thus making it possible to include the tests
2086 used for 19 above in the standard set of tests.
2087
2088 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2089 reference to a subpattern the other side of a comment that contains an
2090 opening parenthesis caused either an internal compiling error, or a
2091 reference to the wrong subpattern.
2092
2093
2094 Version 8.02 19-Mar-2010
2095 ------------------------
2096
2097 1. The Unicode data tables have been updated to Unicode 5.2.0.
2098
2099 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2100 configured.
2101
2102 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2103 original author of that file, following a query about its status.
2104
2105 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2106 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2107
2108 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2109 quantifier applied to a forward-referencing subroutine call, could compile
2110 incorrect code or give the error "internal error: previously-checked
2111 referenced subpattern not found".
2112
2113 6. Both MS Visual Studio and Symbian OS have problems with initializing
2114 variables to point to external functions. For these systems, therefore,
2115 pcre_malloc etc. are now initialized to local functions that call the
2116 relevant global functions.
2117
2118 7. There were two entries missing in the vectors called coptable and poptable
2119 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2120 I've fixed the data, and added a kludgy way of testing at compile time that
2121 the lengths are correct (equal to the number of opcodes).
2122
2123 8. Following on from 7, I added a similar kludge to check the length of the
2124 eint vector in pcreposix.c.
2125
2126 9. Error texts for pcre_compile() are held as one long string to avoid too
2127 much relocation at load time. To find a text, the string is searched,
2128 counting zeros. There was no check for running off the end of the string,
2129 which could happen if a new error number was added without updating the
2130 string.
2131
2132 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2133
2134 11. \K was not working if it appeared in an atomic group or in a group that
2135 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2136 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2137 the assertion is positive, but not if it is negative.
2138
2139 12. Change 11 fortuitously reduced the size of the stack frame used in the
2140 "match()" function of pcre_exec.c by one pointer. Forthcoming
2141 implementation of support for (*MARK) will need an extra pointer on the
2142 stack; I have reserved it now, so that the stack frame size does not
2143 decrease.
2144
2145 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2146 item in branch that calls a recursion is a subroutine call - as in the
2147 second branch in the above example - was incorrectly given the compile-
2148 time error "recursive call could loop indefinitely" because pcre_compile()
2149 was not correctly checking the subroutine for matching a non-empty string.
2150
2151 14. The checks for overrunning compiling workspace could trigger after an
2152 overrun had occurred. This is a "should never occur" error, but it can be
2153 triggered by pathological patterns such as hundreds of nested parentheses.
2154 The checks now trigger 100 bytes before the end of the workspace.
2155
2156 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2157
2158
2159 Version 8.01 19-Jan-2010
2160 ------------------------
2161
2162 1. If a pattern contained a conditional subpattern with only one branch (in
2163 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2164 computed the wrong minimum data length (which is of course zero for such
2165 subpatterns). This could cause incorrect "no match" results.
2166
2167 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2168 the pattern is reset in the first branch, pcre_compile() failed with
2169 "internal error: code overflow at offset...". This happened only when
2170 the reset was to the original external option setting. (An optimization
2171 abstracts leading options settings into an external setting, which was the
2172 cause of this.)
2173
2174 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2175 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2176 assertion pattern did not match (meaning that the assertion was true), it
2177 was incorrectly treated as false if the SKIP had been reached during the
2178 matching. This also applied to assertions used as conditions.
2179
2180 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2181 assertion subpattern, including such a pattern used as a condition,
2182 unpredictable results occurred, instead of the error return
2183 PCRE_ERROR_DFA_UITEM.
2184
2185 5. The C++ GlobalReplace function was not working like Perl for the special
2186 situation when an empty string is matched. It now does the fancy magic
2187 stuff that is necessary.
2188
2189 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2190 removed. (These were left over from very, very early versions of PCRE.)
2191
2192 7. Some cosmetic changes to the code to make life easier when compiling it
2193 as part of something else:
2194
2195 (a) Change DEBUG to PCRE_DEBUG.
2196
2197 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2198 called "current" as "current_branch", to prevent a collision with the
2199 Linux macro when compiled as a kernel module.
2200
2201 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2202 prevent a collision with the Linux macro when compiled as a kernel
2203 module.
2204
2205 8. In pcre_compile() there are some checks for integer overflows that used to
2206 cast potentially large values to (double). This has been changed to that
2207 when building, a check for int64_t is made, and if it is found, it is used
2208 instead, thus avoiding the use of floating point arithmetic. (There is no
2209 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2210 double.
2211
2212 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2213 2005 (difference between two addresses compared to an unsigned value).
2214
2215 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2216 custom one, because of the following reported problem in Windows:
2217
2218 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2219 under Win32.
2220 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2221 therefore missing the function definition.
2222 - The compiler thus generates a "C" signature for the test function.
2223 - The linker fails to find the "C" function.
2224 - PCRE fails to configure if asked to do so against libbz2.
2225
2226 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2227 messages were output:
2228
2229 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2230 rerunning libtoolize, to keep the correct libtool macros in-tree.
2231 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2232
2233 I have done both of these things.
2234
2235 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2236 most of the time, it *can* run out if it is given a pattern that contains a
2237 runaway infinite recursion. I updated the discussion in the pcrestack man
2238 page.
2239
2240 13. Now that we have gone to the x.xx style of version numbers, the minor
2241 version may start with zero. Using 08 or 09 is a bad idea because users
2242 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2243 interpreted as invalid octal numbers. I've updated the previous comment in
2244 configure.ac, and also added a check that gives an error if 08 or 09 are
2245 used.
2246
2247 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2248 causing partial matching to fail when the end of the subject matched \W
2249 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2250
2251 15. There were some discrepancies between the declarations in pcre_internal.h
2252 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2253 their definitions. The declarations used "const uschar *" and the
2254 definitions used USPTR. Even though USPTR is normally defined as "const
2255 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2256 reported that: "This difference in casting confuses some C++ compilers, for
2257 example, SunCC recognizes above declarations as different functions and
2258 generates broken code for hbpcre." I have changed the declarations to use
2259 USPTR.
2260
2261 16. GNU libtool is named differently on some systems. The autogen.sh script now
2262 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2263 (FreeBSD).
2264
2265 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2266 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2267 comment: "Figure out how to create a longlong from a string: strtoll and
2268 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2269 instance, but it only takes 2 args instead of 3!"
2270
2271 18. A subtle bug concerned with back references has been fixed by a change of
2272 specification, with a corresponding code fix. A pattern such as
2273 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2274 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2275 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2276 same bug. Such groups have to be quantified to be useful, or contained
2277 inside another quantified group. (If there's no repetition, the reference
2278 can never match.) The problem arises because, having left the group and
2279 moved on to the rest of the pattern, a later failure that backtracks into
2280 the group uses the captured value from the final iteration of the group
2281 rather than the correct earlier one. I have fixed this in PCRE by forcing
2282 any group that contains a reference to itself to be an atomic group; that
2283 is, there cannot be any backtracking into it once it has completed. This is
2284 similar to recursive and subroutine calls.
2285
2286
2287 Version 8.00 19-Oct-09
2288 ----------------------
2289
2290 1. The table for translating pcre_compile() error codes into POSIX error codes
2291 was out-of-date, and there was no check on the pcre_compile() error code
2292 being within the table. This could lead to an OK return being given in
2293 error.
2294
2295 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2296 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2297 in a Windows environment.
2298
2299 3. The pcregrep --count option prints the count for each file even when it is
2300 zero, as does GNU grep. However, pcregrep was also printing all files when
2301 --files-with-matches was added. Now, when both options are given, it prints
2302 counts only for those files that have at least one match. (GNU grep just
2303 prints the file name in this circumstance, but including the count seems
2304 more useful - otherwise, why use --count?) Also ensured that the
2305 combination -clh just lists non-zero counts, with no names.
2306
2307 4. The long form of the pcregrep -F option was incorrectly implemented as
2308 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2309 but it seems right to fix it, and I didn't think it was worth preserving
2310 the old behaviour.
2311
2312 5. The command line items --regex=pattern and --regexp=pattern were not
2313 recognized by pcregrep, which required --regex pattern or --regexp pattern
2314 (with a space rather than an '='). The man page documented the '=' forms,
2315 which are compatible with GNU grep; these now work.
2316
2317 6. No libpcreposix.pc file was created for pkg-config; there was just
2318 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2319
2320 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2321 when UCP support is not needed, by modifying the Python script that
2322 generates it from Unicode data files. This should not matter if the module
2323 is correctly used as a library, but I received one complaint about 50K of
2324 unwanted data. My guess is that the person linked everything into his
2325 program rather than using a library. Anyway, it does no harm.
2326
2327 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2328 was a minimum greater than 1 for a wide character in a possessive
2329 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2330 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2331 character. Chaos in the form of incorrect output or a compiling loop could
2332 result.
2333
2334 9. The restrictions on what a pattern can contain when partial matching is
2335 requested for pcre_exec() have been removed. All patterns can now be
2336 partially matched by this function. In addition, if there are at least two
2337 slots in the offset vector, the offset of the earliest inspected character
2338 for the match and the offset of the end of the subject are set in them when
2339 PCRE_ERROR_PARTIAL is returned.
2340
2341 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2342 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2343 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2344 and may be more useful for multi-segment matching.
2345
2346 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2347 used to be given if ever the end of the subject was reached; now it is
2348 given only if matching could not proceed because another character was
2349 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2350 string "Z", which now yields "no match" instead of "partial match". In the
2351 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2352 final character ended with (*FAIL).
2353
2354 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2355 if the pattern had a "must contain" character that was already found in the
2356 earlier partial match, unless partial matching was again requested. For
2357 example, with the pattern /dog.(body)?/, the "must contain" character is
2358 "g". If the first part-match was for the string "dog", restarting with
2359 "sbody" failed. This bug has been fixed.
2360
2361 13. The string returned by pcre_dfa_exec() after a partial match has been
2362 changed so that it starts at the first inspected character rather than the
2363 first character of the match. This makes a difference only if the pattern
2364 starts with a lookbehind assertion or \b or \B (\K is not supported by
2365 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2366 matching functions compatible, and I think it's the right thing to do.
2367
2368 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2369 so that the demonstration program is easily available in environments where
2370 PCRE has not been installed from source.
2371
2372 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2373 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2374 library.
2375
2376 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2377 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2378 is not the first non-POSIX option to be added. Clearly some people find
2379 these options useful.
2380
2381 17. If a caller to the POSIX matching function regexec() passes a non-zero
2382 value for nmatch with a NULL value for pmatch, the value of
2383 nmatch is forced to zero.
2384
2385 18. RunGrepTest did not have a test for the availability of the -u option of
2386 the diff command, as RunTest does. It now checks in the same way as
2387 RunTest, and also checks for the -b option.
2388
2389 19. If an odd number of negated classes containing just a single character
2390 interposed, within parentheses, between a forward reference to a named
2391 subpattern and the definition of the subpattern, compilation crashed with
2392 an internal error, complaining that it could not find the referenced
2393 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2394 [The bug was that it was starting one character too far in when skipping
2395 over the character class, thus treating the ] as data rather than
2396 terminating the class. This meant it could skip too much.]
2397
2398 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2399 /g option in pcretest when the pattern contains \K, which makes it possible
2400 to have an empty string match not at the start, even when the pattern is
2401 anchored. Updated pcretest and pcredemo to use this option.
2402
2403 21. If the maximum number of capturing subpatterns in a recursion was greater
2404 than the maximum at the outer level, the higher number was returned, but
2405 with unset values at the outer level. The correct (outer level) value is
2406 now given.
2407
2408 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2409 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2410 make it do so. The string so far is captured, making this feature
2411 compatible with Perl.
2412
2413 23. The tests have been re-organized, adding tests 11 and 12, to make it
2414 possible to check the Perl 5.10 features against Perl 5.10.
2415
2416 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2417 pattern matches a fixed length string. PCRE did not allow this; now it
2418 does. Neither allows recursion.
2419
2420 25. I finally figured out how to implement a request to provide the minimum
2421 length of subject string that was needed in order to match a given pattern.
2422 (It was back references and recursion that I had previously got hung up
2423 on.) This code has now been added to pcre_study(); it finds a lower bound
2424 to the length of subject needed. It is not necessarily the greatest lower
2425 bound, but using it to avoid searching strings that are too short does give
2426 some useful speed-ups. The value is available to calling programs via
2427 pcre_fullinfo().
2428
2429 26. While implementing 25, I discovered to my embarrassment that pcretest had
2430 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2431 study optimizations had never been tested with that matching function.
2432 Oops. What is worse, even when it was passed study data, there was a bug in
2433 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2434 were also very few tests of studied patterns with pcre_dfa_exec().
2435
2436 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2437 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2438 on the other side of the coin, they are no longer allowed to have different
2439 names, because these cannot be distinguished in PCRE, and this has caused
2440 confusion. (This is a difference from Perl.)
2441
2442 28. When duplicate subpattern names are present (necessarily with different
2443 numbers, as required by 27 above), and a test is made by name in a
2444 conditional pattern, either for a subpattern having been matched, or for
2445 recursion in such a pattern, all the associated numbered subpatterns are
2446 tested, and the overall condition is true if the condition is true for any
2447 one of them. This is the way Perl works, and is also more like the way
2448 testing by number works.
2449
2450
2451 Version 7.9 11-Apr-09
2452 ---------------------
2453
2454 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2455 (pcretest), all targets were linked against these libraries. This included
2456 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2457 libraries. This caused unwanted dependencies to be created. This problem
2458 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2459 pcretest is linked with readline.
2460
2461 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2462 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2463 moved outside it again, because FALSE and TRUE are already defined in AIX,
2464 but BOOL is not.
2465
2466 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2467 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2468
2469 4. The pcregrep documentation said spaces were inserted as well as colons (or
2470 hyphens) following file names and line numbers when outputting matching
2471 lines. This is not true; no spaces are inserted. I have also clarified the
2472 wording for the --colour (or --color) option.
2473
2474 5. In pcregrep, when --colour was used with -o, the list of matching strings
2475 was not coloured; this is different to GNU grep, so I have changed it to be
2476 the same.
2477
2478 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2479 each matching line was coloured. Now it goes on to look for further matches
2480 of any of the test patterns, which is the same behaviour as GNU grep.
2481
2482 7. A pattern that could match an empty string could cause pcregrep to loop; it
2483 doesn't make sense to accept an empty string match in pcregrep, so I have
2484 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2485 seems to be how GNU grep behaves. [But see later change 40 for release
2486 8.33.]
2487
2488 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2489 start or after a newline", because the conditional assertion was not being
2490 correctly handled. The rule now is that both the assertion and what follows
2491 in the first alternative must satisfy the test.
2492
2493 9. If auto-callout was enabled in a pattern with a conditional group whose
2494 condition was an assertion, PCRE could crash during matching, both with
2495 pcre_exec() and pcre_dfa_exec().
2496
2497 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2498 used for matching.
2499
2500 11. Unicode property support in character classes was not working for
2501 characters (bytes) greater than 127 when not in UTF-8 mode.
2502
2503 12. Added the -M command line option to pcretest.
2504
2505 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2506
2507 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2508
2509 16. Added comments and documentation about mis-use of no_arg in the C++
2510 wrapper.
2511
2512 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2513 from Martin Jerabek that uses macro names for all relevant character and
2514 string constants.
2515
2516 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2517 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2518 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2519 these, but not everybody uses configure.
2520
2521 19. A conditional group that had only one branch was not being correctly
2522 recognized as an item that could match an empty string. This meant that an
2523 enclosing group might also not be so recognized, causing infinite looping
2524 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2525 with the subject "ab", where knowledge that the repeated group can match
2526 nothing is needed in order to break the loop.
2527
2528 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2529 exec(), but without supplying a callout function, matching went wrong.
2530
2531 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2532 leak if the size of the offset vector was greater than 30. When the vector
2533 is smaller, the saved offsets during recursion go onto a local stack
2534 vector, but for larger vectors malloc() is used. It was failing to free
2535 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2536 error, in fact).
2537
2538 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2539 heapframe that is used only when UTF-8 support is enabled. This caused no
2540 problem, but was untidy.
2541
2542 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2543 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2544 included within another project.
2545
2546 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2547 slightly modified by me:
2548
2549 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2550 not building pcregrep.
2551
2552 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2553 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2554
2555 25. Forward references, both numeric and by name, in patterns that made use of
2556 duplicate group numbers, could behave incorrectly or give incorrect errors,
2557 because when scanning forward to find the reference group, PCRE was not
2558 taking into account the duplicate group numbers. A pattern such as
2559 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2560
2561 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2562 the feature of a custom pointer more persuasive (as requested by a user).
2563
2564 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2565 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2566 pre-defined.
2567
2568 28. Added support for (*UTF8) at the start of a pattern.
2569
2570 29. Arrange for flags added by the "release type" setting in CMake to be shown
2571 in the configuration summary.
2572
2573
2574 Version 7.8 05-Sep-08
2575 ---------------------
2576
2577 1. Replaced UCP searching code with optimized version as implemented for Ad
2578 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2579 stage table and inline lookup instead of a function, giving speed ups of 2
2580 to 5 times on some simple patterns that I tested. Permission was given to
2581 distribute the MultiStage2.py script that generates the tables (it's not in
2582 the tarball, but is in the Subversion repository).
2583
2584 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2585 scripts.
2586
2587 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2588 a group with a zero qualifier. The result of the study could be incorrect,
2589 or the function might crash, depending on the pattern.
2590
2591 4. Caseless matching was not working for non-ASCII characters in back
2592 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2593 It now works when Unicode Property Support is available.
2594
2595 5. In pcretest, an escape such as \x{de} in the data was always generating
2596 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2597 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2598 truncation.
2599
2600 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2601
2602 7. Added two (int) casts to pcregrep when printing the difference of two
2603 pointers, in case they are 64-bit values.
2604
2605 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2606 test 2 if it fails.
2607
2608 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2609 and a #define of that name to empty if it is not externally set. This is to
2610 allow users of MSVC to set it if necessary.
2611
2612 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2613 the convenience functions in the pcre_get.c source file.
2614
2615 11. An option change at the start of a pattern that had top-level alternatives
2616 could cause overwriting and/or a crash. This command provoked a crash in
2617 some environments:
2618
2619 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2620
2621 This potential security problem was recorded as CVE-2008-2371.
2622
2623 12. For a pattern where the match had to start at the beginning or immediately
2624 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2625 pcre_dfa_exec() could read past the end of the passed subject if there was
2626 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2627 pcretest so that it places the subject at the end of its malloc-ed buffer.
2628
2629 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2630 exec() might read past the end of the data buffer in UTF-8 mode.
2631
2632 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2633 the data contained the byte 0x85 as part of a UTF-8 character within its
2634 first line. This applied both to normal and DFA matching.
2635
2636 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2637 /^[^d]*?$/8 failed to match "abc".
2638
2639 16. Added a missing copyright notice to pcrecpp_internal.h.
2640
2641 17. Make it more clear in the documentation that values returned from
2642 pcre_exec() in ovector are byte offsets, not character counts.
2643
2644 18. Tidied a few places to stop certain compilers from issuing warnings.
2645
2646 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2647 supplied by Stefan Weber. I made a further small update for 7.8 because
2648 there is a change of source arrangements: the pcre_searchfuncs.c module is
2649 replaced by pcre_ucd.c.
2650
2651
2652 Version 7.7 07-May-08
2653 ---------------------
2654
2655 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2656 a string to a long long, pretend we don't even have a long long." This is
2657 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2658
2659 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2660 pre-7.6 versions, which defined a global no_arg variable instead of putting
2661 it in the RE class. (See also #8 below.)
2662
2663 3. Remove a line of dead code, identified by coverity and reported by Nuno
2664 Lopes.
2665
2666 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2667
2668 (1) The include/exclude patterns were being applied to the whole pathnames
2669 of files, instead of just to the final components.
2670
2671 (2) If there was more than one level of directory, the subdirectories were
2672 skipped unless they satisfied the include/exclude conditions. This is
2673 inconsistent with GNU grep (and could even be seen as contrary to the
2674 pcregrep specification - which I improved to make it absolutely clear).
2675 The action now is always to scan all levels of directory, and just
2676 apply the include/exclude patterns to regular files.
2677
2678 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2679 --exclude_dir in the tests to avoid scanning .svn directories.
2680
2681 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2682 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2683 doesn't support NULs in patterns.
2684
2685 7. Added some missing "const"s to declarations of static tables in
2686 pcre_compile.c and pcre_dfa_exec.c.
2687
2688 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2689 caused by fix #2 above. (Subsequently also a second patch to fix the
2690 first patch. And a third patch - this was a messy problem.)
2691
2692 9. Applied Craig's patch to remove the use of push_back().
2693
2694 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2695 matching function regexec().
2696
2697 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2698 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2699 references. PCRE supports relative numbers with this syntax (I don't think
2700 Oniguruma does).
2701
2702 12. Previously, a group with a zero repeat such as (...){0} was completely
2703 omitted from the compiled regex. However, this means that if the group
2704 was called as a subroutine from elsewhere in the pattern, things went wrong
2705 (an internal error was given). Such groups are now left in the compiled
2706 pattern, with a new opcode that causes them to be skipped at execution
2707 time.
2708
2709 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2710 to the way PCRE behaves:
2711
2712 (a) A lone ] character is dis-allowed (Perl treats it as data).
2713
2714 (b) A back reference to an unmatched subpattern matches an empty string
2715 (Perl fails the current match path).
2716
2717 (c) A data ] in a character class must be notated as \] because if the
2718 first data character in a class is ], it defines an empty class. (In
2719 Perl it is not possible to have an empty class.) The empty class []
2720 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2721 The negative empty class [^] matches any one character, independently
2722 of the DOTALL setting.
2723
2724 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2725 non-existent subpattern following a character class starting with ']' and
2726 containing () gave an internal compiling error instead of "reference to
2727 non-existent subpattern". Fortunately, when the pattern did exist, the
2728 compiled code was correct. (When scanning forwards to check for the
2729 existence of the subpattern, it was treating the data ']' as terminating
2730 the class, so got the count wrong. When actually compiling, the reference
2731 was subsequently set up correctly.)
2732
2733 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2734 it was being rejected as not supported by pcre_dfa_exec(), even though
2735 other assertions are supported. I have made pcre_dfa_exec() support
2736 (*FAIL).
2737
2738 16. The implementation of 13c above involved the invention of a new opcode,
2739 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2740 cannot be changed at match time, I realized I could make a small
2741 improvement to matching performance by compiling OP_ALLANY instead of
2742 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2743 on the OP_ANY path.
2744
2745 17. Compiling pcretest on Windows with readline support failed without the
2746 following two fixes: (1) Make the unistd.h include conditional on
2747 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2748
2749 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2750 ncurses library to be included for pcretest when ReadLine support is
2751 requested, but also to allow for it to be overridden. This patch came from
2752 Daniel Bergström.
2753
2754 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2755 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2756 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2757 spotting this.
2758
2759
2760 Version 7.6 28-Jan-08
2761 ---------------------
2762
2763 1. A character class containing a very large number of characters with
2764 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2765 overflow.
2766
2767 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2768 HAVE_LONG_LONG is not defined.
2769
2770 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2771 bring them up to date and include new features. This patch includes:
2772
2773 - Fixed PH's badly added libz and libbz2 support.
2774 - Fixed a problem with static linking.
2775 - Added pcredemo. [But later removed - see 7 below.]
2776 - Fixed dftables problem and added an option.
2777 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2778 HAVE_LONG_LONG.
2779 - Added readline support for pcretest.
2780 - Added an listing of the option settings after cmake has run.
2781
2782 4. A user submitted a patch to Makefile that makes it easy to create
2783 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2784 Makefile.am that cause it to include this special target, without
2785 affecting anything else. Note that the same mingw target plus all
2786 the other distribution libraries and programs are now supported
2787 when configuring with CMake (see 6 below) instead of with
2788 Configure/Make.
2789
2790 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2791 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2792 exported in the Windows port". It has not yet been confirmed that the patch
2793 solves the problem, but it does no harm.
2794
2795 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2796 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2797 with CMake, and also correct the comment about stack recursion.
2798
2799 7. Remove the automatic building of pcredemo from the ./configure system and
2800 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2801 of a program that users should build themselves after PCRE is installed, so
2802 building it automatically is not really right. What is more, it gave
2803 trouble in some build environments.
2804
2805 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2806
2807
2808 Version 7.5 10-Jan-08
2809 ---------------------
2810
2811 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2812 values in parens when parsing an RE using the C++ wrapper."
2813
2814 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2815 Characters greater than 255 were excluded from the class instead of being
2816 included.
2817
2818 3. The same bug as (2) above applied to negated POSIX classes such as
2819 [:^space:].
2820
2821 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2822 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2823 I have changed it.
2824
2825 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2826 first named subpattern) and a construct such as (?&a) would reference the
2827 first named subpattern whose name started with "a" (in other words, the
2828 length check was missing). Both these problems are fixed. "Subpattern name
2829 expected" is now given for (?&) (a zero-length name), and this patch also
2830 makes it give the same error for \k'' (previously it complained that that
2831 was a reference to a non-existent subpattern).
2832
2833 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2834 this is right because (?- can be followed by option settings as well as by
2835 digits. I have, however, made the messages clearer.
2836
2837 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2838 than the number used in the conditional) now cause a compile-time error.
2839 This is actually not compatible with Perl, which accepts such patterns, but
2840 treats the conditional as always being FALSE (as PCRE used to), but it
2841 seems to me that giving a diagnostic is better.
2842
2843 8. Change "alphameric" to the more common word "alphanumeric" in comments
2844 and messages.
2845
2846 9. Fix two occurrences of "backslash" in comments that should have been
2847 "backspace".
2848
2849 10. Remove two redundant lines of code that can never be obeyed (their function
2850 was moved elsewhere).
2851
2852 11. The program that makes PCRE's Unicode character property table had a bug
2853 which caused it to generate incorrect table entries for sequences of
2854 characters that have the same character type, but are in different scripts.
2855 It amalgamated them into a single range, with the script of the first of
2856 them. In other words, some characters were in the wrong script. There were
2857 thirteen such cases, affecting characters in the following ranges:
2858
2859 U+002b0 - U+002c1
2860 U+0060c - U+0060d
2861 U+0061e - U+00612
2862 U+0064b - U+0065e
2863 U+0074d - U+0076d
2864 U+01800 - U+01805
2865 U+01d00 - U+01d77
2866 U+01d9b - U+01dbf
2867 U+0200b - U+0200f
2868 U+030fc - U+030fe
2869 U+03260 - U+0327f
2870 U+0fb46 - U+0fbb1
2871 U+10450 - U+1049d
2872
2873 12. The -o option (show only the matching part of a line) for pcregrep was not
2874 compatible with GNU grep in that, if there was more than one match in a
2875 line, it showed only the first of them. It now behaves in the same way as
2876 GNU grep.
2877
2878 13. If the -o and -v options were combined for pcregrep, it printed a blank
2879 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2880 does the same. The return code can be used to tell if there were any
2881 non-matching lines.
2882
2883 14. Added --file-offsets and --line-offsets to pcregrep.
2884
2885 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2886 infinitely looping recursion. The bug was that positive lookaheads were not
2887 being skipped when checking for a possible empty match (negative lookaheads
2888 and both kinds of lookbehind were skipped).
2889
2890 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2891 inclusion of <windows.h> to before rather than after the definition of
2892 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2893
2894 17. Specifying a possessive quantifier with a specific limit for a Unicode
2895 character property caused pcre_compile() to compile bad code, which led at
2896 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2897 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2898 caused the error; without that there was no problem.
2899
2900 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2901
2902 19. Added --enable-pcretest-libreadline.
2903
2904 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2905 RE::GlobalReplace(). As a result, the number of replacements returned was
2906 double what it should be. I removed one of the increments, but Craig sent a
2907 later patch that removed the other one (the right fix) and added unit tests
2908 that check the return values (which was not done before).
2909
2910 21. Several CMake things:
2911
2912 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2913 the names libpcre and libpcreposix, not just pcre and pcreposix.
2914
2915 (2) The above change means that pcretest and pcregrep are now correctly
2916 linked with the newly-built libraries, not previously installed ones.
2917
2918 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2919
2920 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2921 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2922 UTF-8 newline character). The key issue is that the pattern starts .*;
2923 this means that the match must be either at the beginning, or after a
2924 newline. The bug was in the code for advancing after a failed match and
2925 checking that the new position followed a newline. It was not taking
2926 account of UTF-8 characters correctly.
2927
2928 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2929 character classes. PCRE was not treating the sequence [:...:] as a
2930 character class unless the ... were all letters. Perl, however, seems to
2931 allow any characters between [: and :], though of course it rejects as
2932 unknown any "names" that contain non-letters, because all the known class
2933 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2934 for example, whereas PCRE did not - it did not recognize a POSIX character
2935 class. This seemed a bit dangerous, so the code has been changed to be
2936 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2937 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2938 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2939 Perl does, and where it didn't before.
2940
2941 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2942 Windows environments %n is disabled by default.
2943
2944
2945 Version 7.4 21-Sep-07
2946 ---------------------
2947
2948 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2949 means that a class such as [\s] counted as "explicit reference to CR or
2950 LF". That isn't really right - the whole point of the change was to try to
2951 help when there was an actual mention of one of the two characters. So now
2952 the change happens only if \r or \n (or a literal CR or LF) character is
2953 encountered.
2954
2955 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2956 of both had grown to the point where there were only 3 bits left.
2957 Fortunately, there was spare space in the data structure, and so I have
2958 moved the internal flags into a new 16-bit field to free up more option
2959 bits.
2960
2961 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2962 but did not set the internal JCHANGED flag - either of these is enough to
2963 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2964 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2965 start sets both bits.
2966
2967 4. Added options (at build time, compile time, exec time) to change \R from
2968 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2969
2970 5. doc/pcresyntax.html was missing from the distribution.
2971
2972 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2973 compatibility, even though it is no longer used.
2974
2975 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2976 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2977 windows.h file is present (where different names are used). [This was
2978 reversed later after testing - see 16 below.]
2979
2980 8. Changed all #include <config.h> to #include "config.h". There were also
2981 some further <pcre.h> cases that I changed to "pcre.h".
2982
2983 9. When pcregrep was used with the --colour option, it missed the line ending
2984 sequence off the lines that it output.
2985
2986 10. It was pointed out to me that arrays of string pointers cause lots of
2987 relocations when a shared library is dynamically loaded. A technique of
2988 using a single long string with a table of offsets can drastically reduce
2989 these. I have refactored PCRE in four places to do this. The result is
2990 dramatic:
2991
2992 Originally: 290
2993 After changing UCP table: 187
2994 After changing error message table: 43
2995 After changing table of "verbs" 36
2996 After changing table of Posix names 22
2997
2998 Thanks to the folks working on Gregex for glib for this insight.
2999
3000 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3001 unicode-properties was also set.
3002
3003 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3004
3005 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3006 checked only for CRLF.
3007
3008 14. Added casts to pcretest.c to avoid compiler warnings.
3009
3010 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3011
3012 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3013 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3014 entirely. This removes changes made in 7 above.
3015
3016 17. The CMake files have been updated, and there is now more information about
3017 building with CMake in the NON-UNIX-USE document.
3018
3019
3020 Version 7.3 28-Aug-07
3021 ---------------------
3022
3023 1. In the rejigging of the build system that eventually resulted in 7.1, the
3024 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3025 brackets there is not right, since it causes compilers to look for an
3026 installed pcre.h, not the version that is in the source that is being
3027 compiled (which of course may be different). I have changed it back to:
3028
3029 #include "pcre.h"
3030
3031 I have a vague recollection that the change was concerned with compiling in
3032 different directories, but in the new build system, that is taken care of
3033 by the VPATH setting the Makefile.
3034
3035 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3036 when the subject happened to end in the byte 0x85 (e.g. if the last
3037 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3038 characters but of course it shouldn't be taken as a newline when it is part
3039 of another character. The bug was that, for an unlimited repeat of . in
3040 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3041 characters when looking for a newline.
3042
3043 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3044
3045 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3046 in debug output.
3047
3048 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3049 long printing in the pcrecpp unittest when running under MinGW.
3050
3051 6. ESC_K was left out of the EBCDIC table.
3052
3053 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3054 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3055 limit also applies to "virtual nesting" when a pattern is recursive, and in
3056 this case 1000 isn't so big. I have been able to remove this limit at the
3057 expense of backing off one optimization in certain circumstances. Normally,
3058 when pcre_exec() would call its internal match() function recursively and
3059 immediately return the result unconditionally, it uses a "tail recursion"
3060 feature to save stack. However, when a subpattern that can match an empty
3061 string has an unlimited repetition quantifier, it no longer makes this
3062 optimization. That gives it a stack frame in which to save the data for
3063 checking that an empty string has been matched. Previously this was taken
3064 from the 1000-entry workspace that had been reserved. So now there is no
3065 explicit limit, but more stack is used.
3066
3067 8. Applied Daniel's patches to solve problems with the import/export magic
3068 syntax that is required for Windows, and which was going wrong for the
3069 pcreposix and pcrecpp parts of the library. These were overlooked when this
3070 problem was solved for the main library.
3071
3072 9. There were some crude static tests to avoid integer overflow when computing
3073 the size of patterns that contain repeated groups with explicit upper
3074 limits. As the maximum quantifier is 65535, the maximum group length was
3075 set at 30,000 so that the product of these two numbers did not overflow a
3076 32-bit integer. However, it turns out that people want to use groups that
3077 are longer than 30,000 bytes (though not repeat them that many times).
3078 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3079 made it possible to implement the integer overflow checks in a much more
3080 dynamic way, which I have now done. The artificial limitation on group
3081 length has been removed - we now have only the limit on the total length of
3082 the compiled pattern, which depends on the LINK_SIZE setting.
3083
3084 10. Fixed a bug in the documentation for get/copy named substring when
3085 duplicate names are permitted. If none of the named substrings are set, the
3086 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3087 empty string.
3088
3089 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3090 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3091 because the ] is interpreted as the first data character and the
3092 terminating ] is not found. PCRE has been made compatible with Perl in this
3093 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3094 cause memory overwriting.
3095
3096 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3097 string has been matched (to stop an infinite loop). It was not recognizing
3098 a conditional subpattern that could match an empty string if that
3099 subpattern was within another subpattern. For example, it looped when
3100 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3101 condition was not nested. This bug has been fixed.
3102
3103 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3104 past the start of the subject in the presence of bytes with the top bit
3105 set, for example "\x8aBCD".
3106
3107 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3108 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3109
3110 14. Optimized (?!) to (*FAIL).
3111
3112 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3113 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3114 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3115 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3116 does: it's just the validity check that is more restrictive.
3117
3118 16. Inserted checks for integer overflows during escape sequence (backslash)
3119 processing, and also fixed erroneous offset values for syntax errors during
3120 backslash processing.
3121
3122 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3123 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3124
3125 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3126 caused an overrun.
3127
3128 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3129 something other than just ASCII characters) inside a group that had an
3130 unlimited repeat caused a loop at compile time (while checking to see
3131 whether the group could match an empty string).
3132
3133 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3134 [\P{Any}] did so. (Error in the code for printing property names.)
3135
3136 21. An orphan \E inside a character class could cause a crash.
3137
3138 22. A repeated capturing bracket such as (A)? could cause a wild memory
3139 reference during compilation.
3140
3141 23. There are several functions in pcre_compile() that scan along a compiled
3142 expression for various reasons (e.g. to see if it's fixed length for look
3143 behind). There were bugs in these functions when a repeated \p or \P was
3144 present in the pattern. These operators have additional parameters compared
3145 with \d, etc, and these were not being taken into account when moving along
3146 the compiled data. Specifically:
3147
3148 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3149 length.
3150
3151 (b) An item such as \pL+ within a repeated group could cause crashes or
3152 loops.
3153
3154 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3155 "reference to non-existent subpattern" error.
3156
3157 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3158
3159 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3160 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3161
3162 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3163
3164 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3165 character were causing crashes (broken optimization).
3166
3167 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3168 \p or \P) caused a compile-time loop.
3169
3170 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3171 break. For example, the unstudied pattern [\r\n]A does not match the string
3172 "\r\nA" because change 7.0/46 below moves the current point on by two
3173 characters after failing to match at the start. However, the pattern \nA
3174 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3175 the same is true. There doesn't seem any very clean way out of this, but
3176 what I have chosen to do makes the common cases work: PCRE now takes note
3177 of whether there can be an explicit match for \r or \n anywhere in the
3178 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3179 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3180 pattern has explicit CR or LF references.
3181
3182 29. Added (*CR) etc for changing newline setting at start of pattern.
3183
3184
3185 Version 7.2 19-Jun-07
3186 ---------------------
3187
3188 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3189 which is apparently normally available under Windows.
3190
3191 2. Re-jig the pcregrep tests with different newline settings in an attempt
3192 to make them independent of the local environment's newline setting.
3193
3194 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3195
3196 4. Some of the "internals" tests were previously cut out when the link size
3197 was not 2, because the output contained actual offsets. The recent new
3198 "Z" feature of pcretest means that these can be cut out, making the tests
3199 usable with all link sizes.
3200
3201 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3202 stack recursion. This gives a massive performance boost under BSD, but just
3203 a small improvement under Linux. However, it saves one field in the frame
3204 in all cases.
3205
3206 6. Added more features from the forthcoming Perl 5.10:
3207
3208 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3209 recursion call. It refers to the nth most recently opened parentheses.
3210
3211 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3212 to be opened parentheses.
3213
3214 (c) Conditions that refer to capturing parentheses can be specified
3215 relatively, for example, (?(-2)... or (?(+3)...
3216
3217 (d) \K resets the start of the current match so that everything before
3218 is not part of it.
3219
3220 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3221
3222 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3223 reference syntax.
3224
3225 (g) (?| introduces a group in which the numbering of parentheses in each
3226 alternative starts with the same number.
3227
3228 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3229
3230 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3231 PCRE_INFO_JCHANGED.
3232
3233 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3234 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3235 for detecting groups that can match an empty string.
3236
3237 9. A pattern with a very large number of alternatives (more than several
3238 hundred) was running out of internal workspace during the pre-compile
3239 phase, where pcre_compile() figures out how much memory will be needed. A
3240 bit of new cunning has reduced the workspace needed for groups with
3241 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3242 workspace instead of running out of the 4096 that are available.
3243
3244 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3245
3246 11. Applied patch from Google to remove an optimization that didn't quite work.
3247 The report of the bug said:
3248
3249 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3250 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3251 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3252
3253 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3254 it matched the wrong number of bytes.
3255
3256
3257 Version 7.1 24-Apr-07
3258 ---------------------
3259
3260 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3261 that is more "standard", making use of automake and other Autotools. There
3262 is some re-arrangement of the files and adjustment of comments consequent
3263 on this.
3264
3265 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3266 for recursive directory scanning broke on some systems because the files
3267 are not scanned in any specific order and on different systems the order
3268 was different. A call to "sort" has been inserted into RunGrepTest for the
3269 approprate test as a short-term fix. In the longer term there may be an
3270 alternative.
3271
3272 3. I had an email from Eric Raymond about problems translating some of PCRE's
3273 man pages to HTML (despite the fact that I distribute HTML pages, some
3274 people do their own conversions for various reasons). The problems
3275 concerned the use of low-level troff macros .br and .in. I have therefore
3276 removed all such uses from the man pages (some were redundant, some could
3277 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3278 HTML has been updated to handle .nf/.fi and to complain if it encounters
3279 .br or .in.
3280
3281 4. Updated comments in configure.ac that get placed in config.h.in and also
3282 arranged for config.h to be included in the distribution, with the name
3283 config.h.generic, for the benefit of those who have to compile without
3284 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3285
3286 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3287 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3288 makevp.bat for the current PCRE, using the additional files
3289 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3290
3291 6. A Windows user reported a minor discrepancy with test 2, which turned out
3292 to be caused by a trailing space on an input line that had got lost in his
3293 copy. The trailing space was an accident, so I've just removed it.
3294
3295 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3296 that is needed.
3297
3298 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3299 as "const" (a) because they are and (b) because it helps the PHP
3300 maintainers who have recently made a script to detect big data structures
3301 in the php code that should be moved to the .rodata section. I remembered
3302 to update Builducptable as well, so it won't revert if ucptable.h is ever
3303 re-created.
3304
3305 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3306 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3307 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3308 support is not required. This saves 1.5-2K of code, which is important in
3309 some applications.
3310
3311 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3312 so as not to refer to the tables, even though these functions will never be
3313 called when UTF-8 support is disabled. Otherwise there are problems with a
3314 shared library.
3315
3316 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3317
3318 (a) It was defining its arguments as char * instead of void *.
3319
3320 (b) It was assuming that all moves were upwards in memory; this was true
3321 a long time ago when I wrote it, but is no longer the case.
3322
3323 The emulated memove() is provided for those environments that have neither
3324 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3325 is clearly not the case, as these two bugs were recently reported.
3326
3327 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3328 and Detrail to create the HTML documentation, the .txt form of the man
3329 pages, and it removes trailing spaces from listed files. It also creates
3330 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3331 case, it wraps all the #defines with #ifndefs. This script should be run
3332 before "make dist".
3333
3334 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3335 with Unicode property support.
3336
3337 (a) For a maximizing quantifier, if the two different cases of the
3338 character were of different lengths in their UTF-8 codings (there are
3339 some cases like this - I found 11), and the matching function had to
3340 back up over a mixture of the two cases, it incorrectly assumed they
3341 were both the same length.
3342
3343 (b) When PCRE was configured to use the heap rather than the stack for
3344 recursion during matching, it was not correctly preserving the data for
3345 the other case of a UTF-8 character when checking ahead for a match
3346 while processing a minimizing repeat. If the check also involved
3347 matching a wide character, but failed, corruption could cause an
3348 erroneous result when trying to check for a repeat of the original
3349 character.
3350
3351 13. Some tidying changes to the testing mechanism:
3352
3353 (a) The RunTest script now detects the internal link size and whether there
3354 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3355 values substituted by "configure". (The RunGrepTest script already did
3356 this for UTF-8.) The configure.ac script no longer substitutes the
3357 relevant variables.
3358
3359 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3360 with length and offset values. This means that the output is different
3361 for different internal link sizes. Test 2 is skipped for link sizes
3362 other than 2 because of this, bypassing the problem. Unfortunately,
3363 there was also a test in test 3 (the locale tests) that used /B and
3364 failed for link sizes other than 2. Rather than cut the whole test out,
3365 I have added a new /Z option to pcretest that replaces the length and
3366 offset values with spaces. This is now used to make test 3 independent
3367 of link size. (Test 2 will be tidied up later.)
3368
3369 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3370 segmentation fault instead of returning the appropriate error message.
3371
3372 15. In multiline mode when the newline sequence was set to "any", the pattern
3373 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3374 This doesn't seem right; it now treats the CRLF combination as the line
3375 ending, and so does not match in that case. It's only a pattern such as ^$
3376 that would hit this one: something like ^ABC$ would have failed after \r
3377 and then tried again after \r\n.
3378
3379 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3380 in an attempt to make files that differ only in their line terminators
3381 compare equal. This works on Linux.
3382
3383 17. Under certain error circumstances pcregrep might try to free random memory
3384 as it exited. This is now fixed, thanks to valgrind.
3385
3386 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3387 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3388 was because its rules for how to advance for /g after matching an empty
3389 string at the end of a line did not allow for this case. They now check for
3390 it specially.
3391
3392 20. pcretest is supposed to handle patterns and data of any length, by
3393 extending its buffers when necessary. It was getting this wrong when the
3394 buffer for a data line had to be extended.
3395
3396 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3397 CRLF as a newline sequence.
3398
3399 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3400 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3401 I have nevertheless tidied it up.
3402
3403 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3404
3405 24. Added a man page for pcre-config.
3406
3407
3408 Version 7.0 19-Dec-06
3409 ---------------------
3410
3411 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3412 moving to gcc 4.1.1.
3413
3414 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3415 sys/time.h, which is documented as needed for this function. It doesn't
3416 seem to matter on Linux, but it showed up on some releases of OS X.
3417
3418 3. It seems that there are systems where bytes whose values are greater than
3419 127 match isprint() in the "C" locale. The "C" locale should be the
3420 default when a C program starts up. In most systems, only ASCII printing
3421 characters match isprint(). This difference caused the output from pcretest
3422 to vary, making some of the tests fail. I have changed pcretest so that:
3423
3424 (a) When it is outputting text in the compiled version of a pattern, bytes
3425 other than 32-126 are always shown as hex escapes.
3426
3427 (b) When it is outputting text that is a matched part of a subject string,
3428 it does the same, unless a different locale has been set for the match
3429 (using the /L modifier). In this case, it uses isprint() to decide.
3430
3431 4. Fixed a major bug that caused incorrect computation of the amount of memory
3432 required for a compiled pattern when options that changed within the
3433 pattern affected the logic of the preliminary scan that determines the
3434 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3435 that the computed length was too small. The symptoms of this bug were
3436 either the PCRE error "internal error: code overflow" from pcre_compile(),
3437 or a glibc crash with a message such as "pcretest: free(): invalid next
3438 size (fast)". Examples of patterns that provoked this bug (shown in
3439 pcretest format) are:
3440
3441 /(?-x: )/x
3442 /(?x)(?-x: \s*#\s*)/
3443 /((?i)[\x{c0}])/8
3444 /(?i:[\x{c0}])/8
3445
3446 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3447 is now done differently.
3448
3449 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3450 wrapper classes; (b) implement a new function in the C++ scanner that is
3451 more efficient than the old way of doing things because it avoids levels of
3452 recursion in the regex matching; (c) add a paragraph to the documentation
3453 for the FullMatch() function.
3454
3455 6. The escape sequence \n was being treated as whatever was defined as
3456 "newline". Not only was this contrary to the documentation, which states
3457 that \n is character 10 (hex 0A), but it also went horribly wrong when
3458 "newline" was defined as CRLF. This has been fixed.
3459
3460 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3461 was being set to -1 for the "end of line" case (supposedly a value that no
3462 character can have). Though this value is never used (the check for end of
3463 line is "zero bytes in current character"), it caused compiler complaints.
3464 I've changed it to 0xffffffff.
3465
3466 8. In pcre_version.c, the version string was being built by a sequence of
3467 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3468 string (as it is for production releases) called a macro with an empty
3469 argument. The C standard says the result of this is undefined. The gcc
3470 compiler treats it as an empty string (which was what was wanted) but it is
3471 reported that Visual C gives an error. The source has been hacked around to
3472 avoid this problem.
3473
3474 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3475 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3476 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3477 of them did).
3478
3479 10. Originally, pcretest opened its input and output without "b"; then I was
3480 told that "b" was needed in some environments, so it was added for release
3481 5.0 to both the input and output. (It makes no difference on Unix-like
3482 systems.) Later I was told that it is wrong for the input on Windows. I've
3483 now abstracted the modes into two macros, to make it easier to fiddle with
3484 them, and removed "b" from the input mode under Windows.
3485
3486 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3487
3488 12. Added -help and --help to pcretest as an official way of being reminded
3489 of the options.
3490
3491 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3492 and pcrecpp.cc because they annoy compilers at high warning levels.
3493
3494 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3495
3496 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3497 scripts are not C programs :-) and which was not noticed because it works
3498 on Linux.
3499
3500 16. pcretest is supposed to handle any length of pattern and data line (as one
3501 line or as a continued sequence of lines) by extending its input buffer if
3502 necessary. This feature was broken for very long pattern lines, leading to
3503 a string of junk being passed to pcre_compile() if the pattern was longer
3504 than about 50K.
3505
3506 17. I have done a major re-factoring of the way pcre_compile() computes the
3507 amount of memory needed for a compiled pattern. Previously, there was code
3508 that made a preliminary scan of the pattern in order to do this. That was
3509 OK when PCRE was new, but as the facilities have expanded, it has become
3510 harder and harder to keep it in step with the real compile phase, and there
3511 have been a number of bugs (see for example, 4 above). I have now found a
3512 cunning way of running the real compile function in a "fake" mode that
3513 enables it to compute how much memory it would need, while actually only
3514 ever using a few hundred bytes of working memory and without too many
3515 tests of the mode. This should make future maintenance and development
3516 easier. A side effect of this work is that the limit of 200 on the nesting
3517 depth of parentheses has been removed (though this was never a serious
3518 limitation, I suspect). However, there is a downside: pcre_compile() now
3519 runs more slowly than before (30% or more, depending on the pattern). I
3520 hope this isn't a big issue. There is no effect on runtime performance.
3521
3522 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3523 newline (only possible for the last line of a file) and it was a
3524 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3525
3526 19. Added additional timing features to pcretest. (1) The -tm option now times
3527 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3528 separate command line item, by a number that specifies the number of
3529 repeats to use when timing. The default is 50000; this gives better
3530 precision, but takes uncomfortably long for very large patterns.
3531
3532 20. Extended pcre_study() to be more clever in cases where a branch of a
3533 subpattern has no definite first character. For example, (a*|b*)[cd] would
3534 previously give no result from pcre_study(). Now it recognizes that the
3535 first character must be a, b, c, or d.
3536
3537 21. There was an incorrect error "recursive call could loop indefinitely" if
3538 a subpattern (or the entire pattern) that was being tested for matching an
3539 empty string contained only one non-empty item after a nested subpattern.
3540 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3541 incorrectly, because the \d was being skipped in the check.
3542
3543 22. The pcretest program now has a new pattern option /B and a command line
3544 option -b, which is equivalent to adding /B to every pattern. This causes
3545 it to show the compiled bytecode, without the additional information that
3546 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3547 is the same as /B/I).
3548
3549 23. A new optimization is now able automatically to treat some sequences such
3550 as a*b as a*+b. More specifically, if something simple (such as a character
3551 or a simple class like \d) has an unlimited quantifier, and is followed by
3552 something that cannot possibly match the quantified thing, the quantifier
3553 is automatically "possessified".
3554
3555 24. A recursive reference to a subpattern whose number was greater than 39
3556 went wrong under certain circumstances in UTF-8 mode. This bug could also
3557 have affected the operation of pcre_study().
3558
3559 25. Realized that a little bit of performance could be had by replacing
3560 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3561
3562 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3563
3564 27. Possessive quantifiers such as a++ were previously implemented by turning
3565 them into atomic groups such as ($>a+). Now they have their own opcodes,
3566 which improves performance. This includes the automatically created ones
3567 from 23 above.
3568
3569 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3570 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3571 the first matched character to be a colon. This applied both to named and
3572 numbered groups.
3573
3574 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3575
3576 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3577 building PCRE on Windows easier, so I have included it in the distribution.
3578
3579 31. There is now a check in pcretest against a ridiculously large number being
3580 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3581 loop, the loop is abandoned.
3582
3583 32. Forward references to subpatterns in conditions such as (?(2)...) where
3584 subpattern 2 is defined later cause pcre_compile() to search forwards in
3585 the pattern for the relevant set of parentheses. This search went wrong
3586 when there were unescaped parentheses in a character class, parentheses
3587 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3588
3589 33. "Subroutine" calls and backreferences were previously restricted to
3590 referencing subpatterns earlier in the regex. This restriction has now
3591 been removed.
3592
3593 34. Added a number of extra features that are going to be in Perl 5.10. On the
3594 whole, these are just syntactic alternatives for features that PCRE had
3595 previously implemented using the Python syntax or my own invention. The
3596 other formats are all retained for compatibility.
3597
3598 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3599 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3600 also .NET compatible.
3601
3602 (b) A recursion or subroutine call to a named group can now be defined as
3603 (?&name) as well as (?P>name).
3604
3605 (c) A backreference to a named group can now be defined as \k<name> or
3606 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3607 5.10, are also .NET compatible.
3608
3609 (d) A conditional reference to a named group can now use the syntax
3610 (?(<name>) or (?('name') as well as (?(name).
3611
3612 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3613 groups (named and numbered) that are never evaluated inline, but can be
3614 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3615 is always false. There may be only one alternative in such a group.
3616
3617 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3618 as the simple (?(R). The condition is true only if the most recent
3619 recursion is that of the given number or name. It does not search out
3620 through the entire recursion stack.
3621
3622 (g) The escape \gN or \g{N} has been added, where N is a positive or
3623 negative number, specifying an absolute or relative reference.
3624
3625 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3626 some "unreachable code" warnings.
3627
3628 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3629 things, this adds five new scripts.
3630
3631 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3632 There were also incompatibilities regarding the handling of \Q..\E inside
3633 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3634 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3635
3636 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3637 matches an empty string, and forcibly breaks the loop. There were bugs in
3638 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3639 against aaaa the result was just "a" rather than "aaaa", for example. Two
3640 separate and independent bugs (that affected different cases) have been
3641 fixed.
3642
3643 39. Refactored the code to abolish the use of different opcodes for small
3644 capturing bracket numbers. This is a tidy that I avoided doing when I
3645 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3646 The new approach is not only tidier, it makes it possible to reduce the
3647 memory needed to fix the previous bug (38).
3648
3649 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3650 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3651 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3652 mode.
3653
3654 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3655 report.
3656
3657 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3658 copy construction and assignment in the C++ wrapper.
3659
3660 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3661 couple of bugs that could have given wrong results in the "--newline=crlf"
3662 case.
3663
3664 44. Added a number of casts and did some reorganization of signed/unsigned int
3665 variables following suggestions from Dair Grant. Also renamed the variable
3666 "this" as "item" because it is a C++ keyword.
3667
3668 45. Arranged for dftables to add
3669
3670 #include "pcre_internal.h"
3671
3672 to pcre_chartables.c because without it, gcc 4.x may remove the array
3673 definition from the final binary if PCRE is built into a static library and
3674 dead code stripping is activated.
3675
3676 46. For an unanchored pattern, if a match attempt fails at the start of a
3677 newline sequence, and the newline setting is CRLF or ANY, and the next two
3678 characters are CRLF, advance by two characters instead of one.
3679
3680
3681 Version 6.7 04-Jul-06
3682 ---------------------
3683
3684 1. In order to handle tests when input lines are enormously long, pcretest has
3685 been re-factored so that it automatically extends its buffers when
3686 necessary. The code is crude, but this _is_ just a test program. The
3687 default size has been increased from 32K to 50K.
3688
3689 2. The code in pcre_study() was using the value of the re argument before
3690 testing it for NULL. (Of course, in any sensible call of the function, it
3691 won't be NULL.)
3692
3693 3. The memmove() emulation function in pcre_internal.h, which is used on
3694 systems that lack both memmove() and bcopy() - that is, hardly ever -
3695 was missing a "static" storage class specifier.
3696
3697 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3698 containing an extended class (one that cannot be represented by a bitmap
3699 because it contains high-valued characters or Unicode property items, e.g.
3700 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3701 pattern, but PCRE should not loop if you do not (it no longer does).
3702 [Detail: two cases were found: (a) a repeated subpattern containing an
3703 extended class; (b) a recursive reference to a subpattern that followed a
3704 previous extended class. It wasn't skipping over the extended class
3705 correctly when UTF-8 mode was not set.]
3706
3707 5. A negated single-character class was not being recognized as fixed-length
3708 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3709 compile error "lookbehind assertion is not fixed length".
3710
3711 6. The RunPerlTest auxiliary script was showing an unexpected difference
3712 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3713 write a Perl script that can interpret lines of an input file either as
3714 byte characters or as UTF-8, which is what "perltest" was being required to
3715 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3716 can't do is switch easily at run time between having the "use utf8;" pragma
3717 or not. In the end, I fudged it by using the RunPerlTest script to insert
3718 "use utf8;" explicitly for the UTF-8 tests.
3719
3720 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3721 the end of the subject string, contrary to the documentation and to what
3722 Perl does. This was true of both matching functions. Now it matches only at
3723 the start of the subject and immediately after *internal* newlines.
3724
3725 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3726 a pointer to an int instead of a pointer to an unsigned long int. This
3727 caused problems on 64-bit systems.
3728
3729 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3730 instance of the 'standard' template library not being so standard".
3731
3732 10. There was no check on the number of named subpatterns nor the maximum
3733 length of a subpattern name. The product of these values is used to compute
3734 the size of the memory block for a compiled pattern. By supplying a very
3735 long subpattern name and a large number of named subpatterns, the size
3736 computation could be caused to overflow. This is now prevented by limiting
3737 the length of names to 32 characters, and the number of named subpatterns
3738 to 10,000.
3739
3740 11. Subpatterns that are repeated with specific counts have to be replicated in
3741 the compiled pattern. The size of memory for this was computed from the
3742 length of the subpattern and the repeat count. The latter is limited to
3743 65535, but there was no limit on the former, meaning that integer overflow
3744 could in principle occur. The compiled length of a repeated subpattern is
3745 now limited to 30,000 bytes in order to prevent this.
3746
3747 12. Added the optional facility to have named substrings with the same name.
3748
3749 13. Added the ability to use a named substring as a condition, using the
3750 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3751 are numbers (not recommended). Forward references are permitted.
3752
3753 14. Added forward references in named backreferences (if you see what I mean).
3754
3755 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3756 pattern could run off the end of the subject. For example, the pattern
3757 "(?s)(.{1,5})"8 did this with the subject "ab".
3758
3759 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3760 PCRE_CASELESS was set when matching characters that were quantified with ?
3761 or *.
3762
3763 17. A character class other than a single negated character that had a minimum
3764 but no maximum quantifier - for example [ab]{6,} - was not handled
3765 correctly by pce_dfa_exec(). It would match only one character.
3766
3767 18. A valid (though odd) pattern that looked like a POSIX character
3768 class but used an invalid character after [ (for example [[,abc,]]) caused
3769 pcre_compile() to give the error "Failed: internal error: code overflow" or
3770 in some cases to crash with a glibc free() error. This could even happen if
3771 the pattern terminated after [[ but there just happened to be a sequence of
3772 letters, a binary zero, and a closing ] in the memory that followed.
3773
3774 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3775 over the years. Originally (before any Unicode support), just the bottom 8
3776 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3777 output from "man perlunicode" includes this:
3778
3779 The regular expression compiler produces polymorphic opcodes. That
3780 is, the pattern adapts to the data and automatically switches to
3781 the Unicode character scheme when presented with Unicode data--or
3782 instead uses a traditional byte scheme when presented with byte
3783 data.
3784
3785 Sadly, a wide octal escape does not cause a switch, and in a string with
3786 no other multibyte characters, these octal escapes are treated as before.
3787 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3788 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3789 Unicode string.
3790
3791 I have not perpetrated such confusion in PCRE. Up till now, it took just
3792 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3793 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3794 translate to the appropriate multibyte character.
3795
3796 29. Applied some refactoring to reduce the number of warnings from Microsoft
3797 and Borland compilers. This has included removing the fudge introduced
3798 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3799 a warning about an unused variable.
3800
3801 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3802 characters since release 4.0, because Perl (from release 5.004) does not.
3803 [Or at least, is documented not to: some releases seem to be in conflict
3804 with the documentation.] However, when a pattern was studied with
3805 pcre_study() and all its branches started with \s, PCRE still included VT
3806 as a possible starting character. Of course, this did no harm; it just
3807 caused an unnecessary match attempt.
3808
3809 22. Removed a now-redundant internal flag bit that recorded the fact that case
3810 dependency changed within the pattern. This was once needed for "required
3811 byte" processing, but is no longer used. This recovers a now-scarce options
3812 bit. Also moved the least significant internal flag bit to the most-
3813 significant bit of the word, which was not previously used (hangover from
3814 the days when it was an int rather than a uint) to free up another bit for
3815 the future.
3816
3817 23. Added support for CRLF line endings as well as CR and LF. As well as the
3818 default being selectable at build time, it can now be changed at runtime
3819 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3820 specify that it is scanning data with non-default line endings.
3821
3822 24. Changed the definition of CXXLINK to make it agree with the definition of
3823 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3824
3825 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3826 recursions. This makes a big different to stack usage for some patterns.
3827
3828 26. If a subpattern containing a named recursion or subroutine reference such
3829 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3830 the space required for the compiled pattern went wrong and gave too small a
3831 value. Depending on the environment, this could lead to "Failed: internal
3832 error: code overflow at offset 49" or "glibc detected double free or
3833 corruption" errors.
3834
3835 27. Applied patches from Google (a) to support the new newline modes and (b) to
3836 advance over multibyte UTF-8 characters in GlobalReplace.
3837
3838 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3839 difference for some implementation of PCRE in some Windows version.
3840
3841 29. Added some extra testing facilities to pcretest:
3842
3843 \q<number> in a data line sets the "match limit" value
3844 \Q<number> in a data line sets the "match recursion limt" value
3845 -S <number> sets the stack size, where <number> is in megabytes
3846
3847 The -S option isn't available for Windows.
3848
3849
3850 Version 6.6 06-Feb-06
3851 ---------------------
3852
3853 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3854 in pcreposix.h. I have copied the definition from pcre.h.
3855
3856 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3857 because pcre.h is no longer a built file.
3858
3859 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3860 not normally included in the compiled code.
3861
3862
3863 Version 6.5 01-Feb-06
3864 ---------------------
3865
3866 1. When using the partial match feature with pcre_dfa_exec(), it was not
3867 anchoring the second and subsequent partial matches at the new starting
3868 point. This could lead to incorrect results. For example, with the pattern
3869 /1234/, partially matching against "123" and then "a4" gave a match.
3870
3871 2. Changes to pcregrep:
3872
3873 (a) All non-match returns from pcre_exec() were being treated as failures
3874 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3875 error message is output. Some extra information is given for the
3876 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3877 probably the only errors that are likely to be caused by users (by
3878 specifying a regex that has nested indefinite repeats, for instance).
3879 If there are more than 20 of these errors, pcregrep is abandoned.
3880
3881 (b) A binary zero was treated as data while matching, but terminated the
3882 output line if it was written out. This has been fixed: binary zeroes
3883 are now no different to any other data bytes.
3884
3885 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3886 used to set a locale for matching. The --locale=xxxx long option has
3887 been added (no short equivalent) to specify a locale explicitly on the
3888 pcregrep command, overriding the environment variables.
3889
3890 (d) When -B was used with -n, some line numbers in the output were one less
3891 than they should have been.
3892
3893 (e) Added the -o (--only-matching) option.
3894
3895 (f) If -A or -C was used with -c (count only), some lines of context were
3896 accidentally printed for the final match.
3897
3898 (g) Added the -H (--with-filename) option.
3899
3900 (h) The combination of options -rh failed to suppress file names for files
3901 that were found from directory arguments.
3902
3903 (i) Added the -D (--devices) and -d (--directories) options.
3904
3905 (j) Added the -F (--fixed-strings) option.
3906
3907 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3908
3909 (l) Added the --colo(u)r option.
3910
3911 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3912 is not present by default.
3913
3914 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3915 items such as (?R) or (?1), when the recursion could match a number of
3916 alternatives. If it matched one of the alternatives, but subsequently,
3917 outside the recursion, there was a failure, the code tried to back up into
3918 the recursion. However, because of the way PCRE is implemented, this is not
3919 possible, and the result was an incorrect result from the match.
3920
3921 In order to prevent this happening, the specification of recursion has
3922 been changed so that all such subpatterns are automatically treated as
3923 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3924
3925 4. I had overlooked the fact that, in some locales, there are characters for
3926 which isalpha() is true but neither isupper() nor islower() are true. In
3927 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3928 and ordfeminine) are like this. This affected the treatment of \w and \W
3929 when they appeared in character classes, but not when they appeared outside
3930 a character class. The bit map for "word" characters is now created
3931 separately from the results of isalnum() instead of just taking it from the
3932 upper, lower, and digit maps. (Plus the underscore character, of course.)
3933
3934 5. The above bug also affected the handling of POSIX character classes such as
3935 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3936 permanent tables. Instead, the bit maps for such a class were previously
3937 created as the appropriate unions of the upper, lower, and digit bitmaps.
3938 Now they are created by subtraction from the [[:word:]] class, which has
3939 its own bitmap.
3940
3941 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3942 It is created by subtracting the vertical space characters (\x09, \x0a,
3943 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3944 subtraction was done in the overall bitmap for a character class, meaning
3945 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3946 be recognized. This bug has been fixed.
3947
3948 7. Patches from the folks at Google:
3949
3950 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3951 real life, but is still worth protecting against".
3952
3953 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3954 regular expressions".
3955
3956 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3957 have it.
3958
3959 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3960 "configure" and the latter not, in order to fix a problem somebody had
3961 with compiling the Arg class on HP-UX.
3962
3963 (e) Improve the error-handling of the C++ wrapper a little bit.
3964
3965 (f) New tests for checking recursion limiting.
3966
3967 8. The pcre_memmove() function, which is used only if the environment does not
3968 have a standard memmove() function (and is therefore rarely compiled),
3969 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3970 returning a result (though PCRE never actually uses the result).
3971
3972 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3973 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3974 returned instead of calling malloc() with an overflowing number that would
3975 most likely cause subsequent chaos.
3976
3977 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3978
3979 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3980 with this option is matched, the nmatch and pmatch options of regexec() are
3981 ignored.
3982
3983 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3984 provided in case anyone wants to the the POSIX interface with UTF-8
3985 strings.
3986
3987 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3988 C++ linking (needed for some HP-UX environments).
3989
3990 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3991 (unused parameter) and in the pcre_printint() function (omitted "default"
3992 switch label when the default is to do nothing).
3993
3994 15. Added some code to make it possible, when PCRE is compiled as a C++
3995 library, to replace subject pointers for pcre_exec() with a smart pointer
3996 class, thus making it possible to process discontinuous strings.
3997
3998 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3999 much the same function. They were added by different people who were trying
4000 to make PCRE easy to compile on non-Unix systems. It has been suggested
4001 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4002 for compiling on Windows systems. I have therefore replaced it with
4003 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4004 defaults to "extern" for C or "extern C" for C++, which works fine on
4005 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4006 SCOPE with something explicit in config.h. In addition:
4007
4008 (a) pcreposix.h still had just "extern" instead of either of these macros;
4009 I have replaced it with PCRE_DATA_SCOPE.
4010
4011 (b) Functions such as _pcre_xclass(), which are internal to the library,
4012 but external in the C sense, all had PCRE_EXPORT in their definitions.
4013 This is apparently wrong for the Windows case, so I have removed it.
4014 (It makes no difference on Unix-like systems.)
4015
4016 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4017 of recursive calls to match(). This is different to MATCH_LIMIT because
4018 that limits the total number of calls to match(), not all of which increase
4019 the depth of recursion. Limiting the recursion depth limits the amount of
4020 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4021 when PCRE is compiled, and changed at run time. A patch from Google adds
4022 this functionality to the C++ interface.
4023
4024 18. Changes to the handling of Unicode character properties:
4025
4026 (a) Updated the table to Unicode 4.1.0.
4027
4028 (b) Recognize characters that are not in the table as "Cn" (undefined).
4029
4030 (c) I revised the way the table is implemented to a much improved format
4031 which includes recognition of ranges. It now supports the ranges that
4032 are defined in UnicodeData.txt, and it also amalgamates other
4033 characters into ranges. This has reduced the number of entries in the
4034 table from around 16,000 to around 3,000, thus reducing its size
4035 considerably. I realized I did not need to use a tree structure after
4036 all - a binary chop search is just as efficient. Having reduced the
4037 number of entries, I extended their size from 6 bytes to 8 bytes to
4038 allow for more data.
4039
4040 (d) Added support for Unicode script names via properties such as \p{Han}.
4041
4042 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4043 matching that character.
4044
4045 20. When matching a repeated Unicode property with a minimum greater than zero,
4046 (for example \pL{2,}), PCRE could look past the end of the subject if it
4047 reached it while seeking the minimum number of characters. This could
4048 happen only if some of the characters were more than one byte long, because
4049 there is a check for at least the minimum number of bytes.
4050
4051 21. Refactored the implementation of \p and \P so as to be more general, to
4052 allow for more different types of property in future. This has changed the
4053 compiled form incompatibly. Anybody with saved compiled patterns that use
4054 \p or \P will have to recompile them.
4055
4056 22. Added "Any" and "L&" to the supported property types.
4057
4058 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4059 but give a compile time error if the value is greater than 0xff.
4060
4061 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4062 accidentally not being installed or uninstalled.
4063
4064 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4065 made were to insert the current release number. This seemed silly, because
4066 it made things harder for people building PCRE on systems that don't run
4067 "configure". I have turned pcre.h into a distributed file, no longer built
4068 by "configure", with the version identification directly included. There is
4069 no longer a pcre.h.in file.
4070
4071 However, this change necessitated a change to the pcre-config script as
4072 well. It is built from pcre-config.in, and one of the substitutions was the
4073 release number. I have updated configure.ac so that ./configure now finds
4074 the release number by grepping pcre.h.
4075
4076 26. Added the ability to run the tests under valgrind.
4077
4078
4079 Version 6.4 05-Sep-05
4080 ---------------------
4081
4082 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4083 "--" to be printed when multiple files were scanned, even when none of the
4084 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4085 consider it to be a bug, and have restored the previous behaviour.
4086
4087 2. A couple of code tidies to get rid of compiler warnings.
4088
4089 3. The pcretest program used to cheat by referring to symbols in the library
4090 whose names begin with _pcre_. These are internal symbols that are not
4091 really supposed to be visible externally, and in some environments it is
4092 possible to suppress them. The cheating is now confined to including
4093 certain files from the library's source, which is a bit cleaner.
4094
4095 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4096 file's purpose clearer.
4097
4098 5. Reorganized pcre_ucp_findchar().
4099
4100
4101 Version 6.3 15-Aug-05
4102 ---------------------
4103
4104 1. The file libpcre.pc.in did not have general read permission in the tarball.
4105
4106 2. There were some problems when building without C++ support:
4107
4108 (a) If C++ support was not built, "make install" and "make test" still
4109 tried to test it.
4110
4111 (b) There were problems when the value of CXX was explicitly set. Some
4112 changes have been made to try to fix these, and ...
4113
4114 (c) --disable-cpp can now be used to explicitly disable C++ support.
4115
4116 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4117 backslash in a target when C++ was disabled. This confuses some
4118 versions of "make", apparently. Using an intermediate variable solves
4119 this. (Same for CPP_LOBJ.)
4120
4121 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4122 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4123 necessary on certain architectures.
4124
4125 4. Added a setting of -export-symbols-regex to the link command to remove
4126 those symbols that are exported in the C sense, but actually are local
4127 within the library, and not documented. Their names all begin with
4128 "_pcre_". This is not a perfect job, because (a) we have to except some
4129 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4130 available (and never for static libraries). I have made a note to try to
4131 find a way round (a) in the future.
4132
4133
4134 Version 6.2 01-Aug-05
4135 ---------------------
4136
4137 1. There was no test for integer overflow of quantifier values. A construction
4138 such as {1111111111111111} would give undefined results. What is worse, if
4139 a minimum quantifier for a parenthesized subpattern overflowed and became
4140 negative, the calculation of the memory size went wrong. This could have
4141 led to memory overwriting.
4142
4143 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4144
4145 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4146 operating environments where this matters.
4147
4148 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4149 PCRE options from within the C++ wrapper.
4150
4151 5. Named capturing subpatterns were not being correctly counted when a pattern
4152 was compiled. This caused two problems: (a) If there were more than 100
4153 such subpatterns, the calculation of the memory needed for the whole
4154 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4155 back references of the form \12, where the number was greater than 9, were
4156 not recognized as back references, even though there were sufficient
4157 previous subpatterns.
4158
4159 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4160 versions of gcc, e.g. 2.95.4.
4161
4162
4163 Version 6.1 21-Jun-05
4164 ---------------------
4165
4166 1. There was one reference to the variable "posix" in pcretest.c that was not
4167 surrounded by "#if !defined NOPOSIX".
4168
4169 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4170 the cross-check on the old pcre_info() function, for the benefit of the
4171 cut-down version of PCRE that is currently imported into Exim.
4172
4173 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4174 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4175 patterns that start (?i) but I don't think that matters. The use of (?i) is
4176 just an example; this all applies to the other options as well.
4177
4178 4. Since libtool seems to echo the compile commands it is issuing, the output
4179 from "make" can be reduced a bit by putting "@" in front of each libtool
4180 compile command.
4181
4182 5. Patch from the folks at Google for configure.in to be a bit more thorough
4183 in checking for a suitable C++ installation before trying to compile the
4184 C++ stuff. This should fix a reported problem when a compiler was present,
4185 but no suitable headers.
4186
4187 6. The man pages all had just "PCRE" as their title. I have changed them to
4188 be the relevant file name. I have also arranged that these names are
4189 retained in the file doc/pcre.txt, which is a concatenation in text format
4190 of all the man pages except the little individual ones for each function.
4191
4192 7. The NON-UNIX-USE file had not been updated for the different set of source
4193 files that come with release 6. I also added a few comments about the C++
4194 wrapper.
4195
4196
4197 Version 6.0 07-Jun-05
4198 ---------------------
4199
4200 1. Some minor internal re-organization to help with my DFA experiments.
4201
4202 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4203 didn't matter for the library itself when fully configured, but did matter
4204 when compiling without UCP support, or within Exim, where the ucp files are
4205 not imported.
4206
4207 3. Refactoring of the library code to split up the various functions into
4208 different source modules. The addition of the new DFA matching code (see
4209 below) to a single monolithic source would have made it really too
4210 unwieldy, quite apart from causing all the code to be include in a
4211 statically linked application, when only some functions are used. This is
4212 relevant even without the DFA addition now that patterns can be compiled in
4213 one application and matched in another.
4214
4215 The downside of splitting up is that there have to be some external
4216 functions and data tables that are used internally in different modules of
4217 the library but which are not part of the API. These have all had their
4218 names changed to start with "_pcre_" so that they are unlikely to clash
4219 with other external names.
4220
4221 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4222 a different (DFA) algorithm. Although it is slower than the original
4223 function, it does have some advantages for certain types of matching
4224 problem.
4225
4226 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4227 including restarting after a partial match.
4228
4229 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4230 defined when compiling for Windows was sent to me. I have put it into the
4231 code, though I have no means of testing or verifying it.
4232
4233 7. Added the pcre_refcount() auxiliary function.
4234
4235 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4236 match before or at the first newline in the subject string. In pcretest,
4237 the /f option on a pattern can be used to set this.
4238
4239 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4240 would behave wrongly. This has been present in PCRE since release 4.0.
4241
4242 10. A number of changes to the pcregrep command:
4243
4244 (a) Refactored how -x works; insert ^(...)$ instead of setting
4245 PCRE_ANCHORED and checking the length, in preparation for adding
4246 something similar for -w.
4247
4248 (b) Added the -w (match as a word) option.
4249
4250 (c) Refactored the way lines are read and buffered so as to have more
4251 than one at a time available.
4252
4253 (d) Implemented a pcregrep test script.
4254
4255 (e) Added the -M (multiline match) option. This allows patterns to match
4256 over several lines of the subject. The buffering ensures that at least
4257 8K, or the rest of the document (whichever is the shorter) is available
4258 for matching (and similarly the previous 8K for lookbehind assertions).
4259
4260 (f) Changed the --help output so that it now says
4261
4262 -w, --word-regex(p)
4263
4264 instead of two lines, one with "regex" and the other with "regexp"
4265 because that confused at least one person since the short forms are the
4266 same. (This required a bit of code, as the output is generated
4267 automatically from a table. It wasn't just a text change.)
4268
4269 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4270 option but starts with a hyphen. Could be a pattern or a path name
4271 starting with a hyphen, for instance.
4272
4273 (h) "-" can be given as a file name to represent stdin.
4274
4275 (i) When file names are being printed, "(standard input)" is used for
4276 the standard input, for compatibility with GNU grep. Previously
4277 "<stdin>" was used.
4278
4279 (j) The option --label=xxx can be used to supply a name to be used for
4280 stdin when file names are being printed. There is no short form.
4281
4282 (k) Re-factored the options decoding logic because we are going to add
4283 two more options that take data. Such options can now be given in four
4284 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4285
4286 (l) Added the -A, -B, and -C options for requesting that lines of context
4287 around matches be printed.
4288
4289 (m) Added the -L option to print the names of files that do not contain
4290 any matching lines, that is, the complement of -l.
4291
4292 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4293 continue to scan other files.
4294
4295 (o) The -s option was incorrectly implemented. For compatibility with other
4296 greps, it now suppresses the error message for a non-existent or non-
4297 accessible file (but not the return code). There is a new option called
4298 -q that suppresses the output of matching lines, which was what -s was
4299 previously doing.
4300
4301 (p) Added --include and --exclude options to specify files for inclusion
4302 and exclusion when recursing.
4303
4304 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4305 Hopefully, it now does.
4306
4307 12. Missing cast in pcre_study().
4308
4309 13. Added an "uninstall" target to the makefile.
4310
4311 14. Replaced "extern" in the function prototypes in Makefile.in with
4312 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4313 world, but is set differently for Windows.
4314
4315 15. Added a second compiling function called pcre_compile2(). The only
4316 difference is that it has an extra argument, which is a pointer to an
4317 integer error code. When there is a compile-time failure, this is set
4318 non-zero, in addition to the error test pointer being set to point to an
4319 error message. The new argument may be NULL if no error number is required
4320 (but then you may as well call pcre_compile(), which is now just a
4321 wrapper). This facility is provided because some applications need a
4322 numeric error indication, but it has also enabled me to tidy up the way
4323 compile-time errors are handled in the POSIX wrapper.
4324
4325 16. Added VPATH=.libs to the makefile; this should help when building with one
4326 prefix path and installing with another. (Or so I'm told by someone who
4327 knows more about this stuff than I do.)
4328
4329 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4330 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4331 match everything, including newlines. This is not POSIX-compatible, but
4332 somebody wanted the feature. From pcretest it can be activated by using
4333 both the P and the s flags.
4334
4335 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4336
4337 19. libpcre.pc was being incorrectly installed as executable.
4338
4339 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4340 it now also looks for '\r' so that it will work unmodified on Windows.
4341
4342 21. Added Google's contributed C++ wrapper to the distribution.
4343
4344 22. Added some untidy missing memory free() calls in pcretest, to keep
4345 Electric Fence happy when testing.
4346
4347
4348
4349 Version 5.0 13-Sep-04
4350 ---------------------
4351
4352 1. Internal change: literal characters are no longer packed up into items
4353 containing multiple characters in a single byte-string. Each character
4354 is now matched using a separate opcode. However, there may be more than one
4355 byte in the character in UTF-8 mode.
4356
4357 2. The pcre_callout_block structure has two new fields: pattern_position and
4358 next_item_length. These contain the offset in the pattern to the next match
4359 item, and its length, respectively.
4360
4361 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4362 insertion of callouts before each pattern item. Added the /C option to
4363 pcretest to make use of this.
4364
4365 4. On the advice of a Windows user, the lines
4366
4367 #if defined(_WIN32) || defined(WIN32)
4368 _setmode( _fileno( stdout ), 0x8000 );
4369 #endif /* defined(_WIN32) || defined(WIN32) */
4370
4371 have been added to the source of pcretest. This apparently does useful
4372 magic in relation to line terminators.
4373
4374 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4375 for the benefit of those environments where the "b" makes a difference.
4376
4377 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4378 to know about it. I have put a hack into configure.in that adds in code
4379 to set GCC=yes if CC=icc. This seems to end up at a point in the
4380 generated configure script that is early enough to affect the setting of
4381 compiler options, which is what is needed, but I have no means of testing
4382 whether it really works. (The user who reported this had patched the
4383 generated configure script, which of course I cannot do.)
4384
4385 LATER: After change 22 below (new libtool files), the configure script
4386 seems to know about icc (and also ecc). Therefore, I have commented out
4387 this hack in configure.in.
4388
4389 7. Added support for pkg-config (2 patches were sent in).
4390
4391 8. Negated POSIX character classes that used a combination of internal tables
4392 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4393 [[:^ascii]]. Typically, they would match almost any characters. The other
4394 POSIX classes were not broken in this way.
4395
4396 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4397 to find the match, as PCRE was deluded into thinking that the match had to
4398 start at the start point or following a newline. The same bug applied to
4399 patterns with negative forward assertions or any backward assertions
4400 preceding ".*" at the start, unless the pattern required a fixed first
4401 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4402
4403 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4404 starting at the last subject character, bytes beyond the end of the subject
4405 string were read.
4406
4407 11. Renamed the variable "class" as "classbits" to make life easier for C++
4408 users. (Previously there was a macro definition, but it apparently wasn't
4409 enough.)
4410
4411 12. Added the new field "tables" to the extra data so that tables can be passed
4412 in at exec time, or the internal tables can be re-selected. This allows
4413 a compiled regex to be saved and re-used at a later time by a different
4414 program that might have everything at different addresses.
4415
4416 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4417 -R library as well as a -L library.
4418
4419 14. The debugging options of pcretest (-d on the command line or D on a
4420 pattern) showed incorrect output for anything following an extended class
4421 that contained multibyte characters and which was followed by a quantifier.
4422
4423 15. Added optional support for general category Unicode character properties
4424 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4425 support. It adds about 90K to the size of the library. The meanings of the
4426 inbuilt class escapes such as \d and \s have NOT been changed.
4427
4428 16. Updated pcredemo.c to include calls to free() to release the memory for the
4429 compiled pattern.
4430
4431 17. The generated file chartables.c was being created in the source directory
4432 instead of in the building directory. This caused the build to fail if the
4433 source directory was different from the building directory, and was
4434 read-only.
4435
4436 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4437 file. No doubt somebody will tell me if they don't make sense... Also added
4438 Dan Mooney's comments about building on OpenVMS.
4439
4440 19. Added support for partial matching via the PCRE_PARTIAL option for
4441 pcre_exec() and the \P data escape in pcretest.
4442
4443 20. Extended pcretest with 3 new pattern features:
4444
4445 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4446 write the compiled pattern to the file whose name is "rest-of-line".
4447 This is a straight binary dump of the data, with the saved pointer to
4448 the character tables forced to be NULL. The study data, if any, is
4449 written too. After writing, pcretest reads a new pattern.
4450
4451 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4452 compiled pattern from the given file. There must not be any
4453 occurrences of "<" in the file name (pretty unlikely); if there are,
4454 pcretest will instead treat the initial "<" as a pattern delimiter.
4455 After reading in the pattern, pcretest goes on to read data lines as
4456 usual.
4457
4458 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4459 and 16-bit fields in a compiled pattern, to simulate a pattern that
4460 was compiled on a host of opposite endianness.
4461
4462 21. The pcre-exec() function can now cope with patterns that were compiled on
4463 hosts of opposite endianness, with this restriction:
4464
4465 As for any compiled expression that is saved and used later, the tables
4466 pointer field cannot be preserved; the extra_data field in the arguments
4467 to pcre_exec() should be used to pass in a tables address if a value
4468 other than the default internal tables were used at compile time.
4469
4470 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4471 now diagnosed as an error. Previously, most of the time, a negative number
4472 would have been treated as zero, but if in addition "ovector" was passed as
4473 NULL, a crash could occur.
4474
4475 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4476 new versions from the libtool 1.5 distribution (the last one is a copy of
4477 a file called libtool.m4). This seems to have fixed the need to patch
4478 "configure" to support Darwin 1.3 (which I used to do). However, I still
4479 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4480 workstation).
4481
4482 24. Changed the PCRE licence to be the more standard "BSD" licence.
4483
4484
4485 Version 4.5 01-Dec-03
4486 ---------------------
4487
4488 1. There has been some re-arrangement of the code for the match() function so
4489 that it can be compiled in a version that does not call itself recursively.
4490 Instead, it keeps those local variables that need separate instances for
4491 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4492 needs to "recurse". Keeping track of where control must go is done by means
4493 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4494 hide most of the details from the main code, and operates only if
4495 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4496 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4497 operating.
4498
4499 To make it easier for callers to provide specially tailored get/free
4500 functions for this usage, two new functions, pcre_stack_malloc, and
4501 pcre_stack_free, are used. They are always called in strict stacking order,
4502 and the size of block requested is always the same.
4503
4504 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4505 PCRE has been compiled to use the stack or the heap for recursion. The
4506 -C option of pcretest uses this to show which version is compiled.
4507
4508 A new data escape \S, is added to pcretest; it causes the amounts of store
4509 obtained and freed by both kinds of malloc/free at match time to be added
4510 to the output.
4511
4512 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4513 what's available on my current Linux desktop machine.
4514
4515 3. When matching a UTF-8 string, the test for a valid string at the start has
4516 been extended. If start_offset is not zero, PCRE now checks that it points
4517 to a byte that is the start of a UTF-8 character. If not, it returns
4518 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4519 this is necessary because there may be backward assertions in the pattern.
4520 When matching the same subject several times, it may save resources to use
4521 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4522
4523 4. The code for checking the validity of UTF-8 strings has been tightened so
4524 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4525 containing "overlong sequences".
4526
4527 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4528 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4529 should have been "&", but it just so happened that all the cases this let
4530 through by mistake were picked up later in the function.
4531
4532 6. I had used a variable called "isblank" - this is a C99 function, causing
4533 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4534
4535 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4536 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4537 so the version is shown; (c) stop "make test" echoing "./RunTest".
4538
4539 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4540
4541 9. The prototype for memmove() for systems that don't have it was using
4542 size_t, but the inclusion of the header that defines size_t was later. I've
4543 moved the #includes for the C headers earlier to avoid this.
4544
4545 10. Added some adjustments to the code to make it easier to compiler on certain
4546 special systems:
4547
4548 (a) Some "const" qualifiers were missing.
4549 (b) Added the macro EXPORT before all exported functions; by default this
4550 is defined to be empty.
4551 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4552 that it reads its output file name as an argument instead of writing
4553 to the standard output and assuming this can be redirected.
4554
4555 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4556 class containing characters with values greater than 255, PCRE compilation
4557 went into a loop.
4558
4559 12. A recursive reference to a subpattern that was within another subpattern
4560 that had a minimum quantifier of zero caused PCRE to crash. For example,
4561 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4562 recursion. If the recursively-called subpattern itself had a zero repeat,
4563 that was OK.
4564
4565 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4566 buffer into which it was copied (for escape processing) was still set at
4567 1024, so long lines caused crashes.
4568
4569 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4570 "internal error: code overflow...". This applied to any character class
4571 that was followed by a possessive quantifier.
4572
4573 15. Modified the Makefile to add libpcre.la as a prerequisite for
4574 libpcreposix.la because I was told this is needed for a parallel build to
4575 work.
4576
4577 16. If a pattern that contained .* following optional items at the start was
4578 studied, the wrong optimizing data was generated, leading to matching
4579 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4580 matching string must start with a or b or c. The correct conclusion for
4581 this pattern is that a match can start with any character.
4582
4583
4584 Version 4.4 13-Aug-03
4585 ---------------------
4586
4587 1. In UTF-8 mode, a character class containing characters with values between
4588 127 and 255 was not handled correctly if the compiled pattern was studied.
4589 In fixing this, I have also improved the studying algorithm for such
4590 classes (slightly).
4591
4592 2. Three internal functions had redundant arguments passed to them. Removal
4593 might give a very teeny performance improvement.
4594
4595 3. Documentation bug: the value of the capture_top field in a callout is *one
4596 more than* the number of the hightest numbered captured substring.
4597
4598 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4599 in incorrectly linking with a previously installed version. They now link
4600 explicitly with libpcre.la.
4601
4602 5. configure.in no longer needs to recognize Cygwin specially.
4603
4604 6. A problem in pcre.in for Windows platforms is fixed.
4605
4606 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4607 pcretest, it used to include the size of the study block as part of its
4608 output. Unfortunately, the structure contains a field that has a different
4609 size on different hardware architectures. This meant that the tests that
4610 showed this size failed. As the block is currently always of a fixed size,
4611 this information isn't actually particularly useful in pcretest output, so
4612 I have just removed it.
4613
4614 8. Three pre-processor statements accidentally did not start in column 1.
4615 Sadly, there are *still* compilers around that complain, even though
4616 standard C has not required this for well over a decade. Sigh.
4617
4618 9. In pcretest, the code for checking callouts passed small integers in the
4619 callout_data field, which is a void * field. However, some picky compilers
4620 complained about the casts involved for this on 64-bit systems. Now
4621 pcretest passes the address of the small integer instead, which should get
4622 rid of the warnings.
4623
4624 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4625 both compile and run time, and gives an error if an invalid UTF-8 sequence
4626 is found. There is a option for disabling this check in cases where the
4627 string is known to be correct and/or the maximum performance is wanted.
4628
4629 11. In response to a bug report, I changed one line in Makefile.in from
4630
4631 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4632 to
4633 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4634
4635 to look similar to other lines, but I have no way of telling whether this
4636 is the right thing to do, as I do not use Windows. No doubt I'll get told
4637 if it's wrong...
4638
4639
4640 Version 4.3 21-May-03
4641 ---------------------
4642
4643 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4644 Makefile.
4645
4646 2. Some refactoring to improve the quality of the code:
4647
4648 (i) The utf8_table... variables are now declared "const".
4649
4650 (ii) The code for \cx, which used the "case flipping" table to upper case
4651 lower case letters, now just substracts 32. This is ASCII-specific,
4652 but the whole concept of \cx is ASCII-specific, so it seems
4653 reasonable.
4654
4655 (iii) PCRE was using its character types table to recognize decimal and
4656 hexadecimal digits in the pattern. This is silly, because it handles
4657 only 0-9, a-f, and A-F, but the character types table is locale-
4658 specific, which means strange things might happen. A private
4659 table is now used for this - though it costs 256 bytes, a table is
4660 much faster than multiple explicit tests. Of course, the standard
4661 character types table is still used for matching digits in subject
4662 strings against \d.
4663
4664 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4665 ending in _t are). So I've renamed it as ESC_tee.
4666
4667 3. The first argument for regexec() in the POSIX wrapper should have been
4668 defined as "const".
4669
4670 4. Changed pcretest to use malloc() for its buffers so that they can be
4671 Electric Fenced for debugging.
4672
4673 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4674 to read one or more bytes before the start of the subject string. Often this
4675 had no effect on PCRE's behaviour, but in some circumstances it could
4676 provoke a segmentation fault.
4677
4678 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4679 to try to read one or more bytes before the start of the subject string.
4680
4681 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4682 UTF-8 support could misbehave in various ways if the subject string
4683 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4684 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4685 back over UTF-8 characters.)
4686
4687
4688 Version 4.2 14-Apr-03
4689 ---------------------
4690
4691 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4692
4693 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4694 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4695 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4696 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4697 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4698 and BUILD_EXEEXT
4699 Note: automatic setting of the BUILD variables is not yet working
4700 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4701 compile-time but not at link-time
4702 [LINK]: use for linking executables only
4703 make different versions for Windows and non-Windows
4704 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4705 libraries
4706 [LINK_FOR_BUILD]: new variable
4707 [OBJEXT]: use throughout
4708 [EXEEXT]: use throughout
4709 <winshared>: new target
4710 <wininstall>: new target
4711 <dftables.o>: use native compiler
4712 <dftables>: use native linker
4713 <install>: handle Windows platform correctly
4714 <clean>: ditto
4715 <check>: ditto
4716 copy DLL to top builddir before testing
4717
4718 As part of these changes, -no-undefined was removed again. This was reported
4719 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4720 in any case.
4721
4722 3. Some tidies to get rid of compiler warnings:
4723
4724 . In the match_data structure, match_limit was an unsigned long int, whereas
4725 match_call_count was an int. I've made them both unsigned long ints.
4726
4727 . In pcretest the fact that a const uschar * doesn't automatically cast to
4728 a void * provoked a warning.
4729
4730 . Turning on some more compiler warnings threw up some "shadow" variables
4731 and a few more missing casts.
4732
4733 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4734 option, a class that contained a single character with a value between 128
4735 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4736
4737 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4738 option, a class that contained several characters, but with at least one
4739 whose value was between 128 and 255 caused PCRE to crash.
4740
4741
4742 Version 4.1 12-Mar-03
4743 ---------------------
4744
4745 1. Compiling with gcc -pedantic found a couple of places where casts were
4746 needed, and a string in dftables.c that was longer than standard compilers are
4747 required to support.
4748
4749 2. Compiling with Sun's compiler found a few more places where the code could
4750 be tidied up in order to avoid warnings.
4751
4752 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4753 first of these names is deprecated in the latest Autoconf in favour of the name
4754 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4755 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4756 analogy I have changed it to CFLAGS_FOR_BUILD.
4757
4758 4. Added -no-undefined to the linking command in the Makefile, because this is
4759 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4760 linking step for the pcreposix library.
4761
4762 5. PCRE was failing to diagnose the case of two named groups with the same
4763 name.
4764
4765 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4766 literal character that is needed in the subject for a match, and scans along to
4767 ensure that it is present before embarking on the full matching process. This
4768 saves time in cases of nested unlimited repeats that are never going to match.
4769 Problem: the scan can take a lot of time if the subject is very long (e.g.
4770 megabytes), thus penalizing straightforward matches. It is now done only if the
4771 amount of subject to be scanned is less than 1000 bytes.
4772
4773 7. A lesser problem with the same optimization is that it was recording the
4774 first character of an anchored pattern as "needed", thus provoking a search
4775 right along the subject, even when the first match of the pattern was going to
4776 fail. The "needed" character is now not set for anchored patterns, unless it
4777 follows something in the pattern that is of non-fixed length. Thus, it still
4778 fulfils its original purpose of finding quick non-matches in cases of nested
4779 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4780
4781
4782 Version 4.0 17-Feb-03
4783 ---------------------
4784
4785 1. If a comment in an extended regex that started immediately after a meta-item
4786 extended to the end of string, PCRE compiled incorrect data. This could lead to
4787 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4788
4789 2. Moved to autoconf 2.53 and libtool 1.4.2.
4790
4791 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4792 the special perltest8 script is no longer needed - all the tests can be run
4793 from a single perltest script.
4794
4795 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4796 by \s. It has now been removed in PCRE. This means it isn't recognized as
4797 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4798 class [:space:] *does* include VT, thereby creating a mess.
4799
4800 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4801 space and tab.
4802
4803 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4804 its new features into the main test script, reducing the number of scripts.
4805
4806 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4807 were backward compatible, and made the (?i) apply to the whole pattern, as if
4808 /i were given. Now it behaves more logically, and applies the option setting
4809 only to what follows. PCRE has been changed to follow suit. However, if it
4810 finds options settings right at the start of the pattern, it extracts them into
4811 the global options, as before. Thus, they show up in the info data.
4812
4813 8. Added support for the \Q...\E escape sequence. Characters in between are
4814 treated as literals. This is slightly different from Perl in that $ and @ are
4815 also handled as literals inside the quotes. In Perl, they will cause variable
4816 interpolation. Note the following examples:
4817
4818 Pattern PCRE matches Perl matches
4819
4820 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4821 \Qabc\$xyz\E abc\$xyz abc\$xyz
4822 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4823
4824 For compatibility with Perl, \Q...\E sequences are recognized inside character
4825 classes as well as outside them.
4826
4827 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4828 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4829 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4830 signed/unsigned warnings.
4831
4832 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4833 option for pcretest, so I've replaced it by a simple function that does just
4834 that job.
4835
4836 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4837 "pcregrep -".
4838
4839 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4840 Java package. This provides some syntactic sugar for simple cases of what my
4841 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4842 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4843 item, you can use this simplified notation. Note that only makes sense with
4844 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4845 greediness, whatever the setting of the PCRE_UNGREEDY option.
4846
4847 13. A change of greediness default within a pattern was not taking effect at
4848 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4849 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4850 was abstracted outside.
4851
4852 14. PCRE now supports the \G assertion. It is true when the current matching
4853 position is at the start point of the match. This differs from \A when the
4854 starting offset is non-zero. Used with the /g option of pcretest (or similar
4855 code), it works in the same way as it does for Perl's /g option. If all
4856 alternatives of a regex begin with \G, the expression is anchored to the start
4857 match position, and the "anchored" flag is set in the compiled expression.
4858
4859 15. Some bugs concerning the handling of certain option changes within patterns
4860 have been fixed. These applied to options other than (?ims). For example,
4861 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4862 been the other way round. Some of this was related to change 7 above.
4863
4864 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4865 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4866 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4867 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4868
4869 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4870 mode. Unlike ".", it always matches newline, whatever the setting of
4871 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4872 assertions. Perl allows it, but it doesn't (in general) work because it can't
4873 calculate the length of the lookbehind. At least, that's the case for Perl
4874 5.8.0 - I've been told they are going to document that it doesn't work in
4875 future.
4876
4877 18. Added an error diagnosis for escapes that PCRE does not support: these are
4878 \L, \l, \N, \P, \p, \U, \u, and \X.
4879
4880 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4881 reading past the end of the pattern in cases such as /[abcd/.
4882
4883 20. PCRE was getting more memory than necessary for patterns with classes that
4884 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4885
4886 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4887 compiling PCRE for use with Virtual Pascal.
4888
4889 22. Small fix to the Makefile to make it work properly if the build is done
4890 outside the source tree.
4891
4892 23. Added a new extension: a condition to go with recursion. If a conditional
4893 subpattern starts with (?(R) the "true" branch is used if recursion has
4894 happened, whereas the "false" branch is used only at the top level.
4895
4896 24. When there was a very long string of literal characters (over 255 bytes
4897 without UTF support, over 250 bytes with UTF support), the computation of how
4898 much memory was required could be incorrect, leading to segfaults or other
4899 strange effects.
4900
4901 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4902 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4903 there was a subsequent back reference to those brackets. This meant that, for
4904 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4905 possible to check for precisely this case. All we can do is abandon the
4906 optimization if .* occurs inside capturing brackets when there are any back
4907 references whatsoever. (See below for a better fix that came later.)
4908
4909 26. The handling of the optimization for finding the first character of a
4910 non-anchored pattern, and for finding a character that is required later in the
4911 match were failing in some cases. This didn't break the matching; it just
4912 failed to optimize when it could. The way this is done has been re-implemented.
4913
4914 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4915
4916 28. Added a new feature that provides some of the functionality that Perl
4917 provides with (?{...}). The facility is termed a "callout". The way it is done
4918 in PCRE is for the caller to provide an optional function, by setting
4919 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4920 global variable. By default it is unset, which disables all calling out. To get
4921 the function called, the regex must include (?C) at appropriate points. This
4922 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4923 This provides a means of identifying different callout points. When PCRE
4924 reaches such a point in the regex, if pcre_callout has been set, the external
4925 function is called. It is provided with data in a structure called
4926 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4927 matching continues; if it returns a non-zero value, the match at the current
4928 point fails. However, backtracking will occur if possible. [This was changed
4929 later and other features added - see item 49 below.]
4930
4931 29. pcretest is upgraded to test the callout functionality. It provides a
4932 callout function that displays information. By default, it shows the start of
4933 the match and the current position in the text. There are some new data escapes
4934 to vary what happens:
4935
4936 \C+ in addition, show current contents of captured substrings
4937 \C- do not supply a callout function
4938 \C!n return 1 when callout number n is reached
4939 \C!n!m return 1 when callout number n is reached for the mth time
4940
4941 30. If pcregrep was called with the -l option and just a single file name, it
4942 output "<stdin>" if a match was found, instead of the file name.
4943
4944 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4945 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4946 pcre_exec(). This saves a malloc/free per call. The default value of
4947 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4948 when configuring.
4949
4950 32. The default maximum size of a compiled pattern is 64K. There have been a
4951 few cases of people hitting this limit. The code now uses macros to handle the
4952 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4953 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4954 configuring. Tests 2 and 5 work only with 2-byte links because they output
4955 debugging information about compiled patterns.
4956
4957 33. Internal code re-arrangements:
4958
4959 (a) Moved the debugging function for printing out a compiled regex into
4960 its own source file (printint.c) and used #include to pull it into
4961 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4962 separate copies.
4963
4964 (b) Defined the list of op-code names for debugging as a macro in
4965 internal.h so that it is next to the definition of the opcodes.
4966
4967 (c) Defined a table of op-code lengths for simpler skipping along compiled
4968 code. This is again a macro in internal.h so that it is next to the
4969 definition of the opcodes.
4970
4971 34. Added support for recursive calls to individual subpatterns, along the
4972 lines of Robin Houston's patch (but implemented somewhat differently).
4973
4974 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4975 allow it to read and process whole directories in Win32. This code was
4976 contributed by Lionel Fourquaux; it has not been tested by me.
4977
4978 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4979 used to name a group. Names consist of alphanumerics and underscores, and must
4980 be unique. Back references use the syntax (?P=name) and recursive calls use
4981 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4982 numbers. The function pcre_fullinfo() can be used after compilation to extract
4983 a name/number map. There are three relevant calls:
4984
4985 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4986 PCRE_INFO_NAMECOUNT yields the number of entries
4987 PCRE_INFO_NAMETABLE yields a pointer to the map.
4988
4989 The map is a vector of fixed-size entries. The size of each entry depends on
4990 the length of the longest name used. The first two bytes of each entry are the
4991 group number, most significant byte first. There follows the corresponding
4992 name, zero terminated. The names are in alphabetical order.
4993
4994 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4995 case instead of 255. Making it the same both with and without UTF-8 support
4996 means that the same test output works with both.
4997
4998 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4999 calling malloc() with a zero argument.
5000
5001 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5002 optimization. I've improved things by keeping a bitmap of backreferences with
5003 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5004 fact referenced, the optimization can be applied. It is unlikely that a
5005 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5006 the match to follow \n) will appear inside brackets with a number greater than
5007 31, but if it does, any back reference > 31 suppresses the optimization.
5008
5009 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5010 of disabling numbered capturing parentheses. Any opening parenthesis that is
5011 not followed by ? behaves as if it were followed by ?: but named parentheses
5012 can still be used for capturing (and they will acquire numbers in the usual
5013 way).
5014
5015 41. Redesigned the return codes from the match() function into yes/no/error so
5016 that errors can be passed back from deep inside the nested calls. A malloc
5017 failure while inside a recursive subpattern call now causes the
5018 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5019
5020 42. It is now possible to set a limit on the number of times the match()
5021 function is called in a call to pcre_exec(). This facility makes it possible to
5022 limit the amount of recursion and backtracking, though not in a directly
5023 obvious way, because the match() function is used in a number of different
5024 circumstances. The count starts from zero for each position in the subject
5025 string (for non-anchored patterns). The default limit is, for compatibility, a
5026 large number, namely 10 000 000. You can change this in two ways:
5027
5028 (a) When configuring PCRE before making, you can use --with-match-limit=n
5029 to set a default value for the compiled library.
5030
5031 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5032 a different value is set. See 45 below.
5033
5034 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5035
5036 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5037 of things that can be changed at compile time. The first argument specifies
5038 what is wanted and the second points to where the information is to be placed.
5039 The current list of available information is:
5040
5041 PCRE_CONFIG_UTF8
5042
5043 The output is an integer that is set to one if UTF-8 support is available;
5044 otherwise it is set to zero.
5045
5046 PCRE_CONFIG_NEWLINE
5047
5048 The output is an integer that it set to the value of the code that is used for
5049 newline. It is either LF (10) or CR (13).
5050
5051 PCRE_CONFIG_LINK_SIZE
5052
5053 The output is an integer that contains the number of bytes used for internal
5054 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5055
5056 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5057
5058 The output is an integer that contains the threshold above which the POSIX
5059 interface uses malloc() for output vectors. See item 31 above.
5060
5061 PCRE_CONFIG_MATCH_LIMIT
5062
5063 The output is an unsigned integer that contains the default limit of the number
5064 of match() calls in a pcre_exec() execution. See 42 above.
5065
5066 44. pcretest has been upgraded by the addition of the -C option. This causes it
5067 to extract all the available output from the new pcre_config() function, and to
5068 output it. The program then exits immediately.
5069
5070 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5071 order to support additional features. One way would have been to define
5072 pcre_exec2() (for example) with extra arguments, but this would not have been
5073 extensible, and would also have required all calls to the original function to
5074 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5075 is used for passing in "extra" data from pcre_study().
5076
5077 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5078 contains the following fields:
5079
5080 flags a bitmap indicating which of the following fields are set
5081 study_data opaque data from pcre_study()
5082 match_limit a way of specifying a limit on match() calls for a specific
5083 call to pcre_exec()
5084 callout_data data for callouts (see 49 below)
5085
5086 The flag bits are also defined in pcre.h, and are
5087
5088 PCRE_EXTRA_STUDY_DATA
5089 PCRE_EXTRA_MATCH_LIMIT
5090 PCRE_EXTRA_CALLOUT_DATA
5091
5092 The pcre_study() function now returns one of these new pcre_extra blocks, with
5093 the actual study data pointed to by the study_data field, and the
5094 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5095 before. That is, this change is entirely upwards-compatible and requires no
5096 change to existing code.
5097
5098 If you want to pass in additional data to pcre_exec(), you can either place it
5099 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5100 block.
5101
5102 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5103 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5104 times with different match limits, until it finds the minimum value needed for
5105 pcre_exec() to complete. The value is then output. This can be instructive; for
5106 most simple matches the number is quite small, but for pathological cases it
5107 gets very large very quickly.
5108
5109 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5110 returns the size of the data block pointed to by the study_data field in a
5111 pcre_extra block, that is, the value that was passed as the argument to
5112 pcre_malloc() when PCRE was getting memory in which to place the information
5113 created by pcre_study(). The fourth argument should point to a size_t variable.
5114 pcretest has been extended so that this information is shown after a successful
5115 pcre_study() call when information about the compiled regex is being displayed.
5116
5117 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5118 because what follows is always an absolute path. (Later: it turns out that this
5119 is more than cosmetic for MinGW, because it doesn't like empty path
5120 components.)
5121
5122 49. Some changes have been made to the callout feature (see 28 above):
5123
5124 (i) A callout function now has three choices for what it returns:
5125
5126 0 => success, carry on matching
5127 > 0 => failure at this point, but backtrack if possible
5128 < 0 => serious error, return this value from pcre_exec()
5129
5130 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5131 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5132 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5133 use by callout functions. It will never be used by PCRE itself.
5134
5135 (ii) The pcre_extra structure (see 45 above) has a void * field called
5136 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5137 pcre_callout_block structure has a field of the same name. The contents of
5138 the field passed in the pcre_extra structure are passed to the callout
5139 function in the corresponding field in the callout block. This makes it
5140 easier to use the same callout-containing regex from multiple threads. For
5141 testing, the pcretest program has a new data escape
5142
5143 \C*n pass the number n (may be negative) as callout_data
5144
5145 If the callout function in pcretest receives a non-zero value as
5146 callout_data, it returns that value.
5147
5148 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5149 there were some redundant $(CFLAGS) in commands that are now specified as
5150 $(LINK), which already includes $(CFLAGS).
5151
5152 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5153 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5154 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5155 one-byte characters throughout. Note that case-insensitive matching applies
5156 only to characters whose values are less than 256. PCRE doesn't support the
5157 notion of cases for higher-valued characters.
5158
5159 (i) A character class whose characters are all within 0-255 is handled as
5160 a bit map, and the map is inverted for negative classes. Previously, a
5161 character > 255 always failed to match such a class; however it should
5162 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5163
5164 (ii) A negated character class with a single character < 255 is coded as
5165 "not this character" (OP_NOT). This wasn't working properly when the test
5166 character was multibyte, either singly or repeated.
5167
5168 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5169 mode, for example: \x{100}{2,3}.
5170
5171 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5172 singly or repeated) now correctly test multibyte characters. However,
5173 PCRE doesn't recognize any characters with values greater than 255 as
5174 digits, spaces, or word characters. Such characters always match \D, \S,
5175 and \W, and never match \d, \s, or \w.
5176
5177 (v) Classes may now contain characters and character ranges with values
5178 greater than 255. For example: [ab\x{100}-\x{400}].
5179
5180 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5181 PCRE in UTF-8 mode.
5182
5183 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5184 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5185 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5186 value.)
5187
5188 53. The single man page has become too large. I have therefore split it up into
5189 a number of separate man pages. These also give rise to individual HTML pages;
5190 these are now put in a separate directory, and there is an index.html page that
5191 lists them all. Some hyperlinking between the pages has been installed.
5192
5193 54. Added convenience functions for handling named capturing parentheses.
5194
5195 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5196 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5197 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5198 are faulted.
5199
5200 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5201 calling configure. These values are used when compiling the dftables.c program
5202 which is run to generate the source of the default character tables. They
5203 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5204 you will need to set these values.
5205
5206 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5207
5208
5209 Version 3.9 02-Jan-02
5210 ---------------------
5211
5212 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5213
5214 2. If --disable-static was given, the building process failed when trying to
5215 build pcretest and pcregrep. (For some reason it was using libtool to compile
5216 them, which is not right, as they aren't part of the library.)
5217
5218
5219 Version 3.8 18-Dec-01
5220 ---------------------
5221
5222 1. The experimental UTF-8 code was completely screwed up. It was packing the
5223 bytes in the wrong order. How dumb can you get?
5224
5225
5226 Version 3.7 29-Oct-01
5227 ---------------------
5228
5229 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5230 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5231 this didn't happen under Solaris 8, where I normally test things.
5232
5233 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5234 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5235 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5236 this fix an hour or so after the initial 3.7 release.)
5237
5238
5239 Version 3.6 23-Oct-01
5240 ---------------------
5241
5242 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5243 offsets passed as NULL with zero offset count.
5244
5245 2. The config.guess and config.sub files had not been updated when I moved to
5246 the latest autoconf.
5247
5248
5249 Version 3.5 15-Aug-01
5250 ---------------------
5251
5252 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5253 had been forgotten.
5254
5255 2. By using declared but undefined structures, we can avoid using "void"
5256 definitions in pcre.h while keeping the internal definitions of the structures
5257 private.
5258
5259 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5260 user point of view, this means that both static and shared libraries are built
5261 by default, but this can be individually controlled. More of the work of
5262 handling this static/shared cases is now inside libtool instead of PCRE's make
5263 file.
5264
5265 4. The pcretest utility is now installed along with pcregrep because it is
5266 useful for users (to test regexs) and by doing this, it automatically gets
5267 relinked by libtool. The documentation has been turned into a man page, so
5268 there are now .1, .txt, and .html versions in /doc.
5269
5270 5. Upgrades to pcregrep:
5271 (i) Added long-form option names like gnu grep.
5272 (ii) Added --help to list all options with an explanatory phrase.
5273 (iii) Added -r, --recursive to recurse into sub-directories.
5274 (iv) Added -f, --file to read patterns from a file.
5275
5276 6. pcre_exec() was referring to its "code" argument before testing that
5277 argument for NULL (and giving an error if it was NULL).
5278
5279 7. Upgraded Makefile.in to allow for compiling in a different directory from
5280 the source directory.
5281
5282 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5283 options bits, the pointer it was passed was to an int instead of to an unsigned
5284 long int. This mattered only on 64-bit systems.
5285
5286 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5287 generated) instead of pcre.in, which it its source. Also made the same change
5288 in several of the .c files.
5289
5290 10. A new release of gcc defines printf() as a macro, which broke pcretest
5291 because it had an ifdef in the middle of a string argument for printf(). Fixed
5292 by using separate calls to printf().
5293
5294 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5295 script, to force use of CR or LF instead of \n in the source. On non-Unix
5296 systems, the value can be set in config.h.
5297
5298 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5299 absolute limit. Changed the text of the error message to make this clear, and
5300 likewise updated the man page.
5301
5302 13. The limit of 99 on the number of capturing subpatterns has been removed.
5303 The new limit is 65535, which I hope will not be a "real" limit.
5304
5305
5306 Version 3.4 22-Aug-00
5307 ---------------------
5308
5309 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5310
5311 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5312
5313
5314 Version 3.3 01-Aug-00
5315 ---------------------
5316
5317 1. If an octal character was given, but the value was greater than \377, it
5318 was not getting masked to the least significant bits, as documented. This could
5319 lead to crashes in some systems.
5320
5321 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5322 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5323
5324 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5325 These just pass their arguments on to (pcre_free)(), but they are provided
5326 because some uses of PCRE bind it to non-C systems that can call its functions,
5327 but cannot call free() or pcre_free() directly.
5328
5329 4. Add "make test" as a synonym for "make check". Corrected some comments in
5330 the Makefile.
5331
5332 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5333 Makefile.
5334
5335 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5336 command called pgrep for grepping around the active processes.
5337
5338 7. Added the beginnings of support for UTF-8 character strings.
5339
5340 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5341 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5342 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5343 out for the ar command.)
5344
5345
5346 Version 3.2 12-May-00
5347 ---------------------
5348
5349 This is purely a bug fixing release.
5350
5351 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5352 of ZA. This was just one example of several cases that could provoke this bug,
5353 which was introduced by change 9 of version 2.00. The code for breaking
5354 infinite loops after an iteration that matches an empty string was't working
5355 correctly.
5356
5357 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5358 when matched against abbab (for example). After matching an empty string, it
5359 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5360 caused it to match further down the string than it should.
5361
5362 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5363 was there because it doesn't seem to be needed, and it causes trouble on some
5364 systems, as it is not a Standard C header. It has been removed.
5365
5366 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5367 were reported on the Macintosh. The changes were from
5368
5369 while ((c = *(++ptr)) != 0 && c != '\n');
5370 to
5371 while ((c = *(++ptr)) != 0 && c != '\n') ;
5372
5373 Totally extraordinary, but if that's what it takes...
5374
5375 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5376 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5377 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5378 assumes the way PCRE uses memmove() (always moving upwards).
5379
5380 6. PCRE is being used in one environment where strchr() is not available. There
5381 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5382 faster code anyway.
5383
5384
5385 Version 3.1 09-Feb-00
5386 ---------------------
5387
5388 The only change in this release is the fixing of some bugs in Makefile.in for
5389 the "install" target:
5390
5391 (1) It was failing to install pcreposix.h.
5392
5393 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5394
5395
5396 Version 3.0 01-Feb-00
5397 ---------------------
5398
5399 1. Add support for the /+ modifier to perltest (to output $` like it does in
5400 pcretest).
5401
5402 2. Add support for the /g modifier to perltest.
5403
5404 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5405 matches null strings.
5406
5407 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5408 pattern. Perl treats empty patterns specially - it reuses the most recent
5409 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5410 effect.
5411
5412 5. The POSIX interface was broken in that it was just handing over the POSIX
5413 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5414 required a bigger vector, with some working space on the end. This means that
5415 the POSIX wrapper now has to get and free some memory, and copy the results.
5416
5417 6. Added some simple autoconf support, placing the test data and the
5418 documentation in separate directories, re-organizing some of the
5419 information files, and making it build pcre-config (a GNU standard). Also added
5420 libtool support for building PCRE as a shared library, which is now the
5421 default.
5422
5423 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5424 09 are not valid octal constants. Single digits will be used for minor values
5425 less than 10.
5426
5427 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5428 existing programs that set these in the POSIX interface can use PCRE without
5429 modification.
5430
5431 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5432 return all that pcre_info() returns, plus additional data. The pcre_info()
5433 function is retained for compatibility, but is considered to be obsolete.
5434
5435 10. Added experimental recursion feature (?R) to handle one common case that
5436 Perl 5.6 will be able to do with (?p{...}).
5437
5438 11. Added support for POSIX character classes like [:alpha:], which Perl is
5439 adopting.
5440
5441
5442 Version 2.08 31-Aug-99
5443 ----------------------
5444
5445 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5446 trying to match at the startoffset position, but instead was moving forward to
5447 the next newline as if a previous match had failed.
5448
5449 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5450 and could get into a loop if a null string was matched other than at the start
5451 of the subject.
5452
5453 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5454 be distinguished at compile time, and for completeness also added PCRE_DATE.
5455
5456 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5457 in GnuWin32 environments.
5458
5459
5460 Version 2.07 29-Jul-99
5461 ----------------------
5462
5463 1. The documentation is now supplied in plain text form and HTML as well as in
5464 the form of man page sources.
5465
5466 2. C++ compilers don't like assigning (void *) values to other pointer types.
5467 In particular this affects malloc(). Although there is no problem in Standard
5468 C, I've put in casts to keep C++ compilers happy.
5469
5470 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5471 should be (const char *).
5472
5473 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5474 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5475 However, I haven't made this a standard facility. The documentation doesn't
5476 mention it, and the Makefile doesn't support it.
5477
5478 5. The Makefile now contains an "install" target, with editable destinations at
5479 the top of the file. The pcretest program is not installed.
5480
5481 6. pgrep -V now gives the PCRE version number and date.
5482
5483 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5484 causing the entire string to be ignored, instead of just the last character.
5485
5486 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5487 non-matching string, it can take a very, very long time, even for strings of
5488 quite modest length, because of the nested recursion. PCRE now does better in
5489 some of these cases. It does this by remembering the last required literal
5490 character in the pattern, and pre-searching the subject to ensure it is present
5491 before running the real match. In other words, it applies a heuristic to detect
5492 some types of certain failure quickly, and in the above example, if presented
5493 with a string that has no trailing " it gives "no match" very quickly.
5494
5495 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5496 other alternatives are tried instead.
5497
5498
5499 Version 2.06 09-Jun-99
5500 ----------------------
5501
5502 1. Change pcretest's output for amount of store used to show just the code
5503 space, because the remainder (the data block) varies in size between 32-bit and
5504 64-bit systems.
5505
5506 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5507 start matching at. This allows lookbehinds to work when searching for multiple
5508 occurrences in a string.
5509
5510 3. Added additional options to pcretest for testing multiple occurrences:
5511
5512 /+ outputs the rest of the string that follows a match
5513 /g loops for multiple occurrences, using the new startoffset argument
5514 /G loops for multiple occurrences by passing an incremented pointer
5515
5516 4. PCRE wasn't doing the "first character" optimization for patterns starting
5517 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5518 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5519 the letter 'x'. On long subject strings, this gives a significant speed-up.
5520
5521
5522 Version 2.05 21-Apr-99
5523 ----------------------
5524
5525 1. Changed the type of magic_number from int to long int so that it works
5526 properly on 16-bit systems.
5527
5528 2. Fixed a bug which caused patterns starting with .* not to work correctly
5529 when the subject string contained newline characters. PCRE was assuming
5530 anchoring for such patterns in all cases, which is not correct because .* will
5531 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5532 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5533 must be retried after every newline in the subject.
5534
5535
5536 Version 2.04 18-Feb-99
5537 ----------------------
5538
5539 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5540 computation of the store needed to hold the pattern was incorrect (too large).
5541 If such patterns were nested a few deep, this could multiply and become a real
5542 problem.
5543
5544 2. Added /M option to pcretest to show the memory requirement of a specific
5545 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5546
5547 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5548 compiled in such a way that the backtracking after subsequent failure was
5549 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5550 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5551
5552
5553 Version 2.03 02-Feb-99
5554 ----------------------
5555
5556 1. Fixed typo and small mistake in man page.
5557
5558 2. Added 4th condition (GPL supersedes if conflict) and created separate
5559 LICENCE file containing the conditions.
5560
5561 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5562 Perl, that is the internal \ allows the delimiter to be included in the
5563 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5564 the final delimiter, add \ to the end of the pattern (to test the error).
5565
5566 4. Added the convenience functions for extracting substrings after a successful
5567 match. Updated pcretest to make it able to test these functions.
5568
5569
5570 Version 2.02 14-Jan-99
5571 ----------------------
5572
5573 1. Initialized the working variables associated with each extraction so that
5574 their saving and restoring doesn't refer to uninitialized store.
5575
5576 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5577 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5578 fix the problem.
5579
5580 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5581 calls, and wasn't printing the correct value for compiling calls. Increased the
5582 default value of LOOPREPEAT, and the number of significant figures in the
5583 times.
5584
5585 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5586
5587 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5588 a building problem on Windows NT with a FAT file system.
5589
5590
5591 Version 2.01 21-Oct-98
5592 ----------------------
5593
5594 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5595 to character tables built by pcre_maketables() in the current locale. If NULL
5596 is passed, the default tables are used.
5597
5598
5599 Version 2.00 24-Sep-98
5600 ----------------------
5601
5602 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5603 it any more.
5604
5605 2. Allow quantification of (?>) groups, and make it work correctly.
5606
5607 3. The first character computation wasn't working for (?>) groups.
5608
5609 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5610 end of the subject) and add 5.005's \z, which really does match only at the
5611 very end of the subject.
5612
5613 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5614
5615 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5616 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5617 localized options. All options to pcre_study() were also removed.
5618
5619 7. Add other new features from 5.005:
5620
5621 $(?<= positive lookbehind
5622 $(?<! negative lookbehind
5623 (?imsx-imsx) added the unsetting capability
5624 such a setting is global if at outer level; local otherwise
5625 (?imsx-imsx:) non-capturing groups with option setting
5626 (?(cond)re|re) conditional pattern matching
5627
5628 A backreference to itself in a repeated group matches the previous
5629 captured string.
5630
5631 8. General tidying up of studying (both automatic and via "study")
5632 consequential on the addition of new assertions.
5633
5634 9. As in 5.005, unlimited repeated groups that could match an empty substring
5635 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5636 runtime if any iteration does actually match an empty substring.
5637
5638 10. Include the RunTest script in the distribution.
5639
5640 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5641 discrepancies, some of which were old and were also with respect to 5.004. They
5642 have now been fixed.
5643
5644
5645 Version 1.09 28-Apr-98
5646 ----------------------
5647
5648 1. A negated single character class followed by a quantifier with a minimum
5649 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5650 program crashes, or just wrong answers. This did not apply to negated classes
5651 containing more than one character, or to minima other than one.
5652
5653
5654 Version 1.08 27-Mar-98
5655 ----------------------
5656
5657 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5658
5659 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5660 latter must appear before anything that relies on it in the pattern.
5661
5662
5663 Version 1.07 16-Feb-98
5664 ----------------------
5665
5666 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5667 repeat of a potentially empty string).
5668
5669
5670 Version 1.06 23-Jan-98
5671 ----------------------
5672
5673 1. Added Markus Oberhumer's little patches for C++.
5674
5675 2. Literal strings longer than 255 characters were broken.
5676
5677
5678 Version 1.05 23-Dec-97
5679 ----------------------
5680
5681 1. Negated character classes containing more than one character were failing if
5682 PCRE_CASELESS was set at run time.
5683
5684
5685 Version 1.04 19-Dec-97
5686 ----------------------
5687
5688 1. Corrected the man page, where some "const" qualifiers had been omitted.
5689
5690 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5691 input syntax.
5692
5693 3. Fixed memory leak which occurred when a regex with back references was
5694 matched with an offsets vector that wasn't big enough. The temporary memory
5695 that is used in this case wasn't being freed if the match failed.
5696
5697 4. Tidied pcretest to ensure it frees memory that it gets.
5698
5699 5. Temporary memory was being obtained in the case where the passed offsets
5700 vector was exactly big enough.
5701
5702 6. Corrected definition of offsetof() from change 5 below.
5703
5704 7. I had screwed up change 6 below and broken the rules for the use of
5705 setjmp(). Now fixed.
5706
5707
5708 Version 1.03 18-Dec-97
5709 ----------------------
5710
5711 1. A erroneous regex with a missing opening parenthesis was correctly
5712 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5713 on some systems.
5714
5715 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5716 it was reported that one broken compiler failed on the former because "code" is
5717 also an independent variable.
5718
5719 3. The erroneous regex a[]b caused an array overrun reference.
5720
5721 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5722 fail on data ending with that character. (It was going on too far, and checking
5723 the next character, typically a binary zero.) This was specific to the
5724 optimized code for single-character negative classes.
5725
5726 5. Added a contributed patch from the TIN world which does the following:
5727
5728 + Add an undef for memmove, in case the the system defines a macro for it.
5729
5730 + Add a definition of offsetof(), in case there isn't one. (I don't know
5731 the reason behind this - offsetof() is part of the ANSI standard - but
5732 it does no harm).
5733
5734 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5735 most of the places where whitespace preceded '#'. I have given up and
5736 allowed the remaining 2 cases to be at the margin.
5737
5738 + Rename some variables in pcre to eliminate shadowing. This seems very
5739 pedantic, but does no harm, of course.
5740
5741 6. Moved the call to setjmp() into its own function, to get rid of warnings
5742 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5743
5744 7. Constructs such as \d{8,} were compiling into the equivalent of
5745 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5746 outcome, but in this particular case used more store than had been allocated,
5747 which caused the bug to be discovered because it threw up an internal error.
5748
5749 8. The debugging code in both pcre and pcretest for outputting the compiled
5750 form of a regex was going wrong in the case of back references followed by
5751 curly-bracketed repeats.
5752
5753
5754 Version 1.02 12-Dec-97
5755 ----------------------
5756
5757 1. Typos in pcre.3 and comments in the source fixed.
5758
5759 2. Applied a contributed patch to get rid of places where it used to remove
5760 'const' from variables, and fixed some signed/unsigned and uninitialized
5761 variable warnings.
5762
5763 3. Added the "runtest" target to Makefile.
5764
5765 4. Set default compiler flag to -O2 rather than just -O.
5766
5767
5768 Version 1.01 19-Nov-97
5769 ----------------------
5770
5771 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5772 like /([ab]*)*/, that is, for classes with more than one character in them.
5773
5774 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5775 as /((?>a*))*/ (a PCRE_EXTRA facility).
5776
5777
5778 Version 1.00 18-Nov-97
5779 ----------------------
5780
5781 1. Added compile-time macros to support systems such as SunOS4 which don't have
5782 memmove() or strerror() but have other things that can be used instead.
5783
5784 2. Arranged that "make clean" removes the executables.
5785
5786
5787 Version 0.99 27-Oct-97
5788 ----------------------
5789
5790 1. Fixed bug in code for optimizing classes with only one character. It was
5791 initializing a 32-byte map regardless, which could cause it to run off the end
5792 of the memory it had got.
5793
5794 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5795
5796
5797 Version 0.98 22-Oct-97
5798 ----------------------
5799
5800 1. Fixed bug in code for handling temporary memory usage when there are more
5801 back references than supplied space in the ovector. This could cause segfaults.
5802
5803
5804 Version 0.97 21-Oct-97
5805 ----------------------
5806
5807 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5808
5809 2. Optimized negated single characters not to use a bit map.
5810
5811 3. Brought error texts together as macro definitions; clarified some of them;
5812 fixed one that was wrong - it said "range out of order" when it meant "invalid
5813 escape sequence".
5814
5815 4. Changed some char * arguments to const char *.
5816
5817 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5818
5819 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5820 pcretest.
5821
5822
5823 Version 0.96 16-Oct-97
5824 ----------------------
5825
5826 1. Added a simple "pgrep" utility to the distribution.
5827
5828 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5829 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5830 where "ddd" means "one or more decimal digits".
5831
5832 3. Fixed serious bug. If a pattern had a back reference, but the call to
5833 pcre_exec() didn't supply a large enough ovector to record the related
5834 identifying subpattern, the match always failed. PCRE now remembers the number
5835 of the largest back reference, and gets some temporary memory in which to save
5836 the offsets during matching if necessary, in order to ensure that
5837 backreferences always work.
5838
5839 4. Increased the compatibility with Perl in a number of ways:
5840
5841 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5842 to request this handling. The option can be set at compile or exec time.
5843
5844 (b) $ matches before a terminating newline by default; an option
5845 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5846 mode). The option can be set at compile or exec time.
5847
5848 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5849 the same as Perl's. If the decimal number it represents is less than 10
5850 or there aren't that many previous left capturing parentheses, an octal
5851 escape is read. Inside a character class, it's always an octal escape,
5852 even if it is a single digit.
5853
5854 (d) An escaped but undefined alphabetic character is taken as a literal,
5855 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5856 escapes.
5857
5858 (e) {0} is now permitted. (The previous item is removed from the compiled
5859 pattern).
5860
5861 5. Changed all the names of code files so that the basic parts are no longer
5862 than 10 characters, and abolished the teeny "globals.c" file.
5863
5864 6. Changed the handling of character classes; they are now done with a 32-byte
5865 bit map always.
5866
5867 7. Added the -d and /D options to pcretest to make it possible to look at the
5868 internals of compilation without having to recompile pcre.
5869
5870
5871 Version 0.95 23-Sep-97
5872 ----------------------
5873
5874 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5875 \x20 at the start of a run of normal characters. These were being treated as
5876 real characters, instead of the source characters being re-checked.
5877
5878
5879 Version 0.94 18-Sep-97
5880 ----------------------
5881
5882 1. The functions are now thread-safe, with the caveat that the global variables
5883 containing pointers to malloc() and free() or alternative functions are the
5884 same for all threads.
5885
5886 2. Get pcre_study() to generate a bitmap of initial characters for non-
5887 anchored patterns when this is possible, and use it if passed to pcre_exec().
5888
5889
5890 Version 0.93 15-Sep-97
5891 ----------------------
5892
5893 1. /(b)|(:+)/ was computing an incorrect first character.
5894
5895 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5896 but not actually doing anything yet.
5897
5898 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5899 as Perl does (e.g. [-az] or [az-]).
5900
5901 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5902 all possible positions.
5903
5904 5. Split up into different modules to avoid including unneeded functions in a
5905 compiled binary. However, compile and exec are still in one module. The "study"
5906 function is split off.
5907
5908 6. The character tables are now in a separate module whose source is generated
5909 by an auxiliary program - but can then be edited by hand if required. There are
5910 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5911 toupper() in the code.
5912
5913 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5914 make them global. Abolish the function for setting them, as the caller can now
5915 set them directly.
5916
5917
5918 Version 0.92 11-Sep-97
5919 ----------------------
5920
5921 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5922 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5923
5924 2. Caseless matching was not working in character classes if the characters in
5925 the pattern were in upper case.
5926
5927 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5928
5929 4. Make PCRE_ANCHORED public and accept as a compile option.
5930
5931 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5932 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5933 pass them.
5934
5935 6. Give an error if bad option bits passed at compile or run time.
5936
5937 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5938 pcretest to cause it to pass that flag.
5939
5940 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5941 options, and the first character, if set.
5942
5943 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5944
5945
5946 Version 0.91 10-Sep-97
5947 ----------------------
5948
5949 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5950 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5951
5952 2. PCRE was looping on encountering an indefinitely repeated back reference to
5953 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5954 Perl does - treats the match as successful.
5955
5956 ****

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

  ViewVC Help
Powered by ViewVC 1.1.5