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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1588 - (show annotations)
Mon Aug 10 14:09:19 2015 UTC (4 years, 4 months ago) by ph10
File size: 282690 byte(s)
Fix reported offsets in \k and \g syntax errors.
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 8.38 xx-xxx-xxxx
8 ------------------------
9
10 1. If a group that contained a recursive back reference also contained a
11 forward reference subroutine call followed by a non-forward-reference
12 subroutine call, for example /.((?2)(?R)\1)()/, pcre2_compile() failed to
13 compile correct code, leading to undefined behaviour or an internally
14 detected error. This bug was discovered by the LLVM fuzzer.
15
16 2. Quantification of certain items (e.g. atomic back references) could cause
17 incorrect code to be compiled when recursive forward references were
18 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
19 This bug was discovered by the LLVM fuzzer.
20
21 3. A repeated conditional group whose condition was a reference by name caused
22 a buffer overflow if there was more than one group with the given name.
23 This bug was discovered by the LLVM fuzzer.
24
25 4. A recursive back reference by name within a group that had the same name as
26 another group caused a buffer overflow. For example:
27 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
28
29 5. A forward reference by name to a group whose number is the same as the
30 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
31 a buffer overflow at compile time. This bug was discovered by the LLVM
32 fuzzer.
33
34 6. A lookbehind assertion within a set of mutually recursive subpatterns could
35 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
36
37 7. Another buffer overflow bug involved duplicate named groups with a
38 reference between their definition, with a group that reset capture
39 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
40 fixed by always allowing for more memory, even if not needed. (A proper fix
41 is implemented in PCRE2, but it involves more refactoring.)
42
43 8. There was no check for integer overflow in subroutine calls such as (?123).
44
45 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
46 being treated as a literal 'l' instead of causing an error.
47
48 10. There was a buffer overflow if pcre_exec() was called with an ovector of
49 size 1. This bug was found by american fuzzy lop.
50
51 11. If a non-capturing group containing a conditional group that could match
52 an empty string was repeated, it was not identified as matching an empty
53 string itself. For example: /^(?:(?(1)x|)+)+$()/.
54
55 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
56 \a and \e in test subject lines.
57
58 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
59 instead of the EBCDIC value.
60
61 14. The handling of \c in an EBCDIC environment has been revised so that it is
62 now compatible with the specification in Perl's perlebcdic page.
63
64 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
65 ASCII/Unicode. This has now been added to the list of characters that are
66 recognized as white space in EBCDIC.
67
68 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
69 error (correctly) when used outside a class, but did not give an error
70 within a class.
71
72 17. \h within a class was incorrectly compiled in EBCDIC environments.
73
74 18. A pattern with an unmatched closing parenthesis that contained a backward
75 assertion which itself contained a forward reference caused buffer
76 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
77
78 19. JIT should return with error when the compiled pattern requires more stack
79 space than the maximum.
80
81 20. A possessively repeated conditional group that could match an empty string,
82 for example, /(?(R))*+/, was incorrectly compiled.
83
84 21. Fix infinite recursion in the JIT compiler when certain patterns such as
85 /(?:|a|){100}x/ are analysed.
86
87 22. Some patterns with character classes involving [: and \\ were incorrectly
88 compiled and could cause reading from uninitialized memory or an incorrect
89 error diagnosis.
90
91 23. Pathological patterns containing many nested occurrences of [: caused
92 pcre_compile() to run for a very long time.
93
94 24. A conditional group with only one branch has an implicit empty alternative
95 branch and must therefore be treated as potentially matching an empty
96 string.
97
98 25. If (?R was followed by - or + incorrect behaviour happened instead of a
99 diagnostic.
100
101 26. Arrange to give up on finding the minimum matching length for overly
102 complex patterns.
103
104 27. Similar to (4) above: in a pattern with duplicated named groups and an
105 occurrence of (?| it is possible for an apparently non-recursive back
106 reference to become recursive if a later named group with the relevant
107 number is encountered. This could lead to a buffer overflow. Wen Guanxing
108 from Venustech ADLAB discovered this bug.
109
110 28. If pcregrep was given the -q option with -c or -l, or when handling a
111 binary file, it incorrectly wrote output to stdout.
112
113 29. The JIT compiler did not restore the control verb head in case of *THEN
114 control verbs. This issue was found by Karl Skomski with a custom LLVM
115 fuzzer.
116
117 30. Error messages for syntax errors following \g and \k were giving inaccurate
118 offsets in the pattern.
119
120
121 Version 8.37 28-April-2015
122 --------------------------
123
124 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
125 for those parentheses to be closed with whatever has been captured so far.
126 However, it was failing to mark any other groups between the hightest
127 capture so far and the currrent group as "unset". Thus, the ovector for
128 those groups contained whatever was previously there. An example is the
129 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
130
131 2. If an assertion condition was quantified with a minimum of zero (an odd
132 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
133
134 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
135 unrecognized modifier, a crash could occur.
136
137 4. An attempt to do global matching in pcretest with a zero-length ovector
138 caused a crash.
139
140 5. Fixed a memory leak during matching that could occur for a subpattern
141 subroutine call (recursive or otherwise) if the number of captured groups
142 that had to be saved was greater than ten.
143
144 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
145 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
146 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
147
148 7. A UTF pattern containing a "not" match of a non-ASCII character and a
149 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
150
151 8. When a pattern is compiled, it remembers the highest back reference so that
152 when matching, if the ovector is too small, extra memory can be obtained to
153 use instead. A conditional subpattern whose condition is a check on a
154 capture having happened, such as, for example in the pattern
155 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
156 setting the highest backreference number. This mattered only if pcre_exec()
157 was called with an ovector that was too small to hold the capture, and there
158 was no other kind of back reference (a situation which is probably quite
159 rare). The effect of the bug was that the condition was always treated as
160 FALSE when the capture could not be consulted, leading to a incorrect
161 behaviour by pcre_exec(). This bug has been fixed.
162
163 9. A reference to a duplicated named group (either a back reference or a test
164 for being set in a conditional) that occurred in a part of the pattern where
165 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
166 to be incorrectly calculated, leading to overwriting.
167
168 10. A mutually recursive set of back references such as (\2)(\1) caused a
169 segfault at study time (while trying to find the minimum matching length).
170 The infinite loop is now broken (with the minimum length unset, that is,
171 zero).
172
173 11. If an assertion that was used as a condition was quantified with a minimum
174 of zero, matching went wrong. In particular, if the whole group had
175 unlimited repetition and could match an empty string, a segfault was
176 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
177 assertions to be quantified, but not if they are being used as conditions,
178 so the above pattern is faulted by Perl. PCRE has now been changed so that
179 it also rejects such patterns.
180
181 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
182 failed to allow the zero-repeat case if pcre2_exec() was called with an
183 ovector too small to capture the group.
184
185 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
186 Red Hat Product Security:
187
188 (a) A crash if /K and /F were both set with the option to save the compiled
189 pattern.
190
191 (b) Another crash if the option to print captured substrings in a callout
192 was combined with setting a null ovector, for example \O\C+ as a subject
193 string.
194
195 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
196 forward reference repeated a large (but limited) number of times within a
197 repeated outer group that has a zero minimum quantifier, caused incorrect
198 code to be compiled, leading to the error "internal error:
199 previously-checked referenced subpattern not found" when an incorrect
200 memory address was read. This bug was reported as "heap overflow",
201 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
202 CVE-2015-2325.
203
204 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
205 call within a group that also contained a recursive back reference caused
206 incorrect code to be compiled. This bug was reported as "heap overflow",
207 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
208 number CVE-2015-2326.
209
210 24. Computing the size of the JIT read-only data in advance has been a source
211 of various issues, and new ones are still appear unfortunately. To fix
212 existing and future issues, size computation is eliminated from the code,
213 and replaced by on-demand memory allocation.
214
215 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
216 adjacent to the end of the range, and the range contained characters with
217 more than one other case, caused incorrect behaviour when compiled in UTF
218 mode. In that example, the range a-j was left out of the class.
219
220 26. Fix JIT compilation of conditional blocks, which assertion
221 is converted to (*FAIL). E.g: /(?(?!))/.
222
223 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
224 discovered by the LLVM fuzzer.
225
226 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
227 when this assertion was used as a condition, for example (?(?!)a|b). In
228 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
229 error about an unsupported item.
230
231 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
232 possessification code could take exponential time to complete. A recursion
233 depth limit of 1000 has been imposed to limit the resources used by this
234 optimization.
235
236 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
237 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
238 because \S ensures they are all in the class. The code for doing this was
239 interacting badly with the code for computing the amount of space needed to
240 compile the pattern, leading to a buffer overflow. This bug was discovered
241 by the LLVM fuzzer.
242
243 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
244 other kinds of group caused stack overflow at compile time. This bug was
245 discovered by the LLVM fuzzer.
246
247 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
248 between a subroutine call and its quantifier was incorrectly compiled,
249 leading to buffer overflow or other errors. This bug was discovered by the
250 LLVM fuzzer.
251
252 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
253 assertion after (?(. The code was failing to check the character after
254 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
255 was discovered by the LLVM fuzzer.
256
257 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
258 a fixed maximum following a group that contains a subroutine reference was
259 incorrectly compiled and could trigger buffer overflow. This bug was
260 discovered by the LLVM fuzzer.
261
262 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
263 caused a stack overflow instead of the diagnosis of a non-fixed length
264 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
265
266 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
267 (e.g. /(?<=\Ka)/) could make pcregrep loop.
268
269 37. There was a similar problem to 36 in pcretest for global matches.
270
271 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
272 and a subsequent item in the pattern caused a non-match, backtracking over
273 the repeated \X did not stop, but carried on past the start of the subject,
274 causing reference to random memory and/or a segfault. There were also some
275 other cases where backtracking after \C could crash. This set of bugs was
276 discovered by the LLVM fuzzer.
277
278 39. The function for finding the minimum length of a matching string could take
279 a very long time if mutual recursion was present many times in a pattern,
280 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
281 method has been implemented. This infelicity was discovered by the LLVM
282 fuzzer.
283
284 40. Static linking against the PCRE library using the pkg-config module was
285 failing on missing pthread symbols.
286
287
288 Version 8.36 26-September-2014
289 ------------------------------
290
291 1. Got rid of some compiler warnings in the C++ modules that were shown up by
292 -Wmissing-field-initializers and -Wunused-parameter.
293
294 2. The tests for quantifiers being too big (greater than 65535) were being
295 applied after reading the number, and stupidly assuming that integer
296 overflow would give a negative number. The tests are now applied as the
297 numbers are read.
298
299 3. Tidy code in pcre_exec.c where two branches that used to be different are
300 now the same.
301
302 4. The JIT compiler did not generate match limit checks for certain
303 bracketed expressions with quantifiers. This may lead to exponential
304 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
305 issue should be resolved now.
306
307 5. Fixed an issue, which occures when nested alternatives are optimized
308 with table jumps.
309
310 6. Inserted two casts and changed some ints to size_t in the light of some
311 reported 64-bit compiler warnings (Bugzilla 1477).
312
313 7. Fixed a bug concerned with zero-minimum possessive groups that could match
314 an empty string, which sometimes were behaving incorrectly in the
315 interpreter (though correctly in the JIT matcher). This pcretest input is
316 an example:
317
318 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
319 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
320
321 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
322 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
323 for an empty string was breaking the inner loop and carrying on at a lower
324 level, when possessive repeated groups should always return to a higher
325 level as they have no backtrack points in them. The empty string test now
326 occurs at the outer level.
327
328 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
329 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
330
331 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
332 doesn't).
333
334 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
335 to hold a pattern to be incorrectly computed (too small) when there were
336 named back references to duplicated names. This could cause "internal
337 error: code overflow" or "double free or corruption" or other memory
338 handling errors.
339
340 11. When named subpatterns had the same prefixes, back references could be
341 confused. For example, in this pattern:
342
343 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
344
345 the reference to 'Name' was incorrectly treated as a reference to a
346 duplicate name.
347
348 12. A pattern such as /^s?c/mi8 where the optional character has more than
349 one "other case" was incorrectly compiled such that it would only try to
350 match starting at "c".
351
352 13. When a pattern starting with \s was studied, VT was not included in the
353 list of possible starting characters; this should have been part of the
354 8.34/18 patch.
355
356 14. If a character class started [\Qx]... where x is any character, the class
357 was incorrectly terminated at the ].
358
359 15. If a pattern that started with a caseless match for a character with more
360 than one "other case" was studied, PCRE did not set up the starting code
361 unit bit map for the list of possible characters. Now it does. This is an
362 optimization improvement, not a bug fix.
363
364 16. The Unicode data tables have been updated to Unicode 7.0.0.
365
366 17. Fixed a number of memory leaks in pcregrep.
367
368 18. Avoid a compiler warning (from some compilers) for a function call with
369 a cast that removes "const" from an lvalue by using an intermediate
370 variable (to which the compiler does not object).
371
372 19. Incorrect code was compiled if a group that contained an internal recursive
373 back reference was optional (had quantifier with a minimum of zero). This
374 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
375 caused segmentation faults because of stack overflows at compile time.
376
377 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
378 group that is quantified with an indefinite repeat, caused a compile-time
379 loop which used up all the system stack and provoked a segmentation fault.
380 This was not the same bug as 19 above.
381
382 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
383 Patch by Mike Frysinger.
384
385
386 Version 8.35 04-April-2014
387 --------------------------
388
389 1. A new flag is set, when property checks are present in an XCLASS.
390 When this flag is not set, PCRE can perform certain optimizations
391 such as studying these XCLASS-es.
392
393 2. The auto-possessification of character sets were improved: a normal
394 and an extended character set can be compared now. Furthermore
395 the JIT compiler optimizes more character set checks.
396
397 3. Got rid of some compiler warnings for potentially uninitialized variables
398 that show up only when compiled with -O2.
399
400 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
401 of a match later then the end of the match. The pcretest program was not
402 handling the case sensibly - it was outputting from the start to the next
403 binary zero. It now reports this situation in a message, and outputs the
404 text from the end to the start.
405
406 5. Fast forward search is improved in JIT. Instead of the first three
407 characters, any three characters with fixed position can be searched.
408 Search order: first, last, middle.
409
410 6. Improve character range checks in JIT. Characters are read by an inprecise
411 function now, which returns with an unknown value if the character code is
412 above a certain threshold (e.g: 256). The only limitation is that the value
413 must be bigger than the threshold as well. This function is useful when
414 the characters above the threshold are handled in the same way.
415
416 7. The macros whose names start with RAWUCHAR are placeholders for a future
417 mode in which only the bottom 21 bits of 32-bit data items are used. To
418 make this more memorable for those maintaining the code, the names have
419 been changed to start with UCHAR21, and an extensive comment has been added
420 to their definition.
421
422 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
423 to the export list in Makefile.am (they were accidentally omitted from the
424 8.34 tarball).
425
426 9. The informational output from pcretest used the phrase "starting byte set"
427 which is inappropriate for the 16-bit and 32-bit libraries. As the output
428 for "first char" and "need char" really means "non-UTF-char", I've changed
429 "byte" to "char", and slightly reworded the output. The documentation about
430 these values has also been (I hope) clarified.
431
432 10. Another JIT related optimization: use table jumps for selecting the correct
433 backtracking path, when more than four alternatives are present inside a
434 bracket.
435
436 11. Empty match is not possible, when the minimum length is greater than zero,
437 and there is no \K in the pattern. JIT should avoid empty match checks in
438 such cases.
439
440 12. In a caseless character class with UCP support, when a character with more
441 than one alternative case was not the first character of a range, not all
442 the alternative cases were added to the class. For example, s and \x{17f}
443 are both alternative cases for S: the class [RST] was handled correctly,
444 but [R-T] was not.
445
446 13. The configure.ac file always checked for pthread support when JIT was
447 enabled. This is not used in Windows, so I have put this test inside a
448 check for the presence of windows.h (which was already tested for).
449
450 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
451 The algorithm provides a way to skip certain starting offsets, and usually
452 faster than linear prefix searches.
453
454 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
455 as for 'fr_FR' and 'french'. For some reason, however, it then used the
456 Windows-specific input and output files, which have 'french' screwed in.
457 So this could never have worked. One of the problems with locales is that
458 they aren't always the same. I have now updated RunTest so that it checks
459 the output of the locale test (test 3) against three different output
460 files, and it allows the test to pass if any one of them matches. With luck
461 this should make the test pass on some versions of Solaris where it was
462 failing. Because of the uncertainty, the script did not used to stop if
463 test 3 failed; it now does. If further versions of a French locale ever
464 come to light, they can now easily be added.
465
466 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
467 there was a message during ./configure, but it did not stop. This now
468 provokes an error. The invalid example in README has been corrected.
469 If a value less than the minimum is given, the minimum value has always
470 been used, but now a warning is given.
471
472 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
473 was a bug in the test system, which is now fixed. Also, the list of various
474 configurations that are tested for each release did not have one with both
475 16/32 bits and --enable-bar-anycrlf. It now does.
476
477 18. pcretest was missing "-C bsr" for displaying the \R default setting.
478
479 19. Little endian PowerPC systems are supported now by the JIT compiler.
480
481 20. The fast forward newline mechanism could enter to an infinite loop on
482 certain invalid UTF-8 input. Although we don't support these cases
483 this issue can be fixed by a performance optimization.
484
485 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
486 not take account if existing stack usage. There is now a new global
487 variable called pcre_stack_guard that can be set to point to an external
488 function to check stack availability. It is called at the start of
489 processing every parenthesized group.
490
491 22. A typo in the code meant that in ungreedy mode the max/min qualifier
492 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
493 not match "ab".
494
495 23. When UTF was disabled, the JIT program reported some incorrect compile
496 errors. These messages are silenced now.
497
498 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
499 compiler.
500
501 25. Change all the temporary files used in RunGrepTest to be different to those
502 used by RunTest so that the tests can be run simultaneously, for example by
503 "make -j check".
504
505
506 Version 8.34 15-December-2013
507 -----------------------------
508
509 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
510 executable memory. Patch inspired by Carsten Klein.
511
512 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
513 this macro is never tested and has no effect, because the work to support
514 coverage involves only compiling and linking options and special targets in
515 the Makefile. The comment in config.h implied that defining the macro would
516 enable coverage support, which is totally false. There was also support for
517 setting this macro in the CMake files (my fault, I just copied it from
518 configure). SUPPORT_GCOV has now been removed.
519
520 3. Make a small performance improvement in strlen16() and strlen32() in
521 pcretest.
522
523 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
524 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
525 these cases). There was also one in pcretest.c.
526
527 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
528
529 6. In UTF mode, the code for checking whether a group could match an empty
530 string (which is used for indefinitely repeated groups to allow for
531 breaking an infinite loop) was broken when the group contained a repeated
532 negated single-character class with a character that occupied more than one
533 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
534 UTF-8 mode). The effect was undefined: the group might or might not be
535 deemed as matching an empty string, or the program might have crashed.
536
537 7. The code for checking whether a group could match an empty string was not
538 recognizing that \h, \H, \v, \V, and \R must match a character.
539
540 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
541 an empty string. If it can, pcretest shows this in its information output.
542
543 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
544 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
545 matched by pcre_exec() without using JIT:
546
547 (a) If the rest of the pattern did not match after a maximal run of
548 grapheme clusters, the code for backing up to try with fewer of them
549 did not always back up over a full grapheme when characters that do not
550 have the modifier quality were involved, e.g. Hangul syllables.
551
552 (b) If the match point in a subject started with modifier character, and
553 there was no match, the code could incorrectly back up beyond the match
554 point, and potentially beyond the first character in the subject,
555 leading to a segfault or an incorrect match result.
556
557 10. A conditional group with an assertion condition could lead to PCRE
558 recording an incorrect first data item for a match if no other first data
559 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
560 first data item, and therefore matched "ca" after "c" instead of at the
561 start.
562
563 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
564 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
565
566 12. The source of pcregrep now includes z/OS-specific code so that it can be
567 compiled for z/OS as part of the special z/OS distribution.
568
569 13. Added the -T and -TM options to pcretest.
570
571 14. The code in pcre_compile.c for creating the table of named capturing groups
572 has been refactored. Instead of creating the table dynamically during the
573 actual compiling pass, the information is remembered during the pre-compile
574 pass (on the stack unless there are more than 20 named groups, in which
575 case malloc() is used) and the whole table is created before the actual
576 compile happens. This has simplified the code (it is now nearly 150 lines
577 shorter) and prepared the way for better handling of references to groups
578 with duplicate names.
579
580 15. A back reference to a named subpattern when there is more than one of the
581 same name now checks them in the order in which they appear in the pattern.
582 The first one that is set is used for the reference. Previously only the
583 first one was inspected. This change makes PCRE more compatible with Perl.
584
585 16. Unicode character properties were updated from Unicode 6.3.0.
586
587 17. The compile-time code for auto-possessification has been refactored, based
588 on a patch by Zoltan Herczeg. It now happens after instead of during
589 compilation. The code is cleaner, and more cases are handled. The option
590 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
591 options in pcretest are provided to set it. It can also be set by
592 (*NO_AUTO_POSSESS) at the start of a pattern.
593
594 18. The character VT has been added to the default ("C" locale) set of
595 characters that match \s and are generally treated as white space,
596 following this same change in Perl 5.18. There is now no difference between
597 "Perl space" and "POSIX space". Whether VT is treated as white space in
598 other locales depends on the locale.
599
600 19. The code for checking named groups as conditions, either for being set or
601 for being recursed, has been refactored (this is related to 14 and 15
602 above). Processing unduplicated named groups should now be as fast at
603 numerical groups, and processing duplicated groups should be faster than
604 before.
605
606 20. Two patches to the CMake build system, by Alexander Barkov:
607
608 (1) Replace the "source" command by "." in CMakeLists.txt because
609 "source" is a bash-ism.
610
611 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
612 without these the CMake build does not work on Solaris.
613
614 21. Perl has changed its handling of \8 and \9. If there is no previously
615 encountered capturing group of those numbers, they are treated as the
616 literal characters 8 and 9 instead of a binary zero followed by the
617 literals. PCRE now does the same.
618
619 22. Following Perl, added \o{} to specify codepoints in octal, making it
620 possible to specify values greater than 0777 and also making them
621 unambiguous.
622
623 23. Perl now gives an error for missing closing braces after \x{... instead of
624 treating the string as literal. PCRE now does the same.
625
626 24. RunTest used to grumble if an inappropriate test was selected explicitly,
627 but just skip it when running all tests. This make it awkward to run ranges
628 of tests when one of them was inappropriate. Now it just skips any
629 inappropriate tests, as it always did when running all tests.
630
631 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
632 character types such as \d or \w, too many callouts were inserted, and the
633 data that they returned was rubbish.
634
635 26. In UCP mode, \s was not matching two of the characters that Perl matches,
636 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
637 were matched by \h. The code has now been refactored so that the lists of
638 the horizontal and vertical whitespace characters used for \h and \v (which
639 are defined only in one place) are now also used for \s.
640
641 27. Add JIT support for the 64 bit TileGX architecture.
642 Patch by Jiong Wang (Tilera Corporation).
643
644 28. Possessive quantifiers for classes (both explicit and automatically
645 generated) now use special opcodes instead of wrapping in ONCE brackets.
646
647 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
648 (because it's meaningless), this was not happening when PCRE_CASELESS was
649 set. Not wrong, but inefficient.
650
651 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
652 properties for \w, \d, etc) is present in a test regex. Otherwise if the
653 test contains no characters greater than 255, Perl doesn't realise it
654 should be using Unicode semantics.
655
656 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
657 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
658 does in Unicode mode.
659
660 32. Added the "forbid" facility to pcretest so that putting tests into the
661 wrong test files can sometimes be quickly detected.
662
663 33. There is now a limit (default 250) on the depth of nesting of parentheses.
664 This limit is imposed to control the amount of system stack used at compile
665 time. It can be changed at build time by --with-parens-nest-limit=xxx or
666 the equivalent in CMake.
667
668 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
669 errors. Perl warns for these when in warning mode, but PCRE has no facility
670 for giving warnings.
671
672 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
673 However, this was not working for (?!) because it is optimized to (*FAIL),
674 for which PCRE does not allow quantifiers. The optimization is now disabled
675 when a quantifier follows (?!). I can't see any use for this, but it makes
676 things uniform.
677
678 36. Perl no longer allows group names to start with digits, so I have made this
679 change also in PCRE. It simplifies the code a bit.
680
681 37. In extended mode, Perl ignores spaces before a + that indicates a
682 possessive quantifier. PCRE allowed a space before the quantifier, but not
683 before the possessive +. It now does.
684
685 38. The use of \K (reset reported match start) within a repeated possessive
686 group such as (a\Kb)*+ was not working.
687
688 40. Document that the same character tables must be used at compile time and
689 run time, and that the facility to pass tables to pcre_exec() and
690 pcre_dfa_exec() is for use only with saved/restored patterns.
691
692 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
693 features for Builds with MSVC:
694
695 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
696 builds). The .rc files can be used to set FileDescription and many other
697 attributes.
698
699 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
700 This allows higher-level build scripts which want .pdb files to avoid
701 hard-coding the exact files needed."
702
703 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
704 mean "start of word" and "end of word", respectively, as a transition aid.
705
706 43. A minimizing repeat of a class containing codepoints greater than 255 in
707 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
708 compiled to use the heap for recursion.
709
710 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
711 is configured.
712
713
714 Version 8.33 28-May-2013
715 ------------------------
716
717 1. Added 'U' to some constants that are compared to unsigned integers, to
718 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
719 variables that are added to signed variables, to ensure the result is
720 signed and can be negated.
721
722 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
723 CMake config files.
724
725 3. Revise the creation of config.h.generic so that all boolean macros are
726 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
727 overriding via -D on the command line possible.
728
729 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
730 to unsigned int is reported to make a quite noticeable speed difference in
731 a specific Windows environment. Testing on Linux did also appear to show
732 some benefit (and it is clearly not harmful). Also fixed the definition of
733 Xop which should be unsigned.
734
735 5. Related to (4), changing the definition of the intermediate variable cc
736 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
737 improvements.
738
739 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
740 unnecessary spaces.
741
742 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
743 and later.
744
745 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
746
747 9. Optimizing fast_forward_start_bits in JIT.
748
749 10. Adding support for callouts in JIT, and fixing some issues revealed
750 during this work. Namely:
751
752 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
753
754 (b) Minimum length was not checked before the matching is started.
755
756 11. The value of capture_last that is passed to callouts was incorrect in some
757 cases when there was a capture on one path that was subsequently abandoned
758 after a backtrack. Also, the capture_last value is now reset after a
759 recursion, since all captures are also reset in this case.
760
761 12. The interpreter no longer returns the "too many substrings" error in the
762 case when an overflowing capture is in a branch that is subsequently
763 abandoned after a backtrack.
764
765 13. In the pathological case when an offset vector of size 2 is used, pcretest
766 now prints out the matched string after a yield of 0 or 1.
767
768 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
769 Only supported by the JIT compiler at the moment.
770
771 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
772
773 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
774 offset of the starting point of the matching process, provided the offsets
775 vector is large enough.
776
777 17. The \A escape now records a lookbehind value of 1, though its execution
778 does not actually inspect the previous character. This is to ensure that,
779 in partial multi-segment matching, at least one character from the old
780 segment is retained when a new segment is processed. Otherwise, if there
781 are no lookbehinds in the pattern, \A might match incorrectly at the start
782 of a new segment.
783
784 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
785
786 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
787 gives some modest performance improvement in 8-bit mode.
788
789 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
790 be expressed in certain programming languages using Universal Character
791 Names.
792
793 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
794 which points out that "non characters" are not "characters that may not
795 appear in Unicode strings" but rather "characters that are reserved for
796 internal use and have only local meaning".
797
798 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
799 there was a conditional group that depended on an assertion, if the
800 assertion was false, the callout that immediately followed the alternation
801 in the condition was skipped when pcre_exec() was used for matching.
802
803 23. Allow an explicit callout to be inserted before an assertion that is the
804 condition for a conditional group, for compatibility with automatic
805 callouts, which always insert a callout at this point.
806
807 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
808 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
809
810 25. (*PRUNE) is now supported by the JIT compiler.
811
812 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
813
814 27. Fix the case where there are two or more SKIPs with arguments that may be
815 ignored.
816
817 28. (*SKIP) is now supported by the JIT compiler.
818
819 29. (*THEN) is now supported by the JIT compiler.
820
821 30. Update RunTest with additional test selector options.
822
823 31. The way PCRE handles backtracking verbs has been changed in two ways.
824
825 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
826 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
827 backtracking. In some cases this makes it more Perl-compatible, but Perl's
828 rather obscure rules do not always do the same thing.
829
830 (2) Previously, backtracking verbs were confined within assertions. This is
831 no longer the case for positive assertions, except for (*ACCEPT). Again,
832 this sometimes improves Perl compatibility, and sometimes does not.
833
834 32. A number of tests that were in test 2 because Perl did things differently
835 have been moved to test 1, because either Perl or PCRE has changed, and
836 these tests are now compatible.
837
838 32. Backtracking control verbs are now handled in the same way in JIT and
839 interpreter.
840
841 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
842 contained a forward subroutine reference caused a compile error.
843
844 34. Auto-detect and optimize limited repetitions in JIT.
845
846 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
847 blocking (*UTF) etc.
848
849 36. In the interpreter, maximizing pattern repetitions for characters and
850 character types now use tail recursion, which reduces stack usage.
851
852 37. The value of the max lookbehind was not correctly preserved if a compiled
853 and saved regex was reloaded on a host of different endianness.
854
855 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
856 of the compiled pattern block, expand the flags field from 16 to 32 bits
857 because it was almost full.
858
859 39. Try madvise first before posix_madvise.
860
861 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
862 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
863 original change locked out all matches of empty strings. This has been
864 changed so that one match of an empty string per line is recognized.
865 Subsequent searches on the same line (for colouring or for --only-matching,
866 for example) do not recognize empty strings.
867
868 41. Applied a user patch to fix a number of spelling mistakes in comments.
869
870 42. Data lines longer than 65536 caused pcretest to crash.
871
872 43. Clarified the data type for length and startoffset arguments for pcre_exec
873 and pcre_dfa_exec in the function-specific man pages, where they were
874 explicitly stated to be in bytes, never having been updated. I also added
875 some clarification to the pcreapi man page.
876
877 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
878 a segmentation fault.
879
880
881 Version 8.32 30-November-2012
882 -----------------------------
883
884 1. Improved JIT compiler optimizations for first character search and single
885 character iterators.
886
887 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
888 Patch by Daniel Richard G.
889
890 3. Single character iterator optimizations in the JIT compiler.
891
892 4. Improved JIT compiler optimizations for character ranges.
893
894 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
895 Reported by Giuseppe D'Angelo.
896
897 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
898 of a line, was being set incorrectly in cases where .* appeared inside
899 atomic brackets at the start of a pattern, or where there was a subsequent
900 *PRUNE or *SKIP.
901
902 7. Improved instruction cache flush for POWER/PowerPC.
903 Patch by Daniel Richard G.
904
905 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
906 grep:
907
908 (a) There is now no limit to the number of patterns to be matched.
909
910 (b) An error is given if a pattern is too long.
911
912 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
913 are now supported.
914
915 (d) --exclude-from and --include-from (multiple use) have been added.
916
917 (e) Exclusions and inclusions now apply to all files and directories, not
918 just to those obtained from scanning a directory recursively.
919
920 (f) Multiple uses of -f and --file-list are now supported.
921
922 (g) In a Windows environment, the default for -d has been changed from
923 "read" (the GNU grep default) to "skip", because otherwise the presence
924 of a directory in the file list provokes an error.
925
926 (h) The documentation has been revised and clarified in places.
927
928 9. Improve the matching speed of capturing brackets.
929
930 10. Changed the meaning of \X so that it now matches a Unicode extended
931 grapheme cluster.
932
933 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
934 out POSIX threads when JIT support is configured.
935
936 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
937
938 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
939 structure could go wrong in environments where size_t is not the same size
940 as int.
941
942 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
943 set.
944
945 15. The EBCDIC support had decayed; later updates to the code had included
946 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
947 general tidy up of EBCDIC-related issues, and the documentation was also
948 not quite right. There is now a test that can be run on ASCII systems to
949 check some of the EBCDIC-related things (but is it not a full test).
950
951 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
952 in a small tidy to the code.
953
954 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
955
956 18. If the --only-matching (-o) option in pcregrep is specified multiple
957 times, each one causes appropriate output. For example, -o1 -o2 outputs the
958 substrings matched by the 1st and 2nd capturing parentheses. A separating
959 string can be specified by --om-separator (default empty).
960
961 19. Improving the first n character searches.
962
963 20. Turn case lists for horizontal and vertical white space into macros so that
964 they are defined only once.
965
966 21. This set of changes together give more compatible Unicode case-folding
967 behaviour for characters that have more than one other case when UCP
968 support is available.
969
970 (a) The Unicode property table now has offsets into a new table of sets of
971 three or more characters that are case-equivalent. The MultiStage2.py
972 script that generates these tables (the pcre_ucd.c file) now scans
973 CaseFolding.txt instead of UnicodeData.txt for character case
974 information.
975
976 (b) The code for adding characters or ranges of characters to a character
977 class has been abstracted into a generalized function that also handles
978 case-independence. In UTF-mode with UCP support, this uses the new data
979 to handle characters with more than one other case.
980
981 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
982 whose other case is greater than 256 are now correctly matched
983 caselessly. Previously, the high codepoint matched the low one, but not
984 vice versa.
985
986 (d) The processing of \h, \H, \v, and \ in character classes now makes use
987 of the new class addition function, using character lists defined as
988 macros alongside the case definitions of 20 above.
989
990 (e) Caseless back references now work with characters that have more than
991 one other case.
992
993 (f) General caseless matching of characters with more than one other case
994 is supported.
995
996 22. Unicode character properties were updated from Unicode 6.2.0
997
998 23. Improved CMake support under Windows. Patch by Daniel Richard G.
999
1000 24. Add support for 32-bit character strings, and UTF-32
1001
1002 25. Major JIT compiler update (code refactoring and bugfixing).
1003 Experimental Sparc 32 support is added.
1004
1005 26. Applied a modified version of Daniel Richard G's patch to create
1006 pcre.h.generic and config.h.generic by "make" instead of in the
1007 PrepareRelease script.
1008
1009 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1010 pcre_compile.c when checking for a zero character.
1011
1012 28. Introducing a native interface for JIT. Through this interface, the compiled
1013 machine code can be directly executed. The purpose of this interface is to
1014 provide fast pattern matching, so several sanity checks are not performed.
1015 However, feature tests are still performed. The new interface provides
1016 1.4x speedup compared to the old one.
1017
1018 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1019 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1020 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1021
1022 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1023 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1024 the "old" RFC 2279). Instead, it ended up passing a negative length to
1025 pcre_exec().
1026
1027 31. Add support for GCC's visibility feature to hide internal functions.
1028
1029 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1030 "unknown -C option" after outputting 0 or 1.
1031
1032 33. There is now support for generating a code coverage report for the test
1033 suite in environments where gcc is the compiler and lcov is installed. This
1034 is mainly for the benefit of the developers.
1035
1036 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1037 unaddressable using valgrind annotations, allowing valgrind to detect
1038 invalid memory accesses. This is mainly for the benefit of the developers.
1039
1040 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1041
1042 26. Give configure error if --enable-cpp but no C++ compiler found.
1043
1044
1045 Version 8.31 06-July-2012
1046 -------------------------
1047
1048 1. Fixing a wrong JIT test case and some compiler warnings.
1049
1050 2. Removed a bashism from the RunTest script.
1051
1052 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1053 to unsigned type, result still unsigned" that was given by an MS compiler
1054 on encountering the code "-sizeof(xxx)".
1055
1056 4. Partial matching support is added to the JIT compiler.
1057
1058 5. Fixed several bugs concerned with partial matching of items that consist
1059 of more than one character:
1060
1061 (a) /^(..)\1/ did not partially match "aba" because checking references was
1062 done on an "all or nothing" basis. This also applied to repeated
1063 references.
1064
1065 (b) \R did not give a hard partial match if \r was found at the end of the
1066 subject.
1067
1068 (c) \X did not give a hard partial match after matching one or more
1069 characters at the end of the subject.
1070
1071 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1072 a partial match for the string "\r".
1073
1074 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1075 a partial match for a CR character at the end of the subject string.
1076
1077 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1078 running pcretest, the text "(JIT)" added to the output whenever JIT is
1079 actually used to run the match.
1080
1081 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1082 or /S+[+] with a digit between 1 and 7.
1083
1084 8. OP_NOT now supports any UTF character not just single-byte ones.
1085
1086 9. (*MARK) control verb is now supported by the JIT compiler.
1087
1088 10. The command "./RunTest list" lists the available tests without actually
1089 running any of them. (Because I keep forgetting what they all are.)
1090
1091 11. Add PCRE_INFO_MAXLOOKBEHIND.
1092
1093 12. Applied a (slightly modified) user-supplied patch that improves performance
1094 when the heap is used for recursion (compiled with --disable-stack-for-
1095 recursion). Instead of malloc and free for each heap frame each time a
1096 logical recursion happens, frames are retained on a chain and re-used where
1097 possible. This sometimes gives as much as 30% improvement.
1098
1099 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1100 call.
1101
1102 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1103
1104 15. It is now possible to link pcretest with libedit as an alternative to
1105 libreadline.
1106
1107 16. (*COMMIT) control verb is now supported by the JIT compiler.
1108
1109 17. The Unicode data tables have been updated to Unicode 6.1.0.
1110
1111 18. Added --file-list option to pcregrep.
1112
1113 19. Added binary file support to pcregrep, including the -a, --binary-files,
1114 -I, and --text options.
1115
1116 20. The madvise function is renamed for posix_madvise for QNX compatibility
1117 reasons. Fixed by Giuseppe D'Angelo.
1118
1119 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1120
1121 22. Changed the option for creating symbolic links for 16-bit man pages from
1122 -s to -sf so that re-installing does not cause issues.
1123
1124 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1125
1126 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1127 matches in certain environments (the workspace was not being correctly
1128 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1129 some of the workspace data at the beginning of a restart.
1130
1131 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1132 was not doing so when it should - probably a typo introduced by SVN 528
1133 (change 8.10/14).
1134
1135 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1136 \w+ when the character tables indicated that \x{c4} was a word character.
1137 There were several related cases, all because the tests for doing a table
1138 lookup were testing for characters less than 127 instead of 255.
1139
1140 27. If a pattern contains capturing parentheses that are not used in a match,
1141 their slots in the ovector are set to -1. For those that are higher than
1142 any matched groups, this happens at the end of processing. In the case when
1143 there were back references that the ovector was too small to contain
1144 (causing temporary malloc'd memory to be used during matching), and the
1145 highest capturing number was not used, memory off the end of the ovector
1146 was incorrectly being set to -1. (It was using the size of the temporary
1147 memory instead of the true size.)
1148
1149 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1150 ovector size, it uses memory at the end of the block that it has got.
1151
1152 29. Check for an overlong MARK name and give an error at compile time. The
1153 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1154
1155 30. JIT compiler update.
1156
1157 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1158 Rill for the patch.
1159
1160 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1161
1162 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1163
1164 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1165 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1166 was enabled.
1167
1168 35. Improve JIT code generation for greedy plus quantifier.
1169
1170 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1171 1 to "aa" instead of to an empty string. The bug affected repeated groups
1172 that could potentially match an empty string.
1173
1174 37. Optimizing single character iterators in JIT.
1175
1176 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1177 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1178 codepoints that are too big for the mode are faulted, and in a UTF mode,
1179 disallowed codepoints are also faulted.
1180
1181 39. If PCRE was compiled with UTF support, in three places in the DFA
1182 matcher there was code that should only have been obeyed in UTF mode, but
1183 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1184 processing when bytes with values greater than 127 were present. In 16-bit
1185 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1186 both cases the values are those that cannot be the first data item in a UTF
1187 character. The three items that might have provoked this were recursions,
1188 possessively repeated groups, and atomic groups.
1189
1190 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1191 and pcregrep, because some OS require shared objects to be explicitly
1192 passed to ld, causing the link step to fail if they are not.
1193
1194 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1195 mode, patterns that started with \h* or \R* might be incorrectly matched.
1196
1197
1198 Version 8.30 04-February-2012
1199 -----------------------------
1200
1201 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1202 name is defined in ctype.h.
1203
1204 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1205 only in quite long subpatterns.
1206
1207 3. Removed the function pcre_info(), which has been obsolete and deprecated
1208 since it was replaced by pcre_fullinfo() in February 2000.
1209
1210 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1211 match a (*MARK), and the match failed at the start of the subject, a
1212 reference to memory before the start of the subject could occur. This bug
1213 was introduced by fix 17 of release 8.21.
1214
1215 5. A reference to an unset group with zero minimum repetition was giving
1216 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1217 /(another)?(\1?)test/ matched against "hello world test". This bug was
1218 introduced in release 8.13.
1219
1220 6. Add support for 16-bit character strings (a large amount of work involving
1221 many changes and refactorings).
1222
1223 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1224 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1225 from a file.
1226
1227 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1228 rounding is not applied in this particular case).
1229
1230 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1231 if they appear, or are escaped, in patterns.
1232
1233 10. Get rid of a number of -Wunused-but-set-variable warnings.
1234
1235 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1236 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1237 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1238 also returns the mark "x". This bug applied to capturing parentheses,
1239 non-capturing parentheses, and atomic parentheses. It also applied to some
1240 assertions.
1241
1242 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1243 information out of configure.ac instead of relying on pcre.h.generic, which
1244 is not stored in the repository.
1245
1246 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1247 -lreadline.
1248
1249 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1250
1251 15. Applied Graycode's patch to put the top-level frame on the stack rather
1252 than the heap when not using the stack for recursion. This gives a
1253 performance improvement in many cases when recursion is not deep.
1254
1255 16. Experimental code added to "pcretest -C" to output the stack frame size.
1256
1257
1258 Version 8.21 12-Dec-2011
1259 ------------------------
1260
1261 1. Updating the JIT compiler.
1262
1263 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1264 are added as well.
1265
1266 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1267 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1268 calling _pcre_jit_exec. Some extra comments are added.
1269
1270 4. (*MARK) settings inside atomic groups that do not contain any capturing
1271 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1272 was introduced by change 18 for 8.20.
1273
1274 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1275 ECMA-262 standard.
1276
1277 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1278 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1279 This bug was probably introduced by change 9 of 8.13.
1280
1281 7. While fixing 6 above, I noticed that a number of other items were being
1282 incorrectly rejected as "not fixed length". This arose partly because newer
1283 opcodes had not been added to the fixed-length checking code. I have (a)
1284 corrected the bug and added tests for these items, and (b) arranged for an
1285 error to occur if an unknown opcode is encountered while checking for fixed
1286 length instead of just assuming "not fixed length". The items that were
1287 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1288 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1289 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1290
1291 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1292 being incorrectly compiled and would have given unpredicatble results.
1293
1294 9. A possessively repeated subpattern with minimum repeat count greater than
1295 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1296 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1297 the first (A) could occur when it should not.
1298
1299 10. Add a cast and remove a redundant test from the code.
1300
1301 11. JIT should use pcre_malloc/pcre_free for allocation.
1302
1303 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1304 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1305 is anything other than /usr, -L is still shown).
1306
1307 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1308
1309 14. Perl does not support \N without a following name in a [] class; PCRE now
1310 also gives an error.
1311
1312 15. If a forward reference was repeated with an upper limit of around 2000,
1313 it caused the error "internal error: overran compiling workspace". The
1314 maximum number of forward references (including repeats) was limited by the
1315 internal workspace, and dependent on the LINK_SIZE. The code has been
1316 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1317 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1318 of around 200,000 forward references. While doing this, I also speeded up
1319 the filling in of repeated forward references.
1320
1321 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1322 incorrectly expecting the subject to contain another "a" after the start.
1323
1324 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1325 in the match, the SKIP should be ignored. This was not happening; instead
1326 the SKIP was being treated as NOMATCH. For patterns such as
1327 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1328 tested.
1329
1330 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1331 now much more compatible with Perl, in particular in cases where the result
1332 is a non-match for a non-anchored pattern. For example, if
1333 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1334 "m", where previously it did not return a name. A side effect of this
1335 change is that for partial matches, the last encountered mark name is
1336 returned, as for non matches. A number of tests that were previously not
1337 Perl-compatible have been moved into the Perl-compatible test files. The
1338 refactoring has had the pleasing side effect of removing one argument from
1339 the match() function, thus reducing its stack requirements.
1340
1341 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1342 subsequent uses of /S (without +) incorrectly behaved like /S+.
1343
1344 21. Retrieve executable code size support for the JIT compiler and fixing
1345 some warnings.
1346
1347 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1348 not work when the shorter character appeared right at the end of the
1349 subject string.
1350
1351 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1352 systems.
1353
1354 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1355 output it when the /M option is used in pcretest.
1356
1357 25. The CheckMan script was not being included in the distribution. Also, added
1358 an explicit "perl" to run Perl scripts from the PrepareRelease script
1359 because this is reportedly needed in Windows.
1360
1361 26. If study data was being save in a file and studying had not found a set of
1362 "starts with" bytes for the pattern, the data written to the file (though
1363 never used) was taken from uninitialized memory and so caused valgrind to
1364 complain.
1365
1366 27. Updated RunTest.bat as provided by Sheri Pierce.
1367
1368 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1369
1370 29. Computation of memory usage for the table of capturing group names was
1371 giving an unnecessarily large value.
1372
1373
1374 Version 8.20 21-Oct-2011
1375 ------------------------
1376
1377 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1378 a POSIX class. After further experiments with Perl, which convinced me that
1379 Perl has bugs and confusions, a closing square bracket is no longer allowed
1380 in a POSIX name. This bug also affected patterns with classes that started
1381 with full stops.
1382
1383 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1384 captured substring, but while checking the failing first alternative,
1385 substring 1 is temporarily captured. If the output vector supplied to
1386 pcre_exec() was not big enough for this capture, the yield of the function
1387 was still zero ("insufficient space for captured substrings"). This cannot
1388 be totally fixed without adding another stack variable, which seems a lot
1389 of expense for a edge case. However, I have improved the situation in cases
1390 such as /(a)(b)x|abc/ matched against "abc", where the return code
1391 indicates that fewer than the maximum number of slots in the ovector have
1392 been set.
1393
1394 3. Related to (2) above: when there are more back references in a pattern than
1395 slots in the output vector, pcre_exec() uses temporary memory during
1396 matching, and copies in the captures as far as possible afterwards. It was
1397 using the entire output vector, but this conflicts with the specification
1398 that only 2/3 is used for passing back captured substrings. Now it uses
1399 only the first 2/3, for compatibility. This is, of course, another edge
1400 case.
1401
1402 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1403 main code base, and can be used by building with --enable-jit. When this is
1404 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1405 runtime --no-jit option is given.
1406
1407 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1408 ovector, the return from the function was zero, implying that there were
1409 other matches that did not fit. The correct "exactly full" value is now
1410 returned.
1411
1412 6. If a subpattern that was called recursively or as a subroutine contained
1413 (*PRUNE) or any other control that caused it to give a non-standard return,
1414 invalid errors such as "Error -26 (nested recursion at the same subject
1415 position)" or even infinite loops could occur.
1416
1417 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1418 computing the minimum length on reaching *ACCEPT, and so ended up with the
1419 wrong value of 1 rather than 0. Further investigation indicates that
1420 computing a minimum subject length in the presence of *ACCEPT is difficult
1421 (think back references, subroutine calls), and so I have changed the code
1422 so that no minimum is registered for a pattern that contains *ACCEPT.
1423
1424 8. If (*THEN) was present in the first (true) branch of a conditional group,
1425 it was not handled as intended. [But see 16 below.]
1426
1427 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1428 Sheri Pierce.
1429
1430 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1431 the first byte in a match must be "a".
1432
1433 11. Change 17 for 8.13 increased the recursion depth for patterns like
1434 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1435 pattern contains any instances of (*THEN). If it does not, the old
1436 optimizations are restored. It would be nice to do this on a per-group
1437 basis, but at the moment that is not feasible.
1438
1439 12. In some environments, the output of pcretest -C is CRLF terminated. This
1440 broke RunTest's code that checks for the link size. A single white space
1441 character after the value is now allowed for.
1442
1443 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1444 For "fr", it uses the Windows-specific input and output files.
1445
1446 14. If (*THEN) appeared in a group that was called recursively or as a
1447 subroutine, it did not work as intended. [But see next item.]
1448
1449 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1450 pattern fragments (but not containing any | characters). If A and B are
1451 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1452 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1453 D. In other words, Perl considers parentheses that do not contain any |
1454 characters to be part of a surrounding alternative, whereas PCRE was
1455 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1456 differently. PCRE now behaves in the same way as Perl, except in the case
1457 of subroutine/recursion calls such as (?1) which have in any case always
1458 been different (but PCRE had them first :-).
1459
1460 16. Related to 15 above: Perl does not treat the | in a conditional group as
1461 creating alternatives. Such a group is treated in the same way as an
1462 ordinary group without any | characters when processing (*THEN). PCRE has
1463 been changed to match Perl's behaviour.
1464
1465 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1466 RunGrepTest script failed.
1467
1468 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1469 inevitable for groups that contain captures, but it can lead to a lot of
1470 stack use in large patterns. The old behaviour has been restored for atomic
1471 groups that do not contain any capturing parentheses.
1472
1473 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1474 suppress the check for a minimum subject length at run time. (If it was
1475 given to pcre_exec() or pcre_dfa_exec() it did work.)
1476
1477 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1478 fail to work when decoding hex characters in data strings in EBCDIC
1479 environments.
1480
1481 21. It appears that in at least one Mac OS environment, the isxdigit() function
1482 is implemented as a macro that evaluates to its argument more than once,
1483 contravening the C 90 Standard (I haven't checked a later standard). There
1484 was an instance in pcretest which caused it to go wrong when processing
1485 \x{...} escapes in subject strings. The has been rewritten to avoid using
1486 things like p++ in the argument of isxdigit().
1487
1488
1489 Version 8.13 16-Aug-2011
1490 ------------------------
1491
1492 1. The Unicode data tables have been updated to Unicode 6.0.0.
1493
1494 2. Two minor typos in pcre_internal.h have been fixed.
1495
1496 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1497 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1498 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1499
1500 4. There were a number of related bugs in the code for matching backrefences
1501 caselessly in UTF-8 mode when codes for the characters concerned were
1502 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1503 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1504 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1505 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1506 2-byte code at the end of the subject (it thought there wasn't enough data
1507 left).
1508
1509 5. Comprehensive information about what went wrong is now returned by
1510 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1511 as the output vector has at least 2 elements. The offset of the start of
1512 the failing character and a reason code are placed in the vector.
1513
1514 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1515 now returned is for the first byte of the failing character, instead of the
1516 last byte inspected. This is an incompatible change, but I hope it is small
1517 enough not to be a problem. It makes the returned offset consistent with
1518 pcre_exec() and pcre_dfa_exec().
1519
1520 7. pcretest now gives a text phrase as well as the error number when
1521 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1522 failure, the offset and reason code are output.
1523
1524 8. When \R was used with a maximizing quantifier it failed to skip backwards
1525 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1526 back over a single character (\n). This seems wrong (because it treated the
1527 two characters as a single entity when going forwards), conflicts with the
1528 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1529 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1530 has been changed.
1531
1532 9. Some internal refactoring has changed the processing so that the handling
1533 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1534 time (the PCRE_DOTALL option was changed this way some time ago: version
1535 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1536 which was always a bit of a fudge. It also means that there is one less
1537 argument for the match() function, which reduces its stack requirements
1538 slightly. This change also fixes an incompatibility with Perl: the pattern
1539 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1540
1541 10. More internal refactoring has drastically reduced the number of recursive
1542 calls to match() for possessively repeated groups such as (abc)++ when
1543 using pcre_exec().
1544
1545 11. While implementing 10, a number of bugs in the handling of groups were
1546 discovered and fixed:
1547
1548 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1549 (a|)*(?1) gave a compile-time internal error.
1550 ((a|)+)+ did not notice that the outer group could match an empty string.
1551 (^a|^)+ was not marked as anchored.
1552 (.*a|.*)+ was not marked as matching at start or after a newline.
1553
1554 12. Yet more internal refactoring has removed another argument from the match()
1555 function. Special calls to this function are now indicated by setting a
1556 value in a variable in the "match data" data block.
1557
1558 13. Be more explicit in pcre_study() instead of relying on "default" for
1559 opcodes that mean there is no starting character; this means that when new
1560 ones are added and accidentally left out of pcre_study(), testing should
1561 pick them up.
1562
1563 14. The -s option of pcretest has been documented for ages as being an old
1564 synonym of -m (show memory usage). I have changed it to mean "force study
1565 for every regex", that is, assume /S for every regex. This is similar to -i
1566 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1567 using it. It makes it easier to run collections of tests with and without
1568 study enabled, and thereby test pcre_study() more easily. All the standard
1569 tests are now run with and without -s (but some patterns can be marked as
1570 "never study" - see 20 below).
1571
1572 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1573 restoration of the capturing data to the outer values was not happening
1574 correctly.
1575
1576 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1577 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1578 pattern had matched an empty string, and so incorrectly returned a no
1579 match.
1580
1581 17. There was optimizing code for the last branch of non-capturing parentheses,
1582 and also for the obeyed branch of a conditional subexpression, which used
1583 tail recursion to cut down on stack usage. Unfortunately, now that there is
1584 the possibility of (*THEN) occurring in these branches, tail recursion is
1585 no longer possible because the return has to be checked for (*THEN). These
1586 two optimizations have therefore been removed. [But see 8.20/11 above.]
1587
1588 18. If a pattern containing \R was studied, it was assumed that \R always
1589 matched two bytes, thus causing the minimum subject length to be
1590 incorrectly computed because \R can also match just one byte.
1591
1592 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1593 was incorrectly computed.
1594
1595 20. If /S is present twice on a test pattern in pcretest input, it now
1596 *disables* studying, thereby overriding the use of -s on the command line
1597 (see 14 above). This is necessary for one or two tests to keep the output
1598 identical in both cases.
1599
1600 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1601 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1602
1603 22. When an atomic group that contained a capturing parenthesis was
1604 successfully matched, but the branch in which it appeared failed, the
1605 capturing was not being forgotten if a higher numbered group was later
1606 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1607 group 1 to "a", when in fact it should be unset. This applied to multi-
1608 branched capturing and non-capturing groups, repeated or not, and also to
1609 positive assertions (capturing in negative assertions does not happen
1610 in PCRE) and also to nested atomic groups.
1611
1612 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1613 subject after a captured substring, to make it easier to tell which of a
1614 number of identical substrings has been captured.
1615
1616 24. The way atomic groups are processed by pcre_exec() has been changed so that
1617 if they are repeated, backtracking one repetition now resets captured
1618 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1619 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1620 "aaa". Previously, it would have been "a". As part of this code
1621 refactoring, the way recursive calls are handled has also been changed.
1622
1623 25. If an assertion condition captured any substrings, they were not passed
1624 back unless some other capturing happened later. For example, if
1625 (?(?=(a))a) was matched against "a", no capturing was returned.
1626
1627 26. When studying a pattern that contained subroutine calls or assertions,
1628 the code for finding the minimum length of a possible match was handling
1629 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1630 group 1 called group 2 while simultaneously a separate group 2 called group
1631 1). A stack overflow occurred in this case. I have fixed this by limiting
1632 the recursion depth to 10.
1633
1634 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1635 Fortmann. This supports explicit test numbers on the command line, and has
1636 argument validation and error reporting.
1637
1638 28. An instance of \X with an unlimited repeat could fail if at any point the
1639 first character it looked at was a mark character.
1640
1641 29. Some minor code refactoring concerning Unicode properties and scripts
1642 should reduce the stack requirement of match() slightly.
1643
1644 30. Added the '=' option to pcretest to check the setting of unused capturing
1645 slots at the end of the pattern, which are documented as being -1, but are
1646 not included in the return count.
1647
1648 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1649 compiled something random. Now it gives a compile-time error (as does
1650 Perl).
1651
1652 32. A *MARK encountered during the processing of a positive assertion is now
1653 recorded and passed back (compatible with Perl).
1654
1655 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1656 had alternative anchored branches, the search for a second match in a line
1657 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1658 matched the line "0102" twice. The same bug affected patterns that started
1659 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1660 twice.
1661
1662 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1663 does, and because of capturing effects, quantifying parenthesized
1664 assertions may at times be useful. Quantifiers are now allowed for
1665 parenthesized assertions.
1666
1667 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1668
1669 36. \g was being checked for fancy things in a character class, when it should
1670 just be a literal "g".
1671
1672 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1673 appearance of a nested POSIX class supersedes an apparent external class.
1674 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1675 unescaped square brackets may also appear as part of class names. For
1676 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1677 more like Perl. (But see 8.20/1 above.)
1678
1679 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1680 was because it thought it was \N{name}, which is not supported).
1681
1682 39. Add minix to OS list not supporting the -S option in pcretest.
1683
1684 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1685 cannot analyze patterns in sufficient detail to catch mutual recursions
1686 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1687 subgroup is called recursively as a subpattern for a second time at the
1688 same position in the subject string. In previous releases this might have
1689 been caught by the recursion limit, or it might have run out of stack.
1690
1691 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1692 happen only once. PCRE was, however incorrectly giving a compile time error
1693 "recursive call could loop indefinitely" because it cannot analyze the
1694 pattern in sufficient detail. The compile time test no longer happens when
1695 PCRE is compiling a conditional subpattern, but actual runaway loops are
1696 now caught at runtime (see 40 above).
1697
1698 42. It seems that Perl allows any characters other than a closing parenthesis
1699 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1700 has been changed to be the same.
1701
1702 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1703 as not to get warnings when autogen.sh is called. Also changed
1704 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1705
1706 44. To help people who use pcregrep to scan files containing exceedingly long
1707 lines, the following changes have been made:
1708
1709 (a) The default value of the buffer size parameter has been increased from
1710 8K to 20K. (The actual buffer used is three times this size.)
1711
1712 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1713 PCRE is built.
1714
1715 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1716 to be set at run time.
1717
1718 (d) Numerical values in pcregrep options can be followed by K or M, for
1719 example --buffer-size=50K.
1720
1721 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1722 given and the return code is set to 2.
1723
1724 45. Add a pointer to the latest mark to the callout data block.
1725
1726 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1727 partial match of an empty string instead of no match. This was specific to
1728 the use of ".".
1729
1730 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1731 complete match instead of a partial match. This bug was dependent on both
1732 the PCRE_UTF8 and PCRE_DOTALL options being set.
1733
1734 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1735 starting byte set, because \b was not being ignored.
1736
1737
1738 Version 8.12 15-Jan-2011
1739 ------------------------
1740
1741 1. Fixed some typos in the markup of the man pages, and wrote a script that
1742 checks for such things as part of the documentation building process.
1743
1744 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1745 --match-limit and --recursion-limit options (added for 8.11). In
1746 particular, this made one of the standard tests fail. (The integer value
1747 went into the wrong half of a long int.)
1748
1749 3. If the --colour option was given to pcregrep with -v (invert match), it
1750 did strange things, either producing crazy output, or crashing. It should,
1751 of course, ignore a request for colour when reporting lines that do not
1752 match.
1753
1754 4. Another pcregrep bug caused similar problems if --colour was specified with
1755 -M (multiline) and the pattern match finished with a line ending.
1756
1757 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1758 matched in multiline mode, the following line was shown as part of the
1759 match. This seems wrong, so I have changed it.
1760
1761 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1762 the check for further matches in the same line (so they could be coloured)
1763 to overrun the end of the current line. If another match was found, it was
1764 incorrectly shown (and then shown again when found in the next line).
1765
1766 7. If pcregrep was compiled under Windows, there was a reference to the
1767 function pcregrep_exit() before it was defined. I am assuming this was
1768 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1769 reported by a user. I've moved the definition above the reference.
1770
1771
1772 Version 8.11 10-Dec-2010
1773 ------------------------
1774
1775 1. (*THEN) was not working properly if there were untried alternatives prior
1776 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1777 backtracked to try for "b" instead of moving to the next alternative branch
1778 at the same level (in this case, to look for "c"). The Perl documentation
1779 is clear that when (*THEN) is backtracked onto, it goes to the "next
1780 alternative in the innermost enclosing group".
1781
1782 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1783 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1784 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1785 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1786 (*THEN).
1787
1788 3. If \s appeared in a character class, it removed the VT character from
1789 the class, even if it had been included by some previous item, for example
1790 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1791 of \s, but is part of the POSIX "space" class.)
1792
1793 4. A partial match never returns an empty string (because you can always
1794 match an empty string at the end of the subject); however the checking for
1795 an empty string was starting at the "start of match" point. This has been
1796 changed to the "earliest inspected character" point, because the returned
1797 data for a partial match starts at this character. This means that, for
1798 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1799 (previously it gave "no match").
1800
1801 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1802 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1803 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1804 has an implication that the given string is incomplete (because a partial
1805 match is preferred over a full match). For this reason, these items now
1806 give a partial match in this situation. [Aside: previously, the one case
1807 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1808 match rather than a full match, which was wrong by the old rules, but is
1809 now correct.]
1810
1811 6. There was a bug in the handling of #-introduced comments, recognized when
1812 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1813 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1814 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1815 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1816 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1817 places in pcre_compile().
1818
1819 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1820 comments when looking ahead for named forward references to subpatterns,
1821 the only newline sequence it recognized was NL. It now handles newlines
1822 according to the set newline convention.
1823
1824 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1825 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1826 cater for a lack of strerror(). These oversights have been fixed.
1827
1828 9. Added --match-limit and --recursion-limit to pcregrep.
1829
1830 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1831
1832 11. When the -o option was used, pcregrep was setting a return code of 1, even
1833 when matches were found, and --line-buffered was not being honoured.
1834
1835 12. Added an optional parentheses number to the -o and --only-matching options
1836 of pcregrep.
1837
1838 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1839 can match an empty string. The code to do it in pcretest and pcredemo
1840 needed fixing:
1841
1842 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1843 only one byte after an empty string match just before CRLF (this case
1844 just got forgotten; "any" and "anycrlf" were OK).
1845
1846 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1847 mode when an empty string match preceded an ASCII character followed by
1848 a non-ASCII character. (The code for advancing by one character rather
1849 than one byte was nonsense.)
1850
1851 (c) The pcredemo.c sample program did not have any code at all to handle
1852 the cases when CRLF is a valid newline sequence.
1853
1854 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1855 as a starting offset was within the subject string. There is now a new
1856 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1857 negative or greater than the length of the string. In order to test this,
1858 pcretest is extended to allow the setting of negative starting offsets.
1859
1860 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1861 starting offset points to the beginning of a UTF-8 character was
1862 unnecessarily clumsy. I tidied it up.
1863
1864 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1865 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1866
1867 17. Nobody had reported that the --include_dir option, which was added in
1868 release 7.7 should have been called --include-dir (hyphen, not underscore)
1869 for compatibility with GNU grep. I have changed it to --include-dir, but
1870 left --include_dir as an undocumented synonym, and the same for
1871 --exclude-dir, though that is not available in GNU grep, at least as of
1872 release 2.5.4.
1873
1874 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1875 characters from a string of bytes) have been redefined so as not to use
1876 loops, in order to improve performance in some environments. At the same
1877 time, I abstracted some of the common code into auxiliary macros to save
1878 repetition (this should not affect the compiled code).
1879
1880 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1881 compile-time error is now given if \c is not followed by an ASCII
1882 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1883 different, and any byte value is allowed.)
1884
1885 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1886 START_OPTIMIZE option, which is now allowed at compile time - but just
1887 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1888 to pcregrep and other applications that have no direct access to PCRE
1889 options. The new /Y option in pcretest sets this option when calling
1890 pcre_compile().
1891
1892 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1893 back references. Groups containing recursive back references were forced to
1894 be atomic by that change, but in the case of named groups, the amount of
1895 memory required was incorrectly computed, leading to "Failed: internal
1896 error: code overflow". This has been fixed.
1897
1898 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1899 pcretest.c, to avoid build problems in some Borland environments.
1900
1901
1902 Version 8.10 25-Jun-2010
1903 ------------------------
1904
1905 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1906 THEN.
1907
1908 2. (*ACCEPT) was not working when inside an atomic group.
1909
1910 3. Inside a character class, \B is treated as a literal by default, but
1911 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1912 causes the error). The code is unchanged, but I tidied the documentation.
1913
1914 4. Inside a character class, PCRE always treated \R and \X as literals,
1915 whereas Perl faults them if its -w option is set. I have changed PCRE so
1916 that it faults them when PCRE_EXTRA is set.
1917
1918 5. Added support for \N, which always matches any character other than
1919 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1920
1921 6. When compiling pcregrep with newer versions of gcc which may have
1922 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1923 declared with attribute warn_unused_result" were given. Just casting the
1924 result to (void) does not stop the warnings; a more elaborate fudge is
1925 needed. I've used a macro to implement this.
1926
1927 7. Minor change to pcretest.c to avoid a compiler warning.
1928
1929 8. Added four artifical Unicode properties to help with an option to make
1930 \s etc use properties (see next item). The new properties are: Xan
1931 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1932
1933 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1934 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1935 this option. Modified pcretest to add /W to test this facility. Added
1936 REG_UCP to make it available via the POSIX interface.
1937
1938 10. Added --line-buffered to pcregrep.
1939
1940 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1941 studied, and the match started with a letter with a code point greater than
1942 127 whose first byte was different to the first byte of the other case of
1943 the letter, the other case of this starting letter was not recognized
1944 (#976).
1945
1946 12. If a pattern that was studied started with a repeated Unicode property
1947 test, for example, \p{Nd}+, there was the theoretical possibility of
1948 setting up an incorrect bitmap of starting bytes, but fortunately it could
1949 not have actually happened in practice until change 8 above was made (it
1950 added property types that matched character-matching opcodes).
1951
1952 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1953 possible starting bytes for non-anchored patterns.
1954
1955 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1956 \R, and also a number of cases that involve Unicode properties, both
1957 explicit and implicit when PCRE_UCP is set.
1958
1959 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1960 input, it could crash or give wrong results if characters with values
1961 greater than 0xc0 were present in the subject string. (Detail: it assumed
1962 UTF-8 input when processing these items.)
1963
1964 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1965 size_t is 64-bit (#991).
1966
1967 17. Added a check for running out of memory when PCRE is compiled with
1968 --disable-stack-for-recursion (#990).
1969
1970 18. If the last data line in a file for pcretest does not have a newline on
1971 the end, a newline was missing in the output.
1972
1973 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1974 less than 128) in its various bitmaps. However, there is a facility for
1975 generating tables according to the current locale when PCRE is compiled. It
1976 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1977 characters, are recognized by isspace() and therefore were getting set in
1978 these tables, and indeed these tables seem to approximate to ISO 8859. This
1979 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1980 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1981 which of course cannot start UTF-8 characters. I have changed the code so
1982 that only real ASCII characters (less than 128) and the correct starting
1983 bytes for UTF-8 encodings are set for characters greater than 127 when in
1984 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1985 altogether.)
1986
1987 20. Added the /T option to pcretest so as to be able to run tests with non-
1988 standard character tables, thus making it possible to include the tests
1989 used for 19 above in the standard set of tests.
1990
1991 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1992 reference to a subpattern the other side of a comment that contains an
1993 opening parenthesis caused either an internal compiling error, or a
1994 reference to the wrong subpattern.
1995
1996
1997 Version 8.02 19-Mar-2010
1998 ------------------------
1999
2000 1. The Unicode data tables have been updated to Unicode 5.2.0.
2001
2002 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2003 configured.
2004
2005 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2006 original author of that file, following a query about its status.
2007
2008 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2009 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2010
2011 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2012 quantifier applied to a forward-referencing subroutine call, could compile
2013 incorrect code or give the error "internal error: previously-checked
2014 referenced subpattern not found".
2015
2016 6. Both MS Visual Studio and Symbian OS have problems with initializing
2017 variables to point to external functions. For these systems, therefore,
2018 pcre_malloc etc. are now initialized to local functions that call the
2019 relevant global functions.
2020
2021 7. There were two entries missing in the vectors called coptable and poptable
2022 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2023 I've fixed the data, and added a kludgy way of testing at compile time that
2024 the lengths are correct (equal to the number of opcodes).
2025
2026 8. Following on from 7, I added a similar kludge to check the length of the
2027 eint vector in pcreposix.c.
2028
2029 9. Error texts for pcre_compile() are held as one long string to avoid too
2030 much relocation at load time. To find a text, the string is searched,
2031 counting zeros. There was no check for running off the end of the string,
2032 which could happen if a new error number was added without updating the
2033 string.
2034
2035 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2036
2037 11. \K was not working if it appeared in an atomic group or in a group that
2038 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2039 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2040 the assertion is positive, but not if it is negative.
2041
2042 12. Change 11 fortuitously reduced the size of the stack frame used in the
2043 "match()" function of pcre_exec.c by one pointer. Forthcoming
2044 implementation of support for (*MARK) will need an extra pointer on the
2045 stack; I have reserved it now, so that the stack frame size does not
2046 decrease.
2047
2048 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2049 item in branch that calls a recursion is a subroutine call - as in the
2050 second branch in the above example - was incorrectly given the compile-
2051 time error "recursive call could loop indefinitely" because pcre_compile()
2052 was not correctly checking the subroutine for matching a non-empty string.
2053
2054 14. The checks for overrunning compiling workspace could trigger after an
2055 overrun had occurred. This is a "should never occur" error, but it can be
2056 triggered by pathological patterns such as hundreds of nested parentheses.
2057 The checks now trigger 100 bytes before the end of the workspace.
2058
2059 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2060
2061
2062 Version 8.01 19-Jan-2010
2063 ------------------------
2064
2065 1. If a pattern contained a conditional subpattern with only one branch (in
2066 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2067 computed the wrong minimum data length (which is of course zero for such
2068 subpatterns). This could cause incorrect "no match" results.
2069
2070 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2071 the pattern is reset in the first branch, pcre_compile() failed with
2072 "internal error: code overflow at offset...". This happened only when
2073 the reset was to the original external option setting. (An optimization
2074 abstracts leading options settings into an external setting, which was the
2075 cause of this.)
2076
2077 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2078 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2079 assertion pattern did not match (meaning that the assertion was true), it
2080 was incorrectly treated as false if the SKIP had been reached during the
2081 matching. This also applied to assertions used as conditions.
2082
2083 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2084 assertion subpattern, including such a pattern used as a condition,
2085 unpredictable results occurred, instead of the error return
2086 PCRE_ERROR_DFA_UITEM.
2087
2088 5. The C++ GlobalReplace function was not working like Perl for the special
2089 situation when an empty string is matched. It now does the fancy magic
2090 stuff that is necessary.
2091
2092 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2093 removed. (These were left over from very, very early versions of PCRE.)
2094
2095 7. Some cosmetic changes to the code to make life easier when compiling it
2096 as part of something else:
2097
2098 (a) Change DEBUG to PCRE_DEBUG.
2099
2100 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2101 called "current" as "current_branch", to prevent a collision with the
2102 Linux macro when compiled as a kernel module.
2103
2104 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2105 prevent a collision with the Linux macro when compiled as a kernel
2106 module.
2107
2108 8. In pcre_compile() there are some checks for integer overflows that used to
2109 cast potentially large values to (double). This has been changed to that
2110 when building, a check for int64_t is made, and if it is found, it is used
2111 instead, thus avoiding the use of floating point arithmetic. (There is no
2112 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2113 double.
2114
2115 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2116 2005 (difference between two addresses compared to an unsigned value).
2117
2118 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2119 custom one, because of the following reported problem in Windows:
2120
2121 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2122 under Win32.
2123 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2124 therefore missing the function definition.
2125 - The compiler thus generates a "C" signature for the test function.
2126 - The linker fails to find the "C" function.
2127 - PCRE fails to configure if asked to do so against libbz2.
2128
2129 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2130 messages were output:
2131
2132 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2133 rerunning libtoolize, to keep the correct libtool macros in-tree.
2134 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2135
2136 I have done both of these things.
2137
2138 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2139 most of the time, it *can* run out if it is given a pattern that contains a
2140 runaway infinite recursion. I updated the discussion in the pcrestack man
2141 page.
2142
2143 13. Now that we have gone to the x.xx style of version numbers, the minor
2144 version may start with zero. Using 08 or 09 is a bad idea because users
2145 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2146 interpreted as invalid octal numbers. I've updated the previous comment in
2147 configure.ac, and also added a check that gives an error if 08 or 09 are
2148 used.
2149
2150 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2151 causing partial matching to fail when the end of the subject matched \W
2152 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2153
2154 15. There were some discrepancies between the declarations in pcre_internal.h
2155 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2156 their definitions. The declarations used "const uschar *" and the
2157 definitions used USPTR. Even though USPTR is normally defined as "const
2158 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2159 reported that: "This difference in casting confuses some C++ compilers, for
2160 example, SunCC recognizes above declarations as different functions and
2161 generates broken code for hbpcre." I have changed the declarations to use
2162 USPTR.
2163
2164 16. GNU libtool is named differently on some systems. The autogen.sh script now
2165 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2166 (FreeBSD).
2167
2168 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2169 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2170 comment: "Figure out how to create a longlong from a string: strtoll and
2171 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2172 instance, but it only takes 2 args instead of 3!"
2173
2174 18. A subtle bug concerned with back references has been fixed by a change of
2175 specification, with a corresponding code fix. A pattern such as
2176 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2177 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2178 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2179 same bug. Such groups have to be quantified to be useful, or contained
2180 inside another quantified group. (If there's no repetition, the reference
2181 can never match.) The problem arises because, having left the group and
2182 moved on to the rest of the pattern, a later failure that backtracks into
2183 the group uses the captured value from the final iteration of the group
2184 rather than the correct earlier one. I have fixed this in PCRE by forcing
2185 any group that contains a reference to itself to be an atomic group; that
2186 is, there cannot be any backtracking into it once it has completed. This is
2187 similar to recursive and subroutine calls.
2188
2189
2190 Version 8.00 19-Oct-09
2191 ----------------------
2192
2193 1. The table for translating pcre_compile() error codes into POSIX error codes
2194 was out-of-date, and there was no check on the pcre_compile() error code
2195 being within the table. This could lead to an OK return being given in
2196 error.
2197
2198 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2199 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2200 in a Windows environment.
2201
2202 3. The pcregrep --count option prints the count for each file even when it is
2203 zero, as does GNU grep. However, pcregrep was also printing all files when
2204 --files-with-matches was added. Now, when both options are given, it prints
2205 counts only for those files that have at least one match. (GNU grep just
2206 prints the file name in this circumstance, but including the count seems
2207 more useful - otherwise, why use --count?) Also ensured that the
2208 combination -clh just lists non-zero counts, with no names.
2209
2210 4. The long form of the pcregrep -F option was incorrectly implemented as
2211 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2212 but it seems right to fix it, and I didn't think it was worth preserving
2213 the old behaviour.
2214
2215 5. The command line items --regex=pattern and --regexp=pattern were not
2216 recognized by pcregrep, which required --regex pattern or --regexp pattern
2217 (with a space rather than an '='). The man page documented the '=' forms,
2218 which are compatible with GNU grep; these now work.
2219
2220 6. No libpcreposix.pc file was created for pkg-config; there was just
2221 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2222
2223 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2224 when UCP support is not needed, by modifying the Python script that
2225 generates it from Unicode data files. This should not matter if the module
2226 is correctly used as a library, but I received one complaint about 50K of
2227 unwanted data. My guess is that the person linked everything into his
2228 program rather than using a library. Anyway, it does no harm.
2229
2230 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2231 was a minimum greater than 1 for a wide character in a possessive
2232 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2233 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2234 character. Chaos in the form of incorrect output or a compiling loop could
2235 result.
2236
2237 9. The restrictions on what a pattern can contain when partial matching is
2238 requested for pcre_exec() have been removed. All patterns can now be
2239 partially matched by this function. In addition, if there are at least two
2240 slots in the offset vector, the offset of the earliest inspected character
2241 for the match and the offset of the end of the subject are set in them when
2242 PCRE_ERROR_PARTIAL is returned.
2243
2244 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2245 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2246 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2247 and may be more useful for multi-segment matching.
2248
2249 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2250 used to be given if ever the end of the subject was reached; now it is
2251 given only if matching could not proceed because another character was
2252 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2253 string "Z", which now yields "no match" instead of "partial match". In the
2254 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2255 final character ended with (*FAIL).
2256
2257 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2258 if the pattern had a "must contain" character that was already found in the
2259 earlier partial match, unless partial matching was again requested. For
2260 example, with the pattern /dog.(body)?/, the "must contain" character is
2261 "g". If the first part-match was for the string "dog", restarting with
2262 "sbody" failed. This bug has been fixed.
2263
2264 13. The string returned by pcre_dfa_exec() after a partial match has been
2265 changed so that it starts at the first inspected character rather than the
2266 first character of the match. This makes a difference only if the pattern
2267 starts with a lookbehind assertion or \b or \B (\K is not supported by
2268 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2269 matching functions compatible, and I think it's the right thing to do.
2270
2271 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2272 so that the demonstration program is easily available in environments where
2273 PCRE has not been installed from source.
2274
2275 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2276 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2277 library.
2278
2279 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2280 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2281 is not the first non-POSIX option to be added. Clearly some people find
2282 these options useful.
2283
2284 17. If a caller to the POSIX matching function regexec() passes a non-zero
2285 value for nmatch with a NULL value for pmatch, the value of
2286 nmatch is forced to zero.
2287
2288 18. RunGrepTest did not have a test for the availability of the -u option of
2289 the diff command, as RunTest does. It now checks in the same way as
2290 RunTest, and also checks for the -b option.
2291
2292 19. If an odd number of negated classes containing just a single character
2293 interposed, within parentheses, between a forward reference to a named
2294 subpattern and the definition of the subpattern, compilation crashed with
2295 an internal error, complaining that it could not find the referenced
2296 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2297 [The bug was that it was starting one character too far in when skipping
2298 over the character class, thus treating the ] as data rather than
2299 terminating the class. This meant it could skip too much.]
2300
2301 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2302 /g option in pcretest when the pattern contains \K, which makes it possible
2303 to have an empty string match not at the start, even when the pattern is
2304 anchored. Updated pcretest and pcredemo to use this option.
2305
2306 21. If the maximum number of capturing subpatterns in a recursion was greater
2307 than the maximum at the outer level, the higher number was returned, but
2308 with unset values at the outer level. The correct (outer level) value is
2309 now given.
2310
2311 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2312 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2313 make it do so. The string so far is captured, making this feature
2314 compatible with Perl.
2315
2316 23. The tests have been re-organized, adding tests 11 and 12, to make it
2317 possible to check the Perl 5.10 features against Perl 5.10.
2318
2319 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2320 pattern matches a fixed length string. PCRE did not allow this; now it
2321 does. Neither allows recursion.
2322
2323 25. I finally figured out how to implement a request to provide the minimum
2324 length of subject string that was needed in order to match a given pattern.
2325 (It was back references and recursion that I had previously got hung up
2326 on.) This code has now been added to pcre_study(); it finds a lower bound
2327 to the length of subject needed. It is not necessarily the greatest lower
2328 bound, but using it to avoid searching strings that are too short does give
2329 some useful speed-ups. The value is available to calling programs via
2330 pcre_fullinfo().
2331
2332 26. While implementing 25, I discovered to my embarrassment that pcretest had
2333 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2334 study optimizations had never been tested with that matching function.
2335 Oops. What is worse, even when it was passed study data, there was a bug in
2336 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2337 were also very few tests of studied patterns with pcre_dfa_exec().
2338
2339 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2340 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2341 on the other side of the coin, they are no longer allowed to have different
2342 names, because these cannot be distinguished in PCRE, and this has caused
2343 confusion. (This is a difference from Perl.)
2344
2345 28. When duplicate subpattern names are present (necessarily with different
2346 numbers, as required by 27 above), and a test is made by name in a
2347 conditional pattern, either for a subpattern having been matched, or for
2348 recursion in such a pattern, all the associated numbered subpatterns are
2349 tested, and the overall condition is true if the condition is true for any
2350 one of them. This is the way Perl works, and is also more like the way
2351 testing by number works.
2352
2353
2354 Version 7.9 11-Apr-09
2355 ---------------------
2356
2357 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2358 (pcretest), all targets were linked against these libraries. This included
2359 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2360 libraries. This caused unwanted dependencies to be created. This problem
2361 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2362 pcretest is linked with readline.
2363
2364 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2365 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2366 moved outside it again, because FALSE and TRUE are already defined in AIX,
2367 but BOOL is not.
2368
2369 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2370 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2371
2372 4. The pcregrep documentation said spaces were inserted as well as colons (or
2373 hyphens) following file names and line numbers when outputting matching
2374 lines. This is not true; no spaces are inserted. I have also clarified the
2375 wording for the --colour (or --color) option.
2376
2377 5. In pcregrep, when --colour was used with -o, the list of matching strings
2378 was not coloured; this is different to GNU grep, so I have changed it to be
2379 the same.
2380
2381 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2382 each matching line was coloured. Now it goes on to look for further matches
2383 of any of the test patterns, which is the same behaviour as GNU grep.
2384
2385 7. A pattern that could match an empty string could cause pcregrep to loop; it
2386 doesn't make sense to accept an empty string match in pcregrep, so I have
2387 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2388 seems to be how GNU grep behaves. [But see later change 40 for release
2389 8.33.]
2390
2391 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2392 start or after a newline", because the conditional assertion was not being
2393 correctly handled. The rule now is that both the assertion and what follows
2394 in the first alternative must satisfy the test.
2395
2396 9. If auto-callout was enabled in a pattern with a conditional group whose
2397 condition was an assertion, PCRE could crash during matching, both with
2398 pcre_exec() and pcre_dfa_exec().
2399
2400 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2401 used for matching.
2402
2403 11. Unicode property support in character classes was not working for
2404 characters (bytes) greater than 127 when not in UTF-8 mode.
2405
2406 12. Added the -M command line option to pcretest.
2407
2408 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2409
2410 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2411
2412 16. Added comments and documentation about mis-use of no_arg in the C++
2413 wrapper.
2414
2415 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2416 from Martin Jerabek that uses macro names for all relevant character and
2417 string constants.
2418
2419 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2420 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2421 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2422 these, but not everybody uses configure.
2423
2424 19. A conditional group that had only one branch was not being correctly
2425 recognized as an item that could match an empty string. This meant that an
2426 enclosing group might also not be so recognized, causing infinite looping
2427 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2428 with the subject "ab", where knowledge that the repeated group can match
2429 nothing is needed in order to break the loop.
2430
2431 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2432 exec(), but without supplying a callout function, matching went wrong.
2433
2434 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2435 leak if the size of the offset vector was greater than 30. When the vector
2436 is smaller, the saved offsets during recursion go onto a local stack
2437 vector, but for larger vectors malloc() is used. It was failing to free
2438 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2439 error, in fact).
2440
2441 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2442 heapframe that is used only when UTF-8 support is enabled. This caused no
2443 problem, but was untidy.
2444
2445 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2446 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2447 included within another project.
2448
2449 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2450 slightly modified by me:
2451
2452 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2453 not building pcregrep.
2454
2455 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2456 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2457
2458 25. Forward references, both numeric and by name, in patterns that made use of
2459 duplicate group numbers, could behave incorrectly or give incorrect errors,
2460 because when scanning forward to find the reference group, PCRE was not
2461 taking into account the duplicate group numbers. A pattern such as
2462 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2463
2464 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2465 the feature of a custom pointer more persuasive (as requested by a user).
2466
2467 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2468 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2469 pre-defined.
2470
2471 28. Added support for (*UTF8) at the start of a pattern.
2472
2473 29. Arrange for flags added by the "release type" setting in CMake to be shown
2474 in the configuration summary.
2475
2476
2477 Version 7.8 05-Sep-08
2478 ---------------------
2479
2480 1. Replaced UCP searching code with optimized version as implemented for Ad
2481 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2482 stage table and inline lookup instead of a function, giving speed ups of 2
2483 to 5 times on some simple patterns that I tested. Permission was given to
2484 distribute the MultiStage2.py script that generates the tables (it's not in
2485 the tarball, but is in the Subversion repository).
2486
2487 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2488 scripts.
2489
2490 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2491 a group with a zero qualifier. The result of the study could be incorrect,
2492 or the function might crash, depending on the pattern.
2493
2494 4. Caseless matching was not working for non-ASCII characters in back
2495 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2496 It now works when Unicode Property Support is available.
2497
2498 5. In pcretest, an escape such as \x{de} in the data was always generating
2499 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2500 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2501 truncation.
2502
2503 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2504
2505 7. Added two (int) casts to pcregrep when printing the difference of two
2506 pointers, in case they are 64-bit values.
2507
2508 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2509 test 2 if it fails.
2510
2511 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2512 and a #define of that name to empty if it is not externally set. This is to
2513 allow users of MSVC to set it if necessary.
2514
2515 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2516 the convenience functions in the pcre_get.c source file.
2517
2518 11. An option change at the start of a pattern that had top-level alternatives
2519 could cause overwriting and/or a crash. This command provoked a crash in
2520 some environments:
2521
2522 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2523
2524 This potential security problem was recorded as CVE-2008-2371.
2525
2526 12. For a pattern where the match had to start at the beginning or immediately
2527 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2528 pcre_dfa_exec() could read past the end of the passed subject if there was
2529 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2530 pcretest so that it places the subject at the end of its malloc-ed buffer.
2531
2532 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2533 exec() might read past the end of the data buffer in UTF-8 mode.
2534
2535 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2536 the data contained the byte 0x85 as part of a UTF-8 character within its
2537 first line. This applied both to normal and DFA matching.
2538
2539 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2540 /^[^d]*?$/8 failed to match "abc".
2541
2542 16. Added a missing copyright notice to pcrecpp_internal.h.
2543
2544 17. Make it more clear in the documentation that values returned from
2545 pcre_exec() in ovector are byte offsets, not character counts.
2546
2547 18. Tidied a few places to stop certain compilers from issuing warnings.
2548
2549 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2550 supplied by Stefan Weber. I made a further small update for 7.8 because
2551 there is a change of source arrangements: the pcre_searchfuncs.c module is
2552 replaced by pcre_ucd.c.
2553
2554
2555 Version 7.7 07-May-08
2556 ---------------------
2557
2558 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2559 a string to a long long, pretend we don't even have a long long." This is
2560 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2561
2562 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2563 pre-7.6 versions, which defined a global no_arg variable instead of putting
2564 it in the RE class. (See also #8 below.)
2565
2566 3. Remove a line of dead code, identified by coverity and reported by Nuno
2567 Lopes.
2568
2569 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2570
2571 (1) The include/exclude patterns were being applied to the whole pathnames
2572 of files, instead of just to the final components.
2573
2574 (2) If there was more than one level of directory, the subdirectories were
2575 skipped unless they satisfied the include/exclude conditions. This is
2576 inconsistent with GNU grep (and could even be seen as contrary to the
2577 pcregrep specification - which I improved to make it absolutely clear).
2578 The action now is always to scan all levels of directory, and just
2579 apply the include/exclude patterns to regular files.
2580
2581 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2582 --exclude_dir in the tests to avoid scanning .svn directories.
2583
2584 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2585 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2586 doesn't support NULs in patterns.
2587
2588 7. Added some missing "const"s to declarations of static tables in
2589 pcre_compile.c and pcre_dfa_exec.c.
2590
2591 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2592 caused by fix #2 above. (Subsequently also a second patch to fix the
2593 first patch. And a third patch - this was a messy problem.)
2594
2595 9. Applied Craig's patch to remove the use of push_back().
2596
2597 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2598 matching function regexec().
2599
2600 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2601 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2602 references. PCRE supports relative numbers with this syntax (I don't think
2603 Oniguruma does).
2604
2605 12. Previously, a group with a zero repeat such as (...){0} was completely
2606 omitted from the compiled regex. However, this means that if the group
2607 was called as a subroutine from elsewhere in the pattern, things went wrong
2608 (an internal error was given). Such groups are now left in the compiled
2609 pattern, with a new opcode that causes them to be skipped at execution
2610 time.
2611
2612 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2613 to the way PCRE behaves:
2614
2615 (a) A lone ] character is dis-allowed (Perl treats it as data).
2616
2617 (b) A back reference to an unmatched subpattern matches an empty string
2618 (Perl fails the current match path).
2619
2620 (c) A data ] in a character class must be notated as \] because if the
2621 first data character in a class is ], it defines an empty class. (In
2622 Perl it is not possible to have an empty class.) The empty class []
2623 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2624 The negative empty class [^] matches any one character, independently
2625 of the DOTALL setting.
2626
2627 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2628 non-existent subpattern following a character class starting with ']' and
2629 containing () gave an internal compiling error instead of "reference to
2630 non-existent subpattern". Fortunately, when the pattern did exist, the
2631 compiled code was correct. (When scanning forwards to check for the
2632 existence of the subpattern, it was treating the data ']' as terminating
2633 the class, so got the count wrong. When actually compiling, the reference
2634 was subsequently set up correctly.)
2635
2636 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2637 it was being rejected as not supported by pcre_dfa_exec(), even though
2638 other assertions are supported. I have made pcre_dfa_exec() support
2639 (*FAIL).
2640
2641 16. The implementation of 13c above involved the invention of a new opcode,
2642 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2643 cannot be changed at match time, I realized I could make a small
2644 improvement to matching performance by compiling OP_ALLANY instead of
2645 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2646 on the OP_ANY path.
2647
2648 17. Compiling pcretest on Windows with readline support failed without the
2649 following two fixes: (1) Make the unistd.h include conditional on
2650 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2651
2652 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2653 ncurses library to be included for pcretest when ReadLine support is
2654 requested, but also to allow for it to be overridden. This patch came from
2655 Daniel Bergström.
2656
2657 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2658 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2659 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2660 spotting this.
2661
2662
2663 Version 7.6 28-Jan-08
2664 ---------------------
2665
2666 1. A character class containing a very large number of characters with
2667 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2668 overflow.
2669
2670 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2671 HAVE_LONG_LONG is not defined.
2672
2673 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2674 bring them up to date and include new features. This patch includes:
2675
2676 - Fixed PH's badly added libz and libbz2 support.
2677 - Fixed a problem with static linking.
2678 - Added pcredemo. [But later removed - see 7 below.]
2679 - Fixed dftables problem and added an option.
2680 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2681 HAVE_LONG_LONG.
2682 - Added readline support for pcretest.
2683 - Added an listing of the option settings after cmake has run.
2684
2685 4. A user submitted a patch to Makefile that makes it easy to create
2686 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2687 Makefile.am that cause it to include this special target, without
2688 affecting anything else. Note that the same mingw target plus all
2689 the other distribution libraries and programs are now supported
2690 when configuring with CMake (see 6 below) instead of with
2691 Configure/Make.
2692
2693 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2694 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2695 exported in the Windows port". It has not yet been confirmed that the patch
2696 solves the problem, but it does no harm.
2697
2698 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2699 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2700 with CMake, and also correct the comment about stack recursion.
2701
2702 7. Remove the automatic building of pcredemo from the ./configure system and
2703 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2704 of a program that users should build themselves after PCRE is installed, so
2705 building it automatically is not really right. What is more, it gave
2706 trouble in some build environments.
2707
2708 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2709
2710
2711 Version 7.5 10-Jan-08
2712 ---------------------
2713
2714 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2715 values in parens when parsing an RE using the C++ wrapper."
2716
2717 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2718 Characters greater than 255 were excluded from the class instead of being
2719 included.
2720
2721 3. The same bug as (2) above applied to negated POSIX classes such as
2722 [:^space:].
2723
2724 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2725 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2726 I have changed it.
2727
2728 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2729 first named subpattern) and a construct such as (?&a) would reference the
2730 first named subpattern whose name started with "a" (in other words, the
2731 length check was missing). Both these problems are fixed. "Subpattern name
2732 expected" is now given for (?&) (a zero-length name), and this patch also
2733 makes it give the same error for \k'' (previously it complained that that
2734 was a reference to a non-existent subpattern).
2735
2736 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2737 this is right because (?- can be followed by option settings as well as by
2738 digits. I have, however, made the messages clearer.
2739
2740 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2741 than the number used in the conditional) now cause a compile-time error.
2742 This is actually not compatible with Perl, which accepts such patterns, but
2743 treats the conditional as always being FALSE (as PCRE used to), but it
2744 seems to me that giving a diagnostic is better.
2745
2746 8. Change "alphameric" to the more common word "alphanumeric" in comments
2747 and messages.
2748
2749 9. Fix two occurrences of "backslash" in comments that should have been
2750 "backspace".
2751
2752 10. Remove two redundant lines of code that can never be obeyed (their function
2753 was moved elsewhere).
2754
2755 11. The program that makes PCRE's Unicode character property table had a bug
2756 which caused it to generate incorrect table entries for sequences of
2757 characters that have the same character type, but are in different scripts.
2758 It amalgamated them into a single range, with the script of the first of
2759 them. In other words, some characters were in the wrong script. There were
2760 thirteen such cases, affecting characters in the following ranges:
2761
2762 U+002b0 - U+002c1
2763 U+0060c - U+0060d
2764 U+0061e - U+00612
2765 U+0064b - U+0065e
2766 U+0074d - U+0076d
2767 U+01800 - U+01805
2768 U+01d00 - U+01d77
2769 U+01d9b - U+01dbf
2770 U+0200b - U+0200f
2771 U+030fc - U+030fe
2772 U+03260 - U+0327f
2773 U+0fb46 - U+0fbb1
2774 U+10450 - U+1049d
2775
2776 12. The -o option (show only the matching part of a line) for pcregrep was not
2777 compatible with GNU grep in that, if there was more than one match in a
2778 line, it showed only the first of them. It now behaves in the same way as
2779 GNU grep.
2780
2781 13. If the -o and -v options were combined for pcregrep, it printed a blank
2782 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2783 does the same. The return code can be used to tell if there were any
2784 non-matching lines.
2785
2786 14. Added --file-offsets and --line-offsets to pcregrep.
2787
2788 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2789 infinitely looping recursion. The bug was that positive lookaheads were not
2790 being skipped when checking for a possible empty match (negative lookaheads
2791 and both kinds of lookbehind were skipped).
2792
2793 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2794 inclusion of <windows.h> to before rather than after the definition of
2795 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2796
2797 17. Specifying a possessive quantifier with a specific limit for a Unicode
2798 character property caused pcre_compile() to compile bad code, which led at
2799 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2800 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2801 caused the error; without that there was no problem.
2802
2803 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2804
2805 19. Added --enable-pcretest-libreadline.
2806
2807 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2808 RE::GlobalReplace(). As a result, the number of replacements returned was
2809 double what it should be. I removed one of the increments, but Craig sent a
2810 later patch that removed the other one (the right fix) and added unit tests
2811 that check the return values (which was not done before).
2812
2813 21. Several CMake things:
2814
2815 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2816 the names libpcre and libpcreposix, not just pcre and pcreposix.
2817
2818 (2) The above change means that pcretest and pcregrep are now correctly
2819 linked with the newly-built libraries, not previously installed ones.
2820
2821 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2822
2823 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2824 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2825 UTF-8 newline character). The key issue is that the pattern starts .*;
2826 this means that the match must be either at the beginning, or after a
2827 newline. The bug was in the code for advancing after a failed match and
2828 checking that the new position followed a newline. It was not taking
2829 account of UTF-8 characters correctly.
2830
2831 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2832 character classes. PCRE was not treating the sequence [:...:] as a
2833 character class unless the ... were all letters. Perl, however, seems to
2834 allow any characters between [: and :], though of course it rejects as
2835 unknown any "names" that contain non-letters, because all the known class
2836 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2837 for example, whereas PCRE did not - it did not recognize a POSIX character
2838 class. This seemed a bit dangerous, so the code has been changed to be
2839 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2840 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2841 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2842 Perl does, and where it didn't before.
2843
2844 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2845 Windows environments %n is disabled by default.
2846
2847
2848 Version 7.4 21-Sep-07
2849 ---------------------
2850
2851 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2852 means that a class such as [\s] counted as "explicit reference to CR or
2853 LF". That isn't really right - the whole point of the change was to try to
2854 help when there was an actual mention of one of the two characters. So now
2855 the change happens only if \r or \n (or a literal CR or LF) character is
2856 encountered.
2857
2858 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2859 of both had grown to the point where there were only 3 bits left.
2860 Fortunately, there was spare space in the data structure, and so I have
2861 moved the internal flags into a new 16-bit field to free up more option
2862 bits.
2863
2864 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2865 but did not set the internal JCHANGED flag - either of these is enough to
2866 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2867 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2868 start sets both bits.
2869
2870 4. Added options (at build time, compile time, exec time) to change \R from
2871 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2872
2873 5. doc/pcresyntax.html was missing from the distribution.
2874
2875 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2876 compatibility, even though it is no longer used.
2877
2878 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2879 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2880 windows.h file is present (where different names are used). [This was
2881 reversed later after testing - see 16 below.]
2882
2883 8. Changed all #include <config.h> to #include "config.h". There were also
2884 some further <pcre.h> cases that I changed to "pcre.h".
2885
2886 9. When pcregrep was used with the --colour option, it missed the line ending
2887 sequence off the lines that it output.
2888
2889 10. It was pointed out to me that arrays of string pointers cause lots of
2890 relocations when a shared library is dynamically loaded. A technique of
2891 using a single long string with a table of offsets can drastically reduce
2892 these. I have refactored PCRE in four places to do this. The result is
2893 dramatic:
2894
2895 Originally: 290
2896 After changing UCP table: 187
2897 After changing error message table: 43
2898 After changing table of "verbs" 36
2899 After changing table of Posix names 22
2900
2901 Thanks to the folks working on Gregex for glib for this insight.
2902
2903 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2904 unicode-properties was also set.
2905
2906 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2907
2908 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2909 checked only for CRLF.
2910
2911 14. Added casts to pcretest.c to avoid compiler warnings.
2912
2913 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2914
2915 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2916 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2917 entirely. This removes changes made in 7 above.
2918
2919 17. The CMake files have been updated, and there is now more information about
2920 building with CMake in the NON-UNIX-USE document.
2921
2922
2923 Version 7.3 28-Aug-07
2924 ---------------------
2925
2926 1. In the rejigging of the build system that eventually resulted in 7.1, the
2927 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2928 brackets there is not right, since it causes compilers to look for an
2929 installed pcre.h, not the version that is in the source that is being
2930 compiled (which of course may be different). I have changed it back to:
2931
2932 #include "pcre.h"
2933
2934 I have a vague recollection that the change was concerned with compiling in
2935 different directories, but in the new build system, that is taken care of
2936 by the VPATH setting the Makefile.
2937
2938 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2939 when the subject happened to end in the byte 0x85 (e.g. if the last
2940 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2941 characters but of course it shouldn't be taken as a newline when it is part
2942 of another character. The bug was that, for an unlimited repeat of . in
2943 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2944 characters when looking for a newline.
2945
2946 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2947
2948 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2949 in debug output.
2950
2951 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2952 long printing in the pcrecpp unittest when running under MinGW.
2953
2954 6. ESC_K was left out of the EBCDIC table.
2955
2956 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2957 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2958 limit also applies to "virtual nesting" when a pattern is recursive, and in
2959 this case 1000 isn't so big. I have been able to remove this limit at the
2960 expense of backing off one optimization in certain circumstances. Normally,
2961 when pcre_exec() would call its internal match() function recursively and
2962 immediately return the result unconditionally, it uses a "tail recursion"
2963 feature to save stack. However, when a subpattern that can match an empty
2964 string has an unlimited repetition quantifier, it no longer makes this
2965 optimization. That gives it a stack frame in which to save the data for
2966 checking that an empty string has been matched. Previously this was taken
2967 from the 1000-entry workspace that had been reserved. So now there is no
2968 explicit limit, but more stack is used.
2969
2970 8. Applied Daniel's patches to solve problems with the import/export magic
2971 syntax that is required for Windows, and which was going wrong for the
2972 pcreposix and pcrecpp parts of the library. These were overlooked when this
2973 problem was solved for the main library.
2974
2975 9. There were some crude static tests to avoid integer overflow when computing
2976 the size of patterns that contain repeated groups with explicit upper
2977 limits. As the maximum quantifier is 65535, the maximum group length was
2978 set at 30,000 so that the product of these two numbers did not overflow a
2979 32-bit integer. However, it turns out that people want to use groups that
2980 are longer than 30,000 bytes (though not repeat them that many times).
2981 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2982 made it possible to implement the integer overflow checks in a much more
2983 dynamic way, which I have now done. The artificial limitation on group
2984 length has been removed - we now have only the limit on the total length of
2985 the compiled pattern, which depends on the LINK_SIZE setting.
2986
2987 10. Fixed a bug in the documentation for get/copy named substring when
2988 duplicate names are permitted. If none of the named substrings are set, the
2989 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2990 empty string.
2991
2992 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2993 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2994 because the ] is interpreted as the first data character and the
2995 terminating ] is not found. PCRE has been made compatible with Perl in this
2996 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2997 cause memory overwriting.
2998
2999 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3000 string has been matched (to stop an infinite loop). It was not recognizing
3001 a conditional subpattern that could match an empty string if that
3002 subpattern was within another subpattern. For example, it looped when
3003 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3004 condition was not nested. This bug has been fixed.
3005
3006 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3007 past the start of the subject in the presence of bytes with the top bit
3008 set, for example "\x8aBCD".
3009
3010 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3011 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3012
3013 14. Optimized (?!) to (*FAIL).
3014
3015 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3016 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3017 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3018 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3019 does: it's just the validity check that is more restrictive.
3020
3021 16. Inserted checks for integer overflows during escape sequence (backslash)
3022 processing, and also fixed erroneous offset values for syntax errors during
3023 backslash processing.
3024
3025 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3026 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3027
3028 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3029 caused an overrun.
3030
3031 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3032 something other than just ASCII characters) inside a group that had an
3033 unlimited repeat caused a loop at compile time (while checking to see
3034 whether the group could match an empty string).
3035
3036 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3037 [\P{Any}] did so. (Error in the code for printing property names.)
3038
3039 21. An orphan \E inside a character class could cause a crash.
3040
3041 22. A repeated capturing bracket such as (A)? could cause a wild memory
3042 reference during compilation.
3043
3044 23. There are several functions in pcre_compile() that scan along a compiled
3045 expression for various reasons (e.g. to see if it's fixed length for look
3046 behind). There were bugs in these functions when a repeated \p or \P was
3047 present in the pattern. These operators have additional parameters compared
3048 with \d, etc, and these were not being taken into account when moving along
3049 the compiled data. Specifically:
3050
3051 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3052 length.
3053
3054 (b) An item such as \pL+ within a repeated group could cause crashes or
3055 loops.
3056
3057 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3058 "reference to non-existent subpattern" error.
3059
3060 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3061
3062 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3063 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3064
3065 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3066
3067 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3068 character were causing crashes (broken optimization).
3069
3070 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3071 \p or \P) caused a compile-time loop.
3072
3073 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3074 break. For example, the unstudied pattern [\r\n]A does not match the string
3075 "\r\nA" because change 7.0/46 below moves the current point on by two
3076 characters after failing to match at the start. However, the pattern \nA
3077 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3078 the same is true. There doesn't seem any very clean way out of this, but
3079 what I have chosen to do makes the common cases work: PCRE now takes note
3080 of whether there can be an explicit match for \r or \n anywhere in the
3081 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3082 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3083 pattern has explicit CR or LF references.
3084
3085 29. Added (*CR) etc for changing newline setting at start of pattern.
3086
3087
3088 Version 7.2 19-Jun-07
3089 ---------------------
3090
3091 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3092 which is apparently normally available under Windows.
3093
3094 2. Re-jig the pcregrep tests with different newline settings in an attempt
3095 to make them independent of the local environment's newline setting.
3096
3097 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3098
3099 4. Some of the "internals" tests were previously cut out when the link size
3100 was not 2, because the output contained actual offsets. The recent new
3101 "Z" feature of pcretest means that these can be cut out, making the tests
3102 usable with all link sizes.
3103
3104 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3105 stack recursion. This gives a massive performance boost under BSD, but just
3106 a small improvement under Linux. However, it saves one field in the frame
3107 in all cases.
3108
3109 6. Added more features from the forthcoming Perl 5.10:
3110
3111 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3112 recursion call. It refers to the nth most recently opened parentheses.
3113
3114 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3115 to be opened parentheses.
3116
3117 (c) Conditions that refer to capturing parentheses can be specified
3118 relatively, for example, (?(-2)... or (?(+3)...
3119
3120 (d) \K resets the start of the current match so that everything before
3121 is not part of it.
3122
3123 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3124
3125 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3126 reference syntax.
3127
3128 (g) (?| introduces a group in which the numbering of parentheses in each
3129 alternative starts with the same number.
3130
3131 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3132
3133 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3134 PCRE_INFO_JCHANGED.
3135
3136 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3137 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3138 for detecting groups that can match an empty string.
3139
3140 9. A pattern with a very large number of alternatives (more than several
3141 hundred) was running out of internal workspace during the pre-compile
3142 phase, where pcre_compile() figures out how much memory will be needed. A
3143 bit of new cunning has reduced the workspace needed for groups with
3144 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3145 workspace instead of running out of the 4096 that are available.
3146
3147 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3148
3149 11. Applied patch from Google to remove an optimization that didn't quite work.
3150 The report of the bug said:
3151
3152 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3153 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3154 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3155
3156 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3157 it matched the wrong number of bytes.
3158
3159
3160 Version 7.1 24-Apr-07
3161 ---------------------
3162
3163 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3164 that is more "standard", making use of automake and other Autotools. There
3165 is some re-arrangement of the files and adjustment of comments consequent
3166 on this.
3167
3168 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3169 for recursive directory scanning broke on some systems because the files
3170 are not scanned in any specific order and on different systems the order
3171 was different. A call to "sort" has been inserted into RunGrepTest for the
3172 approprate test as a short-term fix. In the longer term there may be an
3173 alternative.
3174
3175 3. I had an email from Eric Raymond about problems translating some of PCRE's
3176 man pages to HTML (despite the fact that I distribute HTML pages, some
3177 people do their own conversions for various reasons). The problems
3178 concerned the use of low-level troff macros .br and .in. I have therefore
3179 removed all such uses from the man pages (some were redundant, some could
3180 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3181 HTML has been updated to handle .nf/.fi and to complain if it encounters
3182 .br or .in.
3183
3184 4. Updated comments in configure.ac that get placed in config.h.in and also
3185 arranged for config.h to be included in the distribution, with the name
3186 config.h.generic, for the benefit of those who have to compile without
3187 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3188
3189 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3190 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3191 makevp.bat for the current PCRE, using the additional files
3192 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3193
3194 6. A Windows user reported a minor discrepancy with test 2, which turned out
3195 to be caused by a trailing space on an input line that had got lost in his
3196 copy. The trailing space was an accident, so I've just removed it.
3197
3198 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3199 that is needed.
3200
3201 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3202 as "const" (a) because they are and (b) because it helps the PHP
3203 maintainers who have recently made a script to detect big data structures
3204 in the php code that should be moved to the .rodata section. I remembered
3205 to update Builducptable as well, so it won't revert if ucptable.h is ever
3206 re-created.
3207
3208 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3209 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3210 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3211 support is not required. This saves 1.5-2K of code, which is important in
3212 some applications.
3213
3214 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3215 so as not to refer to the tables, even though these functions will never be
3216 called when UTF-8 support is disabled. Otherwise there are problems with a
3217 shared library.
3218
3219 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3220
3221 (a) It was defining its arguments as char * instead of void *.
3222
3223 (b) It was assuming that all moves were upwards in memory; this was true
3224 a long time ago when I wrote it, but is no longer the case.
3225
3226 The emulated memove() is provided for those environments that have neither
3227 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3228 is clearly not the case, as these two bugs were recently reported.
3229
3230 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3231 and Detrail to create the HTML documentation, the .txt form of the man
3232 pages, and it removes trailing spaces from listed files. It also creates
3233 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3234 case, it wraps all the #defines with #ifndefs. This script should be run
3235 before "make dist".
3236
3237 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3238 with Unicode property support.
3239
3240 (a) For a maximizing quantifier, if the two different cases of the
3241 character were of different lengths in their UTF-8 codings (there are
3242 some cases like this - I found 11), and the matching function had to
3243 back up over a mixture of the two cases, it incorrectly assumed they
3244 were both the same length.
3245
3246 (b) When PCRE was configured to use the heap rather than the stack for
3247 recursion during matching, it was not correctly preserving the data for
3248 the other case of a UTF-8 character when checking ahead for a match
3249 while processing a minimizing repeat. If the check also involved
3250 matching a wide character, but failed, corruption could cause an
3251 erroneous result when trying to check for a repeat of the original
3252 character.
3253
3254 13. Some tidying changes to the testing mechanism:
3255
3256 (a) The RunTest script now detects the internal link size and whether there
3257 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3258 values substituted by "configure". (The RunGrepTest script already did
3259 this for UTF-8.) The configure.ac script no longer substitutes the
3260 relevant variables.
3261
3262 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3263 with length and offset values. This means that the output is different
3264 for different internal link sizes. Test 2 is skipped for link sizes
3265 other than 2 because of this, bypassing the problem. Unfortunately,
3266 there was also a test in test 3 (the locale tests) that used /B and
3267 failed for link sizes other than 2. Rather than cut the whole test out,
3268 I have added a new /Z option to pcretest that replaces the length and
3269 offset values with spaces. This is now used to make test 3 independent
3270 of link size. (Test 2 will be tidied up later.)
3271
3272 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3273 segmentation fault instead of returning the appropriate error message.
3274
3275 15. In multiline mode when the newline sequence was set to "any", the pattern
3276 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3277 This doesn't seem right; it now treats the CRLF combination as the line
3278 ending, and so does not match in that case. It's only a pattern such as ^$
3279 that would hit this one: something like ^ABC$ would have failed after \r
3280 and then tried again after \r\n.
3281
3282 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3283 in an attempt to make files that differ only in their line terminators
3284 compare equal. This works on Linux.
3285
3286 17. Under certain error circumstances pcregrep might try to free random memory
3287 as it exited. This is now fixed, thanks to valgrind.
3288
3289 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3290 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3291 was because its rules for how to advance for /g after matching an empty
3292 string at the end of a line did not allow for this case. They now check for
3293 it specially.
3294
3295 20. pcretest is supposed to handle patterns and data of any length, by
3296 extending its buffers when necessary. It was getting this wrong when the
3297 buffer for a data line had to be extended.
3298
3299 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3300 CRLF as a newline sequence.
3301
3302 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3303 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3304 I have nevertheless tidied it up.
3305
3306 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3307
3308 24. Added a man page for pcre-config.
3309
3310
3311 Version 7.0 19-Dec-06
3312 ---------------------
3313
3314 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3315 moving to gcc 4.1.1.
3316
3317 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3318 sys/time.h, which is documented as needed for this function. It doesn't
3319 seem to matter on Linux, but it showed up on some releases of OS X.
3320
3321 3. It seems that there are systems where bytes whose values are greater than
3322 127 match isprint() in the "C" locale. The "C" locale should be the
3323 default when a C program starts up. In most systems, only ASCII printing
3324 characters match isprint(). This difference caused the output from pcretest
3325 to vary, making some of the tests fail. I have changed pcretest so that:
3326
3327 (a) When it is outputting text in the compiled version of a pattern, bytes
3328 other than 32-126 are always shown as hex escapes.
3329
3330 (b) When it is outputting text that is a matched part of a subject string,
3331 it does the same, unless a different locale has been set for the match
3332 (using the /L modifier). In this case, it uses isprint() to decide.
3333
3334 4. Fixed a major bug that caused incorrect computation of the amount of memory
3335 required for a compiled pattern when options that changed within the
3336 pattern affected the logic of the preliminary scan that determines the
3337 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3338 that the computed length was too small. The symptoms of this bug were
3339 either the PCRE error "internal error: code overflow" from pcre_compile(),
3340 or a glibc crash with a message such as "pcretest: free(): invalid next
3341 size (fast)". Examples of patterns that provoked this bug (shown in
3342 pcretest format) are:
3343
3344 /(?-x: )/x
3345 /(?x)(?-x: \s*#\s*)/
3346 /((?i)[\x{c0}])/8
3347 /(?i:[\x{c0}])/8
3348
3349 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3350 is now done differently.
3351
3352 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3353 wrapper classes; (b) implement a new function in the C++ scanner that is
3354 more efficient than the old way of doing things because it avoids levels of
3355 recursion in the regex matching; (c) add a paragraph to the documentation
3356 for the FullMatch() function.
3357
3358 6. The escape sequence \n was being treated as whatever was defined as
3359 "newline". Not only was this contrary to the documentation, which states
3360 that \n is character 10 (hex 0A), but it also went horribly wrong when
3361 "newline" was defined as CRLF. This has been fixed.
3362
3363 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3364 was being set to -1 for the "end of line" case (supposedly a value that no
3365 character can have). Though this value is never used (the check for end of
3366 line is "zero bytes in current character"), it caused compiler complaints.
3367 I've changed it to 0xffffffff.
3368
3369 8. In pcre_version.c, the version string was being built by a sequence of
3370 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3371 string (as it is for production releases) called a macro with an empty
3372 argument. The C standard says the result of this is undefined. The gcc
3373 compiler treats it as an empty string (which was what was wanted) but it is
3374 reported that Visual C gives an error. The source has been hacked around to
3375 avoid this problem.
3376
3377 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3378 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3379 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3380 of them did).
3381
3382 10. Originally, pcretest opened its input and output without "b"; then I was
3383 told that "b" was needed in some environments, so it was added for release
3384 5.0 to both the input and output. (It makes no difference on Unix-like
3385 systems.) Later I was told that it is wrong for the input on Windows. I've
3386 now abstracted the modes into two macros, to make it easier to fiddle with
3387 them, and removed "b" from the input mode under Windows.
3388
3389 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3390
3391 12. Added -help and --help to pcretest as an official way of being reminded
3392 of the options.
3393
3394 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3395 and pcrecpp.cc because they annoy compilers at high warning levels.
3396
3397 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3398
3399 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3400 scripts are not C programs :-) and which was not noticed because it works
3401 on Linux.
3402
3403 16. pcretest is supposed to handle any length of pattern and data line (as one
3404 line or as a continued sequence of lines) by extending its input buffer if
3405 necessary. This feature was broken for very long pattern lines, leading to
3406 a string of junk being passed to pcre_compile() if the pattern was longer
3407 than about 50K.
3408
3409 17. I have done a major re-factoring of the way pcre_compile() computes the
3410 amount of memory needed for a compiled pattern. Previously, there was code
3411 that made a preliminary scan of the pattern in order to do this. That was
3412 OK when PCRE was new, but as the facilities have expanded, it has become
3413 harder and harder to keep it in step with the real compile phase, and there
3414 have been a number of bugs (see for example, 4 above). I have now found a
3415 cunning way of running the real compile function in a "fake" mode that
3416 enables it to compute how much memory it would need, while actually only
3417 ever using a few hundred bytes of working memory and without too many
3418 tests of the mode. This should make future maintenance and development
3419 easier. A side effect of this work is that the limit of 200 on the nesting
3420 depth of parentheses has been removed (though this was never a serious
3421 limitation, I suspect). However, there is a downside: pcre_compile() now
3422 runs more slowly than before (30% or more, depending on the pattern). I
3423 hope this isn't a big issue. There is no effect on runtime performance.
3424
3425 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3426 newline (only possible for the last line of a file) and it was a
3427 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3428
3429 19. Added additional timing features to pcretest. (1) The -tm option now times
3430 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3431 separate command line item, by a number that specifies the number of
3432 repeats to use when timing. The default is 50000; this gives better
3433 precision, but takes uncomfortably long for very large patterns.
3434
3435 20. Extended pcre_study() to be more clever in cases where a branch of a
3436 subpattern has no definite first character. For example, (a*|b*)[cd] would
3437 previously give no result from pcre_study(). Now it recognizes that the
3438 first character must be a, b, c, or d.
3439
3440 21. There was an incorrect error "recursive call could loop indefinitely" if
3441 a subpattern (or the entire pattern) that was being tested for matching an
3442 empty string contained only one non-empty item after a nested subpattern.
3443 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3444 incorrectly, because the \d was being skipped in the check.
3445
3446 22. The pcretest program now has a new pattern option /B and a command line
3447 option -b, which is equivalent to adding /B to every pattern. This causes
3448 it to show the compiled bytecode, without the additional information that
3449 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3450 is the same as /B/I).
3451
3452 23. A new optimization is now able automatically to treat some sequences such
3453 as a*b as a*+b. More specifically, if something simple (such as a character
3454 or a simple class like \d) has an unlimited quantifier, and is followed by
3455 something that cannot possibly match the quantified thing, the quantifier
3456 is automatically "possessified".
3457
3458 24. A recursive reference to a subpattern whose number was greater than 39
3459 went wrong under certain circumstances in UTF-8 mode. This bug could also
3460 have affected the operation of pcre_study().
3461
3462 25. Realized that a little bit of performance could be had by replacing
3463 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3464
3465 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3466
3467 27. Possessive quantifiers such as a++ were previously implemented by turning
3468 them into atomic groups such as ($>a+). Now they have their own opcodes,
3469 which improves performance. This includes the automatically created ones
3470 from 23 above.
3471
3472 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3473 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3474 the first matched character to be a colon. This applied both to named and
3475 numbered groups.
3476
3477 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3478
3479 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3480 building PCRE on Windows easier, so I have included it in the distribution.
3481
3482 31. There is now a check in pcretest against a ridiculously large number being
3483 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3484 loop, the loop is abandoned.
3485
3486 32. Forward references to subpatterns in conditions such as (?(2)...) where
3487 subpattern 2 is defined later cause pcre_compile() to search forwards in
3488 the pattern for the relevant set of parentheses. This search went wrong
3489 when there were unescaped parentheses in a character class, parentheses
3490 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3491
3492 33. "Subroutine" calls and backreferences were previously restricted to
3493 referencing subpatterns earlier in the regex. This restriction has now
3494 been removed.
3495
3496 34. Added a number of extra features that are going to be in Perl 5.10. On the
3497 whole, these are just syntactic alternatives for features that PCRE had
3498 previously implemented using the Python syntax or my own invention. The
3499 other formats are all retained for compatibility.
3500
3501 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3502 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3503 also .NET compatible.
3504
3505 (b) A recursion or subroutine call to a named group can now be defined as
3506 (?&name) as well as (?P>name).
3507
3508 (c) A backreference to a named group can now be defined as \k<name> or
3509 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3510 5.10, are also .NET compatible.
3511
3512 (d) A conditional reference to a named group can now use the syntax
3513 (?(<name>) or (?('name') as well as (?(name).
3514
3515 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3516 groups (named and numbered) that are never evaluated inline, but can be
3517 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3518 is always false. There may be only one alternative in such a group.
3519
3520 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3521 as the simple (?(R). The condition is true only if the most recent
3522 recursion is that of the given number or name. It does not search out
3523 through the entire recursion stack.
3524
3525 (g) The escape \gN or \g{N} has been added, where N is a positive or
3526 negative number, specifying an absolute or relative reference.
3527
3528 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3529 some "unreachable code" warnings.
3530
3531 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3532 things, this adds five new scripts.
3533
3534 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3535 There were also incompatibilities regarding the handling of \Q..\E inside
3536 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3537 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3538
3539 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3540 matches an empty string, and forcibly breaks the loop. There were bugs in
3541 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3542 against aaaa the result was just "a" rather than "aaaa", for example. Two
3543 separate and independent bugs (that affected different cases) have been
3544 fixed.
3545
3546 39. Refactored the code to abolish the use of different opcodes for small
3547 capturing bracket numbers. This is a tidy that I avoided doing when I
3548 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3549 The new approach is not only tidier, it makes it possible to reduce the
3550 memory needed to fix the previous bug (38).
3551
3552 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3553 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3554 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3555 mode.
3556
3557 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3558 report.
3559
3560 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3561 copy construction and assignment in the C++ wrapper.
3562
3563 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3564 couple of bugs that could have given wrong results in the "--newline=crlf"
3565 case.
3566
3567 44. Added a number of casts and did some reorganization of signed/unsigned int
3568 variables following suggestions from Dair Grant. Also renamed the variable
3569 "this" as "item" because it is a C++ keyword.
3570
3571 45. Arranged for dftables to add
3572
3573 #include "pcre_internal.h"
3574
3575 to pcre_chartables.c because without it, gcc 4.x may remove the array
3576 definition from the final binary if PCRE is built into a static library and
3577 dead code stripping is activated.
3578
3579 46. For an unanchored pattern, if a match attempt fails at the start of a
3580 newline sequence, and the newline setting is CRLF or ANY, and the next two
3581 characters are CRLF, advance by two characters instead of one.
3582
3583
3584 Version 6.7 04-Jul-06
3585 ---------------------
3586
3587 1. In order to handle tests when input lines are enormously long, pcretest has
3588 been re-factored so that it automatically extends its buffers when
3589 necessary. The code is crude, but this _is_ just a test program. The
3590 default size has been increased from 32K to 50K.
3591
3592 2. The code in pcre_study() was using the value of the re argument before
3593 testing it for NULL. (Of course, in any sensible call of the function, it
3594 won't be NULL.)
3595
3596 3. The memmove() emulation function in pcre_internal.h, which is used on
3597 systems that lack both memmove() and bcopy() - that is, hardly ever -
3598 was missing a "static" storage class specifier.
3599
3600 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3601 containing an extended class (one that cannot be represented by a bitmap
3602 because it contains high-valued characters or Unicode property items, e.g.
3603 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3604 pattern, but PCRE should not loop if you do not (it no longer does).
3605 [Detail: two cases were found: (a) a repeated subpattern containing an
3606 extended class; (b) a recursive reference to a subpattern that followed a
3607 previous extended class. It wasn't skipping over the extended class
3608 correctly when UTF-8 mode was not set.]
3609
3610 5. A negated single-character class was not being recognized as fixed-length
3611 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3612 compile error "lookbehind assertion is not fixed length".
3613
3614 6. The RunPerlTest auxiliary script was showing an unexpected difference
3615 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3616 write a Perl script that can interpret lines of an input file either as
3617 byte characters or as UTF-8, which is what "perltest" was being required to
3618 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3619 can't do is switch easily at run time between having the "use utf8;" pragma
3620 or not. In the end, I fudged it by using the RunPerlTest script to insert
3621 "use utf8;" explicitly for the UTF-8 tests.
3622
3623 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3624 the end of the subject string, contrary to the documentation and to what
3625 Perl does. This was true of both matching functions. Now it matches only at
3626 the start of the subject and immediately after *internal* newlines.
3627
3628 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3629 a pointer to an int instead of a pointer to an unsigned long int. This
3630 caused problems on 64-bit systems.
3631
3632 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3633 instance of the 'standard' template library not being so standard".
3634
3635 10. There was no check on the number of named subpatterns nor the maximum
3636 length of a subpattern name. The product of these values is used to compute
3637 the size of the memory block for a compiled pattern. By supplying a very
3638 long subpattern name and a large number of named subpatterns, the size
3639 computation could be caused to overflow. This is now prevented by limiting
3640 the length of names to 32 characters, and the number of named subpatterns
3641 to 10,000.
3642
3643 11. Subpatterns that are repeated with specific counts have to be replicated in
3644 the compiled pattern. The size of memory for this was computed from the
3645 length of the subpattern and the repeat count. The latter is limited to
3646 65535, but there was no limit on the former, meaning that integer overflow
3647 could in principle occur. The compiled length of a repeated subpattern is
3648 now limited to 30,000 bytes in order to prevent this.
3649
3650 12. Added the optional facility to have named substrings with the same name.
3651
3652 13. Added the ability to use a named substring as a condition, using the
3653 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3654 are numbers (not recommended). Forward references are permitted.
3655
3656 14. Added forward references in named backreferences (if you see what I mean).
3657
3658 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3659 pattern could run off the end of the subject. For example, the pattern
3660 "(?s)(.{1,5})"8 did this with the subject "ab".
3661
3662 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3663 PCRE_CASELESS was set when matching characters that were quantified with ?
3664 or *.
3665
3666 17. A character class other than a single negated character that had a minimum
3667 but no maximum quantifier - for example [ab]{6,} - was not handled
3668 correctly by pce_dfa_exec(). It would match only one character.
3669
3670 18. A valid (though odd) pattern that looked like a POSIX character
3671 class but used an invalid character after [ (for example [[,abc,]]) caused
3672 pcre_compile() to give the error "Failed: internal error: code overflow" or
3673 in some cases to crash with a glibc free() error. This could even happen if
3674 the pattern terminated after [[ but there just happened to be a sequence of
3675 letters, a binary zero, and a closing ] in the memory that followed.
3676
3677 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3678 over the years. Originally (before any Unicode support), just the bottom 8
3679 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3680 output from "man perlunicode" includes this:
3681
3682 The regular expression compiler produces polymorphic opcodes. That
3683 is, the pattern adapts to the data and automatically switches to
3684 the Unicode character scheme when presented with Unicode data--or
3685 instead uses a traditional byte scheme when presented with byte
3686 data.
3687
3688 Sadly, a wide octal escape does not cause a switch, and in a string with
3689 no other multibyte characters, these octal escapes are treated as before.
3690 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3691 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3692 Unicode string.
3693
3694 I have not perpetrated such confusion in PCRE. Up till now, it took just
3695 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3696 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3697 translate to the appropriate multibyte character.
3698
3699 29. Applied some refactoring to reduce the number of warnings from Microsoft
3700 and Borland compilers. This has included removing the fudge introduced
3701 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3702 a warning about an unused variable.
3703
3704 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3705 characters since release 4.0, because Perl (from release 5.004) does not.
3706 [Or at least, is documented not to: some releases seem to be in conflict
3707 with the documentation.] However, when a pattern was studied with
3708 pcre_study() and all its branches started with \s, PCRE still included VT
3709 as a possible starting character. Of course, this did no harm; it just
3710 caused an unnecessary match attempt.
3711
3712 22. Removed a now-redundant internal flag bit that recorded the fact that case
3713 dependency changed within the pattern. This was once needed for "required
3714 byte" processing, but is no longer used. This recovers a now-scarce options
3715 bit. Also moved the least significant internal flag bit to the most-
3716 significant bit of the word, which was not previously used (hangover from
3717 the days when it was an int rather than a uint) to free up another bit for
3718 the future.
3719
3720 23. Added support for CRLF line endings as well as CR and LF. As well as the
3721 default being selectable at build time, it can now be changed at runtime
3722 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3723 specify that it is scanning data with non-default line endings.
3724
3725 24. Changed the definition of CXXLINK to make it agree with the definition of
3726 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3727
3728 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3729 recursions. This makes a big different to stack usage for some patterns.
3730
3731 26. If a subpattern containing a named recursion or subroutine reference such
3732 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3733 the space required for the compiled pattern went wrong and gave too small a
3734 value. Depending on the environment, this could lead to "Failed: internal
3735 error: code overflow at offset 49" or "glibc detected double free or
3736 corruption" errors.
3737
3738 27. Applied patches from Google (a) to support the new newline modes and (b) to
3739 advance over multibyte UTF-8 characters in GlobalReplace.
3740
3741 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3742 difference for some implementation of PCRE in some Windows version.
3743
3744 29. Added some extra testing facilities to pcretest:
3745
3746 \q<number> in a data line sets the "match limit" value
3747 \Q<number> in a data line sets the "match recursion limt" value
3748 -S <number> sets the stack size, where <number> is in megabytes
3749
3750 The -S option isn't available for Windows.
3751
3752
3753 Version 6.6 06-Feb-06
3754 ---------------------
3755
3756 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3757 in pcreposix.h. I have copied the definition from pcre.h.
3758
3759 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3760 because pcre.h is no longer a built file.
3761
3762 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3763 not normally included in the compiled code.
3764
3765
3766 Version 6.5 01-Feb-06
3767 ---------------------
3768
3769 1. When using the partial match feature with pcre_dfa_exec(), it was not
3770 anchoring the second and subsequent partial matches at the new starting
3771 point. This could lead to incorrect results. For example, with the pattern
3772 /1234/, partially matching against "123" and then "a4" gave a match.
3773
3774 2. Changes to pcregrep:
3775
3776 (a) All non-match returns from pcre_exec() were being treated as failures
3777 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3778 error message is output. Some extra information is given for the
3779 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3780 probably the only errors that are likely to be caused by users (by
3781 specifying a regex that has nested indefinite repeats, for instance).
3782 If there are more than 20 of these errors, pcregrep is abandoned.
3783
3784 (b) A binary zero was treated as data while matching, but terminated the
3785 output line if it was written out. This has been fixed: binary zeroes
3786 are now no different to any other data bytes.
3787
3788 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3789 used to set a locale for matching. The --locale=xxxx long option has
3790 been added (no short equivalent) to specify a locale explicitly on the
3791 pcregrep command, overriding the environment variables.
3792
3793 (d) When -B was used with -n, some line numbers in the output were one less
3794 than they should have been.
3795
3796 (e) Added the -o (--only-matching) option.
3797
3798 (f) If -A or -C was used with -c (count only), some lines of context were
3799 accidentally printed for the final match.
3800
3801 (g) Added the -H (--with-filename) option.
3802
3803 (h) The combination of options -rh failed to suppress file names for files
3804 that were found from directory arguments.
3805
3806 (i) Added the -D (--devices) and -d (--directories) options.
3807
3808 (j) Added the -F (--fixed-strings) option.
3809
3810 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3811
3812 (l) Added the --colo(u)r option.
3813
3814 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3815 is not present by default.
3816
3817 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3818 items such as (?R) or (?1), when the recursion could match a number of
3819 alternatives. If it matched one of the alternatives, but subsequently,
3820 outside the recursion, there was a failure, the code tried to back up into
3821 the recursion. However, because of the way PCRE is implemented, this is not
3822 possible, and the result was an incorrect result from the match.
3823
3824 In order to prevent this happening, the specification of recursion has
3825 been changed so that all such subpatterns are automatically treated as
3826 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3827
3828 4. I had overlooked the fact that, in some locales, there are characters for
3829 which isalpha() is true but neither isupper() nor islower() are true. In
3830 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3831 and ordfeminine) are like this. This affected the treatment of \w and \W
3832 when they appeared in character classes, but not when they appeared outside
3833 a character class. The bit map for "word" characters is now created
3834 separately from the results of isalnum() instead of just taking it from the
3835 upper, lower, and digit maps. (Plus the underscore character, of course.)
3836
3837 5. The above bug also affected the handling of POSIX character classes such as
3838 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3839 permanent tables. Instead, the bit maps for such a class were previously
3840 created as the appropriate unions of the upper, lower, and digit bitmaps.
3841 Now they are created by subtraction from the [[:word:]] class, which has
3842 its own bitmap.
3843
3844 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3845 It is created by subtracting the vertical space characters (\x09, \x0a,
3846 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3847 subtraction was done in the overall bitmap for a character class, meaning
3848 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3849 be recognized. This bug has been fixed.
3850
3851 7. Patches from the folks at Google:
3852
3853 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3854 real life, but is still worth protecting against".
3855
3856 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3857 regular expressions".
3858
3859 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3860 have it.
3861
3862 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3863 "configure" and the latter not, in order to fix a problem somebody had
3864 with compiling the Arg class on HP-UX.
3865
3866 (e) Improve the error-handling of the C++ wrapper a little bit.
3867
3868 (f) New tests for checking recursion limiting.
3869
3870 8. The pcre_memmove() function, which is used only if the environment does not
3871 have a standard memmove() function (and is therefore rarely compiled),
3872 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3873 returning a result (though PCRE never actually uses the result).
3874
3875 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3876 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3877 returned instead of calling malloc() with an overflowing number that would
3878 most likely cause subsequent chaos.
3879
3880 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3881
3882 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3883 with this option is matched, the nmatch and pmatch options of regexec() are
3884 ignored.
3885
3886 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3887 provided in case anyone wants to the the POSIX interface with UTF-8
3888 strings.
3889
3890 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3891 C++ linking (needed for some HP-UX environments).
3892
3893 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3894 (unused parameter) and in the pcre_printint() function (omitted "default"
3895 switch label when the default is to do nothing).
3896
3897 15. Added some code to make it possible, when PCRE is compiled as a C++
3898 library, to replace subject pointers for pcre_exec() with a smart pointer
3899 class, thus making it possible to process discontinuous strings.
3900
3901 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3902 much the same function. They were added by different people who were trying
3903 to make PCRE easy to compile on non-Unix systems. It has been suggested
3904 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3905 for compiling on Windows systems. I have therefore replaced it with
3906 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3907 defaults to "extern" for C or "extern C" for C++, which works fine on
3908 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3909 SCOPE with something explicit in config.h. In addition:
3910
3911 (a) pcreposix.h still had just "extern" instead of either of these macros;
3912 I have replaced it with PCRE_DATA_SCOPE.
3913
3914 (b) Functions such as _pcre_xclass(), which are internal to the library,
3915 but external in the C sense, all had PCRE_EXPORT in their definitions.
3916 This is apparently wrong for the Windows case, so I have removed it.
3917 (It makes no difference on Unix-like systems.)
3918
3919 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3920 of recursive calls to match(). This is different to MATCH_LIMIT because
3921 that limits the total number of calls to match(), not all of which increase
3922 the depth of recursion. Limiting the recursion depth limits the amount of
3923 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3924 when PCRE is compiled, and changed at run time. A patch from Google adds
3925 this functionality to the C++ interface.
3926
3927 18. Changes to the handling of Unicode character properties:
3928
3929 (a) Updated the table to Unicode 4.1.0.
3930
3931 (b) Recognize characters that are not in the table as "Cn" (undefined).
3932
3933 (c) I revised the way the table is implemented to a much improved format
3934 which includes recognition of ranges. It now supports the ranges that
3935 are defined in UnicodeData.txt, and it also amalgamates other
3936 characters into ranges. This has reduced the number of entries in the
3937 table from around 16,000 to around 3,000, thus reducing its size
3938 considerably. I realized I did not need to use a tree structure after
3939 all - a binary chop search is just as efficient. Having reduced the
3940 number of entries, I extended their size from 6 bytes to 8 bytes to
3941 allow for more data.
3942
3943 (d) Added support for Unicode script names via properties such as \p{Han}.
3944
3945 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3946 matching that character.
3947
3948 20. When matching a repeated Unicode property with a minimum greater than zero,
3949 (for example \pL{2,}), PCRE could look past the end of the subject if it
3950 reached it while seeking the minimum number of characters. This could
3951 happen only if some of the characters were more than one byte long, because
3952 there is a check for at least the minimum number of bytes.
3953
3954 21. Refactored the implementation of \p and \P so as to be more general, to
3955 allow for more different types of property in future. This has changed the
3956 compiled form incompatibly. Anybody with saved compiled patterns that use
3957 \p or \P will have to recompile them.
3958
3959 22. Added "Any" and "L&" to the supported property types.
3960
3961 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3962 but give a compile time error if the value is greater than 0xff.
3963
3964 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3965 accidentally not being installed or uninstalled.
3966
3967 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3968 made were to insert the current release number. This seemed silly, because
3969 it made things harder for people building PCRE on systems that don't run
3970 "configure". I have turned pcre.h into a distributed file, no longer built
3971 by "configure", with the version identification directly included. There is
3972 no longer a pcre.h.in file.
3973
3974 However, this change necessitated a change to the pcre-config script as
3975 well. It is built from pcre-config.in, and one of the substitutions was the
3976 release number. I have updated configure.ac so that ./configure now finds
3977 the release number by grepping pcre.h.
3978
3979 26. Added the ability to run the tests under valgrind.
3980
3981
3982 Version 6.4 05-Sep-05
3983 ---------------------
3984
3985 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3986 "--" to be printed when multiple files were scanned, even when none of the
3987 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3988 consider it to be a bug, and have restored the previous behaviour.
3989
3990 2. A couple of code tidies to get rid of compiler warnings.
3991
3992 3. The pcretest program used to cheat by referring to symbols in the library
3993 whose names begin with _pcre_. These are internal symbols that are not
3994 really supposed to be visible externally, and in some environments it is
3995 possible to suppress them. The cheating is now confined to including
3996 certain files from the library's source, which is a bit cleaner.
3997
3998 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3999 file's purpose clearer.
4000
4001 5. Reorganized pcre_ucp_findchar().
4002
4003
4004 Version 6.3 15-Aug-05
4005 ---------------------
4006
4007 1. The file libpcre.pc.in did not have general read permission in the tarball.
4008
4009 2. There were some problems when building without C++ support:
4010
4011 (a) If C++ support was not built, "make install" and "make test" still
4012 tried to test it.
4013
4014 (b) There were problems when the value of CXX was explicitly set. Some
4015 changes have been made to try to fix these, and ...
4016
4017 (c) --disable-cpp can now be used to explicitly disable C++ support.
4018
4019 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4020 backslash in a target when C++ was disabled. This confuses some
4021 versions of "make", apparently. Using an intermediate variable solves
4022 this. (Same for CPP_LOBJ.)
4023
4024 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4025 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4026 necessary on certain architectures.
4027
4028 4. Added a setting of -export-symbols-regex to the link command to remove
4029 those symbols that are exported in the C sense, but actually are local
4030 within the library, and not documented. Their names all begin with
4031 "_pcre_". This is not a perfect job, because (a) we have to except some
4032 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4033 available (and never for static libraries). I have made a note to try to
4034 find a way round (a) in the future.
4035
4036
4037 Version 6.2 01-Aug-05
4038 ---------------------
4039
4040 1. There was no test for integer overflow of quantifier values. A construction
4041 such as {1111111111111111} would give undefined results. What is worse, if
4042 a minimum quantifier for a parenthesized subpattern overflowed and became
4043 negative, the calculation of the memory size went wrong. This could have
4044 led to memory overwriting.
4045
4046 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4047
4048 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4049 operating environments where this matters.
4050
4051 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4052 PCRE options from within the C++ wrapper.
4053
4054 5. Named capturing subpatterns were not being correctly counted when a pattern
4055 was compiled. This caused two problems: (a) If there were more than 100
4056 such subpatterns, the calculation of the memory needed for the whole
4057 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4058 back references of the form \12, where the number was greater than 9, were
4059 not recognized as back references, even though there were sufficient
4060 previous subpatterns.
4061
4062 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4063 versions of gcc, e.g. 2.95.4.
4064
4065
4066 Version 6.1 21-Jun-05
4067 ---------------------
4068
4069 1. There was one reference to the variable "posix" in pcretest.c that was not
4070 surrounded by "#if !defined NOPOSIX".
4071
4072 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4073 the cross-check on the old pcre_info() function, for the benefit of the
4074 cut-down version of PCRE that is currently imported into Exim.
4075
4076 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4077 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4078 patterns that start (?i) but I don't think that matters. The use of (?i) is
4079 just an example; this all applies to the other options as well.
4080
4081 4. Since libtool seems to echo the compile commands it is issuing, the output
4082 from "make" can be reduced a bit by putting "@" in front of each libtool
4083 compile command.
4084
4085 5. Patch from the folks at Google for configure.in to be a bit more thorough
4086 in checking for a suitable C++ installation before trying to compile the
4087 C++ stuff. This should fix a reported problem when a compiler was present,
4088 but no suitable headers.
4089
4090 6. The man pages all had just "PCRE" as their title. I have changed them to
4091 be the relevant file name. I have also arranged that these names are
4092 retained in the file doc/pcre.txt, which is a concatenation in text format
4093 of all the man pages except the little individual ones for each function.
4094
4095 7. The NON-UNIX-USE file had not been updated for the different set of source
4096 files that come with release 6. I also added a few comments about the C++
4097 wrapper.
4098
4099
4100 Version 6.0 07-Jun-05
4101 ---------------------
4102
4103 1. Some minor internal re-organization to help with my DFA experiments.
4104
4105 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4106 didn't matter for the library itself when fully configured, but did matter
4107 when compiling without UCP support, or within Exim, where the ucp files are
4108 not imported.
4109
4110 3. Refactoring of the library code to split up the various functions into
4111 different source modules. The addition of the new DFA matching code (see
4112 below) to a single monolithic source would have made it really too
4113 unwieldy, quite apart from causing all the code to be include in a
4114 statically linked application, when only some functions are used. This is
4115 relevant even without the DFA addition now that patterns can be compiled in
4116 one application and matched in another.
4117
4118 The downside of splitting up is that there have to be some external
4119 functions and data tables that are used internally in different modules of
4120 the library but which are not part of the API. These have all had their
4121 names changed to start with "_pcre_" so that they are unlikely to clash
4122 with other external names.
4123
4124 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4125 a different (DFA) algorithm. Although it is slower than the original
4126 function, it does have some advantages for certain types of matching
4127 problem.
4128
4129 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4130 including restarting after a partial match.
4131
4132 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4133 defined when compiling for Windows was sent to me. I have put it into the
4134 code, though I have no means of testing or verifying it.
4135
4136 7. Added the pcre_refcount() auxiliary function.
4137
4138 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4139 match before or at the first newline in the subject string. In pcretest,
4140 the /f option on a pattern can be used to set this.
4141
4142 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4143 would behave wrongly. This has been present in PCRE since release 4.0.
4144
4145 10. A number of changes to the pcregrep command:
4146
4147 (a) Refactored how -x works; insert ^(...)$ instead of setting
4148 PCRE_ANCHORED and checking the length, in preparation for adding
4149 something similar for -w.
4150
4151 (b) Added the -w (match as a word) option.
4152
4153 (c) Refactored the way lines are read and buffered so as to have more
4154 than one at a time available.
4155
4156 (d) Implemented a pcregrep test script.
4157
4158 (e) Added the -M (multiline match) option. This allows patterns to match
4159 over several lines of the subject. The buffering ensures that at least
4160 8K, or the rest of the document (whichever is the shorter) is available
4161 for matching (and similarly the previous 8K for lookbehind assertions).
4162
4163 (f) Changed the --help output so that it now says
4164
4165 -w, --word-regex(p)
4166
4167 instead of two lines, one with "regex" and the other with "regexp"
4168 because that confused at least one person since the short forms are the
4169 same. (This required a bit of code, as the output is generated
4170 automatically from a table. It wasn't just a text change.)
4171
4172 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4173 option but starts with a hyphen. Could be a pattern or a path name
4174 starting with a hyphen, for instance.
4175
4176 (h) "-" can be given as a file name to represent stdin.
4177
4178 (i) When file names are being printed, "(standard input)" is used for
4179 the standard input, for compatibility with GNU grep. Previously
4180 "<stdin>" was used.
4181
4182 (j) The option --label=xxx can be used to supply a name to be used for
4183 stdin when file names are being printed. There is no short form.
4184
4185 (k) Re-factored the options decoding logic because we are going to add
4186 two more options that take data. Such options can now be given in four
4187 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4188
4189 (l) Added the -A, -B, and -C options for requesting that lines of context
4190 around matches be printed.
4191
4192 (m) Added the -L option to print the names of files that do not contain
4193 any matching lines, that is, the complement of -l.
4194
4195 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4196 continue to scan other files.
4197
4198 (o) The -s option was incorrectly implemented. For compatibility with other
4199 greps, it now suppresses the error message for a non-existent or non-
4200 accessible file (but not the return code). There is a new option called
4201 -q that suppresses the output of matching lines, which was what -s was
4202 previously doing.
4203
4204 (p) Added --include and --exclude options to specify files for inclusion
4205 and exclusion when recursing.
4206
4207 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4208 Hopefully, it now does.
4209
4210 12. Missing cast in pcre_study().
4211
4212 13. Added an "uninstall" target to the makefile.
4213
4214 14. Replaced "extern" in the function prototypes in Makefile.in with
4215 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4216 world, but is set differently for Windows.
4217
4218 15. Added a second compiling function called pcre_compile2(). The only
4219 difference is that it has an extra argument, which is a pointer to an
4220 integer error code. When there is a compile-time failure, this is set
4221 non-zero, in addition to the error test pointer being set to point to an
4222 error message. The new argument may be NULL if no error number is required
4223 (but then you may as well call pcre_compile(), which is now just a
4224 wrapper). This facility is provided because some applications need a
4225 numeric error indication, but it has also enabled me to tidy up the way
4226 compile-time errors are handled in the POSIX wrapper.
4227
4228 16. Added VPATH=.libs to the makefile; this should help when building with one
4229 prefix path and installing with another. (Or so I'm told by someone who
4230 knows more about this stuff than I do.)
4231
4232 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4233 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4234 match everything, including newlines. This is not POSIX-compatible, but
4235 somebody wanted the feature. From pcretest it can be activated by using
4236 both the P and the s flags.
4237
4238 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4239
4240 19. libpcre.pc was being incorrectly installed as executable.
4241
4242 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4243 it now also looks for '\r' so that it will work unmodified on Windows.
4244
4245 21. Added Google's contributed C++ wrapper to the distribution.
4246
4247 22. Added some untidy missing memory free() calls in pcretest, to keep
4248 Electric Fence happy when testing.
4249
4250
4251
4252 Version 5.0 13-Sep-04
4253 ---------------------
4254
4255 1. Internal change: literal characters are no longer packed up into items
4256 containing multiple characters in a single byte-string. Each character
4257 is now matched using a separate opcode. However, there may be more than one
4258 byte in the character in UTF-8 mode.
4259
4260 2. The pcre_callout_block structure has two new fields: pattern_position and
4261 next_item_length. These contain the offset in the pattern to the next match
4262 item, and its length, respectively.
4263
4264 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4265 insertion of callouts before each pattern item. Added the /C option to
4266 pcretest to make use of this.
4267
4268 4. On the advice of a Windows user, the lines
4269
4270 #if defined(_WIN32) || defined(WIN32)
4271 _setmode( _fileno( stdout ), 0x8000 );
4272 #endif /* defined(_WIN32) || defined(WIN32) */
4273
4274 have been added to the source of pcretest. This apparently does useful
4275 magic in relation to line terminators.
4276
4277 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4278 for the benefit of those environments where the "b" makes a difference.
4279
4280 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4281 to know about it. I have put a hack into configure.in that adds in code
4282 to set GCC=yes if CC=icc. This seems to end up at a point in the
4283 generated configure script that is early enough to affect the setting of
4284 compiler options, which is what is needed, but I have no means of testing
4285 whether it really works. (The user who reported this had patched the
4286 generated configure script, which of course I cannot do.)
4287
4288 LATER: After change 22 below (new libtool files), the configure script
4289 seems to know about icc (and also ecc). Therefore, I have commented out
4290 this hack in configure.in.
4291
4292 7. Added support for pkg-config (2 patches were sent in).
4293
4294 8. Negated POSIX character classes that used a combination of internal tables
4295 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4296 [[:^ascii]]. Typically, they would match almost any characters. The other
4297 POSIX classes were not broken in this way.
4298
4299 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4300 to find the match, as PCRE was deluded into thinking that the match had to
4301 start at the start point or following a newline. The same bug applied to
4302 patterns with negative forward assertions or any backward assertions
4303 preceding ".*" at the start, unless the pattern required a fixed first
4304 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4305
4306 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4307 starting at the last subject character, bytes beyond the end of the subject
4308 string were read.
4309
4310 11. Renamed the variable "class" as "classbits" to make life easier for C++
4311 users. (Previously there was a macro definition, but it apparently wasn't
4312 enough.)
4313
4314 12. Added the new field "tables" to the extra data so that tables can be passed
4315 in at exec time, or the internal tables can be re-selected. This allows
4316 a compiled regex to be saved and re-used at a later time by a different
4317 program that might have everything at different addresses.
4318
4319 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4320 -R library as well as a -L library.
4321
4322 14. The debugging options of pcretest (-d on the command line or D on a
4323 pattern) showed incorrect output for anything following an extended class
4324 that contained multibyte characters and which was followed by a quantifier.
4325
4326 15. Added optional support for general category Unicode character properties
4327 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4328 support. It adds about 90K to the size of the library. The meanings of the
4329 inbuilt class escapes such as \d and \s have NOT been changed.
4330
4331 16. Updated pcredemo.c to include calls to free() to release the memory for the
4332 compiled pattern.
4333
4334 17. The generated file chartables.c was being created in the source directory
4335 instead of in the building directory. This caused the build to fail if the
4336 source directory was different from the building directory, and was
4337 read-only.
4338
4339 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4340 file. No doubt somebody will tell me if they don't make sense... Also added
4341 Dan Mooney's comments about building on OpenVMS.
4342
4343 19. Added support for partial matching via the PCRE_PARTIAL option for
4344 pcre_exec() and the \P data escape in pcretest.
4345
4346 20. Extended pcretest with 3 new pattern features:
4347
4348 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4349 write the compiled pattern to the file whose name is "rest-of-line".
4350 This is a straight binary dump of the data, with the saved pointer to
4351 the character tables forced to be NULL. The study data, if any, is
4352 written too. After writing, pcretest reads a new pattern.
4353
4354 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4355 compiled pattern from the given file. There must not be any
4356 occurrences of "<" in the file name (pretty unlikely); if there are,
4357 pcretest will instead treat the initial "<" as a pattern delimiter.
4358 After reading in the pattern, pcretest goes on to read data lines as
4359 usual.
4360
4361 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4362 and 16-bit fields in a compiled pattern, to simulate a pattern that
4363 was compiled on a host of opposite endianness.
4364
4365 21. The pcre-exec() function can now cope with patterns that were compiled on
4366 hosts of opposite endianness, with this restriction:
4367
4368 As for any compiled expression that is saved and used later, the tables
4369 pointer field cannot be preserved; the extra_data field in the arguments
4370 to pcre_exec() should be used to pass in a tables address if a value
4371 other than the default internal tables were used at compile time.
4372
4373 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4374 now diagnosed as an error. Previously, most of the time, a negative number
4375 would have been treated as zero, but if in addition "ovector" was passed as
4376 NULL, a crash could occur.
4377
4378 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4379 new versions from the libtool 1.5 distribution (the last one is a copy of
4380 a file called libtool.m4). This seems to have fixed the need to patch
4381 "configure" to support Darwin 1.3 (which I used to do). However, I still
4382 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4383 workstation).
4384
4385 24. Changed the PCRE licence to be the more standard "BSD" licence.
4386
4387
4388 Version 4.5 01-Dec-03
4389 ---------------------
4390
4391 1. There has been some re-arrangement of the code for the match() function so
4392 that it can be compiled in a version that does not call itself recursively.
4393 Instead, it keeps those local variables that need separate instances for
4394 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4395 needs to "recurse". Keeping track of where control must go is done by means
4396 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4397 hide most of the details from the main code, and operates only if
4398 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4399 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4400 operating.
4401
4402 To make it easier for callers to provide specially tailored get/free
4403 functions for this usage, two new functions, pcre_stack_malloc, and
4404 pcre_stack_free, are used. They are always called in strict stacking order,
4405 and the size of block requested is always the same.
4406
4407 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4408 PCRE has been compiled to use the stack or the heap for recursion. The
4409 -C option of pcretest uses this to show which version is compiled.
4410
4411 A new data escape \S, is added to pcretest; it causes the amounts of store
4412 obtained and freed by both kinds of malloc/free at match time to be added
4413 to the output.
4414
4415 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4416 what's available on my current Linux desktop machine.
4417
4418 3. When matching a UTF-8 string, the test for a valid string at the start has
4419 been extended. If start_offset is not zero, PCRE now checks that it points
4420 to a byte that is the start of a UTF-8 character. If not, it returns
4421 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4422 this is necessary because there may be backward assertions in the pattern.
4423 When matching the same subject several times, it may save resources to use
4424 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4425
4426 4. The code for checking the validity of UTF-8 strings has been tightened so
4427 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4428 containing "overlong sequences".
4429
4430 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4431 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4432 should have been "&", but it just so happened that all the cases this let
4433 through by mistake were picked up later in the function.
4434
4435 6. I had used a variable called "isblank" - this is a C99 function, causing
4436 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4437
4438 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4439 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4440 so the version is shown; (c) stop "make test" echoing "./RunTest".
4441
4442 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4443
4444 9. The prototype for memmove() for systems that don't have it was using
4445 size_t, but the inclusion of the header that defines size_t was later. I've
4446 moved the #includes for the C headers earlier to avoid this.
4447
4448 10. Added some adjustments to the code to make it easier to compiler on certain
4449 special systems:
4450
4451 (a) Some "const" qualifiers were missing.
4452 (b) Added the macro EXPORT before all exported functions; by default this
4453 is defined to be empty.
4454 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4455 that it reads its output file name as an argument instead of writing
4456 to the standard output and assuming this can be redirected.
4457
4458 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4459 class containing characters with values greater than 255, PCRE compilation
4460 went into a loop.
4461
4462 12. A recursive reference to a subpattern that was within another subpattern
4463 that had a minimum quantifier of zero caused PCRE to crash. For example,
4464 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4465 recursion. If the recursively-called subpattern itself had a zero repeat,
4466 that was OK.
4467
4468 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4469 buffer into which it was copied (for escape processing) was still set at
4470 1024, so long lines caused crashes.
4471
4472 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4473 "internal error: code overflow...". This applied to any character class
4474 that was followed by a possessive quantifier.
4475
4476 15. Modified the Makefile to add libpcre.la as a prerequisite for
4477 libpcreposix.la because I was told this is needed for a parallel build to
4478 work.
4479
4480 16. If a pattern that contained .* following optional items at the start was
4481 studied, the wrong optimizing data was generated, leading to matching
4482 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4483 matching string must start with a or b or c. The correct conclusion for
4484 this pattern is that a match can start with any character.
4485
4486
4487 Version 4.4 13-Aug-03
4488 ---------------------
4489
4490 1. In UTF-8 mode, a character class containing characters with values between
4491 127 and 255 was not handled correctly if the compiled pattern was studied.
4492 In fixing this, I have also improved the studying algorithm for such
4493 classes (slightly).
4494
4495 2. Three internal functions had redundant arguments passed to them. Removal
4496 might give a very teeny performance improvement.
4497
4498 3. Documentation bug: the value of the capture_top field in a callout is *one
4499 more than* the number of the hightest numbered captured substring.
4500
4501 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4502 in incorrectly linking with a previously installed version. They now link
4503 explicitly with libpcre.la.
4504
4505 5. configure.in no longer needs to recognize Cygwin specially.
4506
4507 6. A problem in pcre.in for Windows platforms is fixed.
4508
4509 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4510 pcretest, it used to include the size of the study block as part of its
4511 output. Unfortunately, the structure contains a field that has a different
4512 size on different hardware architectures. This meant that the tests that
4513 showed this size failed. As the block is currently always of a fixed size,
4514 this information isn't actually particularly useful in pcretest output, so
4515 I have just removed it.
4516
4517 8. Three pre-processor statements accidentally did not start in column 1.
4518 Sadly, there are *still* compilers around that complain, even though
4519 standard C has not required this for well over a decade. Sigh.
4520
4521 9. In pcretest, the code for checking callouts passed small integers in the
4522 callout_data field, which is a void * field. However, some picky compilers
4523 complained about the casts involved for this on 64-bit systems. Now
4524 pcretest passes the address of the small integer instead, which should get
4525 rid of the warnings.
4526
4527 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4528 both compile and run time, and gives an error if an invalid UTF-8 sequence
4529 is found. There is a option for disabling this check in cases where the
4530 string is known to be correct and/or the maximum performance is wanted.
4531
4532 11. In response to a bug report, I changed one line in Makefile.in from
4533
4534 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4535 to
4536 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4537
4538 to look similar to other lines, but I have no way of telling whether this
4539 is the right thing to do, as I do not use Windows. No doubt I'll get told
4540 if it's wrong...
4541
4542
4543 Version 4.3 21-May-03
4544 ---------------------
4545
4546 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4547 Makefile.
4548
4549 2. Some refactoring to improve the quality of the code:
4550
4551 (i) The utf8_table... variables are now declared "const".
4552
4553 (ii) The code for \cx, which used the "case flipping" table to upper case
4554 lower case letters, now just substracts 32. This is ASCII-specific,
4555 but the whole concept of \cx is ASCII-specific, so it seems
4556 reasonable.
4557
4558 (iii) PCRE was using its character types table to recognize decimal and
4559 hexadecimal digits in the pattern. This is silly, because it handles
4560 only 0-9, a-f, and A-F, but the character types table is locale-
4561 specific, which means strange things might happen. A private
4562 table is now used for this - though it costs 256 bytes, a table is
4563 much faster than multiple explicit tests. Of course, the standard
4564 character types table is still used for matching digits in subject
4565 strings against \d.
4566
4567 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4568 ending in _t are). So I've renamed it as ESC_tee.
4569
4570 3. The first argument for regexec() in the POSIX wrapper should have been
4571 defined as "const".
4572
4573 4. Changed pcretest to use malloc() for its buffers so that they can be
4574 Electric Fenced for debugging.
4575
4576 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4577 to read one or more bytes before the start of the subject string. Often this
4578 had no effect on PCRE's behaviour, but in some circumstances it could
4579 provoke a segmentation fault.
4580
4581 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4582 to try to read one or more bytes before the start of the subject string.
4583
4584 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4585 UTF-8 support could misbehave in various ways if the subject string
4586 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4587 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4588 back over UTF-8 characters.)
4589
4590
4591 Version 4.2 14-Apr-03
4592 ---------------------
4593
4594 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4595
4596 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4597 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4598 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4599 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4600 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4601 and BUILD_EXEEXT
4602 Note: automatic setting of the BUILD variables is not yet working
4603 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4604 compile-time but not at link-time
4605 [LINK]: use for linking executables only
4606 make different versions for Windows and non-Windows
4607 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4608 libraries
4609 [LINK_FOR_BUILD]: new variable
4610 [OBJEXT]: use throughout
4611 [EXEEXT]: use throughout
4612 <winshared>: new target
4613 <wininstall>: new target
4614 <dftables.o>: use native compiler
4615 <dftables>: use native linker
4616 <install>: handle Windows platform correctly
4617 <clean>: ditto
4618 <check>: ditto
4619 copy DLL to top builddir before testing
4620
4621 As part of these changes, -no-undefined was removed again. This was reported
4622 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4623 in any case.
4624
4625 3. Some tidies to get rid of compiler warnings:
4626
4627 . In the match_data structure, match_limit was an unsigned long int, whereas
4628 match_call_count was an int. I've made them both unsigned long ints.
4629
4630 . In pcretest the fact that a const uschar * doesn't automatically cast to
4631 a void * provoked a warning.
4632
4633 . Turning on some more compiler warnings threw up some "shadow" variables
4634 and a few more missing casts.
4635
4636 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4637 option, a class that contained a single character with a value between 128
4638 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4639
4640 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4641 option, a class that contained several characters, but with at least one
4642 whose value was between 128 and 255 caused PCRE to crash.
4643
4644
4645 Version 4.1 12-Mar-03
4646 ---------------------
4647
4648 1. Compiling with gcc -pedantic found a couple of places where casts were
4649 needed, and a string in dftables.c that was longer than standard compilers are
4650 required to support.
4651
4652 2. Compiling with Sun's compiler found a few more places where the code could
4653 be tidied up in order to avoid warnings.
4654
4655 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4656 first of these names is deprecated in the latest Autoconf in favour of the name
4657 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4658 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4659 analogy I have changed it to CFLAGS_FOR_BUILD.
4660
4661 4. Added -no-undefined to the linking command in the Makefile, because this is
4662 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4663 linking step for the pcreposix library.
4664
4665 5. PCRE was failing to diagnose the case of two named groups with the same
4666 name.
4667
4668 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4669 literal character that is needed in the subject for a match, and scans along to
4670 ensure that it is present before embarking on the full matching process. This
4671 saves time in cases of nested unlimited repeats that are never going to match.
4672 Problem: the scan can take a lot of time if the subject is very long (e.g.
4673 megabytes), thus penalizing straightforward matches. It is now done only if the
4674 amount of subject to be scanned is less than 1000 bytes.
4675
4676 7. A lesser problem with the same optimization is that it was recording the
4677 first character of an anchored pattern as "needed", thus provoking a search
4678 right along the subject, even when the first match of the pattern was going to
4679 fail. The "needed" character is now not set for anchored patterns, unless it
4680 follows something in the pattern that is of non-fixed length. Thus, it still
4681 fulfils its original purpose of finding quick non-matches in cases of nested
4682 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4683
4684
4685 Version 4.0 17-Feb-03
4686 ---------------------
4687
4688 1. If a comment in an extended regex that started immediately after a meta-item
4689 extended to the end of string, PCRE compiled incorrect data. This could lead to
4690 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4691
4692 2. Moved to autoconf 2.53 and libtool 1.4.2.
4693
4694 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4695 the special perltest8 script is no longer needed - all the tests can be run
4696 from a single perltest script.
4697
4698 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4699 by \s. It has now been removed in PCRE. This means it isn't recognized as
4700 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4701 class [:space:] *does* include VT, thereby creating a mess.
4702
4703 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4704 space and tab.
4705
4706 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4707 its new features into the main test script, reducing the number of scripts.
4708
4709 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4710 were backward compatible, and made the (?i) apply to the whole pattern, as if
4711 /i were given. Now it behaves more logically, and applies the option setting
4712 only to what follows. PCRE has been changed to follow suit. However, if it
4713 finds options settings right at the start of the pattern, it extracts them into
4714 the global options, as before. Thus, they show up in the info data.
4715
4716 8. Added support for the \Q...\E escape sequence. Characters in between are
4717 treated as literals. This is slightly different from Perl in that $ and @ are
4718 also handled as literals inside the quotes. In Perl, they will cause variable
4719 interpolation. Note the following examples:
4720
4721 Pattern PCRE matches Perl matches
4722
4723 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4724 \Qabc\$xyz\E abc\$xyz abc\$xyz
4725 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4726
4727 For compatibility with Perl, \Q...\E sequences are recognized inside character
4728 classes as well as outside them.
4729
4730 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4731 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4732 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4733 signed/unsigned warnings.
4734
4735 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4736 option for pcretest, so I've replaced it by a simple function that does just
4737 that job.
4738
4739 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4740 "pcregrep -".
4741
4742 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4743 Java package. This provides some syntactic sugar for simple cases of what my
4744 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4745 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4746 item, you can use this simplified notation. Note that only makes sense with
4747 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4748 greediness, whatever the setting of the PCRE_UNGREEDY option.
4749
4750 13. A change of greediness default within a pattern was not taking effect at
4751 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4752 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4753 was abstracted outside.
4754
4755 14. PCRE now supports the \G assertion. It is true when the current matching
4756 position is at the start point of the match. This differs from \A when the
4757 starting offset is non-zero. Used with the /g option of pcretest (or similar
4758 code), it works in the same way as it does for Perl's /g option. If all
4759 alternatives of a regex begin with \G, the expression is anchored to the start
4760 match position, and the "anchored" flag is set in the compiled expression.
4761
4762 15. Some bugs concerning the handling of certain option changes within patterns
4763 have been fixed. These applied to options other than (?ims). For example,
4764 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4765 been the other way round. Some of this was related to change 7 above.
4766
4767 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4768 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4769 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4770 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4771
4772 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4773 mode. Unlike ".", it always matches newline, whatever the setting of
4774 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4775 assertions. Perl allows it, but it doesn't (in general) work because it can't
4776 calculate the length of the lookbehind. At least, that's the case for Perl
4777 5.8.0 - I've been told they are going to document that it doesn't work in
4778 future.
4779
4780 18. Added an error diagnosis for escapes that PCRE does not support: these are
4781 \L, \l, \N, \P, \p, \U, \u, and \X.
4782
4783 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4784 reading past the end of the pattern in cases such as /[abcd/.
4785
4786 20. PCRE was getting more memory than necessary for patterns with classes that
4787 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4788
4789 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4790 compiling PCRE for use with Virtual Pascal.
4791
4792 22. Small fix to the Makefile to make it work properly if the build is done
4793 outside the source tree.
4794
4795 23. Added a new extension: a condition to go with recursion. If a conditional
4796 subpattern starts with (?(R) the "true" branch is used if recursion has
4797 happened, whereas the "false" branch is used only at the top level.
4798
4799 24. When there was a very long string of literal characters (over 255 bytes
4800 without UTF support, over 250 bytes with UTF support), the computation of how
4801 much memory was required could be incorrect, leading to segfaults or other
4802 strange effects.
4803
4804 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4805 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4806 there was a subsequent back reference to those brackets. This meant that, for
4807 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4808 possible to check for precisely this case. All we can do is abandon the
4809 optimization if .* occurs inside capturing brackets when there are any back
4810 references whatsoever. (See below for a better fix that came later.)
4811
4812 26. The handling of the optimization for finding the first character of a
4813 non-anchored pattern, and for finding a character that is required later in the
4814 match were failing in some cases. This didn't break the matching; it just
4815 failed to optimize when it could. The way this is done has been re-implemented.
4816
4817 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4818
4819 28. Added a new feature that provides some of the functionality that Perl
4820 provides with (?{...}). The facility is termed a "callout". The way it is done
4821 in PCRE is for the caller to provide an optional function, by setting
4822 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4823 global variable. By default it is unset, which disables all calling out. To get
4824 the function called, the regex must include (?C) at appropriate points. This
4825 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4826 This provides a means of identifying different callout points. When PCRE
4827 reaches such a point in the regex, if pcre_callout has been set, the external
4828 function is called. It is provided with data in a structure called
4829 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4830 matching continues; if it returns a non-zero value, the match at the current
4831 point fails. However, backtracking will occur if possible. [This was changed
4832 later and other features added - see item 49 below.]
4833
4834 29. pcretest is upgraded to test the callout functionality. It provides a
4835 callout function that displays information. By default, it shows the start of
4836 the match and the current position in the text. There are some new data escapes
4837 to vary what happens:
4838
4839 \C+ in addition, show current contents of captured substrings
4840 \C- do not supply a callout function
4841 \C!n return 1 when callout number n is reached
4842 \C!n!m return 1 when callout number n is reached for the mth time
4843
4844 30. If pcregrep was called with the -l option and just a single file name, it
4845 output "<stdin>" if a match was found, instead of the file name.
4846
4847 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4848 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4849 pcre_exec(). This saves a malloc/free per call. The default value of
4850 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4851 when configuring.
4852
4853 32. The default maximum size of a compiled pattern is 64K. There have been a
4854 few cases of people hitting this limit. The code now uses macros to handle the
4855 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4856 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4857 configuring. Tests 2 and 5 work only with 2-byte links because they output
4858 debugging information about compiled patterns.
4859
4860 33. Internal code re-arrangements:
4861
4862 (a) Moved the debugging function for printing out a compiled regex into
4863 its own source file (printint.c) and used #include to pull it into
4864 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4865 separate copies.
4866
4867 (b) Defined the list of op-code names for debugging as a macro in
4868 internal.h so that it is next to the definition of the opcodes.
4869
4870 (c) Defined a table of op-code lengths for simpler skipping along compiled
4871 code. This is again a macro in internal.h so that it is next to the
4872 definition of the opcodes.
4873
4874 34. Added support for recursive calls to individual subpatterns, along the
4875 lines of Robin Houston's patch (but implemented somewhat differently).
4876
4877 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4878 allow it to read and process whole directories in Win32. This code was
4879 contributed by Lionel Fourquaux; it has not been tested by me.
4880
4881 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4882 used to name a group. Names consist of alphanumerics and underscores, and must
4883 be unique. Back references use the syntax (?P=name) and recursive calls use
4884 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4885 numbers. The function pcre_fullinfo() can be used after compilation to extract
4886 a name/number map. There are three relevant calls:
4887
4888 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4889 PCRE_INFO_NAMECOUNT yields the number of entries
4890 PCRE_INFO_NAMETABLE yields a pointer to the map.
4891
4892 The map is a vector of fixed-size entries. The size of each entry depends on
4893 the length of the longest name used. The first two bytes of each entry are the
4894 group number, most significant byte first. There follows the corresponding
4895 name, zero terminated. The names are in alphabetical order.
4896
4897 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4898 case instead of 255. Making it the same both with and without UTF-8 support
4899 means that the same test output works with both.
4900
4901 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4902 calling malloc() with a zero argument.
4903
4904 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4905 optimization. I've improved things by keeping a bitmap of backreferences with
4906 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4907 fact referenced, the optimization can be applied. It is unlikely that a
4908 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4909 the match to follow \n) will appear inside brackets with a number greater than
4910 31, but if it does, any back reference > 31 suppresses the optimization.
4911
4912 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4913 of disabling numbered capturing parentheses. Any opening parenthesis that is
4914 not followed by ? behaves as if it were followed by ?: but named parentheses
4915 can still be used for capturing (and they will acquire numbers in the usual
4916 way).
4917
4918 41. Redesigned the return codes from the match() function into yes/no/error so
4919 that errors can be passed back from deep inside the nested calls. A malloc
4920 failure while inside a recursive subpattern call now causes the
4921 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4922
4923 42. It is now possible to set a limit on the number of times the match()
4924 function is called in a call to pcre_exec(). This facility makes it possible to
4925 limit the amount of recursion and backtracking, though not in a directly
4926 obvious way, because the match() function is used in a number of different
4927 circumstances. The count starts from zero for each position in the subject
4928 string (for non-anchored patterns). The default limit is, for compatibility, a
4929 large number, namely 10 000 000. You can change this in two ways:
4930
4931 (a) When configuring PCRE before making, you can use --with-match-limit=n
4932 to set a default value for the compiled library.
4933
4934 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4935 a different value is set. See 45 below.
4936
4937 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4938
4939 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4940 of things that can be changed at compile time. The first argument specifies
4941 what is wanted and the second points to where the information is to be placed.
4942 The current list of available information is:
4943
4944 PCRE_CONFIG_UTF8
4945
4946 The output is an integer that is set to one if UTF-8 support is available;
4947 otherwise it is set to zero.
4948
4949 PCRE_CONFIG_NEWLINE
4950
4951 The output is an integer that it set to the value of the code that is used for
4952 newline. It is either LF (10) or CR (13).
4953
4954 PCRE_CONFIG_LINK_SIZE
4955
4956 The output is an integer that contains the number of bytes used for internal
4957 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4958
4959 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4960
4961 The output is an integer that contains the threshold above which the POSIX
4962 interface uses malloc() for output vectors. See item 31 above.
4963
4964 PCRE_CONFIG_MATCH_LIMIT
4965
4966 The output is an unsigned integer that contains the default limit of the number
4967 of match() calls in a pcre_exec() execution. See 42 above.
4968
4969 44. pcretest has been upgraded by the addition of the -C option. This causes it
4970 to extract all the available output from the new pcre_config() function, and to
4971 output it. The program then exits immediately.
4972
4973 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4974 order to support additional features. One way would have been to define
4975 pcre_exec2() (for example) with extra arguments, but this would not have been
4976 extensible, and would also have required all calls to the original function to
4977 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4978 is used for passing in "extra" data from pcre_study().
4979
4980 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4981 contains the following fields:
4982
4983 flags a bitmap indicating which of the following fields are set
4984 study_data opaque data from pcre_study()
4985 match_limit a way of specifying a limit on match() calls for a specific
4986 call to pcre_exec()
4987 callout_data data for callouts (see 49 below)
4988
4989 The flag bits are also defined in pcre.h, and are
4990
4991 PCRE_EXTRA_STUDY_DATA
4992 PCRE_EXTRA_MATCH_LIMIT
4993 PCRE_EXTRA_CALLOUT_DATA
4994
4995 The pcre_study() function now returns one of these new pcre_extra blocks, with
4996 the actual study data pointed to by the study_data field, and the
4997 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4998 before. That is, this change is entirely upwards-compatible and requires no
4999 change to existing code.
5000
5001 If you want to pass in additional data to pcre_exec(), you can either place it
5002 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5003 block.
5004
5005 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5006 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5007 times with different match limits, until it finds the minimum value needed for
5008 pcre_exec() to complete. The value is then output. This can be instructive; for
5009 most simple matches the number is quite small, but for pathological cases it
5010 gets very large very quickly.
5011
5012 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5013 returns the size of the data block pointed to by the study_data field in a
5014 pcre_extra block, that is, the value that was passed as the argument to
5015 pcre_malloc() when PCRE was getting memory in which to place the information
5016 created by pcre_study(). The fourth argument should point to a size_t variable.
5017 pcretest has been extended so that this information is shown after a successful
5018 pcre_study() call when information about the compiled regex is being displayed.
5019
5020 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5021 because what follows is always an absolute path. (Later: it turns out that this
5022 is more than cosmetic for MinGW, because it doesn't like empty path
5023 components.)
5024
5025 49. Some changes have been made to the callout feature (see 28 above):
5026
5027 (i) A callout function now has three choices for what it returns:
5028
5029 0 => success, carry on matching
5030 > 0 => failure at this point, but backtrack if possible
5031 < 0 => serious error, return this value from pcre_exec()
5032
5033 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5034 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5035 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5036 use by callout functions. It will never be used by PCRE itself.
5037
5038 (ii) The pcre_extra structure (see 45 above) has a void * field called
5039 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5040 pcre_callout_block structure has a field of the same name. The contents of
5041 the field passed in the pcre_extra structure are passed to the callout
5042 function in the corresponding field in the callout block. This makes it
5043 easier to use the same callout-containing regex from multiple threads. For
5044 testing, the pcretest program has a new data escape
5045
5046 \C*n pass the number n (may be negative) as callout_data
5047
5048 If the callout function in pcretest receives a non-zero value as
5049 callout_data, it returns that value.
5050
5051 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5052 there were some redundant $(CFLAGS) in commands that are now specified as
5053 $(LINK), which already includes $(CFLAGS).
5054
5055 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5056 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5057 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5058 one-byte characters throughout. Note that case-insensitive matching applies
5059 only to characters whose values are less than 256. PCRE doesn't support the
5060 notion of cases for higher-valued characters.
5061
5062 (i) A character class whose characters are all within 0-255 is handled as
5063 a bit map, and the map is inverted for negative classes. Previously, a
5064 character > 255 always failed to match such a class; however it should
5065 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5066
5067 (ii) A negated character class with a single character < 255 is coded as
5068 "not this character" (OP_NOT). This wasn't working properly when the test
5069 character was multibyte, either singly or repeated.
5070
5071 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5072 mode, for example: \x{100}{2,3}.
5073
5074 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5075 singly or repeated) now correctly test multibyte characters. However,
5076 PCRE doesn't recognize any characters with values greater than 255 as
5077 digits, spaces, or word characters. Such characters always match \D, \S,
5078 and \W, and never match \d, \s, or \w.
5079
5080 (v) Classes may now contain characters and character ranges with values
5081 greater than 255. For example: [ab\x{100}-\x{400}].
5082
5083 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5084 PCRE in UTF-8 mode.
5085
5086 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5087 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5088 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5089 value.)
5090
5091 53. The single man page has become too large. I have therefore split it up into
5092 a number of separate man pages. These also give rise to individual HTML pages;
5093 these are now put in a separate directory, and there is an index.html page that
5094 lists them all. Some hyperlinking between the pages has been installed.
5095
5096 54. Added convenience functions for handling named capturing parentheses.
5097
5098 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5099 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5100 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5101 are faulted.
5102
5103 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5104 calling configure. These values are used when compiling the dftables.c program
5105 which is run to generate the source of the default character tables. They
5106 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5107 you will need to set these values.
5108
5109 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5110
5111
5112 Version 3.9 02-Jan-02
5113 ---------------------
5114
5115 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5116
5117 2. If --disable-static was given, the building process failed when trying to
5118 build pcretest and pcregrep. (For some reason it was using libtool to compile
5119 them, which is not right, as they aren't part of the library.)
5120
5121
5122 Version 3.8 18-Dec-01
5123 ---------------------
5124
5125 1. The experimental UTF-8 code was completely screwed up. It was packing the
5126 bytes in the wrong order. How dumb can you get?
5127
5128
5129 Version 3.7 29-Oct-01
5130 ---------------------
5131
5132 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5133 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5134 this didn't happen under Solaris 8, where I normally test things.
5135
5136 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5137 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5138 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5139 this fix an hour or so after the initial 3.7 release.)
5140
5141
5142 Version 3.6 23-Oct-01
5143 ---------------------
5144
5145 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5146 offsets passed as NULL with zero offset count.
5147
5148 2. The config.guess and config.sub files had not been updated when I moved to
5149 the latest autoconf.
5150
5151
5152 Version 3.5 15-Aug-01
5153 ---------------------
5154
5155 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5156 had been forgotten.
5157
5158 2. By using declared but undefined structures, we can avoid using "void"
5159 definitions in pcre.h while keeping the internal definitions of the structures
5160 private.
5161
5162 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5163 user point of view, this means that both static and shared libraries are built
5164 by default, but this can be individually controlled. More of the work of
5165 handling this static/shared cases is now inside libtool instead of PCRE's make
5166 file.
5167
5168 4. The pcretest utility is now installed along with pcregrep because it is
5169 useful for users (to test regexs) and by doing this, it automatically gets
5170 relinked by libtool. The documentation has been turned into a man page, so
5171 there are now .1, .txt, and .html versions in /doc.
5172
5173 5. Upgrades to pcregrep:
5174 (i) Added long-form option names like gnu grep.
5175 (ii) Added --help to list all options with an explanatory phrase.
5176 (iii) Added -r, --recursive to recurse into sub-directories.
5177 (iv) Added -f, --file to read patterns from a file.
5178
5179 6. pcre_exec() was referring to its "code" argument before testing that
5180 argument for NULL (and giving an error if it was NULL).
5181
5182 7. Upgraded Makefile.in to allow for compiling in a different directory from
5183 the source directory.
5184
5185 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5186 options bits, the pointer it was passed was to an int instead of to an unsigned
5187 long int. This mattered only on 64-bit systems.
5188
5189 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5190 generated) instead of pcre.in, which it its source. Also made the same change
5191 in several of the .c files.
5192
5193 10. A new release of gcc defines printf() as a macro, which broke pcretest
5194 because it had an ifdef in the middle of a string argument for printf(). Fixed
5195 by using separate calls to printf().
5196
5197 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5198 script, to force use of CR or LF instead of \n in the source. On non-Unix
5199 systems, the value can be set in config.h.
5200
5201 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5202 absolute limit. Changed the text of the error message to make this clear, and
5203 likewise updated the man page.
5204
5205 13. The limit of 99 on the number of capturing subpatterns has been removed.
5206 The new limit is 65535, which I hope will not be a "real" limit.
5207
5208
5209 Version 3.4 22-Aug-00
5210 ---------------------
5211
5212 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5213
5214 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5215
5216
5217 Version 3.3 01-Aug-00
5218 ---------------------
5219
5220 1. If an octal character was given, but the value was greater than \377, it
5221 was not getting masked to the least significant bits, as documented. This could
5222 lead to crashes in some systems.
5223
5224 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5225 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5226
5227 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5228 These just pass their arguments on to (pcre_free)(), but they are provided
5229 because some uses of PCRE bind it to non-C systems that can call its functions,
5230 but cannot call free() or pcre_free() directly.
5231
5232 4. Add "make test" as a synonym for "make check". Corrected some comments in
5233 the Makefile.
5234
5235 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5236 Makefile.
5237
5238 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5239 command called pgrep for grepping around the active processes.
5240
5241 7. Added the beginnings of support for UTF-8 character strings.
5242
5243 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5244 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5245 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5246 out for the ar command.)
5247
5248
5249 Version 3.2 12-May-00
5250 ---------------------
5251
5252 This is purely a bug fixing release.
5253
5254 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5255 of ZA. This was just one example of several cases that could provoke this bug,
5256 which was introduced by change 9 of version 2.00. The code for breaking
5257 infinite loops after an iteration that matches an empty string was't working
5258 correctly.
5259
5260 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5261 when matched against abbab (for example). After matching an empty string, it
5262 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5263 caused it to match further down the string than it should.
5264
5265 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5266 was there because it doesn't seem to be needed, and it causes trouble on some
5267 systems, as it is not a Standard C header. It has been removed.
5268
5269 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5270 were reported on the Macintosh. The changes were from
5271
5272 while ((c = *(++ptr)) != 0 && c != '\n');
5273 to
5274 while ((c = *(++ptr)) != 0 && c != '\n') ;
5275
5276 Totally extraordinary, but if that's what it takes...
5277
5278 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5279 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5280 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5281 assumes the way PCRE uses memmove() (always moving upwards).
5282
5283 6. PCRE is being used in one environment where strchr() is not available. There
5284 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5285 faster code anyway.
5286
5287
5288 Version 3.1 09-Feb-00
5289 ---------------------
5290
5291 The only change in this release is the fixing of some bugs in Makefile.in for
5292 the "install" target:
5293
5294 (1) It was failing to install pcreposix.h.
5295
5296 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5297
5298
5299 Version 3.0 01-Feb-00
5300 ---------------------
5301
5302 1. Add support for the /+ modifier to perltest (to output $` like it does in
5303 pcretest).
5304
5305 2. Add support for the /g modifier to perltest.
5306
5307 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5308 matches null strings.
5309
5310 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5311 pattern. Perl treats empty patterns specially - it reuses the most recent
5312 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5313 effect.
5314
5315 5. The POSIX interface was broken in that it was just handing over the POSIX
5316 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5317 required a bigger vector, with some working space on the end. This means that
5318 the POSIX wrapper now has to get and free some memory, and copy the results.
5319
5320 6. Added some simple autoconf support, placing the test data and the
5321 documentation in separate directories, re-organizing some of the
5322 information files, and making it build pcre-config (a GNU standard). Also added
5323 libtool support for building PCRE as a shared library, which is now the
5324 default.
5325
5326 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5327 09 are not valid octal constants. Single digits will be used for minor values
5328 less than 10.
5329
5330 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5331 existing programs that set these in the POSIX interface can use PCRE without
5332 modification.
5333
5334 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5335 return all that pcre_info() returns, plus additional data. The pcre_info()
5336 function is retained for compatibility, but is considered to be obsolete.
5337
5338 10. Added experimental recursion feature (?R) to handle one common case that
5339 Perl 5.6 will be able to do with (?p{...}).
5340
5341 11. Added support for POSIX character classes like [:alpha:], which Perl is
5342 adopting.
5343
5344
5345 Version 2.08 31-Aug-99
5346 ----------------------
5347
5348 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5349 trying to match at the startoffset position, but instead was moving forward to
5350 the next newline as if a previous match had failed.
5351
5352 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5353 and could get into a loop if a null string was matched other than at the start
5354 of the subject.
5355
5356 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5357 be distinguished at compile time, and for completeness also added PCRE_DATE.
5358
5359 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5360 in GnuWin32 environments.
5361
5362
5363 Version 2.07 29-Jul-99
5364 ----------------------
5365
5366 1. The documentation is now supplied in plain text form and HTML as well as in
5367 the form of man page sources.
5368
5369 2. C++ compilers don't like assigning (void *) values to other pointer types.
5370 In particular this affects malloc(). Although there is no problem in Standard
5371 C, I've put in casts to keep C++ compilers happy.
5372
5373 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5374 should be (const char *).
5375
5376 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5377 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5378 However, I haven't made this a standard facility. The documentation doesn't
5379 mention it, and the Makefile doesn't support it.
5380
5381 5. The Makefile now contains an "install" target, with editable destinations at
5382 the top of the file. The pcretest program is not installed.
5383
5384 6. pgrep -V now gives the PCRE version number and date.
5385
5386 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5387 causing the entire string to be ignored, instead of just the last character.
5388
5389 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5390 non-matching string, it can take a very, very long time, even for strings of
5391 quite modest length, because of the nested recursion. PCRE now does better in
5392 some of these cases. It does this by remembering the last required literal
5393 character in the pattern, and pre-searching the subject to ensure it is present
5394 before running the real match. In other words, it applies a heuristic to detect
5395 some types of certain failure quickly, and in the above example, if presented
5396 with a string that has no trailing " it gives "no match" very quickly.
5397
5398 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5399 other alternatives are tried instead.
5400
5401
5402 Version 2.06 09-Jun-99
5403 ----------------------
5404
5405 1. Change pcretest's output for amount of store used to show just the code
5406 space, because the remainder (the data block) varies in size between 32-bit and
5407 64-bit systems.
5408
5409 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5410 start matching at. This allows lookbehinds to work when searching for multiple
5411 occurrences in a string.
5412
5413 3. Added additional options to pcretest for testing multiple occurrences:
5414
5415 /+ outputs the rest of the string that follows a match
5416 /g loops for multiple occurrences, using the new startoffset argument
5417 /G loops for multiple occurrences by passing an incremented pointer
5418
5419 4. PCRE wasn't doing the "first character" optimization for patterns starting
5420 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5421 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5422 the letter 'x'. On long subject strings, this gives a significant speed-up.
5423
5424
5425 Version 2.05 21-Apr-99
5426 ----------------------
5427
5428 1. Changed the type of magic_number from int to long int so that it works
5429 properly on 16-bit systems.
5430
5431 2. Fixed a bug which caused patterns starting with .* not to work correctly
5432 when the subject string contained newline characters. PCRE was assuming
5433 anchoring for such patterns in all cases, which is not correct because .* will
5434 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5435 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5436 must be retried after every newline in the subject.
5437
5438
5439 Version 2.04 18-Feb-99
5440 ----------------------
5441
5442 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5443 computation of the store needed to hold the pattern was incorrect (too large).
5444 If such patterns were nested a few deep, this could multiply and become a real
5445 problem.
5446
5447 2. Added /M option to pcretest to show the memory requirement of a specific
5448 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5449
5450 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5451 compiled in such a way that the backtracking after subsequent failure was
5452 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5453 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5454
5455
5456 Version 2.03 02-Feb-99
5457 ----------------------
5458
5459 1. Fixed typo and small mistake in man page.
5460
5461 2. Added 4th condition (GPL supersedes if conflict) and created separate
5462 LICENCE file containing the conditions.
5463
5464 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5465 Perl, that is the internal \ allows the delimiter to be included in the
5466 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5467 the final delimiter, add \ to the end of the pattern (to test the error).
5468
5469 4. Added the convenience functions for extracting substrings after a successful
5470 match. Updated pcretest to make it able to test these functions.
5471
5472
5473 Version 2.02 14-Jan-99
5474 ----------------------
5475
5476 1. Initialized the working variables associated with each extraction so that
5477 their saving and restoring doesn't refer to uninitialized store.
5478
5479 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5480 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5481 fix the problem.
5482
5483 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5484 calls, and wasn't printing the correct value for compiling calls. Increased the
5485 default value of LOOPREPEAT, and the number of significant figures in the
5486 times.
5487
5488 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5489
5490 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5491 a building problem on Windows NT with a FAT file system.
5492
5493
5494 Version 2.01 21-Oct-98
5495 ----------------------
5496
5497 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5498 to character tables built by pcre_maketables() in the current locale. If NULL
5499 is passed, the default tables are used.
5500
5501
5502 Version 2.00 24-Sep-98
5503 ----------------------
5504
5505 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5506 it any more.
5507
5508 2. Allow quantification of (?>) groups, and make it work correctly.
5509
5510 3. The first character computation wasn't working for (?>) groups.
5511
5512 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5513 end of the subject) and add 5.005's \z, which really does match only at the
5514 very end of the subject.
5515
5516 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5517
5518 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5519 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5520 localized options. All options to pcre_study() were also removed.
5521
5522 7. Add other new features from 5.005:
5523
5524 $(?<= positive lookbehind
5525 $(?<! negative lookbehind
5526 (?imsx-imsx) added the unsetting capability
5527 such a setting is global if at outer level; local otherwise
5528 (?imsx-imsx:) non-capturing groups with option setting
5529 (?(cond)re|re) conditional pattern matching
5530
5531 A backreference to itself in a repeated group matches the previous
5532 captured string.
5533
5534 8. General tidying up of studying (both automatic and via "study")
5535 consequential on the addition of new assertions.
5536
5537 9. As in 5.005, unlimited repeated groups that could match an empty substring
5538 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5539 runtime if any iteration does actually match an empty substring.
5540
5541 10. Include the RunTest script in the distribution.
5542
5543 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5544 discrepancies, some of which were old and were also with respect to 5.004. They
5545 have now been fixed.
5546
5547
5548 Version 1.09 28-Apr-98
5549 ----------------------
5550
5551 1. A negated single character class followed by a quantifier with a minimum
5552 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5553 program crashes, or just wrong answers. This did not apply to negated classes
5554 containing more than one character, or to minima other than one.
5555
5556
5557 Version 1.08 27-Mar-98
5558 ----------------------
5559
5560 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5561
5562 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5563 latter must appear before anything that relies on it in the pattern.
5564
5565
5566 Version 1.07 16-Feb-98
5567 ----------------------
5568
5569 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5570 repeat of a potentially empty string).
5571
5572
5573 Version 1.06 23-Jan-98
5574 ----------------------
5575
5576 1. Added Markus Oberhumer's little patches for C++.
5577
5578 2. Literal strings longer than 255 characters were broken.
5579
5580
5581 Version 1.05 23-Dec-97
5582 ----------------------
5583
5584 1. Negated character classes containing more than one character were failing if
5585 PCRE_CASELESS was set at run time.
5586
5587
5588 Version 1.04 19-Dec-97
5589 ----------------------
5590
5591 1. Corrected the man page, where some "const" qualifiers had been omitted.
5592
5593 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5594 input syntax.
5595
5596 3. Fixed memory leak which occurred when a regex with back references was
5597 matched with an offsets vector that wasn't big enough. The temporary memory
5598 that is used in this case wasn't being freed if the match failed.
5599
5600 4. Tidied pcretest to ensure it frees memory that it gets.
5601
5602 5. Temporary memory was being obtained in the case where the passed offsets
5603 vector was exactly big enough.
5604
5605 6. Corrected definition of offsetof() from change 5 below.
5606
5607 7. I had screwed up change 6 below and broken the rules for the use of
5608 setjmp(). Now fixed.
5609
5610
5611 Version 1.03 18-Dec-97
5612 ----------------------
5613
5614 1. A erroneous regex with a missing opening parenthesis was correctly
5615 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5616 on some systems.
5617
5618 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5619 it was reported that one broken compiler failed on the former because "code" is
5620 also an independent variable.
5621
5622 3. The erroneous regex a[]b caused an array overrun reference.
5623
5624 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5625 fail on data ending with that character. (It was going on too far, and checking
5626 the next character, typically a binary zero.) This was specific to the
5627 optimized code for single-character negative classes.
5628
5629 5. Added a contributed patch from the TIN world which does the following:
5630
5631 + Add an undef for memmove, in case the the system defines a macro for it.
5632
5633 + Add a definition of offsetof(), in case there isn't one. (I don't know
5634 the reason behind this - offsetof() is part of the ANSI standard - but
5635 it does no harm).
5636
5637 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5638 most of the places where whitespace preceded '#'. I have given up and
5639 allowed the remaining 2 cases to be at the margin.
5640
5641 + Rename some variables in pcre to eliminate shadowing. This seems very
5642 pedantic, but does no harm, of course.
5643
5644 6. Moved the call to setjmp() into its own function, to get rid of warnings
5645 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5646
5647 7. Constructs such as \d{8,} were compiling into the equivalent of
5648 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5649 outcome, but in this particular case used more store than had been allocated,
5650 which caused the bug to be discovered because it threw up an internal error.
5651
5652 8. The debugging code in both pcre and pcretest for outputting the compiled
5653 form of a regex was going wrong in the case of back references followed by
5654 curly-bracketed repeats.
5655
5656
5657 Version 1.02 12-Dec-97
5658 ----------------------
5659
5660 1. Typos in pcre.3 and comments in the source fixed.
5661
5662 2. Applied a contributed patch to get rid of places where it used to remove
5663 'const' from variables, and fixed some signed/unsigned and uninitialized
5664 variable warnings.
5665
5666 3. Added the "runtest" target to Makefile.
5667
5668 4. Set default compiler flag to -O2 rather than just -O.
5669
5670
5671 Version 1.01 19-Nov-97
5672 ----------------------
5673
5674 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5675 like /([ab]*)*/, that is, for classes with more than one character in them.
5676
5677 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5678 as /((?>a*))*/ (a PCRE_EXTRA facility).
5679
5680
5681 Version 1.00 18-Nov-97
5682 ----------------------
5683
5684 1. Added compile-time macros to support systems such as SunOS4 which don't have
5685 memmove() or strerror() but have other things that can be used instead.
5686
5687 2. Arranged that "make clean" removes the executables.
5688
5689
5690 Version 0.99 27-Oct-97
5691 ----------------------
5692
5693 1. Fixed bug in code for optimizing classes with only one character. It was
5694 initializing a 32-byte map regardless, which could cause it to run off the end
5695 of the memory it had got.
5696
5697 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5698
5699
5700 Version 0.98 22-Oct-97
5701 ----------------------
5702
5703 1. Fixed bug in code for handling temporary memory usage when there are more
5704 back references than supplied space in the ovector. This could cause segfaults.
5705
5706
5707 Version 0.97 21-Oct-97
5708 ----------------------
5709
5710 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5711
5712 2. Optimized negated single characters not to use a bit map.
5713
5714 3. Brought error texts together as macro definitions; clarified some of them;
5715 fixed one that was wrong - it said "range out of order" when it meant "invalid
5716 escape sequence".
5717
5718 4. Changed some char * arguments to const char *.
5719
5720 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5721
5722 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5723 pcretest.
5724
5725
5726 Version 0.96 16-Oct-97
5727 ----------------------
5728
5729 1. Added a simple "pgrep" utility to the distribution.
5730
5731 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5732 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5733 where "ddd" means "one or more decimal digits".
5734
5735 3. Fixed serious bug. If a pattern had a back reference, but the call to
5736 pcre_exec() didn't supply a large enough ovector to record the related
5737 identifying subpattern, the match always failed. PCRE now remembers the number
5738 of the largest back reference, and gets some temporary memory in which to save
5739 the offsets during matching if necessary, in order to ensure that
5740 backreferences always work.
5741
5742 4. Increased the compatibility with Perl in a number of ways:
5743
5744 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5745 to request this handling. The option can be set at compile or exec time.
5746
5747 (b) $ matches before a terminating newline by default; an option
5748 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5749 mode). The option can be set at compile or exec time.
5750
5751 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5752 the same as Perl's. If the decimal number it represents is less than 10
5753 or there aren't that many previous left capturing parentheses, an octal
5754 escape is read. Inside a character class, it's always an octal escape,
5755 even if it is a single digit.
5756
5757 (d) An escaped but undefined alphabetic character is taken as a literal,
5758 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5759 escapes.
5760
5761 (e) {0} is now permitted. (The previous item is removed from the compiled
5762 pattern).
5763
5764 5. Changed all the names of code files so that the basic parts are no longer
5765 than 10 characters, and abolished the teeny "globals.c" file.
5766
5767 6. Changed the handling of character classes; they are now done with a 32-byte
5768 bit map always.
5769
5770 7. Added the -d and /D options to pcretest to make it possible to look at the
5771 internals of compilation without having to recompile pcre.
5772
5773
5774 Version 0.95 23-Sep-97
5775 ----------------------
5776
5777 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5778 \x20 at the start of a run of normal characters. These were being treated as
5779 real characters, instead of the source characters being re-checked.
5780
5781
5782 Version 0.94 18-Sep-97
5783 ----------------------
5784
5785 1. The functions are now thread-safe, with the caveat that the global variables
5786 containing pointers to malloc() and free() or alternative functions are the
5787 same for all threads.
5788
5789 2. Get pcre_study() to generate a bitmap of initial characters for non-
5790 anchored patterns when this is possible, and use it if passed to pcre_exec().
5791
5792
5793 Version 0.93 15-Sep-97
5794 ----------------------
5795
5796 1. /(b)|(:+)/ was computing an incorrect first character.
5797
5798 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5799 but not actually doing anything yet.
5800
5801 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5802 as Perl does (e.g. [-az] or [az-]).
5803
5804 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5805 all possible positions.
5806
5807 5. Split up into different modules to avoid including unneeded functions in a
5808 compiled binary. However, compile and exec are still in one module. The "study"
5809 function is split off.
5810
5811 6. The character tables are now in a separate module whose source is generated
5812 by an auxiliary program - but can then be edited by hand if required. There are
5813 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5814 toupper() in the code.
5815
5816 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5817 make them global. Abolish the function for setting them, as the caller can now
5818 set them directly.
5819
5820
5821 Version 0.92 11-Sep-97
5822 ----------------------
5823
5824 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5825 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5826
5827 2. Caseless matching was not working in character classes if the characters in
5828 the pattern were in upper case.
5829
5830 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5831
5832 4. Make PCRE_ANCHORED public and accept as a compile option.
5833
5834 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5835 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5836 pass them.
5837
5838 6. Give an error if bad option bits passed at compile or run time.
5839
5840 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5841 pcretest to cause it to pass that flag.
5842
5843 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5844 options, and the first character, if set.
5845
5846 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5847
5848
5849 Version 0.91 10-Sep-97
5850 ----------------------
5851
5852 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5853 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5854
5855 2. PCRE was looping on encountering an indefinitely repeated back reference to
5856 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5857 Perl does - treats the match as successful.
5858
5859 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5