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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1565 - (show annotations)
Tue Jun 9 16:46:52 2015 UTC (4 years, 5 months ago) by ph10
File size: 279627 byte(s)
Error occurred while calculating annotation data.
Fix overflow when ovector has size 1.
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
52 Version 8.37 28-April-2015
53 --------------------------
54
55 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
56 for those parentheses to be closed with whatever has been captured so far.
57 However, it was failing to mark any other groups between the hightest
58 capture so far and the currrent group as "unset". Thus, the ovector for
59 those groups contained whatever was previously there. An example is the
60 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
61
62 2. If an assertion condition was quantified with a minimum of zero (an odd
63 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
64
65 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
66 unrecognized modifier, a crash could occur.
67
68 4. An attempt to do global matching in pcretest with a zero-length ovector
69 caused a crash.
70
71 5. Fixed a memory leak during matching that could occur for a subpattern
72 subroutine call (recursive or otherwise) if the number of captured groups
73 that had to be saved was greater than ten.
74
75 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
76 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
77 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
78
79 7. A UTF pattern containing a "not" match of a non-ASCII character and a
80 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
81
82 8. When a pattern is compiled, it remembers the highest back reference so that
83 when matching, if the ovector is too small, extra memory can be obtained to
84 use instead. A conditional subpattern whose condition is a check on a
85 capture having happened, such as, for example in the pattern
86 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
87 setting the highest backreference number. This mattered only if pcre_exec()
88 was called with an ovector that was too small to hold the capture, and there
89 was no other kind of back reference (a situation which is probably quite
90 rare). The effect of the bug was that the condition was always treated as
91 FALSE when the capture could not be consulted, leading to a incorrect
92 behaviour by pcre_exec(). This bug has been fixed.
93
94 9. A reference to a duplicated named group (either a back reference or a test
95 for being set in a conditional) that occurred in a part of the pattern where
96 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
97 to be incorrectly calculated, leading to overwriting.
98
99 10. A mutually recursive set of back references such as (\2)(\1) caused a
100 segfault at study time (while trying to find the minimum matching length).
101 The infinite loop is now broken (with the minimum length unset, that is,
102 zero).
103
104 11. If an assertion that was used as a condition was quantified with a minimum
105 of zero, matching went wrong. In particular, if the whole group had
106 unlimited repetition and could match an empty string, a segfault was
107 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
108 assertions to be quantified, but not if they are being used as conditions,
109 so the above pattern is faulted by Perl. PCRE has now been changed so that
110 it also rejects such patterns.
111
112 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
113 failed to allow the zero-repeat case if pcre2_exec() was called with an
114 ovector too small to capture the group.
115
116 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
117 Red Hat Product Security:
118
119 (a) A crash if /K and /F were both set with the option to save the compiled
120 pattern.
121
122 (b) Another crash if the option to print captured substrings in a callout
123 was combined with setting a null ovector, for example \O\C+ as a subject
124 string.
125
126 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
127 forward reference repeated a large (but limited) number of times within a
128 repeated outer group that has a zero minimum quantifier, caused incorrect
129 code to be compiled, leading to the error "internal error:
130 previously-checked referenced subpattern not found" when an incorrect
131 memory address was read. This bug was reported as "heap overflow",
132 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
133 CVE-2015-2325.
134
135 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
136 call within a group that also contained a recursive back reference caused
137 incorrect code to be compiled. This bug was reported as "heap overflow",
138 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
139 number CVE-2015-2326.
140
141 24. Computing the size of the JIT read-only data in advance has been a source
142 of various issues, and new ones are still appear unfortunately. To fix
143 existing and future issues, size computation is eliminated from the code,
144 and replaced by on-demand memory allocation.
145
146 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
147 adjacent to the end of the range, and the range contained characters with
148 more than one other case, caused incorrect behaviour when compiled in UTF
149 mode. In that example, the range a-j was left out of the class.
150
151 26. Fix JIT compilation of conditional blocks, which assertion
152 is converted to (*FAIL). E.g: /(?(?!))/.
153
154 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
155 discovered by the LLVM fuzzer.
156
157 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
158 when this assertion was used as a condition, for example (?(?!)a|b). In
159 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
160 error about an unsupported item.
161
162 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
163 possessification code could take exponential time to complete. A recursion
164 depth limit of 1000 has been imposed to limit the resources used by this
165 optimization.
166
167 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
168 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
169 because \S ensures they are all in the class. The code for doing this was
170 interacting badly with the code for computing the amount of space needed to
171 compile the pattern, leading to a buffer overflow. This bug was discovered
172 by the LLVM fuzzer.
173
174 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
175 other kinds of group caused stack overflow at compile time. This bug was
176 discovered by the LLVM fuzzer.
177
178 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
179 between a subroutine call and its quantifier was incorrectly compiled,
180 leading to buffer overflow or other errors. This bug was discovered by the
181 LLVM fuzzer.
182
183 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
184 assertion after (?(. The code was failing to check the character after
185 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
186 was discovered by the LLVM fuzzer.
187
188 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
189 a fixed maximum following a group that contains a subroutine reference was
190 incorrectly compiled and could trigger buffer overflow. This bug was
191 discovered by the LLVM fuzzer.
192
193 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
194 caused a stack overflow instead of the diagnosis of a non-fixed length
195 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
196
197 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
198 (e.g. /(?<=\Ka)/) could make pcregrep loop.
199
200 37. There was a similar problem to 36 in pcretest for global matches.
201
202 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
203 and a subsequent item in the pattern caused a non-match, backtracking over
204 the repeated \X did not stop, but carried on past the start of the subject,
205 causing reference to random memory and/or a segfault. There were also some
206 other cases where backtracking after \C could crash. This set of bugs was
207 discovered by the LLVM fuzzer.
208
209 39. The function for finding the minimum length of a matching string could take
210 a very long time if mutual recursion was present many times in a pattern,
211 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
212 method has been implemented. This infelicity was discovered by the LLVM
213 fuzzer.
214
215 40. Static linking against the PCRE library using the pkg-config module was
216 failing on missing pthread symbols.
217
218
219 Version 8.36 26-September-2014
220 ------------------------------
221
222 1. Got rid of some compiler warnings in the C++ modules that were shown up by
223 -Wmissing-field-initializers and -Wunused-parameter.
224
225 2. The tests for quantifiers being too big (greater than 65535) were being
226 applied after reading the number, and stupidly assuming that integer
227 overflow would give a negative number. The tests are now applied as the
228 numbers are read.
229
230 3. Tidy code in pcre_exec.c where two branches that used to be different are
231 now the same.
232
233 4. The JIT compiler did not generate match limit checks for certain
234 bracketed expressions with quantifiers. This may lead to exponential
235 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
236 issue should be resolved now.
237
238 5. Fixed an issue, which occures when nested alternatives are optimized
239 with table jumps.
240
241 6. Inserted two casts and changed some ints to size_t in the light of some
242 reported 64-bit compiler warnings (Bugzilla 1477).
243
244 7. Fixed a bug concerned with zero-minimum possessive groups that could match
245 an empty string, which sometimes were behaving incorrectly in the
246 interpreter (though correctly in the JIT matcher). This pcretest input is
247 an example:
248
249 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
250 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
251
252 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
253 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
254 for an empty string was breaking the inner loop and carrying on at a lower
255 level, when possessive repeated groups should always return to a higher
256 level as they have no backtrack points in them. The empty string test now
257 occurs at the outer level.
258
259 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
260 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
261
262 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
263 doesn't).
264
265 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
266 to hold a pattern to be incorrectly computed (too small) when there were
267 named back references to duplicated names. This could cause "internal
268 error: code overflow" or "double free or corruption" or other memory
269 handling errors.
270
271 11. When named subpatterns had the same prefixes, back references could be
272 confused. For example, in this pattern:
273
274 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
275
276 the reference to 'Name' was incorrectly treated as a reference to a
277 duplicate name.
278
279 12. A pattern such as /^s?c/mi8 where the optional character has more than
280 one "other case" was incorrectly compiled such that it would only try to
281 match starting at "c".
282
283 13. When a pattern starting with \s was studied, VT was not included in the
284 list of possible starting characters; this should have been part of the
285 8.34/18 patch.
286
287 14. If a character class started [\Qx]... where x is any character, the class
288 was incorrectly terminated at the ].
289
290 15. If a pattern that started with a caseless match for a character with more
291 than one "other case" was studied, PCRE did not set up the starting code
292 unit bit map for the list of possible characters. Now it does. This is an
293 optimization improvement, not a bug fix.
294
295 16. The Unicode data tables have been updated to Unicode 7.0.0.
296
297 17. Fixed a number of memory leaks in pcregrep.
298
299 18. Avoid a compiler warning (from some compilers) for a function call with
300 a cast that removes "const" from an lvalue by using an intermediate
301 variable (to which the compiler does not object).
302
303 19. Incorrect code was compiled if a group that contained an internal recursive
304 back reference was optional (had quantifier with a minimum of zero). This
305 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
306 caused segmentation faults because of stack overflows at compile time.
307
308 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
309 group that is quantified with an indefinite repeat, caused a compile-time
310 loop which used up all the system stack and provoked a segmentation fault.
311 This was not the same bug as 19 above.
312
313 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
314 Patch by Mike Frysinger.
315
316
317 Version 8.35 04-April-2014
318 --------------------------
319
320 1. A new flag is set, when property checks are present in an XCLASS.
321 When this flag is not set, PCRE can perform certain optimizations
322 such as studying these XCLASS-es.
323
324 2. The auto-possessification of character sets were improved: a normal
325 and an extended character set can be compared now. Furthermore
326 the JIT compiler optimizes more character set checks.
327
328 3. Got rid of some compiler warnings for potentially uninitialized variables
329 that show up only when compiled with -O2.
330
331 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
332 of a match later then the end of the match. The pcretest program was not
333 handling the case sensibly - it was outputting from the start to the next
334 binary zero. It now reports this situation in a message, and outputs the
335 text from the end to the start.
336
337 5. Fast forward search is improved in JIT. Instead of the first three
338 characters, any three characters with fixed position can be searched.
339 Search order: first, last, middle.
340
341 6. Improve character range checks in JIT. Characters are read by an inprecise
342 function now, which returns with an unknown value if the character code is
343 above a certain threshold (e.g: 256). The only limitation is that the value
344 must be bigger than the threshold as well. This function is useful when
345 the characters above the threshold are handled in the same way.
346
347 7. The macros whose names start with RAWUCHAR are placeholders for a future
348 mode in which only the bottom 21 bits of 32-bit data items are used. To
349 make this more memorable for those maintaining the code, the names have
350 been changed to start with UCHAR21, and an extensive comment has been added
351 to their definition.
352
353 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
354 to the export list in Makefile.am (they were accidentally omitted from the
355 8.34 tarball).
356
357 9. The informational output from pcretest used the phrase "starting byte set"
358 which is inappropriate for the 16-bit and 32-bit libraries. As the output
359 for "first char" and "need char" really means "non-UTF-char", I've changed
360 "byte" to "char", and slightly reworded the output. The documentation about
361 these values has also been (I hope) clarified.
362
363 10. Another JIT related optimization: use table jumps for selecting the correct
364 backtracking path, when more than four alternatives are present inside a
365 bracket.
366
367 11. Empty match is not possible, when the minimum length is greater than zero,
368 and there is no \K in the pattern. JIT should avoid empty match checks in
369 such cases.
370
371 12. In a caseless character class with UCP support, when a character with more
372 than one alternative case was not the first character of a range, not all
373 the alternative cases were added to the class. For example, s and \x{17f}
374 are both alternative cases for S: the class [RST] was handled correctly,
375 but [R-T] was not.
376
377 13. The configure.ac file always checked for pthread support when JIT was
378 enabled. This is not used in Windows, so I have put this test inside a
379 check for the presence of windows.h (which was already tested for).
380
381 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
382 The algorithm provides a way to skip certain starting offsets, and usually
383 faster than linear prefix searches.
384
385 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
386 as for 'fr_FR' and 'french'. For some reason, however, it then used the
387 Windows-specific input and output files, which have 'french' screwed in.
388 So this could never have worked. One of the problems with locales is that
389 they aren't always the same. I have now updated RunTest so that it checks
390 the output of the locale test (test 3) against three different output
391 files, and it allows the test to pass if any one of them matches. With luck
392 this should make the test pass on some versions of Solaris where it was
393 failing. Because of the uncertainty, the script did not used to stop if
394 test 3 failed; it now does. If further versions of a French locale ever
395 come to light, they can now easily be added.
396
397 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
398 there was a message during ./configure, but it did not stop. This now
399 provokes an error. The invalid example in README has been corrected.
400 If a value less than the minimum is given, the minimum value has always
401 been used, but now a warning is given.
402
403 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
404 was a bug in the test system, which is now fixed. Also, the list of various
405 configurations that are tested for each release did not have one with both
406 16/32 bits and --enable-bar-anycrlf. It now does.
407
408 18. pcretest was missing "-C bsr" for displaying the \R default setting.
409
410 19. Little endian PowerPC systems are supported now by the JIT compiler.
411
412 20. The fast forward newline mechanism could enter to an infinite loop on
413 certain invalid UTF-8 input. Although we don't support these cases
414 this issue can be fixed by a performance optimization.
415
416 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
417 not take account if existing stack usage. There is now a new global
418 variable called pcre_stack_guard that can be set to point to an external
419 function to check stack availability. It is called at the start of
420 processing every parenthesized group.
421
422 22. A typo in the code meant that in ungreedy mode the max/min qualifier
423 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
424 not match "ab".
425
426 23. When UTF was disabled, the JIT program reported some incorrect compile
427 errors. These messages are silenced now.
428
429 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
430 compiler.
431
432 25. Change all the temporary files used in RunGrepTest to be different to those
433 used by RunTest so that the tests can be run simultaneously, for example by
434 "make -j check".
435
436
437 Version 8.34 15-December-2013
438 -----------------------------
439
440 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
441 executable memory. Patch inspired by Carsten Klein.
442
443 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
444 this macro is never tested and has no effect, because the work to support
445 coverage involves only compiling and linking options and special targets in
446 the Makefile. The comment in config.h implied that defining the macro would
447 enable coverage support, which is totally false. There was also support for
448 setting this macro in the CMake files (my fault, I just copied it from
449 configure). SUPPORT_GCOV has now been removed.
450
451 3. Make a small performance improvement in strlen16() and strlen32() in
452 pcretest.
453
454 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
455 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
456 these cases). There was also one in pcretest.c.
457
458 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
459
460 6. In UTF mode, the code for checking whether a group could match an empty
461 string (which is used for indefinitely repeated groups to allow for
462 breaking an infinite loop) was broken when the group contained a repeated
463 negated single-character class with a character that occupied more than one
464 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
465 UTF-8 mode). The effect was undefined: the group might or might not be
466 deemed as matching an empty string, or the program might have crashed.
467
468 7. The code for checking whether a group could match an empty string was not
469 recognizing that \h, \H, \v, \V, and \R must match a character.
470
471 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
472 an empty string. If it can, pcretest shows this in its information output.
473
474 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
475 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
476 matched by pcre_exec() without using JIT:
477
478 (a) If the rest of the pattern did not match after a maximal run of
479 grapheme clusters, the code for backing up to try with fewer of them
480 did not always back up over a full grapheme when characters that do not
481 have the modifier quality were involved, e.g. Hangul syllables.
482
483 (b) If the match point in a subject started with modifier character, and
484 there was no match, the code could incorrectly back up beyond the match
485 point, and potentially beyond the first character in the subject,
486 leading to a segfault or an incorrect match result.
487
488 10. A conditional group with an assertion condition could lead to PCRE
489 recording an incorrect first data item for a match if no other first data
490 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
491 first data item, and therefore matched "ca" after "c" instead of at the
492 start.
493
494 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
495 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
496
497 12. The source of pcregrep now includes z/OS-specific code so that it can be
498 compiled for z/OS as part of the special z/OS distribution.
499
500 13. Added the -T and -TM options to pcretest.
501
502 14. The code in pcre_compile.c for creating the table of named capturing groups
503 has been refactored. Instead of creating the table dynamically during the
504 actual compiling pass, the information is remembered during the pre-compile
505 pass (on the stack unless there are more than 20 named groups, in which
506 case malloc() is used) and the whole table is created before the actual
507 compile happens. This has simplified the code (it is now nearly 150 lines
508 shorter) and prepared the way for better handling of references to groups
509 with duplicate names.
510
511 15. A back reference to a named subpattern when there is more than one of the
512 same name now checks them in the order in which they appear in the pattern.
513 The first one that is set is used for the reference. Previously only the
514 first one was inspected. This change makes PCRE more compatible with Perl.
515
516 16. Unicode character properties were updated from Unicode 6.3.0.
517
518 17. The compile-time code for auto-possessification has been refactored, based
519 on a patch by Zoltan Herczeg. It now happens after instead of during
520 compilation. The code is cleaner, and more cases are handled. The option
521 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
522 options in pcretest are provided to set it. It can also be set by
523 (*NO_AUTO_POSSESS) at the start of a pattern.
524
525 18. The character VT has been added to the default ("C" locale) set of
526 characters that match \s and are generally treated as white space,
527 following this same change in Perl 5.18. There is now no difference between
528 "Perl space" and "POSIX space". Whether VT is treated as white space in
529 other locales depends on the locale.
530
531 19. The code for checking named groups as conditions, either for being set or
532 for being recursed, has been refactored (this is related to 14 and 15
533 above). Processing unduplicated named groups should now be as fast at
534 numerical groups, and processing duplicated groups should be faster than
535 before.
536
537 20. Two patches to the CMake build system, by Alexander Barkov:
538
539 (1) Replace the "source" command by "." in CMakeLists.txt because
540 "source" is a bash-ism.
541
542 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
543 without these the CMake build does not work on Solaris.
544
545 21. Perl has changed its handling of \8 and \9. If there is no previously
546 encountered capturing group of those numbers, they are treated as the
547 literal characters 8 and 9 instead of a binary zero followed by the
548 literals. PCRE now does the same.
549
550 22. Following Perl, added \o{} to specify codepoints in octal, making it
551 possible to specify values greater than 0777 and also making them
552 unambiguous.
553
554 23. Perl now gives an error for missing closing braces after \x{... instead of
555 treating the string as literal. PCRE now does the same.
556
557 24. RunTest used to grumble if an inappropriate test was selected explicitly,
558 but just skip it when running all tests. This make it awkward to run ranges
559 of tests when one of them was inappropriate. Now it just skips any
560 inappropriate tests, as it always did when running all tests.
561
562 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
563 character types such as \d or \w, too many callouts were inserted, and the
564 data that they returned was rubbish.
565
566 26. In UCP mode, \s was not matching two of the characters that Perl matches,
567 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
568 were matched by \h. The code has now been refactored so that the lists of
569 the horizontal and vertical whitespace characters used for \h and \v (which
570 are defined only in one place) are now also used for \s.
571
572 27. Add JIT support for the 64 bit TileGX architecture.
573 Patch by Jiong Wang (Tilera Corporation).
574
575 28. Possessive quantifiers for classes (both explicit and automatically
576 generated) now use special opcodes instead of wrapping in ONCE brackets.
577
578 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
579 (because it's meaningless), this was not happening when PCRE_CASELESS was
580 set. Not wrong, but inefficient.
581
582 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
583 properties for \w, \d, etc) is present in a test regex. Otherwise if the
584 test contains no characters greater than 255, Perl doesn't realise it
585 should be using Unicode semantics.
586
587 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
588 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
589 does in Unicode mode.
590
591 32. Added the "forbid" facility to pcretest so that putting tests into the
592 wrong test files can sometimes be quickly detected.
593
594 33. There is now a limit (default 250) on the depth of nesting of parentheses.
595 This limit is imposed to control the amount of system stack used at compile
596 time. It can be changed at build time by --with-parens-nest-limit=xxx or
597 the equivalent in CMake.
598
599 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
600 errors. Perl warns for these when in warning mode, but PCRE has no facility
601 for giving warnings.
602
603 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
604 However, this was not working for (?!) because it is optimized to (*FAIL),
605 for which PCRE does not allow quantifiers. The optimization is now disabled
606 when a quantifier follows (?!). I can't see any use for this, but it makes
607 things uniform.
608
609 36. Perl no longer allows group names to start with digits, so I have made this
610 change also in PCRE. It simplifies the code a bit.
611
612 37. In extended mode, Perl ignores spaces before a + that indicates a
613 possessive quantifier. PCRE allowed a space before the quantifier, but not
614 before the possessive +. It now does.
615
616 38. The use of \K (reset reported match start) within a repeated possessive
617 group such as (a\Kb)*+ was not working.
618
619 40. Document that the same character tables must be used at compile time and
620 run time, and that the facility to pass tables to pcre_exec() and
621 pcre_dfa_exec() is for use only with saved/restored patterns.
622
623 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
624 features for Builds with MSVC:
625
626 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
627 builds). The .rc files can be used to set FileDescription and many other
628 attributes.
629
630 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
631 This allows higher-level build scripts which want .pdb files to avoid
632 hard-coding the exact files needed."
633
634 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
635 mean "start of word" and "end of word", respectively, as a transition aid.
636
637 43. A minimizing repeat of a class containing codepoints greater than 255 in
638 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
639 compiled to use the heap for recursion.
640
641 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
642 is configured.
643
644
645 Version 8.33 28-May-2013
646 ------------------------
647
648 1. Added 'U' to some constants that are compared to unsigned integers, to
649 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
650 variables that are added to signed variables, to ensure the result is
651 signed and can be negated.
652
653 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
654 CMake config files.
655
656 3. Revise the creation of config.h.generic so that all boolean macros are
657 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
658 overriding via -D on the command line possible.
659
660 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
661 to unsigned int is reported to make a quite noticeable speed difference in
662 a specific Windows environment. Testing on Linux did also appear to show
663 some benefit (and it is clearly not harmful). Also fixed the definition of
664 Xop which should be unsigned.
665
666 5. Related to (4), changing the definition of the intermediate variable cc
667 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
668 improvements.
669
670 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
671 unnecessary spaces.
672
673 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
674 and later.
675
676 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
677
678 9. Optimizing fast_forward_start_bits in JIT.
679
680 10. Adding support for callouts in JIT, and fixing some issues revealed
681 during this work. Namely:
682
683 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
684
685 (b) Minimum length was not checked before the matching is started.
686
687 11. The value of capture_last that is passed to callouts was incorrect in some
688 cases when there was a capture on one path that was subsequently abandoned
689 after a backtrack. Also, the capture_last value is now reset after a
690 recursion, since all captures are also reset in this case.
691
692 12. The interpreter no longer returns the "too many substrings" error in the
693 case when an overflowing capture is in a branch that is subsequently
694 abandoned after a backtrack.
695
696 13. In the pathological case when an offset vector of size 2 is used, pcretest
697 now prints out the matched string after a yield of 0 or 1.
698
699 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
700 Only supported by the JIT compiler at the moment.
701
702 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
703
704 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
705 offset of the starting point of the matching process, provided the offsets
706 vector is large enough.
707
708 17. The \A escape now records a lookbehind value of 1, though its execution
709 does not actually inspect the previous character. This is to ensure that,
710 in partial multi-segment matching, at least one character from the old
711 segment is retained when a new segment is processed. Otherwise, if there
712 are no lookbehinds in the pattern, \A might match incorrectly at the start
713 of a new segment.
714
715 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
716
717 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
718 gives some modest performance improvement in 8-bit mode.
719
720 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
721 be expressed in certain programming languages using Universal Character
722 Names.
723
724 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
725 which points out that "non characters" are not "characters that may not
726 appear in Unicode strings" but rather "characters that are reserved for
727 internal use and have only local meaning".
728
729 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
730 there was a conditional group that depended on an assertion, if the
731 assertion was false, the callout that immediately followed the alternation
732 in the condition was skipped when pcre_exec() was used for matching.
733
734 23. Allow an explicit callout to be inserted before an assertion that is the
735 condition for a conditional group, for compatibility with automatic
736 callouts, which always insert a callout at this point.
737
738 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
739 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
740
741 25. (*PRUNE) is now supported by the JIT compiler.
742
743 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
744
745 27. Fix the case where there are two or more SKIPs with arguments that may be
746 ignored.
747
748 28. (*SKIP) is now supported by the JIT compiler.
749
750 29. (*THEN) is now supported by the JIT compiler.
751
752 30. Update RunTest with additional test selector options.
753
754 31. The way PCRE handles backtracking verbs has been changed in two ways.
755
756 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
757 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
758 backtracking. In some cases this makes it more Perl-compatible, but Perl's
759 rather obscure rules do not always do the same thing.
760
761 (2) Previously, backtracking verbs were confined within assertions. This is
762 no longer the case for positive assertions, except for (*ACCEPT). Again,
763 this sometimes improves Perl compatibility, and sometimes does not.
764
765 32. A number of tests that were in test 2 because Perl did things differently
766 have been moved to test 1, because either Perl or PCRE has changed, and
767 these tests are now compatible.
768
769 32. Backtracking control verbs are now handled in the same way in JIT and
770 interpreter.
771
772 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
773 contained a forward subroutine reference caused a compile error.
774
775 34. Auto-detect and optimize limited repetitions in JIT.
776
777 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
778 blocking (*UTF) etc.
779
780 36. In the interpreter, maximizing pattern repetitions for characters and
781 character types now use tail recursion, which reduces stack usage.
782
783 37. The value of the max lookbehind was not correctly preserved if a compiled
784 and saved regex was reloaded on a host of different endianness.
785
786 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
787 of the compiled pattern block, expand the flags field from 16 to 32 bits
788 because it was almost full.
789
790 39. Try madvise first before posix_madvise.
791
792 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
793 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
794 original change locked out all matches of empty strings. This has been
795 changed so that one match of an empty string per line is recognized.
796 Subsequent searches on the same line (for colouring or for --only-matching,
797 for example) do not recognize empty strings.
798
799 41. Applied a user patch to fix a number of spelling mistakes in comments.
800
801 42. Data lines longer than 65536 caused pcretest to crash.
802
803 43. Clarified the data type for length and startoffset arguments for pcre_exec
804 and pcre_dfa_exec in the function-specific man pages, where they were
805 explicitly stated to be in bytes, never having been updated. I also added
806 some clarification to the pcreapi man page.
807
808 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
809 a segmentation fault.
810
811
812 Version 8.32 30-November-2012
813 -----------------------------
814
815 1. Improved JIT compiler optimizations for first character search and single
816 character iterators.
817
818 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
819 Patch by Daniel Richard G.
820
821 3. Single character iterator optimizations in the JIT compiler.
822
823 4. Improved JIT compiler optimizations for character ranges.
824
825 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
826 Reported by Giuseppe D'Angelo.
827
828 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
829 of a line, was being set incorrectly in cases where .* appeared inside
830 atomic brackets at the start of a pattern, or where there was a subsequent
831 *PRUNE or *SKIP.
832
833 7. Improved instruction cache flush for POWER/PowerPC.
834 Patch by Daniel Richard G.
835
836 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
837 grep:
838
839 (a) There is now no limit to the number of patterns to be matched.
840
841 (b) An error is given if a pattern is too long.
842
843 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
844 are now supported.
845
846 (d) --exclude-from and --include-from (multiple use) have been added.
847
848 (e) Exclusions and inclusions now apply to all files and directories, not
849 just to those obtained from scanning a directory recursively.
850
851 (f) Multiple uses of -f and --file-list are now supported.
852
853 (g) In a Windows environment, the default for -d has been changed from
854 "read" (the GNU grep default) to "skip", because otherwise the presence
855 of a directory in the file list provokes an error.
856
857 (h) The documentation has been revised and clarified in places.
858
859 9. Improve the matching speed of capturing brackets.
860
861 10. Changed the meaning of \X so that it now matches a Unicode extended
862 grapheme cluster.
863
864 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
865 out POSIX threads when JIT support is configured.
866
867 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
868
869 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
870 structure could go wrong in environments where size_t is not the same size
871 as int.
872
873 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
874 set.
875
876 15. The EBCDIC support had decayed; later updates to the code had included
877 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
878 general tidy up of EBCDIC-related issues, and the documentation was also
879 not quite right. There is now a test that can be run on ASCII systems to
880 check some of the EBCDIC-related things (but is it not a full test).
881
882 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
883 in a small tidy to the code.
884
885 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
886
887 18. If the --only-matching (-o) option in pcregrep is specified multiple
888 times, each one causes appropriate output. For example, -o1 -o2 outputs the
889 substrings matched by the 1st and 2nd capturing parentheses. A separating
890 string can be specified by --om-separator (default empty).
891
892 19. Improving the first n character searches.
893
894 20. Turn case lists for horizontal and vertical white space into macros so that
895 they are defined only once.
896
897 21. This set of changes together give more compatible Unicode case-folding
898 behaviour for characters that have more than one other case when UCP
899 support is available.
900
901 (a) The Unicode property table now has offsets into a new table of sets of
902 three or more characters that are case-equivalent. The MultiStage2.py
903 script that generates these tables (the pcre_ucd.c file) now scans
904 CaseFolding.txt instead of UnicodeData.txt for character case
905 information.
906
907 (b) The code for adding characters or ranges of characters to a character
908 class has been abstracted into a generalized function that also handles
909 case-independence. In UTF-mode with UCP support, this uses the new data
910 to handle characters with more than one other case.
911
912 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
913 whose other case is greater than 256 are now correctly matched
914 caselessly. Previously, the high codepoint matched the low one, but not
915 vice versa.
916
917 (d) The processing of \h, \H, \v, and \ in character classes now makes use
918 of the new class addition function, using character lists defined as
919 macros alongside the case definitions of 20 above.
920
921 (e) Caseless back references now work with characters that have more than
922 one other case.
923
924 (f) General caseless matching of characters with more than one other case
925 is supported.
926
927 22. Unicode character properties were updated from Unicode 6.2.0
928
929 23. Improved CMake support under Windows. Patch by Daniel Richard G.
930
931 24. Add support for 32-bit character strings, and UTF-32
932
933 25. Major JIT compiler update (code refactoring and bugfixing).
934 Experimental Sparc 32 support is added.
935
936 26. Applied a modified version of Daniel Richard G's patch to create
937 pcre.h.generic and config.h.generic by "make" instead of in the
938 PrepareRelease script.
939
940 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
941 pcre_compile.c when checking for a zero character.
942
943 28. Introducing a native interface for JIT. Through this interface, the compiled
944 machine code can be directly executed. The purpose of this interface is to
945 provide fast pattern matching, so several sanity checks are not performed.
946 However, feature tests are still performed. The new interface provides
947 1.4x speedup compared to the old one.
948
949 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
950 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
951 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
952
953 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
954 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
955 the "old" RFC 2279). Instead, it ended up passing a negative length to
956 pcre_exec().
957
958 31. Add support for GCC's visibility feature to hide internal functions.
959
960 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
961 "unknown -C option" after outputting 0 or 1.
962
963 33. There is now support for generating a code coverage report for the test
964 suite in environments where gcc is the compiler and lcov is installed. This
965 is mainly for the benefit of the developers.
966
967 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
968 unaddressable using valgrind annotations, allowing valgrind to detect
969 invalid memory accesses. This is mainly for the benefit of the developers.
970
971 25. (*UTF) can now be used to start a pattern in any of the three libraries.
972
973 26. Give configure error if --enable-cpp but no C++ compiler found.
974
975
976 Version 8.31 06-July-2012
977 -------------------------
978
979 1. Fixing a wrong JIT test case and some compiler warnings.
980
981 2. Removed a bashism from the RunTest script.
982
983 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
984 to unsigned type, result still unsigned" that was given by an MS compiler
985 on encountering the code "-sizeof(xxx)".
986
987 4. Partial matching support is added to the JIT compiler.
988
989 5. Fixed several bugs concerned with partial matching of items that consist
990 of more than one character:
991
992 (a) /^(..)\1/ did not partially match "aba" because checking references was
993 done on an "all or nothing" basis. This also applied to repeated
994 references.
995
996 (b) \R did not give a hard partial match if \r was found at the end of the
997 subject.
998
999 (c) \X did not give a hard partial match after matching one or more
1000 characters at the end of the subject.
1001
1002 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1003 a partial match for the string "\r".
1004
1005 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1006 a partial match for a CR character at the end of the subject string.
1007
1008 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1009 running pcretest, the text "(JIT)" added to the output whenever JIT is
1010 actually used to run the match.
1011
1012 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1013 or /S+[+] with a digit between 1 and 7.
1014
1015 8. OP_NOT now supports any UTF character not just single-byte ones.
1016
1017 9. (*MARK) control verb is now supported by the JIT compiler.
1018
1019 10. The command "./RunTest list" lists the available tests without actually
1020 running any of them. (Because I keep forgetting what they all are.)
1021
1022 11. Add PCRE_INFO_MAXLOOKBEHIND.
1023
1024 12. Applied a (slightly modified) user-supplied patch that improves performance
1025 when the heap is used for recursion (compiled with --disable-stack-for-
1026 recursion). Instead of malloc and free for each heap frame each time a
1027 logical recursion happens, frames are retained on a chain and re-used where
1028 possible. This sometimes gives as much as 30% improvement.
1029
1030 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1031 call.
1032
1033 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1034
1035 15. It is now possible to link pcretest with libedit as an alternative to
1036 libreadline.
1037
1038 16. (*COMMIT) control verb is now supported by the JIT compiler.
1039
1040 17. The Unicode data tables have been updated to Unicode 6.1.0.
1041
1042 18. Added --file-list option to pcregrep.
1043
1044 19. Added binary file support to pcregrep, including the -a, --binary-files,
1045 -I, and --text options.
1046
1047 20. The madvise function is renamed for posix_madvise for QNX compatibility
1048 reasons. Fixed by Giuseppe D'Angelo.
1049
1050 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1051
1052 22. Changed the option for creating symbolic links for 16-bit man pages from
1053 -s to -sf so that re-installing does not cause issues.
1054
1055 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1056
1057 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1058 matches in certain environments (the workspace was not being correctly
1059 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1060 some of the workspace data at the beginning of a restart.
1061
1062 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1063 was not doing so when it should - probably a typo introduced by SVN 528
1064 (change 8.10/14).
1065
1066 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1067 \w+ when the character tables indicated that \x{c4} was a word character.
1068 There were several related cases, all because the tests for doing a table
1069 lookup were testing for characters less than 127 instead of 255.
1070
1071 27. If a pattern contains capturing parentheses that are not used in a match,
1072 their slots in the ovector are set to -1. For those that are higher than
1073 any matched groups, this happens at the end of processing. In the case when
1074 there were back references that the ovector was too small to contain
1075 (causing temporary malloc'd memory to be used during matching), and the
1076 highest capturing number was not used, memory off the end of the ovector
1077 was incorrectly being set to -1. (It was using the size of the temporary
1078 memory instead of the true size.)
1079
1080 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1081 ovector size, it uses memory at the end of the block that it has got.
1082
1083 29. Check for an overlong MARK name and give an error at compile time. The
1084 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1085
1086 30. JIT compiler update.
1087
1088 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1089 Rill for the patch.
1090
1091 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1092
1093 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1094
1095 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1096 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1097 was enabled.
1098
1099 35. Improve JIT code generation for greedy plus quantifier.
1100
1101 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1102 1 to "aa" instead of to an empty string. The bug affected repeated groups
1103 that could potentially match an empty string.
1104
1105 37. Optimizing single character iterators in JIT.
1106
1107 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1108 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1109 codepoints that are too big for the mode are faulted, and in a UTF mode,
1110 disallowed codepoints are also faulted.
1111
1112 39. If PCRE was compiled with UTF support, in three places in the DFA
1113 matcher there was code that should only have been obeyed in UTF mode, but
1114 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1115 processing when bytes with values greater than 127 were present. In 16-bit
1116 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1117 both cases the values are those that cannot be the first data item in a UTF
1118 character. The three items that might have provoked this were recursions,
1119 possessively repeated groups, and atomic groups.
1120
1121 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1122 and pcregrep, because some OS require shared objects to be explicitly
1123 passed to ld, causing the link step to fail if they are not.
1124
1125 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1126 mode, patterns that started with \h* or \R* might be incorrectly matched.
1127
1128
1129 Version 8.30 04-February-2012
1130 -----------------------------
1131
1132 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1133 name is defined in ctype.h.
1134
1135 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1136 only in quite long subpatterns.
1137
1138 3. Removed the function pcre_info(), which has been obsolete and deprecated
1139 since it was replaced by pcre_fullinfo() in February 2000.
1140
1141 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1142 match a (*MARK), and the match failed at the start of the subject, a
1143 reference to memory before the start of the subject could occur. This bug
1144 was introduced by fix 17 of release 8.21.
1145
1146 5. A reference to an unset group with zero minimum repetition was giving
1147 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1148 /(another)?(\1?)test/ matched against "hello world test". This bug was
1149 introduced in release 8.13.
1150
1151 6. Add support for 16-bit character strings (a large amount of work involving
1152 many changes and refactorings).
1153
1154 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1155 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1156 from a file.
1157
1158 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1159 rounding is not applied in this particular case).
1160
1161 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1162 if they appear, or are escaped, in patterns.
1163
1164 10. Get rid of a number of -Wunused-but-set-variable warnings.
1165
1166 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1167 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1168 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1169 also returns the mark "x". This bug applied to capturing parentheses,
1170 non-capturing parentheses, and atomic parentheses. It also applied to some
1171 assertions.
1172
1173 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1174 information out of configure.ac instead of relying on pcre.h.generic, which
1175 is not stored in the repository.
1176
1177 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1178 -lreadline.
1179
1180 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1181
1182 15. Applied Graycode's patch to put the top-level frame on the stack rather
1183 than the heap when not using the stack for recursion. This gives a
1184 performance improvement in many cases when recursion is not deep.
1185
1186 16. Experimental code added to "pcretest -C" to output the stack frame size.
1187
1188
1189 Version 8.21 12-Dec-2011
1190 ------------------------
1191
1192 1. Updating the JIT compiler.
1193
1194 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1195 are added as well.
1196
1197 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1198 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1199 calling _pcre_jit_exec. Some extra comments are added.
1200
1201 4. (*MARK) settings inside atomic groups that do not contain any capturing
1202 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1203 was introduced by change 18 for 8.20.
1204
1205 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1206 ECMA-262 standard.
1207
1208 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1209 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1210 This bug was probably introduced by change 9 of 8.13.
1211
1212 7. While fixing 6 above, I noticed that a number of other items were being
1213 incorrectly rejected as "not fixed length". This arose partly because newer
1214 opcodes had not been added to the fixed-length checking code. I have (a)
1215 corrected the bug and added tests for these items, and (b) arranged for an
1216 error to occur if an unknown opcode is encountered while checking for fixed
1217 length instead of just assuming "not fixed length". The items that were
1218 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1219 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1220 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1221
1222 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1223 being incorrectly compiled and would have given unpredicatble results.
1224
1225 9. A possessively repeated subpattern with minimum repeat count greater than
1226 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1227 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1228 the first (A) could occur when it should not.
1229
1230 10. Add a cast and remove a redundant test from the code.
1231
1232 11. JIT should use pcre_malloc/pcre_free for allocation.
1233
1234 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1235 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1236 is anything other than /usr, -L is still shown).
1237
1238 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1239
1240 14. Perl does not support \N without a following name in a [] class; PCRE now
1241 also gives an error.
1242
1243 15. If a forward reference was repeated with an upper limit of around 2000,
1244 it caused the error "internal error: overran compiling workspace". The
1245 maximum number of forward references (including repeats) was limited by the
1246 internal workspace, and dependent on the LINK_SIZE. The code has been
1247 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1248 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1249 of around 200,000 forward references. While doing this, I also speeded up
1250 the filling in of repeated forward references.
1251
1252 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1253 incorrectly expecting the subject to contain another "a" after the start.
1254
1255 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1256 in the match, the SKIP should be ignored. This was not happening; instead
1257 the SKIP was being treated as NOMATCH. For patterns such as
1258 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1259 tested.
1260
1261 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1262 now much more compatible with Perl, in particular in cases where the result
1263 is a non-match for a non-anchored pattern. For example, if
1264 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1265 "m", where previously it did not return a name. A side effect of this
1266 change is that for partial matches, the last encountered mark name is
1267 returned, as for non matches. A number of tests that were previously not
1268 Perl-compatible have been moved into the Perl-compatible test files. The
1269 refactoring has had the pleasing side effect of removing one argument from
1270 the match() function, thus reducing its stack requirements.
1271
1272 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1273 subsequent uses of /S (without +) incorrectly behaved like /S+.
1274
1275 21. Retrieve executable code size support for the JIT compiler and fixing
1276 some warnings.
1277
1278 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1279 not work when the shorter character appeared right at the end of the
1280 subject string.
1281
1282 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1283 systems.
1284
1285 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1286 output it when the /M option is used in pcretest.
1287
1288 25. The CheckMan script was not being included in the distribution. Also, added
1289 an explicit "perl" to run Perl scripts from the PrepareRelease script
1290 because this is reportedly needed in Windows.
1291
1292 26. If study data was being save in a file and studying had not found a set of
1293 "starts with" bytes for the pattern, the data written to the file (though
1294 never used) was taken from uninitialized memory and so caused valgrind to
1295 complain.
1296
1297 27. Updated RunTest.bat as provided by Sheri Pierce.
1298
1299 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1300
1301 29. Computation of memory usage for the table of capturing group names was
1302 giving an unnecessarily large value.
1303
1304
1305 Version 8.20 21-Oct-2011
1306 ------------------------
1307
1308 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1309 a POSIX class. After further experiments with Perl, which convinced me that
1310 Perl has bugs and confusions, a closing square bracket is no longer allowed
1311 in a POSIX name. This bug also affected patterns with classes that started
1312 with full stops.
1313
1314 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1315 captured substring, but while checking the failing first alternative,
1316 substring 1 is temporarily captured. If the output vector supplied to
1317 pcre_exec() was not big enough for this capture, the yield of the function
1318 was still zero ("insufficient space for captured substrings"). This cannot
1319 be totally fixed without adding another stack variable, which seems a lot
1320 of expense for a edge case. However, I have improved the situation in cases
1321 such as /(a)(b)x|abc/ matched against "abc", where the return code
1322 indicates that fewer than the maximum number of slots in the ovector have
1323 been set.
1324
1325 3. Related to (2) above: when there are more back references in a pattern than
1326 slots in the output vector, pcre_exec() uses temporary memory during
1327 matching, and copies in the captures as far as possible afterwards. It was
1328 using the entire output vector, but this conflicts with the specification
1329 that only 2/3 is used for passing back captured substrings. Now it uses
1330 only the first 2/3, for compatibility. This is, of course, another edge
1331 case.
1332
1333 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1334 main code base, and can be used by building with --enable-jit. When this is
1335 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1336 runtime --no-jit option is given.
1337
1338 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1339 ovector, the return from the function was zero, implying that there were
1340 other matches that did not fit. The correct "exactly full" value is now
1341 returned.
1342
1343 6. If a subpattern that was called recursively or as a subroutine contained
1344 (*PRUNE) or any other control that caused it to give a non-standard return,
1345 invalid errors such as "Error -26 (nested recursion at the same subject
1346 position)" or even infinite loops could occur.
1347
1348 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1349 computing the minimum length on reaching *ACCEPT, and so ended up with the
1350 wrong value of 1 rather than 0. Further investigation indicates that
1351 computing a minimum subject length in the presence of *ACCEPT is difficult
1352 (think back references, subroutine calls), and so I have changed the code
1353 so that no minimum is registered for a pattern that contains *ACCEPT.
1354
1355 8. If (*THEN) was present in the first (true) branch of a conditional group,
1356 it was not handled as intended. [But see 16 below.]
1357
1358 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1359 Sheri Pierce.
1360
1361 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1362 the first byte in a match must be "a".
1363
1364 11. Change 17 for 8.13 increased the recursion depth for patterns like
1365 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1366 pattern contains any instances of (*THEN). If it does not, the old
1367 optimizations are restored. It would be nice to do this on a per-group
1368 basis, but at the moment that is not feasible.
1369
1370 12. In some environments, the output of pcretest -C is CRLF terminated. This
1371 broke RunTest's code that checks for the link size. A single white space
1372 character after the value is now allowed for.
1373
1374 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1375 For "fr", it uses the Windows-specific input and output files.
1376
1377 14. If (*THEN) appeared in a group that was called recursively or as a
1378 subroutine, it did not work as intended. [But see next item.]
1379
1380 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1381 pattern fragments (but not containing any | characters). If A and B are
1382 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1383 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1384 D. In other words, Perl considers parentheses that do not contain any |
1385 characters to be part of a surrounding alternative, whereas PCRE was
1386 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1387 differently. PCRE now behaves in the same way as Perl, except in the case
1388 of subroutine/recursion calls such as (?1) which have in any case always
1389 been different (but PCRE had them first :-).
1390
1391 16. Related to 15 above: Perl does not treat the | in a conditional group as
1392 creating alternatives. Such a group is treated in the same way as an
1393 ordinary group without any | characters when processing (*THEN). PCRE has
1394 been changed to match Perl's behaviour.
1395
1396 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1397 RunGrepTest script failed.
1398
1399 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1400 inevitable for groups that contain captures, but it can lead to a lot of
1401 stack use in large patterns. The old behaviour has been restored for atomic
1402 groups that do not contain any capturing parentheses.
1403
1404 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1405 suppress the check for a minimum subject length at run time. (If it was
1406 given to pcre_exec() or pcre_dfa_exec() it did work.)
1407
1408 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1409 fail to work when decoding hex characters in data strings in EBCDIC
1410 environments.
1411
1412 21. It appears that in at least one Mac OS environment, the isxdigit() function
1413 is implemented as a macro that evaluates to its argument more than once,
1414 contravening the C 90 Standard (I haven't checked a later standard). There
1415 was an instance in pcretest which caused it to go wrong when processing
1416 \x{...} escapes in subject strings. The has been rewritten to avoid using
1417 things like p++ in the argument of isxdigit().
1418
1419
1420 Version 8.13 16-Aug-2011
1421 ------------------------
1422
1423 1. The Unicode data tables have been updated to Unicode 6.0.0.
1424
1425 2. Two minor typos in pcre_internal.h have been fixed.
1426
1427 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1428 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1429 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1430
1431 4. There were a number of related bugs in the code for matching backrefences
1432 caselessly in UTF-8 mode when codes for the characters concerned were
1433 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1434 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1435 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1436 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1437 2-byte code at the end of the subject (it thought there wasn't enough data
1438 left).
1439
1440 5. Comprehensive information about what went wrong is now returned by
1441 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1442 as the output vector has at least 2 elements. The offset of the start of
1443 the failing character and a reason code are placed in the vector.
1444
1445 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1446 now returned is for the first byte of the failing character, instead of the
1447 last byte inspected. This is an incompatible change, but I hope it is small
1448 enough not to be a problem. It makes the returned offset consistent with
1449 pcre_exec() and pcre_dfa_exec().
1450
1451 7. pcretest now gives a text phrase as well as the error number when
1452 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1453 failure, the offset and reason code are output.
1454
1455 8. When \R was used with a maximizing quantifier it failed to skip backwards
1456 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1457 back over a single character (\n). This seems wrong (because it treated the
1458 two characters as a single entity when going forwards), conflicts with the
1459 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1460 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1461 has been changed.
1462
1463 9. Some internal refactoring has changed the processing so that the handling
1464 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1465 time (the PCRE_DOTALL option was changed this way some time ago: version
1466 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1467 which was always a bit of a fudge. It also means that there is one less
1468 argument for the match() function, which reduces its stack requirements
1469 slightly. This change also fixes an incompatibility with Perl: the pattern
1470 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1471
1472 10. More internal refactoring has drastically reduced the number of recursive
1473 calls to match() for possessively repeated groups such as (abc)++ when
1474 using pcre_exec().
1475
1476 11. While implementing 10, a number of bugs in the handling of groups were
1477 discovered and fixed:
1478
1479 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1480 (a|)*(?1) gave a compile-time internal error.
1481 ((a|)+)+ did not notice that the outer group could match an empty string.
1482 (^a|^)+ was not marked as anchored.
1483 (.*a|.*)+ was not marked as matching at start or after a newline.
1484
1485 12. Yet more internal refactoring has removed another argument from the match()
1486 function. Special calls to this function are now indicated by setting a
1487 value in a variable in the "match data" data block.
1488
1489 13. Be more explicit in pcre_study() instead of relying on "default" for
1490 opcodes that mean there is no starting character; this means that when new
1491 ones are added and accidentally left out of pcre_study(), testing should
1492 pick them up.
1493
1494 14. The -s option of pcretest has been documented for ages as being an old
1495 synonym of -m (show memory usage). I have changed it to mean "force study
1496 for every regex", that is, assume /S for every regex. This is similar to -i
1497 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1498 using it. It makes it easier to run collections of tests with and without
1499 study enabled, and thereby test pcre_study() more easily. All the standard
1500 tests are now run with and without -s (but some patterns can be marked as
1501 "never study" - see 20 below).
1502
1503 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1504 restoration of the capturing data to the outer values was not happening
1505 correctly.
1506
1507 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1508 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1509 pattern had matched an empty string, and so incorrectly returned a no
1510 match.
1511
1512 17. There was optimizing code for the last branch of non-capturing parentheses,
1513 and also for the obeyed branch of a conditional subexpression, which used
1514 tail recursion to cut down on stack usage. Unfortunately, now that there is
1515 the possibility of (*THEN) occurring in these branches, tail recursion is
1516 no longer possible because the return has to be checked for (*THEN). These
1517 two optimizations have therefore been removed. [But see 8.20/11 above.]
1518
1519 18. If a pattern containing \R was studied, it was assumed that \R always
1520 matched two bytes, thus causing the minimum subject length to be
1521 incorrectly computed because \R can also match just one byte.
1522
1523 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1524 was incorrectly computed.
1525
1526 20. If /S is present twice on a test pattern in pcretest input, it now
1527 *disables* studying, thereby overriding the use of -s on the command line
1528 (see 14 above). This is necessary for one or two tests to keep the output
1529 identical in both cases.
1530
1531 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1532 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1533
1534 22. When an atomic group that contained a capturing parenthesis was
1535 successfully matched, but the branch in which it appeared failed, the
1536 capturing was not being forgotten if a higher numbered group was later
1537 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1538 group 1 to "a", when in fact it should be unset. This applied to multi-
1539 branched capturing and non-capturing groups, repeated or not, and also to
1540 positive assertions (capturing in negative assertions does not happen
1541 in PCRE) and also to nested atomic groups.
1542
1543 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1544 subject after a captured substring, to make it easier to tell which of a
1545 number of identical substrings has been captured.
1546
1547 24. The way atomic groups are processed by pcre_exec() has been changed so that
1548 if they are repeated, backtracking one repetition now resets captured
1549 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1550 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1551 "aaa". Previously, it would have been "a". As part of this code
1552 refactoring, the way recursive calls are handled has also been changed.
1553
1554 25. If an assertion condition captured any substrings, they were not passed
1555 back unless some other capturing happened later. For example, if
1556 (?(?=(a))a) was matched against "a", no capturing was returned.
1557
1558 26. When studying a pattern that contained subroutine calls or assertions,
1559 the code for finding the minimum length of a possible match was handling
1560 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1561 group 1 called group 2 while simultaneously a separate group 2 called group
1562 1). A stack overflow occurred in this case. I have fixed this by limiting
1563 the recursion depth to 10.
1564
1565 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1566 Fortmann. This supports explicit test numbers on the command line, and has
1567 argument validation and error reporting.
1568
1569 28. An instance of \X with an unlimited repeat could fail if at any point the
1570 first character it looked at was a mark character.
1571
1572 29. Some minor code refactoring concerning Unicode properties and scripts
1573 should reduce the stack requirement of match() slightly.
1574
1575 30. Added the '=' option to pcretest to check the setting of unused capturing
1576 slots at the end of the pattern, which are documented as being -1, but are
1577 not included in the return count.
1578
1579 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1580 compiled something random. Now it gives a compile-time error (as does
1581 Perl).
1582
1583 32. A *MARK encountered during the processing of a positive assertion is now
1584 recorded and passed back (compatible with Perl).
1585
1586 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1587 had alternative anchored branches, the search for a second match in a line
1588 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1589 matched the line "0102" twice. The same bug affected patterns that started
1590 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1591 twice.
1592
1593 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1594 does, and because of capturing effects, quantifying parenthesized
1595 assertions may at times be useful. Quantifiers are now allowed for
1596 parenthesized assertions.
1597
1598 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1599
1600 36. \g was being checked for fancy things in a character class, when it should
1601 just be a literal "g".
1602
1603 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1604 appearance of a nested POSIX class supersedes an apparent external class.
1605 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1606 unescaped square brackets may also appear as part of class names. For
1607 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1608 more like Perl. (But see 8.20/1 above.)
1609
1610 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1611 was because it thought it was \N{name}, which is not supported).
1612
1613 39. Add minix to OS list not supporting the -S option in pcretest.
1614
1615 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1616 cannot analyze patterns in sufficient detail to catch mutual recursions
1617 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1618 subgroup is called recursively as a subpattern for a second time at the
1619 same position in the subject string. In previous releases this might have
1620 been caught by the recursion limit, or it might have run out of stack.
1621
1622 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1623 happen only once. PCRE was, however incorrectly giving a compile time error
1624 "recursive call could loop indefinitely" because it cannot analyze the
1625 pattern in sufficient detail. The compile time test no longer happens when
1626 PCRE is compiling a conditional subpattern, but actual runaway loops are
1627 now caught at runtime (see 40 above).
1628
1629 42. It seems that Perl allows any characters other than a closing parenthesis
1630 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1631 has been changed to be the same.
1632
1633 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1634 as not to get warnings when autogen.sh is called. Also changed
1635 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1636
1637 44. To help people who use pcregrep to scan files containing exceedingly long
1638 lines, the following changes have been made:
1639
1640 (a) The default value of the buffer size parameter has been increased from
1641 8K to 20K. (The actual buffer used is three times this size.)
1642
1643 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1644 PCRE is built.
1645
1646 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1647 to be set at run time.
1648
1649 (d) Numerical values in pcregrep options can be followed by K or M, for
1650 example --buffer-size=50K.
1651
1652 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1653 given and the return code is set to 2.
1654
1655 45. Add a pointer to the latest mark to the callout data block.
1656
1657 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1658 partial match of an empty string instead of no match. This was specific to
1659 the use of ".".
1660
1661 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1662 complete match instead of a partial match. This bug was dependent on both
1663 the PCRE_UTF8 and PCRE_DOTALL options being set.
1664
1665 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1666 starting byte set, because \b was not being ignored.
1667
1668
1669 Version 8.12 15-Jan-2011
1670 ------------------------
1671
1672 1. Fixed some typos in the markup of the man pages, and wrote a script that
1673 checks for such things as part of the documentation building process.
1674
1675 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1676 --match-limit and --recursion-limit options (added for 8.11). In
1677 particular, this made one of the standard tests fail. (The integer value
1678 went into the wrong half of a long int.)
1679
1680 3. If the --colour option was given to pcregrep with -v (invert match), it
1681 did strange things, either producing crazy output, or crashing. It should,
1682 of course, ignore a request for colour when reporting lines that do not
1683 match.
1684
1685 4. Another pcregrep bug caused similar problems if --colour was specified with
1686 -M (multiline) and the pattern match finished with a line ending.
1687
1688 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1689 matched in multiline mode, the following line was shown as part of the
1690 match. This seems wrong, so I have changed it.
1691
1692 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1693 the check for further matches in the same line (so they could be coloured)
1694 to overrun the end of the current line. If another match was found, it was
1695 incorrectly shown (and then shown again when found in the next line).
1696
1697 7. If pcregrep was compiled under Windows, there was a reference to the
1698 function pcregrep_exit() before it was defined. I am assuming this was
1699 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1700 reported by a user. I've moved the definition above the reference.
1701
1702
1703 Version 8.11 10-Dec-2010
1704 ------------------------
1705
1706 1. (*THEN) was not working properly if there were untried alternatives prior
1707 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1708 backtracked to try for "b" instead of moving to the next alternative branch
1709 at the same level (in this case, to look for "c"). The Perl documentation
1710 is clear that when (*THEN) is backtracked onto, it goes to the "next
1711 alternative in the innermost enclosing group".
1712
1713 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1714 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1715 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1716 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1717 (*THEN).
1718
1719 3. If \s appeared in a character class, it removed the VT character from
1720 the class, even if it had been included by some previous item, for example
1721 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1722 of \s, but is part of the POSIX "space" class.)
1723
1724 4. A partial match never returns an empty string (because you can always
1725 match an empty string at the end of the subject); however the checking for
1726 an empty string was starting at the "start of match" point. This has been
1727 changed to the "earliest inspected character" point, because the returned
1728 data for a partial match starts at this character. This means that, for
1729 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1730 (previously it gave "no match").
1731
1732 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1733 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1734 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1735 has an implication that the given string is incomplete (because a partial
1736 match is preferred over a full match). For this reason, these items now
1737 give a partial match in this situation. [Aside: previously, the one case
1738 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1739 match rather than a full match, which was wrong by the old rules, but is
1740 now correct.]
1741
1742 6. There was a bug in the handling of #-introduced comments, recognized when
1743 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1744 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1745 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1746 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1747 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1748 places in pcre_compile().
1749
1750 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1751 comments when looking ahead for named forward references to subpatterns,
1752 the only newline sequence it recognized was NL. It now handles newlines
1753 according to the set newline convention.
1754
1755 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1756 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1757 cater for a lack of strerror(). These oversights have been fixed.
1758
1759 9. Added --match-limit and --recursion-limit to pcregrep.
1760
1761 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1762
1763 11. When the -o option was used, pcregrep was setting a return code of 1, even
1764 when matches were found, and --line-buffered was not being honoured.
1765
1766 12. Added an optional parentheses number to the -o and --only-matching options
1767 of pcregrep.
1768
1769 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1770 can match an empty string. The code to do it in pcretest and pcredemo
1771 needed fixing:
1772
1773 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1774 only one byte after an empty string match just before CRLF (this case
1775 just got forgotten; "any" and "anycrlf" were OK).
1776
1777 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1778 mode when an empty string match preceded an ASCII character followed by
1779 a non-ASCII character. (The code for advancing by one character rather
1780 than one byte was nonsense.)
1781
1782 (c) The pcredemo.c sample program did not have any code at all to handle
1783 the cases when CRLF is a valid newline sequence.
1784
1785 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1786 as a starting offset was within the subject string. There is now a new
1787 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1788 negative or greater than the length of the string. In order to test this,
1789 pcretest is extended to allow the setting of negative starting offsets.
1790
1791 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1792 starting offset points to the beginning of a UTF-8 character was
1793 unnecessarily clumsy. I tidied it up.
1794
1795 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1796 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1797
1798 17. Nobody had reported that the --include_dir option, which was added in
1799 release 7.7 should have been called --include-dir (hyphen, not underscore)
1800 for compatibility with GNU grep. I have changed it to --include-dir, but
1801 left --include_dir as an undocumented synonym, and the same for
1802 --exclude-dir, though that is not available in GNU grep, at least as of
1803 release 2.5.4.
1804
1805 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1806 characters from a string of bytes) have been redefined so as not to use
1807 loops, in order to improve performance in some environments. At the same
1808 time, I abstracted some of the common code into auxiliary macros to save
1809 repetition (this should not affect the compiled code).
1810
1811 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1812 compile-time error is now given if \c is not followed by an ASCII
1813 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1814 different, and any byte value is allowed.)
1815
1816 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1817 START_OPTIMIZE option, which is now allowed at compile time - but just
1818 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1819 to pcregrep and other applications that have no direct access to PCRE
1820 options. The new /Y option in pcretest sets this option when calling
1821 pcre_compile().
1822
1823 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1824 back references. Groups containing recursive back references were forced to
1825 be atomic by that change, but in the case of named groups, the amount of
1826 memory required was incorrectly computed, leading to "Failed: internal
1827 error: code overflow". This has been fixed.
1828
1829 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1830 pcretest.c, to avoid build problems in some Borland environments.
1831
1832
1833 Version 8.10 25-Jun-2010
1834 ------------------------
1835
1836 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1837 THEN.
1838
1839 2. (*ACCEPT) was not working when inside an atomic group.
1840
1841 3. Inside a character class, \B is treated as a literal by default, but
1842 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1843 causes the error). The code is unchanged, but I tidied the documentation.
1844
1845 4. Inside a character class, PCRE always treated \R and \X as literals,
1846 whereas Perl faults them if its -w option is set. I have changed PCRE so
1847 that it faults them when PCRE_EXTRA is set.
1848
1849 5. Added support for \N, which always matches any character other than
1850 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1851
1852 6. When compiling pcregrep with newer versions of gcc which may have
1853 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1854 declared with attribute warn_unused_result" were given. Just casting the
1855 result to (void) does not stop the warnings; a more elaborate fudge is
1856 needed. I've used a macro to implement this.
1857
1858 7. Minor change to pcretest.c to avoid a compiler warning.
1859
1860 8. Added four artifical Unicode properties to help with an option to make
1861 \s etc use properties (see next item). The new properties are: Xan
1862 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1863
1864 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1865 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1866 this option. Modified pcretest to add /W to test this facility. Added
1867 REG_UCP to make it available via the POSIX interface.
1868
1869 10. Added --line-buffered to pcregrep.
1870
1871 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1872 studied, and the match started with a letter with a code point greater than
1873 127 whose first byte was different to the first byte of the other case of
1874 the letter, the other case of this starting letter was not recognized
1875 (#976).
1876
1877 12. If a pattern that was studied started with a repeated Unicode property
1878 test, for example, \p{Nd}+, there was the theoretical possibility of
1879 setting up an incorrect bitmap of starting bytes, but fortunately it could
1880 not have actually happened in practice until change 8 above was made (it
1881 added property types that matched character-matching opcodes).
1882
1883 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1884 possible starting bytes for non-anchored patterns.
1885
1886 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1887 \R, and also a number of cases that involve Unicode properties, both
1888 explicit and implicit when PCRE_UCP is set.
1889
1890 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1891 input, it could crash or give wrong results if characters with values
1892 greater than 0xc0 were present in the subject string. (Detail: it assumed
1893 UTF-8 input when processing these items.)
1894
1895 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1896 size_t is 64-bit (#991).
1897
1898 17. Added a check for running out of memory when PCRE is compiled with
1899 --disable-stack-for-recursion (#990).
1900
1901 18. If the last data line in a file for pcretest does not have a newline on
1902 the end, a newline was missing in the output.
1903
1904 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1905 less than 128) in its various bitmaps. However, there is a facility for
1906 generating tables according to the current locale when PCRE is compiled. It
1907 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1908 characters, are recognized by isspace() and therefore were getting set in
1909 these tables, and indeed these tables seem to approximate to ISO 8859. This
1910 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1911 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1912 which of course cannot start UTF-8 characters. I have changed the code so
1913 that only real ASCII characters (less than 128) and the correct starting
1914 bytes for UTF-8 encodings are set for characters greater than 127 when in
1915 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1916 altogether.)
1917
1918 20. Added the /T option to pcretest so as to be able to run tests with non-
1919 standard character tables, thus making it possible to include the tests
1920 used for 19 above in the standard set of tests.
1921
1922 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1923 reference to a subpattern the other side of a comment that contains an
1924 opening parenthesis caused either an internal compiling error, or a
1925 reference to the wrong subpattern.
1926
1927
1928 Version 8.02 19-Mar-2010
1929 ------------------------
1930
1931 1. The Unicode data tables have been updated to Unicode 5.2.0.
1932
1933 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1934 configured.
1935
1936 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1937 original author of that file, following a query about its status.
1938
1939 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1940 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1941
1942 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1943 quantifier applied to a forward-referencing subroutine call, could compile
1944 incorrect code or give the error "internal error: previously-checked
1945 referenced subpattern not found".
1946
1947 6. Both MS Visual Studio and Symbian OS have problems with initializing
1948 variables to point to external functions. For these systems, therefore,
1949 pcre_malloc etc. are now initialized to local functions that call the
1950 relevant global functions.
1951
1952 7. There were two entries missing in the vectors called coptable and poptable
1953 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1954 I've fixed the data, and added a kludgy way of testing at compile time that
1955 the lengths are correct (equal to the number of opcodes).
1956
1957 8. Following on from 7, I added a similar kludge to check the length of the
1958 eint vector in pcreposix.c.
1959
1960 9. Error texts for pcre_compile() are held as one long string to avoid too
1961 much relocation at load time. To find a text, the string is searched,
1962 counting zeros. There was no check for running off the end of the string,
1963 which could happen if a new error number was added without updating the
1964 string.
1965
1966 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1967
1968 11. \K was not working if it appeared in an atomic group or in a group that
1969 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1970 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1971 the assertion is positive, but not if it is negative.
1972
1973 12. Change 11 fortuitously reduced the size of the stack frame used in the
1974 "match()" function of pcre_exec.c by one pointer. Forthcoming
1975 implementation of support for (*MARK) will need an extra pointer on the
1976 stack; I have reserved it now, so that the stack frame size does not
1977 decrease.
1978
1979 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1980 item in branch that calls a recursion is a subroutine call - as in the
1981 second branch in the above example - was incorrectly given the compile-
1982 time error "recursive call could loop indefinitely" because pcre_compile()
1983 was not correctly checking the subroutine for matching a non-empty string.
1984
1985 14. The checks for overrunning compiling workspace could trigger after an
1986 overrun had occurred. This is a "should never occur" error, but it can be
1987 triggered by pathological patterns such as hundreds of nested parentheses.
1988 The checks now trigger 100 bytes before the end of the workspace.
1989
1990 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1991
1992
1993 Version 8.01 19-Jan-2010
1994 ------------------------
1995
1996 1. If a pattern contained a conditional subpattern with only one branch (in
1997 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1998 computed the wrong minimum data length (which is of course zero for such
1999 subpatterns). This could cause incorrect "no match" results.
2000
2001 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2002 the pattern is reset in the first branch, pcre_compile() failed with
2003 "internal error: code overflow at offset...". This happened only when
2004 the reset was to the original external option setting. (An optimization
2005 abstracts leading options settings into an external setting, which was the
2006 cause of this.)
2007
2008 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2009 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2010 assertion pattern did not match (meaning that the assertion was true), it
2011 was incorrectly treated as false if the SKIP had been reached during the
2012 matching. This also applied to assertions used as conditions.
2013
2014 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2015 assertion subpattern, including such a pattern used as a condition,
2016 unpredictable results occurred, instead of the error return
2017 PCRE_ERROR_DFA_UITEM.
2018
2019 5. The C++ GlobalReplace function was not working like Perl for the special
2020 situation when an empty string is matched. It now does the fancy magic
2021 stuff that is necessary.
2022
2023 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2024 removed. (These were left over from very, very early versions of PCRE.)
2025
2026 7. Some cosmetic changes to the code to make life easier when compiling it
2027 as part of something else:
2028
2029 (a) Change DEBUG to PCRE_DEBUG.
2030
2031 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2032 called "current" as "current_branch", to prevent a collision with the
2033 Linux macro when compiled as a kernel module.
2034
2035 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2036 prevent a collision with the Linux macro when compiled as a kernel
2037 module.
2038
2039 8. In pcre_compile() there are some checks for integer overflows that used to
2040 cast potentially large values to (double). This has been changed to that
2041 when building, a check for int64_t is made, and if it is found, it is used
2042 instead, thus avoiding the use of floating point arithmetic. (There is no
2043 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2044 double.
2045
2046 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2047 2005 (difference between two addresses compared to an unsigned value).
2048
2049 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2050 custom one, because of the following reported problem in Windows:
2051
2052 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2053 under Win32.
2054 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2055 therefore missing the function definition.
2056 - The compiler thus generates a "C" signature for the test function.
2057 - The linker fails to find the "C" function.
2058 - PCRE fails to configure if asked to do so against libbz2.
2059
2060 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2061 messages were output:
2062
2063 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2064 rerunning libtoolize, to keep the correct libtool macros in-tree.
2065 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2066
2067 I have done both of these things.
2068
2069 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2070 most of the time, it *can* run out if it is given a pattern that contains a
2071 runaway infinite recursion. I updated the discussion in the pcrestack man
2072 page.
2073
2074 13. Now that we have gone to the x.xx style of version numbers, the minor
2075 version may start with zero. Using 08 or 09 is a bad idea because users
2076 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2077 interpreted as invalid octal numbers. I've updated the previous comment in
2078 configure.ac, and also added a check that gives an error if 08 or 09 are
2079 used.
2080
2081 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2082 causing partial matching to fail when the end of the subject matched \W
2083 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2084
2085 15. There were some discrepancies between the declarations in pcre_internal.h
2086 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2087 their definitions. The declarations used "const uschar *" and the
2088 definitions used USPTR. Even though USPTR is normally defined as "const
2089 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2090 reported that: "This difference in casting confuses some C++ compilers, for
2091 example, SunCC recognizes above declarations as different functions and
2092 generates broken code for hbpcre." I have changed the declarations to use
2093 USPTR.
2094
2095 16. GNU libtool is named differently on some systems. The autogen.sh script now
2096 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2097 (FreeBSD).
2098
2099 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2100 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2101 comment: "Figure out how to create a longlong from a string: strtoll and
2102 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2103 instance, but it only takes 2 args instead of 3!"
2104
2105 18. A subtle bug concerned with back references has been fixed by a change of
2106 specification, with a corresponding code fix. A pattern such as
2107 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2108 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2109 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2110 same bug. Such groups have to be quantified to be useful, or contained
2111 inside another quantified group. (If there's no repetition, the reference
2112 can never match.) The problem arises because, having left the group and
2113 moved on to the rest of the pattern, a later failure that backtracks into
2114 the group uses the captured value from the final iteration of the group
2115 rather than the correct earlier one. I have fixed this in PCRE by forcing
2116 any group that contains a reference to itself to be an atomic group; that
2117 is, there cannot be any backtracking into it once it has completed. This is
2118 similar to recursive and subroutine calls.
2119
2120
2121 Version 8.00 19-Oct-09
2122 ----------------------
2123
2124 1. The table for translating pcre_compile() error codes into POSIX error codes
2125 was out-of-date, and there was no check on the pcre_compile() error code
2126 being within the table. This could lead to an OK return being given in
2127 error.
2128
2129 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2130 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2131 in a Windows environment.
2132
2133 3. The pcregrep --count option prints the count for each file even when it is
2134 zero, as does GNU grep. However, pcregrep was also printing all files when
2135 --files-with-matches was added. Now, when both options are given, it prints
2136 counts only for those files that have at least one match. (GNU grep just
2137 prints the file name in this circumstance, but including the count seems
2138 more useful - otherwise, why use --count?) Also ensured that the
2139 combination -clh just lists non-zero counts, with no names.
2140
2141 4. The long form of the pcregrep -F option was incorrectly implemented as
2142 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2143 but it seems right to fix it, and I didn't think it was worth preserving
2144 the old behaviour.
2145
2146 5. The command line items --regex=pattern and --regexp=pattern were not
2147 recognized by pcregrep, which required --regex pattern or --regexp pattern
2148 (with a space rather than an '='). The man page documented the '=' forms,
2149 which are compatible with GNU grep; these now work.
2150
2151 6. No libpcreposix.pc file was created for pkg-config; there was just
2152 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2153
2154 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2155 when UCP support is not needed, by modifying the Python script that
2156 generates it from Unicode data files. This should not matter if the module
2157 is correctly used as a library, but I received one complaint about 50K of
2158 unwanted data. My guess is that the person linked everything into his
2159 program rather than using a library. Anyway, it does no harm.
2160
2161 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2162 was a minimum greater than 1 for a wide character in a possessive
2163 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2164 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2165 character. Chaos in the form of incorrect output or a compiling loop could
2166 result.
2167
2168 9. The restrictions on what a pattern can contain when partial matching is
2169 requested for pcre_exec() have been removed. All patterns can now be
2170 partially matched by this function. In addition, if there are at least two
2171 slots in the offset vector, the offset of the earliest inspected character
2172 for the match and the offset of the end of the subject are set in them when
2173 PCRE_ERROR_PARTIAL is returned.
2174
2175 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2176 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2177 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2178 and may be more useful for multi-segment matching.
2179
2180 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2181 used to be given if ever the end of the subject was reached; now it is
2182 given only if matching could not proceed because another character was
2183 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2184 string "Z", which now yields "no match" instead of "partial match". In the
2185 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2186 final character ended with (*FAIL).
2187
2188 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2189 if the pattern had a "must contain" character that was already found in the
2190 earlier partial match, unless partial matching was again requested. For
2191 example, with the pattern /dog.(body)?/, the "must contain" character is
2192 "g". If the first part-match was for the string "dog", restarting with
2193 "sbody" failed. This bug has been fixed.
2194
2195 13. The string returned by pcre_dfa_exec() after a partial match has been
2196 changed so that it starts at the first inspected character rather than the
2197 first character of the match. This makes a difference only if the pattern
2198 starts with a lookbehind assertion or \b or \B (\K is not supported by
2199 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2200 matching functions compatible, and I think it's the right thing to do.
2201
2202 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2203 so that the demonstration program is easily available in environments where
2204 PCRE has not been installed from source.
2205
2206 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2207 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2208 library.
2209
2210 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2211 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2212 is not the first non-POSIX option to be added. Clearly some people find
2213 these options useful.
2214
2215 17. If a caller to the POSIX matching function regexec() passes a non-zero
2216 value for nmatch with a NULL value for pmatch, the value of
2217 nmatch is forced to zero.
2218
2219 18. RunGrepTest did not have a test for the availability of the -u option of
2220 the diff command, as RunTest does. It now checks in the same way as
2221 RunTest, and also checks for the -b option.
2222
2223 19. If an odd number of negated classes containing just a single character
2224 interposed, within parentheses, between a forward reference to a named
2225 subpattern and the definition of the subpattern, compilation crashed with
2226 an internal error, complaining that it could not find the referenced
2227 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2228 [The bug was that it was starting one character too far in when skipping
2229 over the character class, thus treating the ] as data rather than
2230 terminating the class. This meant it could skip too much.]
2231
2232 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2233 /g option in pcretest when the pattern contains \K, which makes it possible
2234 to have an empty string match not at the start, even when the pattern is
2235 anchored. Updated pcretest and pcredemo to use this option.
2236
2237 21. If the maximum number of capturing subpatterns in a recursion was greater
2238 than the maximum at the outer level, the higher number was returned, but
2239 with unset values at the outer level. The correct (outer level) value is
2240 now given.
2241
2242 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2243 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2244 make it do so. The string so far is captured, making this feature
2245 compatible with Perl.
2246
2247 23. The tests have been re-organized, adding tests 11 and 12, to make it
2248 possible to check the Perl 5.10 features against Perl 5.10.
2249
2250 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2251 pattern matches a fixed length string. PCRE did not allow this; now it
2252 does. Neither allows recursion.
2253
2254 25. I finally figured out how to implement a request to provide the minimum
2255 length of subject string that was needed in order to match a given pattern.
2256 (It was back references and recursion that I had previously got hung up
2257 on.) This code has now been added to pcre_study(); it finds a lower bound
2258 to the length of subject needed. It is not necessarily the greatest lower
2259 bound, but using it to avoid searching strings that are too short does give
2260 some useful speed-ups. The value is available to calling programs via
2261 pcre_fullinfo().
2262
2263 26. While implementing 25, I discovered to my embarrassment that pcretest had
2264 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2265 study optimizations had never been tested with that matching function.
2266 Oops. What is worse, even when it was passed study data, there was a bug in
2267 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2268 were also very few tests of studied patterns with pcre_dfa_exec().
2269
2270 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2271 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2272 on the other side of the coin, they are no longer allowed to have different
2273 names, because these cannot be distinguished in PCRE, and this has caused
2274 confusion. (This is a difference from Perl.)
2275
2276 28. When duplicate subpattern names are present (necessarily with different
2277 numbers, as required by 27 above), and a test is made by name in a
2278 conditional pattern, either for a subpattern having been matched, or for
2279 recursion in such a pattern, all the associated numbered subpatterns are
2280 tested, and the overall condition is true if the condition is true for any
2281 one of them. This is the way Perl works, and is also more like the way
2282 testing by number works.
2283
2284
2285 Version 7.9 11-Apr-09
2286 ---------------------
2287
2288 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2289 (pcretest), all targets were linked against these libraries. This included
2290 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2291 libraries. This caused unwanted dependencies to be created. This problem
2292 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2293 pcretest is linked with readline.
2294
2295 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2296 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2297 moved outside it again, because FALSE and TRUE are already defined in AIX,
2298 but BOOL is not.
2299
2300 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2301 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2302
2303 4. The pcregrep documentation said spaces were inserted as well as colons (or
2304 hyphens) following file names and line numbers when outputting matching
2305 lines. This is not true; no spaces are inserted. I have also clarified the
2306 wording for the --colour (or --color) option.
2307
2308 5. In pcregrep, when --colour was used with -o, the list of matching strings
2309 was not coloured; this is different to GNU grep, so I have changed it to be
2310 the same.
2311
2312 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2313 each matching line was coloured. Now it goes on to look for further matches
2314 of any of the test patterns, which is the same behaviour as GNU grep.
2315
2316 7. A pattern that could match an empty string could cause pcregrep to loop; it
2317 doesn't make sense to accept an empty string match in pcregrep, so I have
2318 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2319 seems to be how GNU grep behaves. [But see later change 40 for release
2320 8.33.]
2321
2322 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2323 start or after a newline", because the conditional assertion was not being
2324 correctly handled. The rule now is that both the assertion and what follows
2325 in the first alternative must satisfy the test.
2326
2327 9. If auto-callout was enabled in a pattern with a conditional group whose
2328 condition was an assertion, PCRE could crash during matching, both with
2329 pcre_exec() and pcre_dfa_exec().
2330
2331 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2332 used for matching.
2333
2334 11. Unicode property support in character classes was not working for
2335 characters (bytes) greater than 127 when not in UTF-8 mode.
2336
2337 12. Added the -M command line option to pcretest.
2338
2339 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2340
2341 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2342
2343 16. Added comments and documentation about mis-use of no_arg in the C++
2344 wrapper.
2345
2346 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2347 from Martin Jerabek that uses macro names for all relevant character and
2348 string constants.
2349
2350 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2351 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2352 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2353 these, but not everybody uses configure.
2354
2355 19. A conditional group that had only one branch was not being correctly
2356 recognized as an item that could match an empty string. This meant that an
2357 enclosing group might also not be so recognized, causing infinite looping
2358 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2359 with the subject "ab", where knowledge that the repeated group can match
2360 nothing is needed in order to break the loop.
2361
2362 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2363 exec(), but without supplying a callout function, matching went wrong.
2364
2365 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2366 leak if the size of the offset vector was greater than 30. When the vector
2367 is smaller, the saved offsets during recursion go onto a local stack
2368 vector, but for larger vectors malloc() is used. It was failing to free
2369 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2370 error, in fact).
2371
2372 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2373 heapframe that is used only when UTF-8 support is enabled. This caused no
2374 problem, but was untidy.
2375
2376 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2377 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2378 included within another project.
2379
2380 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2381 slightly modified by me:
2382
2383 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2384 not building pcregrep.
2385
2386 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2387 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2388
2389 25. Forward references, both numeric and by name, in patterns that made use of
2390 duplicate group numbers, could behave incorrectly or give incorrect errors,
2391 because when scanning forward to find the reference group, PCRE was not
2392 taking into account the duplicate group numbers. A pattern such as
2393 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2394
2395 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2396 the feature of a custom pointer more persuasive (as requested by a user).
2397
2398 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2399 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2400 pre-defined.
2401
2402 28. Added support for (*UTF8) at the start of a pattern.
2403
2404 29. Arrange for flags added by the "release type" setting in CMake to be shown
2405 in the configuration summary.
2406
2407
2408 Version 7.8 05-Sep-08
2409 ---------------------
2410
2411 1. Replaced UCP searching code with optimized version as implemented for Ad
2412 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2413 stage table and inline lookup instead of a function, giving speed ups of 2
2414 to 5 times on some simple patterns that I tested. Permission was given to
2415 distribute the MultiStage2.py script that generates the tables (it's not in
2416 the tarball, but is in the Subversion repository).
2417
2418 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2419 scripts.
2420
2421 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2422 a group with a zero qualifier. The result of the study could be incorrect,
2423 or the function might crash, depending on the pattern.
2424
2425 4. Caseless matching was not working for non-ASCII characters in back
2426 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2427 It now works when Unicode Property Support is available.
2428
2429 5. In pcretest, an escape such as \x{de} in the data was always generating
2430 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2431 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2432 truncation.
2433
2434 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2435
2436 7. Added two (int) casts to pcregrep when printing the difference of two
2437 pointers, in case they are 64-bit values.
2438
2439 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2440 test 2 if it fails.
2441
2442 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2443 and a #define of that name to empty if it is not externally set. This is to
2444 allow users of MSVC to set it if necessary.
2445
2446 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2447 the convenience functions in the pcre_get.c source file.
2448
2449 11. An option change at the start of a pattern that had top-level alternatives
2450 could cause overwriting and/or a crash. This command provoked a crash in
2451 some environments:
2452
2453 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2454
2455 This potential security problem was recorded as CVE-2008-2371.
2456
2457 12. For a pattern where the match had to start at the beginning or immediately
2458 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2459 pcre_dfa_exec() could read past the end of the passed subject if there was
2460 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2461 pcretest so that it places the subject at the end of its malloc-ed buffer.
2462
2463 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2464 exec() might read past the end of the data buffer in UTF-8 mode.
2465
2466 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2467 the data contained the byte 0x85 as part of a UTF-8 character within its
2468 first line. This applied both to normal and DFA matching.
2469
2470 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2471 /^[^d]*?$/8 failed to match "abc".
2472
2473 16. Added a missing copyright notice to pcrecpp_internal.h.
2474
2475 17. Make it more clear in the documentation that values returned from
2476 pcre_exec() in ovector are byte offsets, not character counts.
2477
2478 18. Tidied a few places to stop certain compilers from issuing warnings.
2479
2480 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2481 supplied by Stefan Weber. I made a further small update for 7.8 because
2482 there is a change of source arrangements: the pcre_searchfuncs.c module is
2483 replaced by pcre_ucd.c.
2484
2485
2486 Version 7.7 07-May-08
2487 ---------------------
2488
2489 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2490 a string to a long long, pretend we don't even have a long long." This is
2491 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2492
2493 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2494 pre-7.6 versions, which defined a global no_arg variable instead of putting
2495 it in the RE class. (See also #8 below.)
2496
2497 3. Remove a line of dead code, identified by coverity and reported by Nuno
2498 Lopes.
2499
2500 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2501
2502 (1) The include/exclude patterns were being applied to the whole pathnames
2503 of files, instead of just to the final components.
2504
2505 (2) If there was more than one level of directory, the subdirectories were
2506 skipped unless they satisfied the include/exclude conditions. This is
2507 inconsistent with GNU grep (and could even be seen as contrary to the
2508 pcregrep specification - which I improved to make it absolutely clear).
2509 The action now is always to scan all levels of directory, and just
2510 apply the include/exclude patterns to regular files.
2511
2512 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2513 --exclude_dir in the tests to avoid scanning .svn directories.
2514
2515 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2516 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2517 doesn't support NULs in patterns.
2518
2519 7. Added some missing "const"s to declarations of static tables in
2520 pcre_compile.c and pcre_dfa_exec.c.
2521
2522 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2523 caused by fix #2 above. (Subsequently also a second patch to fix the
2524 first patch. And a third patch - this was a messy problem.)
2525
2526 9. Applied Craig's patch to remove the use of push_back().
2527
2528 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2529 matching function regexec().
2530
2531 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2532 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2533 references. PCRE supports relative numbers with this syntax (I don't think
2534 Oniguruma does).
2535
2536 12. Previously, a group with a zero repeat such as (...){0} was completely
2537 omitted from the compiled regex. However, this means that if the group
2538 was called as a subroutine from elsewhere in the pattern, things went wrong
2539 (an internal error was given). Such groups are now left in the compiled
2540 pattern, with a new opcode that causes them to be skipped at execution
2541 time.
2542
2543 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2544 to the way PCRE behaves:
2545
2546 (a) A lone ] character is dis-allowed (Perl treats it as data).
2547
2548 (b) A back reference to an unmatched subpattern matches an empty string
2549 (Perl fails the current match path).
2550
2551 (c) A data ] in a character class must be notated as \] because if the
2552 first data character in a class is ], it defines an empty class. (In
2553 Perl it is not possible to have an empty class.) The empty class []
2554 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2555 The negative empty class [^] matches any one character, independently
2556 of the DOTALL setting.
2557
2558 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2559 non-existent subpattern following a character class starting with ']' and
2560 containing () gave an internal compiling error instead of "reference to
2561 non-existent subpattern". Fortunately, when the pattern did exist, the
2562 compiled code was correct. (When scanning forwards to check for the
2563 existence of the subpattern, it was treating the data ']' as terminating
2564 the class, so got the count wrong. When actually compiling, the reference
2565 was subsequently set up correctly.)
2566
2567 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2568 it was being rejected as not supported by pcre_dfa_exec(), even though
2569 other assertions are supported. I have made pcre_dfa_exec() support
2570 (*FAIL).
2571
2572 16. The implementation of 13c above involved the invention of a new opcode,
2573 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2574 cannot be changed at match time, I realized I could make a small
2575 improvement to matching performance by compiling OP_ALLANY instead of
2576 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2577 on the OP_ANY path.
2578
2579 17. Compiling pcretest on Windows with readline support failed without the
2580 following two fixes: (1) Make the unistd.h include conditional on
2581 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2582
2583 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2584 ncurses library to be included for pcretest when ReadLine support is
2585 requested, but also to allow for it to be overridden. This patch came from
2586 Daniel Bergström.
2587
2588 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2589 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2590 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2591 spotting this.
2592
2593
2594 Version 7.6 28-Jan-08
2595 ---------------------
2596
2597 1. A character class containing a very large number of characters with
2598 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2599 overflow.
2600
2601 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2602 HAVE_LONG_LONG is not defined.
2603
2604 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2605 bring them up to date and include new features. This patch includes:
2606
2607 - Fixed PH's badly added libz and libbz2 support.
2608 - Fixed a problem with static linking.
2609 - Added pcredemo. [But later removed - see 7 below.]
2610 - Fixed dftables problem and added an option.
2611 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2612 HAVE_LONG_LONG.
2613 - Added readline support for pcretest.
2614 - Added an listing of the option settings after cmake has run.
2615
2616 4. A user submitted a patch to Makefile that makes it easy to create
2617 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2618 Makefile.am that cause it to include this special target, without
2619 affecting anything else. Note that the same mingw target plus all
2620 the other distribution libraries and programs are now supported
2621 when configuring with CMake (see 6 below) instead of with
2622 Configure/Make.
2623
2624 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2625 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2626 exported in the Windows port". It has not yet been confirmed that the patch
2627 solves the problem, but it does no harm.
2628
2629 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2630 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2631 with CMake, and also correct the comment about stack recursion.
2632
2633 7. Remove the automatic building of pcredemo from the ./configure system and
2634 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2635 of a program that users should build themselves after PCRE is installed, so
2636 building it automatically is not really right. What is more, it gave
2637 trouble in some build environments.
2638
2639 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2640
2641
2642 Version 7.5 10-Jan-08
2643 ---------------------
2644
2645 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2646 values in parens when parsing an RE using the C++ wrapper."
2647
2648 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2649 Characters greater than 255 were excluded from the class instead of being
2650 included.
2651
2652 3. The same bug as (2) above applied to negated POSIX classes such as
2653 [:^space:].
2654
2655 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2656 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2657 I have changed it.
2658
2659 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2660 first named subpattern) and a construct such as (?&a) would reference the
2661 first named subpattern whose name started with "a" (in other words, the
2662 length check was missing). Both these problems are fixed. "Subpattern name
2663 expected" is now given for (?&) (a zero-length name), and this patch also
2664 makes it give the same error for \k'' (previously it complained that that
2665 was a reference to a non-existent subpattern).
2666
2667 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2668 this is right because (?- can be followed by option settings as well as by
2669 digits. I have, however, made the messages clearer.
2670
2671 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2672 than the number used in the conditional) now cause a compile-time error.
2673 This is actually not compatible with Perl, which accepts such patterns, but
2674 treats the conditional as always being FALSE (as PCRE used to), but it
2675 seems to me that giving a diagnostic is better.
2676
2677 8. Change "alphameric" to the more common word "alphanumeric" in comments
2678 and messages.
2679
2680 9. Fix two occurrences of "backslash" in comments that should have been
2681 "backspace".
2682
2683 10. Remove two redundant lines of code that can never be obeyed (their function
2684 was moved elsewhere).
2685
2686 11. The program that makes PCRE's Unicode character property table had a bug
2687 which caused it to generate incorrect table entries for sequences of
2688 characters that have the same character type, but are in different scripts.
2689 It amalgamated them into a single range, with the script of the first of
2690 them. In other words, some characters were in the wrong script. There were
2691 thirteen such cases, affecting characters in the following ranges:
2692
2693 U+002b0 - U+002c1
2694 U+0060c - U+0060d
2695 U+0061e - U+00612
2696 U+0064b - U+0065e
2697 U+0074d - U+0076d
2698 U+01800 - U+01805
2699 U+01d00 - U+01d77
2700 U+01d9b - U+01dbf
2701 U+0200b - U+0200f
2702 U+030fc - U+030fe
2703 U+03260 - U+0327f
2704 U+0fb46 - U+0fbb1
2705 U+10450 - U+1049d
2706
2707 12. The -o option (show only the matching part of a line) for pcregrep was not
2708 compatible with GNU grep in that, if there was more than one match in a
2709 line, it showed only the first of them. It now behaves in the same way as
2710 GNU grep.
2711
2712 13. If the -o and -v options were combined for pcregrep, it printed a blank
2713 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2714 does the same. The return code can be used to tell if there were any
2715 non-matching lines.
2716
2717 14. Added --file-offsets and --line-offsets to pcregrep.
2718
2719 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2720 infinitely looping recursion. The bug was that positive lookaheads were not
2721 being skipped when checking for a possible empty match (negative lookaheads
2722 and both kinds of lookbehind were skipped).
2723
2724 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2725 inclusion of <windows.h> to before rather than after the definition of
2726 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2727
2728 17. Specifying a possessive quantifier with a specific limit for a Unicode
2729 character property caused pcre_compile() to compile bad code, which led at
2730 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2731 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2732 caused the error; without that there was no problem.
2733
2734 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2735
2736 19. Added --enable-pcretest-libreadline.
2737
2738 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2739 RE::GlobalReplace(). As a result, the number of replacements returned was
2740 double what it should be. I removed one of the increments, but Craig sent a
2741 later patch that removed the other one (the right fix) and added unit tests
2742 that check the return values (which was not done before).
2743
2744 21. Several CMake things:
2745
2746 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2747 the names libpcre and libpcreposix, not just pcre and pcreposix.
2748
2749 (2) The above change means that pcretest and pcregrep are now correctly
2750 linked with the newly-built libraries, not previously installed ones.
2751
2752 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2753
2754 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2755 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2756 UTF-8 newline character). The key issue is that the pattern starts .*;
2757 this means that the match must be either at the beginning, or after a
2758 newline. The bug was in the code for advancing after a failed match and
2759 checking that the new position followed a newline. It was not taking
2760 account of UTF-8 characters correctly.
2761
2762 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2763 character classes. PCRE was not treating the sequence [:...:] as a
2764 character class unless the ... were all letters. Perl, however, seems to
2765 allow any characters between [: and :], though of course it rejects as
2766 unknown any "names" that contain non-letters, because all the known class
2767 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2768 for example, whereas PCRE did not - it did not recognize a POSIX character
2769 class. This seemed a bit dangerous, so the code has been changed to be
2770 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2771 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2772 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2773 Perl does, and where it didn't before.
2774
2775 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2776 Windows environments %n is disabled by default.
2777
2778
2779 Version 7.4 21-Sep-07
2780 ---------------------
2781
2782 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2783 means that a class such as [\s] counted as "explicit reference to CR or
2784 LF". That isn't really right - the whole point of the change was to try to
2785 help when there was an actual mention of one of the two characters. So now
2786 the change happens only if \r or \n (or a literal CR or LF) character is
2787 encountered.
2788
2789 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2790 of both had grown to the point where there were only 3 bits left.
2791 Fortunately, there was spare space in the data structure, and so I have
2792 moved the internal flags into a new 16-bit field to free up more option
2793 bits.
2794
2795 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2796 but did not set the internal JCHANGED flag - either of these is enough to
2797 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2798 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2799 start sets both bits.
2800
2801 4. Added options (at build time, compile time, exec time) to change \R from
2802 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2803
2804 5. doc/pcresyntax.html was missing from the distribution.
2805
2806 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2807 compatibility, even though it is no longer used.
2808
2809 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2810 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2811 windows.h file is present (where different names are used). [This was
2812 reversed later after testing - see 16 below.]
2813
2814 8. Changed all #include <config.h> to #include "config.h". There were also
2815 some further <pcre.h> cases that I changed to "pcre.h".
2816
2817 9. When pcregrep was used with the --colour option, it missed the line ending
2818 sequence off the lines that it output.
2819
2820 10. It was pointed out to me that arrays of string pointers cause lots of
2821 relocations when a shared library is dynamically loaded. A technique of
2822 using a single long string with a table of offsets can drastically reduce
2823 these. I have refactored PCRE in four places to do this. The result is
2824 dramatic:
2825
2826 Originally: 290
2827 After changing UCP table: 187
2828 After changing error message table: 43
2829 After changing table of "verbs" 36
2830 After changing table of Posix names 22
2831
2832 Thanks to the folks working on Gregex for glib for this insight.
2833
2834 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2835 unicode-properties was also set.
2836
2837 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2838
2839 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2840 checked only for CRLF.
2841
2842 14. Added casts to pcretest.c to avoid compiler warnings.
2843
2844 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2845
2846 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2847 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2848 entirely. This removes changes made in 7 above.
2849
2850 17. The CMake files have been updated, and there is now more information about
2851 building with CMake in the NON-UNIX-USE document.
2852
2853
2854 Version 7.3 28-Aug-07
2855 ---------------------
2856
2857 1. In the rejigging of the build system that eventually resulted in 7.1, the
2858 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2859 brackets there is not right, since it causes compilers to look for an
2860 installed pcre.h, not the version that is in the source that is being
2861 compiled (which of course may be different). I have changed it back to:
2862
2863 #include "pcre.h"
2864
2865 I have a vague recollection that the change was concerned with compiling in
2866 different directories, but in the new build system, that is taken care of
2867 by the VPATH setting the Makefile.
2868
2869 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2870 when the subject happened to end in the byte 0x85 (e.g. if the last
2871 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2872 characters but of course it shouldn't be taken as a newline when it is part
2873 of another character. The bug was that, for an unlimited repeat of . in
2874 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2875 characters when looking for a newline.
2876
2877 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2878
2879 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2880 in debug output.
2881
2882 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2883 long printing in the pcrecpp unittest when running under MinGW.
2884
2885 6. ESC_K was left out of the EBCDIC table.
2886
2887 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2888 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2889 limit also applies to "virtual nesting" when a pattern is recursive, and in
2890 this case 1000 isn't so big. I have been able to remove this limit at the
2891 expense of backing off one optimization in certain circumstances. Normally,
2892 when pcre_exec() would call its internal match() function recursively and
2893 immediately return the result unconditionally, it uses a "tail recursion"
2894 feature to save stack. However, when a subpattern that can match an empty
2895 string has an unlimited repetition quantifier, it no longer makes this
2896 optimization. That gives it a stack frame in which to save the data for
2897 checking that an empty string has been matched. Previously this was taken
2898 from the 1000-entry workspace that had been reserved. So now there is no
2899 explicit limit, but more stack is used.
2900
2901 8. Applied Daniel's patches to solve problems with the import/export magic
2902 syntax that is required for Windows, and which was going wrong for the
2903 pcreposix and pcrecpp parts of the library. These were overlooked when this
2904 problem was solved for the main library.
2905
2906 9. There were some crude static tests to avoid integer overflow when computing
2907 the size of patterns that contain repeated groups with explicit upper
2908 limits. As the maximum quantifier is 65535, the maximum group length was
2909 set at 30,000 so that the product of these two numbers did not overflow a
2910 32-bit integer. However, it turns out that people want to use groups that
2911 are longer than 30,000 bytes (though not repeat them that many times).
2912 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2913 made it possible to implement the integer overflow checks in a much more
2914 dynamic way, which I have now done. The artificial limitation on group
2915 length has been removed - we now have only the limit on the total length of
2916 the compiled pattern, which depends on the LINK_SIZE setting.
2917
2918 10. Fixed a bug in the documentation for get/copy named substring when
2919 duplicate names are permitted. If none of the named substrings are set, the
2920 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2921 empty string.
2922
2923 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2924 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2925 because the ] is interpreted as the first data character and the
2926 terminating ] is not found. PCRE has been made compatible with Perl in this
2927 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2928 cause memory overwriting.
2929
2930 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2931 string has been matched (to stop an infinite loop). It was not recognizing
2932 a conditional subpattern that could match an empty string if that
2933 subpattern was within another subpattern. For example, it looped when
2934 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2935 condition was not nested. This bug has been fixed.
2936
2937 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2938 past the start of the subject in the presence of bytes with the top bit
2939 set, for example "\x8aBCD".
2940
2941 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2942 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2943
2944 14. Optimized (?!) to (*FAIL).
2945
2946 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2947 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2948 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2949 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2950 does: it's just the validity check that is more restrictive.
2951
2952 16. Inserted checks for integer overflows during escape sequence (backslash)
2953 processing, and also fixed erroneous offset values for syntax errors during
2954 backslash processing.
2955
2956 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2957 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2958
2959 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2960 caused an overrun.
2961
2962 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2963 something other than just ASCII characters) inside a group that had an
2964 unlimited repeat caused a loop at compile time (while checking to see
2965 whether the group could match an empty string).
2966
2967 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2968 [\P{Any}] did so. (Error in the code for printing property names.)
2969
2970 21. An orphan \E inside a character class could cause a crash.
2971
2972 22. A repeated capturing bracket such as (A)? could cause a wild memory
2973 reference during compilation.
2974
2975 23. There are several functions in pcre_compile() that scan along a compiled
2976 expression for various reasons (e.g. to see if it's fixed length for look
2977 behind). There were bugs in these functions when a repeated \p or \P was
2978 present in the pattern. These operators have additional parameters compared
2979 with \d, etc, and these were not being taken into account when moving along
2980 the compiled data. Specifically:
2981
2982 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2983 length.
2984
2985 (b) An item such as \pL+ within a repeated group could cause crashes or
2986 loops.
2987
2988 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2989 "reference to non-existent subpattern" error.
2990
2991 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2992
2993 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2994 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2995
2996 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2997
2998 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2999 character were causing crashes (broken optimization).
3000
3001 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3002 \p or \P) caused a compile-time loop.
3003
3004 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3005 break. For example, the unstudied pattern [\r\n]A does not match the string
3006 "\r\nA" because change 7.0/46 below moves the current point on by two
3007 characters after failing to match at the start. However, the pattern \nA
3008 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3009 the same is true. There doesn't seem any very clean way out of this, but
3010 what I have chosen to do makes the common cases work: PCRE now takes note
3011 of whether there can be an explicit match for \r or \n anywhere in the
3012 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3013 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3014 pattern has explicit CR or LF references.
3015
3016 29. Added (*CR) etc for changing newline setting at start of pattern.
3017
3018
3019 Version 7.2 19-Jun-07
3020 ---------------------
3021
3022 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3023 which is apparently normally available under Windows.
3024
3025 2. Re-jig the pcregrep tests with different newline settings in an attempt
3026 to make them independent of the local environment's newline setting.
3027
3028 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3029
3030 4. Some of the "internals" tests were previously cut out when the link size
3031 was not 2, because the output contained actual offsets. The recent new
3032 "Z" feature of pcretest means that these can be cut out, making the tests
3033 usable with all link sizes.
3034
3035 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3036 stack recursion. This gives a massive performance boost under BSD, but just
3037 a small improvement under Linux. However, it saves one field in the frame
3038 in all cases.
3039
3040 6. Added more features from the forthcoming Perl 5.10:
3041
3042 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3043 recursion call. It refers to the nth most recently opened parentheses.
3044
3045 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3046 to be opened parentheses.
3047
3048 (c) Conditions that refer to capturing parentheses can be specified
3049 relatively, for example, (?(-2)... or (?(+3)...
3050
3051 (d) \K resets the start of the current match so that everything before
3052 is not part of it.
3053
3054 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3055
3056 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3057 reference syntax.
3058
3059 (g) (?| introduces a group in which the numbering of parentheses in each
3060 alternative starts with the same number.
3061
3062 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3063
3064 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3065 PCRE_INFO_JCHANGED.
3066
3067 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3068 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3069 for detecting groups that can match an empty string.
3070
3071 9. A pattern with a very large number of alternatives (more than several
3072 hundred) was running out of internal workspace during the pre-compile
3073 phase, where pcre_compile() figures out how much memory will be needed. A
3074 bit of new cunning has reduced the workspace needed for groups with
3075 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3076 workspace instead of running out of the 4096 that are available.
3077
3078 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3079
3080 11. Applied patch from Google to remove an optimization that didn't quite work.
3081 The report of the bug said:
3082
3083 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3084 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3085 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3086
3087 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3088 it matched the wrong number of bytes.
3089
3090
3091 Version 7.1 24-Apr-07
3092 ---------------------
3093
3094 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3095 that is more "standard", making use of automake and other Autotools. There
3096 is some re-arrangement of the files and adjustment of comments consequent
3097 on this.
3098
3099 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3100 for recursive directory scanning broke on some systems because the files
3101 are not scanned in any specific order and on different systems the order
3102 was different. A call to "sort" has been inserted into RunGrepTest for the
3103 approprate test as a short-term fix. In the longer term there may be an
3104 alternative.
3105
3106 3. I had an email from Eric Raymond about problems translating some of PCRE's
3107 man pages to HTML (despite the fact that I distribute HTML pages, some
3108 people do their own conversions for various reasons). The problems
3109 concerned the use of low-level troff macros .br and .in. I have therefore
3110 removed all such uses from the man pages (some were redundant, some could
3111 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3112 HTML has been updated to handle .nf/.fi and to complain if it encounters
3113 .br or .in.
3114
3115 4. Updated comments in configure.ac that get placed in config.h.in and also
3116 arranged for config.h to be included in the distribution, with the name
3117 config.h.generic, for the benefit of those who have to compile without
3118 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3119
3120 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3121 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3122 makevp.bat for the current PCRE, using the additional files
3123 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3124
3125 6. A Windows user reported a minor discrepancy with test 2, which turned out
3126 to be caused by a trailing space on an input line that had got lost in his
3127 copy. The trailing space was an accident, so I've just removed it.
3128
3129 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3130 that is needed.
3131
3132 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3133 as "const" (a) because they are and (b) because it helps the PHP
3134 maintainers who have recently made a script to detect big data structures
3135 in the php code that should be moved to the .rodata section. I remembered
3136 to update Builducptable as well, so it won't revert if ucptable.h is ever
3137 re-created.
3138
3139 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3140 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3141 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3142 support is not required. This saves 1.5-2K of code, which is important in
3143 some applications.
3144
3145 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3146 so as not to refer to the tables, even though these functions will never be
3147 called when UTF-8 support is disabled. Otherwise there are problems with a
3148 shared library.
3149
3150 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3151
3152 (a) It was defining its arguments as char * instead of void *.
3153
3154 (b) It was assuming that all moves were upwards in memory; this was true
3155 a long time ago when I wrote it, but is no longer the case.
3156
3157 The emulated memove() is provided for those environments that have neither
3158 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3159 is clearly not the case, as these two bugs were recently reported.
3160
3161 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3162 and Detrail to create the HTML documentation, the .txt form of the man
3163 pages, and it removes trailing spaces from listed files. It also creates
3164 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3165 case, it wraps all the #defines with #ifndefs. This script should be run
3166 before "make dist".
3167
3168 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3169 with Unicode property support.
3170
3171 (a) For a maximizing quantifier, if the two different cases of the
3172 character were of different lengths in their UTF-8 codings (there are
3173 some cases like this - I found 11), and the matching function had to
3174 back up over a mixture of the two cases, it incorrectly assumed they
3175 were both the same length.
3176
3177 (b) When PCRE was configured to use the heap rather than the stack for
3178 recursion during matching, it was not correctly preserving the data for
3179 the other case of a UTF-8 character when checking ahead for a match
3180 while processing a minimizing repeat. If the check also involved
3181 matching a wide character, but failed, corruption could cause an
3182 erroneous result when trying to check for a repeat of the original
3183 character.
3184
3185 13. Some tidying changes to the testing mechanism:
3186
3187 (a) The RunTest script now detects the internal link size and whether there
3188 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3189 values substituted by "configure". (The RunGrepTest script already did
3190 this for UTF-8.) The configure.ac script no longer substitutes the
3191 relevant variables.
3192
3193 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3194 with length and offset values. This means that the output is different
3195 for different internal link sizes. Test 2 is skipped for link sizes
3196 other than 2 because of this, bypassing the problem. Unfortunately,
3197 there was also a test in test 3 (the locale tests) that used /B and
3198 failed for link sizes other than 2. Rather than cut the whole test out,
3199 I have added a new /Z option to pcretest that replaces the length and
3200 offset values with spaces. This is now used to make test 3 independent
3201 of link size. (Test 2 will be tidied up later.)
3202
3203 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3204 segmentation fault instead of returning the appropriate error message.
3205
3206 15. In multiline mode when the newline sequence was set to "any", the pattern
3207 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3208 This doesn't seem right; it now treats the CRLF combination as the line
3209 ending, and so does not match in that case. It's only a pattern such as ^$
3210 that would hit this one: something like ^ABC$ would have failed after \r
3211 and then tried again after \r\n.
3212
3213 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3214 in an attempt to make files that differ only in their line terminators
3215 compare equal. This works on Linux.
3216
3217 17. Under certain error circumstances pcregrep might try to free random memory
3218 as it exited. This is now fixed, thanks to valgrind.
3219
3220 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3221 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3222 was because its rules for how to advance for /g after matching an empty
3223 string at the end of a line did not allow for this case. They now check for
3224 it specially.
3225
3226 20. pcretest is supposed to handle patterns and data of any length, by
3227 extending its buffers when necessary. It was getting this wrong when the
3228 buffer for a data line had to be extended.
3229
3230 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3231 CRLF as a newline sequence.
3232
3233 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3234 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3235 I have nevertheless tidied it up.
3236
3237 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3238
3239 24. Added a man page for pcre-config.
3240
3241
3242 Version 7.0 19-Dec-06
3243 ---------------------
3244
3245 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3246 moving to gcc 4.1.1.
3247
3248 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3249 sys/time.h, which is documented as needed for this function. It doesn't
3250 seem to matter on Linux, but it showed up on some releases of OS X.
3251
3252 3. It seems that there are systems where bytes whose values are greater than
3253 127 match isprint() in the "C" locale. The "C" locale should be the
3254 default when a C program starts up. In most systems, only ASCII printing
3255 characters match isprint(). This difference caused the output from pcretest
3256 to vary, making some of the tests fail. I have changed pcretest so that:
3257
3258 (a) When it is outputting text in the compiled version of a pattern, bytes
3259 other than 32-126 are always shown as hex escapes.
3260
3261 (b) When it is outputting text that is a matched part of a subject string,
3262 it does the same, unless a different locale has been set for the match
3263 (using the /L modifier). In this case, it uses isprint() to decide.
3264
3265 4. Fixed a major bug that caused incorrect computation of the amount of memory
3266 required for a compiled pattern when options that changed within the
3267 pattern affected the logic of the preliminary scan that determines the
3268 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3269 that the computed length was too small. The symptoms of this bug were
3270 either the PCRE error "internal error: code overflow" from pcre_compile(),
3271 or a glibc crash with a message such as "pcretest: free(): invalid next
3272 size (fast)". Examples of patterns that provoked this bug (shown in
3273 pcretest format) are:
3274
3275 /(?-x: )/x
3276 /(?x)(?-x: \s*#\s*)/
3277 /((?i)[\x{c0}])/8
3278 /(?i:[\x{c0}])/8
3279
3280 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3281 is now done differently.
3282
3283 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3284 wrapper classes; (b) implement a new function in the C++ scanner that is
3285 more efficient than the old way of doing things because it avoids levels of
3286 recursion in the regex matching; (c) add a paragraph to the documentation
3287 for the FullMatch() function.
3288
3289 6. The escape sequence \n was being treated as whatever was defined as
3290 "newline". Not only was this contrary to the documentation, which states
3291 that \n is character 10 (hex 0A), but it also went horribly wrong when
3292 "newline" was defined as CRLF. This has been fixed.
3293
3294 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3295 was being set to -1 for the "end of line" case (supposedly a value that no
3296 character can have). Though this value is never used (the check for end of
3297 line is "zero bytes in current character"), it caused compiler complaints.
3298 I've changed it to 0xffffffff.
3299
3300 8. In pcre_version.c, the version string was being built by a sequence of
3301 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3302 string (as it is for production releases) called a macro with an empty
3303 argument. The C standard says the result of this is undefined. The gcc
3304 compiler treats it as an empty string (which was what was wanted) but it is
3305 reported that Visual C gives an error. The source has been hacked around to
3306 avoid this problem.
3307
3308 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3309 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3310 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3311 of them did).
3312
3313 10. Originally, pcretest opened its input and output without "b"; then I was
3314 told that "b" was needed in some environments, so it was added for release
3315 5.0 to both the input and output. (It makes no difference on Unix-like
3316 systems.) Later I was told that it is wrong for the input on Windows. I've
3317 now abstracted the modes into two macros, to make it easier to fiddle with
3318 them, and removed "b" from the input mode under Windows.
3319
3320 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3321
3322 12. Added -help and --help to pcretest as an official way of being reminded
3323 of the options.
3324
3325 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3326 and pcrecpp.cc because they annoy compilers at high warning levels.
3327
3328 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3329
3330 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3331 scripts are not C programs :-) and which was not noticed because it works
3332 on Linux.
3333
3334 16. pcretest is supposed to handle any length of pattern and data line (as one
3335 line or as a continued sequence of lines) by extending its input buffer if
3336 necessary. This feature was broken for very long pattern lines, leading to
3337 a string of junk being passed to pcre_compile() if the pattern was longer
3338 than about 50K.
3339
3340 17. I have done a major re-factoring of the way pcre_compile() computes the
3341 amount of memory needed for a compiled pattern. Previously, there was code
3342 that made a preliminary scan of the pattern in order to do this. That was
3343 OK when PCRE was new, but as the facilities have expanded, it has become
3344 harder and harder to keep it in step with the real compile phase, and there
3345 have been a number of bugs (see for example, 4 above). I have now found a
3346 cunning way of running the real compile function in a "fake" mode that
3347 enables it to compute how much memory it would need, while actually only
3348 ever using a few hundred bytes of working memory and without too many
3349 tests of the mode. This should make future maintenance and development
3350 easier. A side effect of this work is that the limit of 200 on the nesting
3351 depth of parentheses has been removed (though this was never a serious
3352 limitation, I suspect). However, there is a downside: pcre_compile() now
3353 runs more slowly than before (30% or more, depending on the pattern). I
3354 hope this isn't a big issue. There is no effect on runtime performance.
3355
3356 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3357 newline (only possible for the last line of a file) and it was a
3358 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3359
3360 19. Added additional timing features to pcretest. (1) The -tm option now times
3361 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3362 separate command line item, by a number that specifies the number of
3363 repeats to use when timing. The default is 50000; this gives better
3364 precision, but takes uncomfortably long for very large patterns.
3365
3366 20. Extended pcre_study() to be more clever in cases where a branch of a
3367 subpattern has no definite first character. For example, (a*|b*)[cd] would
3368 previously give no result from pcre_study(). Now it recognizes that the
3369 first character must be a, b, c, or d.
3370
3371 21. There was an incorrect error "recursive call could loop indefinitely" if
3372 a subpattern (or the entire pattern) that was being tested for matching an
3373 empty string contained only one non-empty item after a nested subpattern.
3374 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3375 incorrectly, because the \d was being skipped in the check.
3376
3377 22. The pcretest program now has a new pattern option /B and a command line
3378 option -b, which is equivalent to adding /B to every pattern. This causes
3379 it to show the compiled bytecode, without the additional information that
3380 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3381 is the same as /B/I).
3382
3383 23. A new optimization is now able automatically to treat some sequences such
3384 as a*b as a*+b. More specifically, if something simple (such as a character
3385 or a simple class like \d) has an unlimited quantifier, and is followed by
3386 something that cannot possibly match the quantified thing, the quantifier
3387 is automatically "possessified".
3388
3389 24. A recursive reference to a subpattern whose number was greater than 39
3390 went wrong under certain circumstances in UTF-8 mode. This bug could also
3391 have affected the operation of pcre_study().
3392
3393 25. Realized that a little bit of performance could be had by replacing
3394 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3395
3396 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3397
3398 27. Possessive quantifiers such as a++ were previously implemented by turning
3399 them into atomic groups such as ($>a+). Now they have their own opcodes,
3400 which improves performance. This includes the automatically created ones
3401 from 23 above.
3402
3403 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3404 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3405 the first matched character to be a colon. This applied both to named and
3406 numbered groups.
3407
3408 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3409
3410 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3411 building PCRE on Windows easier, so I have included it in the distribution.
3412
3413 31. There is now a check in pcretest against a ridiculously large number being
3414 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3415 loop, the loop is abandoned.
3416
3417 32. Forward references to subpatterns in conditions such as (?(2)...) where
3418 subpattern 2 is defined later cause pcre_compile() to search forwards in
3419 the pattern for the relevant set of parentheses. This search went wrong
3420 when there were unescaped parentheses in a character class, parentheses
3421 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3422
3423 33. "Subroutine" calls and backreferences were previously restricted to
3424 referencing subpatterns earlier in the regex. This restriction has now
3425 been removed.
3426
3427 34. Added a number of extra features that are going to be in Perl 5.10. On the
3428 whole, these are just syntactic alternatives for features that PCRE had
3429 previously implemented using the Python syntax or my own invention. The
3430 other formats are all retained for compatibility.
3431
3432 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3433 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3434 also .NET compatible.
3435
3436 (b) A recursion or subroutine call to a named group can now be defined as
3437 (?&name) as well as (?P>name).
3438
3439 (c) A backreference to a named group can now be defined as \k<name> or
3440 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3441 5.10, are also .NET compatible.
3442
3443 (d) A conditional reference to a named group can now use the syntax
3444 (?(<name>) or (?('name') as well as (?(name).
3445
3446 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3447 groups (named and numbered) that are never evaluated inline, but can be
3448 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3449 is always false. There may be only one alternative in such a group.
3450
3451 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3452 as the simple (?(R). The condition is true only if the most recent
3453 recursion is that of the given number or name. It does not search out
3454 through the entire recursion stack.
3455
3456 (g) The escape \gN or \g{N} has been added, where N is a positive or
3457 negative number, specifying an absolute or relative reference.
3458
3459 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3460 some "unreachable code" warnings.
3461
3462 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3463 things, this adds five new scripts.
3464
3465 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3466 There were also incompatibilities regarding the handling of \Q..\E inside
3467 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3468 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3469
3470 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3471 matches an empty string, and forcibly breaks the loop. There were bugs in
3472 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3473 against aaaa the result was just "a" rather than "aaaa", for example. Two
3474 separate and independent bugs (that affected different cases) have been
3475 fixed.
3476
3477 39. Refactored the code to abolish the use of different opcodes for small
3478 capturing bracket numbers. This is a tidy that I avoided doing when I
3479 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3480 The new approach is not only tidier, it makes it possible to reduce the
3481 memory needed to fix the previous bug (38).
3482
3483 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3484 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3485 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3486 mode.
3487
3488 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3489 report.
3490
3491 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3492 copy construction and assignment in the C++ wrapper.
3493
3494 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3495 couple of bugs that could have given wrong results in the "--newline=crlf"
3496 case.
3497
3498 44. Added a number of casts and did some reorganization of signed/unsigned int
3499 variables following suggestions from Dair Grant. Also renamed the variable
3500 "this" as "item" because it is a C++ keyword.
3501
3502 45. Arranged for dftables to add
3503
3504 #include "pcre_internal.h"
3505
3506 to pcre_chartables.c because without it, gcc 4.x may remove the array
3507 definition from the final binary if PCRE is built into a static library and
3508 dead code stripping is activated.
3509
3510 46. For an unanchored pattern, if a match attempt fails at the start of a
3511 newline sequence, and the newline setting is CRLF or ANY, and the next two
3512 characters are CRLF, advance by two characters instead of one.
3513
3514
3515 Version 6.7 04-Jul-06
3516 ---------------------
3517
3518 1. In order to handle tests when input lines are enormously long, pcretest has
3519 been re-factored so that it automatically extends its buffers when
3520 necessary. The code is crude, but this _is_ just a test program. The
3521 default size has been increased from 32K to 50K.
3522
3523 2. The code in pcre_study() was using the value of the re argument before
3524 testing it for NULL. (Of course, in any sensible call of the function, it
3525 won't be NULL.)
3526
3527 3. The memmove() emulation function in pcre_internal.h, which is used on
3528 systems that lack both memmove() and bcopy() - that is, hardly ever -
3529 was missing a "static" storage class specifier.
3530
3531 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3532 containing an extended class (one that cannot be represented by a bitmap
3533 because it contains high-valued characters or Unicode property items, e.g.
3534 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3535 pattern, but PCRE should not loop if you do not (it no longer does).
3536 [Detail: two cases were found: (a) a repeated subpattern containing an
3537 extended class; (b) a recursive reference to a subpattern that followed a
3538 previous extended class. It wasn't skipping over the extended class
3539 correctly when UTF-8 mode was not set.]
3540
3541 5. A negated single-character class was not being recognized as fixed-length
3542 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3543 compile error "lookbehind assertion is not fixed length".
3544
3545 6. The RunPerlTest auxiliary script was showing an unexpected difference
3546 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3547 write a Perl script that can interpret lines of an input file either as
3548 byte characters or as UTF-8, which is what "perltest" was being required to
3549 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3550 can't do is switch easily at run time between having the "use utf8;" pragma
3551 or not. In the end, I fudged it by using the RunPerlTest script to insert
3552 "use utf8;" explicitly for the UTF-8 tests.
3553
3554 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3555 the end of the subject string, contrary to the documentation and to what
3556 Perl does. This was true of both matching functions. Now it matches only at
3557 the start of the subject and immediately after *internal* newlines.
3558
3559 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3560 a pointer to an int instead of a pointer to an unsigned long int. This
3561 caused problems on 64-bit systems.
3562
3563 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3564 instance of the 'standard' template library not being so standard".
3565
3566 10. There was no check on the number of named subpatterns nor the maximum
3567 length of a subpattern name. The product of these values is used to compute
3568 the size of the memory block for a compiled pattern. By supplying a very
3569 long subpattern name and a large number of named subpatterns, the size
3570 computation could be caused to overflow. This is now prevented by limiting
3571 the length of names to 32 characters, and the number of named subpatterns
3572 to 10,000.
3573
3574 11. Subpatterns that are repeated with specific counts have to be replicated in
3575 the compiled pattern. The size of memory for this was computed from the
3576 length of the subpattern and the repeat count. The latter is limited to
3577 65535, but there was no limit on the former, meaning that integer overflow
3578 could in principle occur. The compiled length of a repeated subpattern is
3579 now limited to 30,000 bytes in order to prevent this.
3580
3581 12. Added the optional facility to have named substrings with the same name.
3582
3583 13. Added the ability to use a named substring as a condition, using the
3584 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3585 are numbers (not recommended). Forward references are permitted.
3586
3587 14. Added forward references in named backreferences (if you see what I mean).
3588
3589 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3590 pattern could run off the end of the subject. For example, the pattern
3591 "(?s)(.{1,5})"8 did this with the subject "ab".
3592
3593 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3594 PCRE_CASELESS was set when matching characters that were quantified with ?
3595 or *.
3596
3597 17. A character class other than a single negated character that had a minimum
3598 but no maximum quantifier - for example [ab]{6,} - was not handled
3599 correctly by pce_dfa_exec(). It would match only one character.
3600
3601 18. A valid (though odd) pattern that looked like a POSIX character
3602 class but used an invalid character after [ (for example [[,abc,]]) caused
3603 pcre_compile() to give the error "Failed: internal error: code overflow" or
3604 in some cases to crash with a glibc free() error. This could even happen if
3605 the pattern terminated after [[ but there just happened to be a sequence of
3606 letters, a binary zero, and a closing ] in the memory that followed.
3607
3608 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3609 over the years. Originally (before any Unicode support), just the bottom 8
3610 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3611 output from "man perlunicode" includes this:
3612
3613 The regular expression compiler produces polymorphic opcodes. That
3614 is, the pattern adapts to the data and automatically switches to
3615 the Unicode character scheme when presented with Unicode data--or
3616 instead uses a traditional byte scheme when presented with byte
3617 data.
3618
3619 Sadly, a wide octal escape does not cause a switch, and in a string with
3620 no other multibyte characters, these octal escapes are treated as before.
3621 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3622 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3623 Unicode string.
3624
3625 I have not perpetrated such confusion in PCRE. Up till now, it took just
3626 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3627 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3628 translate to the appropriate multibyte character.
3629
3630 29. Applied some refactoring to reduce the number of warnings from Microsoft
3631 and Borland compilers. This has included removing the fudge introduced
3632 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3633 a warning about an unused variable.
3634
3635 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3636 characters since release 4.0, because Perl (from release 5.004) does not.
3637 [Or at least, is documented not to: some releases seem to be in conflict
3638 with the documentation.] However, when a pattern was studied with
3639 pcre_study() and all its branches started with \s, PCRE still included VT
3640 as a possible starting character. Of course, this did no harm; it just
3641 caused an unnecessary match attempt.
3642
3643 22. Removed a now-redundant internal flag bit that recorded the fact that case
3644 dependency changed within the pattern. This was once needed for "required
3645 byte" processing, but is no longer used. This recovers a now-scarce options
3646 bit. Also moved the least significant internal flag bit to the most-
3647 significant bit of the word, which was not previously used (hangover from
3648 the days when it was an int rather than a uint) to free up another bit for
3649 the future.
3650
3651 23. Added support for CRLF line endings as well as CR and LF. As well as the
3652 default being selectable at build time, it can now be changed at runtime
3653 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3654 specify that it is scanning data with non-default line endings.
3655
3656 24. Changed the definition of CXXLINK to make it agree with the definition of
3657 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3658
3659 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3660 recursions. This makes a big different to stack usage for some patterns.
3661
3662 26. If a subpattern containing a named recursion or subroutine reference such
3663 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3664 the space required for the compiled pattern went wrong and gave too small a
3665 value. Depending on the environment, this could lead to "Failed: internal
3666 error: code overflow at offset 49" or "glibc detected double free or
3667 corruption" errors.
3668
3669 27. Applied patches from Google (a) to support the new newline modes and (b) to
3670 advance over multibyte UTF-8 characters in GlobalReplace.
3671
3672 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3673 difference for some implementation of PCRE in some Windows version.
3674
3675 29. Added some extra testing facilities to pcretest:
3676
3677 \q<number> in a data line sets the "match limit" value
3678 \Q<number> in a data line sets the "match recursion limt" value
3679 -S <number> sets the stack size, where <number> is in megabytes
3680
3681 The -S option isn't available for Windows.
3682
3683
3684 Version 6.6 06-Feb-06
3685 ---------------------
3686
3687 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3688 in pcreposix.h. I have copied the definition from pcre.h.
3689
3690 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3691 because pcre.h is no longer a built file.
3692
3693 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3694 not normally included in the compiled code.
3695
3696
3697 Version 6.5 01-Feb-06
3698 ---------------------
3699
3700 1. When using the partial match feature with pcre_dfa_exec(), it was not
3701 anchoring the second and subsequent partial matches at the new starting
3702 point. This could lead to incorrect results. For example, with the pattern
3703 /1234/, partially matching against "123" and then "a4" gave a match.
3704
3705 2. Changes to pcregrep:
3706
3707 (a) All non-match returns from pcre_exec() were being treated as failures
3708 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3709 error message is output. Some extra information is given for the
3710 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3711 probably the only errors that are likely to be caused by users (by
3712 specifying a regex that has nested indefinite repeats, for instance).
3713 If there are more than 20 of these errors, pcregrep is abandoned.
3714
3715 (b) A binary zero was treated as data while matching, but terminated the
3716 output line if it was written out. This has been fixed: binary zeroes
3717 are now no different to any other data bytes.
3718
3719 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3720 used to set a locale for matching. The --locale=xxxx long option has
3721 been added (no short equivalent) to specify a locale explicitly on the
3722 pcregrep command, overriding the environment variables.
3723
3724 (d) When -B was used with -n, some line numbers in the output were one less
3725 than they should have been.
3726
3727 (e) Added the -o (--only-matching) option.
3728
3729 (f) If -A or -C was used with -c (count only), some lines of context were
3730 accidentally printed for the final match.
3731
3732 (g) Added the -H (--with-filename) option.
3733
3734 (h) The combination of options -rh failed to suppress file names for files
3735 that were found from directory arguments.
3736
3737 (i) Added the -D (--devices) and -d (--directories) options.
3738
3739 (j) Added the -F (--fixed-strings) option.
3740
3741 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3742
3743 (l) Added the --colo(u)r option.
3744
3745 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3746 is not present by default.
3747
3748 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3749 items such as (?R) or (?1), when the recursion could match a number of
3750 alternatives. If it matched one of the alternatives, but subsequently,
3751 outside the recursion, there was a failure, the code tried to back up into
3752 the recursion. However, because of the way PCRE is implemented, this is not
3753 possible, and the result was an incorrect result from the match.
3754
3755 In order to prevent this happening, the specification of recursion has
3756 been changed so that all such subpatterns are automatically treated as
3757 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3758
3759 4. I had overlooked the fact that, in some locales, there are characters for
3760 which isalpha() is true but neither isupper() nor islower() are true. In
3761 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3762 and ordfeminine) are like this. This affected the treatment of \w and \W
3763 when they appeared in character classes, but not when they appeared outside
3764 a character class. The bit map for "word" characters is now created
3765 separately from the results of isalnum() instead of just taking it from the
3766 upper, lower, and digit maps. (Plus the underscore character, of course.)
3767
3768 5. The above bug also affected the handling of POSIX character classes such as
3769 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3770 permanent tables. Instead, the bit maps for such a class were previously
3771 created as the appropriate unions of the upper, lower, and digit bitmaps.
3772 Now they are created by subtraction from the [[:word:]] class, which has
3773 its own bitmap.
3774
3775 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3776 It is created by subtracting the vertical space characters (\x09, \x0a,
3777 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3778 subtraction was done in the overall bitmap for a character class, meaning
3779 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3780 be recognized. This bug has been fixed.
3781
3782 7. Patches from the folks at Google:
3783
3784 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3785 real life, but is still worth protecting against".
3786
3787 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3788 regular expressions".
3789
3790 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3791 have it.
3792
3793 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3794 "configure" and the latter not, in order to fix a problem somebody had
3795 with compiling the Arg class on HP-UX.
3796
3797 (e) Improve the error-handling of the C++ wrapper a little bit.
3798
3799 (f) New tests for checking recursion limiting.
3800
3801 8. The pcre_memmove() function, which is used only if the environment does not
3802 have a standard memmove() function (and is therefore rarely compiled),
3803 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3804 returning a result (though PCRE never actually uses the result).
3805
3806 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3807 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3808 returned instead of calling malloc() with an overflowing number that would
3809 most likely cause subsequent chaos.
3810
3811 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3812
3813 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3814 with this option is matched, the nmatch and pmatch options of regexec() are
3815 ignored.
3816
3817 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3818 provided in case anyone wants to the the POSIX interface with UTF-8
3819 strings.
3820
3821 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3822 C++ linking (needed for some HP-UX environments).
3823
3824 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3825 (unused parameter) and in the pcre_printint() function (omitted "default"
3826 switch label when the default is to do nothing).
3827
3828 15. Added some code to make it possible, when PCRE is compiled as a C++
3829 library, to replace subject pointers for pcre_exec() with a smart pointer
3830 class, thus making it possible to process discontinuous strings.
3831
3832 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3833 much the same function. They were added by different people who were trying
3834 to make PCRE easy to compile on non-Unix systems. It has been suggested
3835 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3836 for compiling on Windows systems. I have therefore replaced it with
3837 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3838 defaults to "extern" for C or "extern C" for C++, which works fine on
3839 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3840 SCOPE with something explicit in config.h. In addition:
3841
3842 (a) pcreposix.h still had just "extern" instead of either of these macros;
3843 I have replaced it with PCRE_DATA_SCOPE.
3844
3845 (b) Functions such as _pcre_xclass(), which are internal to the library,
3846 but external in the C sense, all had PCRE_EXPORT in their definitions.
3847 This is apparently wrong for the Windows case, so I have removed it.
3848 (It makes no difference on Unix-like systems.)
3849
3850 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3851 of recursive calls to match(). This is different to MATCH_LIMIT because
3852 that limits the total number of calls to match(), not all of which increase
3853 the depth of recursion. Limiting the recursion depth limits the amount of
3854 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3855 when PCRE is compiled, and changed at run time. A patch from Google adds
3856 this functionality to the C++ interface.
3857
3858 18. Changes to the handling of Unicode character properties:
3859
3860 (a) Updated the table to Unicode 4.1.0.
3861
3862 (b) Recognize characters that are not in the table as "Cn" (undefined).
3863
3864 (c) I revised the way the table is implemented to a much improved format
3865 which includes recognition of ranges. It now supports the ranges that
3866 are defined in UnicodeData.txt, and it also amalgamates other
3867 characters into ranges. This has reduced the number of entries in the
3868 table from around 16,000 to around 3,000, thus reducing its size
3869 considerably. I realized I did not need to use a tree structure after
3870 all - a binary chop search is just as efficient. Having reduced the
3871 number of entries, I extended their size from 6 bytes to 8 bytes to
3872 allow for more data.
3873
3874 (d) Added support for Unicode script names via properties such as \p{Han}.
3875
3876 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3877 matching that character.
3878
3879 20. When matching a repeated Unicode property with a minimum greater than zero,
3880 (for example \pL{2,}), PCRE could look past the end of the subject if it
3881 reached it while seeking the minimum number of characters. This could
3882 happen only if some of the characters were more than one byte long, because
3883 there is a check for at least the minimum number of bytes.
3884
3885 21. Refactored the implementation of \p and \P so as to be more general, to
3886 allow for more different types of property in future. This has changed the
3887 compiled form incompatibly. Anybody with saved compiled patterns that use
3888 \p or \P will have to recompile them.
3889
3890 22. Added "Any" and "L&" to the supported property types.
3891
3892 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3893 but give a compile time error if the value is greater than 0xff.
3894
3895 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3896 accidentally not being installed or uninstalled.
3897
3898 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3899 made were to insert the current release number. This seemed silly, because
3900 it made things harder for people building PCRE on systems that don't run
3901 "configure". I have turned pcre.h into a distributed file, no longer built
3902 by "configure", with the version identification directly included. There is
3903 no longer a pcre.h.in file.
3904
3905 However, this change necessitated a change to the pcre-config script as
3906 well. It is built from pcre-config.in, and one of the substitutions was the
3907 release number. I have updated configure.ac so that ./configure now finds
3908 the release number by grepping pcre.h.
3909
3910 26. Added the ability to run the tests under valgrind.
3911
3912
3913 Version 6.4 05-Sep-05
3914 ---------------------
3915
3916 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3917 "--" to be printed when multiple files were scanned, even when none of the
3918 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3919 consider it to be a bug, and have restored the previous behaviour.
3920
3921 2. A couple of code tidies to get rid of compiler warnings.
3922
3923 3. The pcretest program used to cheat by referring to symbols in the library
3924 whose names begin with _pcre_. These are internal symbols that are not
3925 really supposed to be visible externally, and in some environments it is
3926 possible to suppress them. The cheating is now confined to including
3927 certain files from the library's source, which is a bit cleaner.
3928
3929 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3930 file's purpose clearer.
3931
3932 5. Reorganized pcre_ucp_findchar().
3933
3934
3935 Version 6.3 15-Aug-05
3936 ---------------------
3937
3938 1. The file libpcre.pc.in did not have general read permission in the tarball.
3939
3940 2. There were some problems when building without C++ support:
3941
3942 (a) If C++ support was not built, "make install" and "make test" still
3943 tried to test it.
3944
3945 (b) There were problems when the value of CXX was explicitly set. Some
3946 changes have been made to try to fix these, and ...
3947
3948 (c) --disable-cpp can now be used to explicitly disable C++ support.
3949
3950 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3951 backslash in a target when C++ was disabled. This confuses some
3952 versions of "make", apparently. Using an intermediate variable solves
3953 this. (Same for CPP_LOBJ.)
3954
3955 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3956 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3957 necessary on certain architectures.
3958
3959 4. Added a setting of -export-symbols-regex to the link command to remove
3960 those symbols that are exported in the C sense, but actually are local
3961 within the library, and not documented. Their names all begin with
3962 "_pcre_". This is not a perfect job, because (a) we have to except some
3963 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3964 available (and never for static libraries). I have made a note to try to
3965 find a way round (a) in the future.
3966
3967
3968 Version 6.2 01-Aug-05
3969 ---------------------
3970
3971 1. There was no test for integer overflow of quantifier values. A construction
3972 such as {1111111111111111} would give undefined results. What is worse, if
3973 a minimum quantifier for a parenthesized subpattern overflowed and became
3974 negative, the calculation of the memory size went wrong. This could have
3975 led to memory overwriting.
3976
3977 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3978
3979 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3980 operating environments where this matters.
3981
3982 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3983 PCRE options from within the C++ wrapper.
3984
3985 5. Named capturing subpatterns were not being correctly counted when a pattern
3986 was compiled. This caused two problems: (a) If there were more than 100
3987 such subpatterns, the calculation of the memory needed for the whole
3988 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3989 back references of the form \12, where the number was greater than 9, were
3990 not recognized as back references, even though there were sufficient
3991 previous subpatterns.
3992
3993 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3994 versions of gcc, e.g. 2.95.4.
3995
3996
3997 Version 6.1 21-Jun-05
3998 ---------------------
3999
4000 1. There was one reference to the variable "posix" in pcretest.c that was not
4001 surrounded by "#if !defined NOPOSIX".
4002
4003 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4004 the cross-check on the old pcre_info() function, for the benefit of the
4005 cut-down version of PCRE that is currently imported into Exim.
4006
4007 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4008 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4009 patterns that start (?i) but I don't think that matters. The use of (?i) is
4010 just an example; this all applies to the other options as well.
4011
4012 4. Since libtool seems to echo the compile commands it is issuing, the output
4013 from "make" can be reduced a bit by putting "@" in front of each libtool
4014 compile command.
4015
4016 5. Patch from the folks at Google for configure.in to be a bit more thorough
4017 in checking for a suitable C++ installation before trying to compile the
4018 C++ stuff. This should fix a reported problem when a compiler was present,
4019 but no suitable headers.
4020
4021 6. The man pages all had just "PCRE" as their title. I have changed them to
4022 be the relevant file name. I have also arranged that these names are
4023 retained in the file doc/pcre.txt, which is a concatenation in text format
4024 of all the man pages except the little individual ones for each function.
4025
4026 7. The NON-UNIX-USE file had not been updated for the different set of source
4027 files that come with release 6. I also added a few comments about the C++
4028 wrapper.
4029
4030
4031 Version 6.0 07-Jun-05
4032 ---------------------
4033
4034 1. Some minor internal re-organization to help with my DFA experiments.
4035
4036 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4037 didn't matter for the library itself when fully configured, but did matter
4038 when compiling without UCP support, or within Exim, where the ucp files are
4039 not imported.
4040
4041 3. Refactoring of the library code to split up the various functions into
4042 different source modules. The addition of the new DFA matching code (see
4043 below) to a single monolithic source would have made it really too
4044 unwieldy, quite apart from causing all the code to be include in a
4045 statically linked application, when only some functions are used. This is
4046 relevant even without the DFA addition now that patterns can be compiled in
4047 one application and matched in another.
4048
4049 The downside of splitting up is that there have to be some external
4050 functions and data tables that are used internally in different modules of
4051 the library but which are not part of the API. These have all had their
4052 names changed to start with "_pcre_" so that they are unlikely to clash
4053 with other external names.
4054
4055 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4056 a different (DFA) algorithm. Although it is slower than the original
4057 function, it does have some advantages for certain types of matching
4058 problem.
4059
4060 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4061 including restarting after a partial match.
4062
4063 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4064 defined when compiling for Windows was sent to me. I have put it into the
4065 code, though I have no means of testing or verifying it.
4066
4067 7. Added the pcre_refcount() auxiliary function.
4068
4069 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4070 match before or at the first newline in the subject string. In pcretest,
4071 the /f option on a pattern can be used to set this.
4072
4073 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4074 would behave wrongly. This has been present in PCRE since release 4.0.
4075
4076 10. A number of changes to the pcregrep command:
4077
4078 (a) Refactored how -x works; insert ^(...)$ instead of setting
4079 PCRE_ANCHORED and checking the length, in preparation for adding
4080 something similar for -w.
4081
4082 (b) Added the -w (match as a word) option.
4083
4084 (c) Refactored the way lines are read and buffered so as to have more
4085 than one at a time available.
4086
4087 (d) Implemented a pcregrep test script.
4088
4089 (e) Added the -M (multiline match) option. This allows patterns to match
4090 over several lines of the subject. The buffering ensures that at least
4091 8K, or the rest of the document (whichever is the shorter) is available
4092 for matching (and similarly the previous 8K for lookbehind assertions).
4093
4094 (f) Changed the --help output so that it now says
4095
4096 -w, --word-regex(p)
4097
4098 instead of two lines, one with "regex" and the other with "regexp"
4099 because that confused at least one person since the short forms are the
4100 same. (This required a bit of code, as the output is generated
4101 automatically from a table. It wasn't just a text change.)
4102
4103 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4104 option but starts with a hyphen. Could be a pattern or a path name
4105 starting with a hyphen, for instance.
4106
4107 (h) "-" can be given as a file name to represent stdin.
4108
4109 (i) When file names are being printed, "(standard input)" is used for
4110 the standard input, for compatibility with GNU grep. Previously
4111 "<stdin>" was used.
4112
4113 (j) The option --label=xxx can be used to supply a name to be used for
4114 stdin when file names are being printed. There is no short form.
4115
4116 (k) Re-factored the options decoding logic because we are going to add
4117 two more options that take data. Such options can now be given in four
4118 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4119
4120 (l) Added the -A, -B, and -C options for requesting that lines of context
4121 around matches be printed.
4122
4123 (m) Added the -L option to print the names of files that do not contain
4124 any matching lines, that is, the complement of -l.
4125
4126 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4127 continue to scan other files.
4128
4129 (o) The -s option was incorrectly implemented. For compatibility with other
4130 greps, it now suppresses the error message for a non-existent or non-
4131 accessible file (but not the return code). There is a new option called
4132 -q that suppresses the output of matching lines, which was what -s was
4133 previously doing.
4134
4135 (p) Added --include and --exclude options to specify files for inclusion
4136 and exclusion when recursing.
4137
4138 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4139 Hopefully, it now does.
4140
4141 12. Missing cast in pcre_study().
4142
4143 13. Added an "uninstall" target to the makefile.
4144
4145 14. Replaced "extern" in the function prototypes in Makefile.in with
4146 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4147 world, but is set differently for Windows.
4148
4149 15. Added a second compiling function called pcre_compile2(). The only
4150 difference is that it has an extra argument, which is a pointer to an
4151 integer error code. When there is a compile-time failure, this is set
4152 non-zero, in addition to the error test pointer being set to point to an
4153 error message. The new argument may be NULL if no error number is required
4154 (but then you may as well call pcre_compile(), which is now just a
4155 wrapper). This facility is provided because some applications need a
4156 numeric error indication, but it has also enabled me to tidy up the way
4157 compile-time errors are handled in the POSIX wrapper.
4158
4159 16. Added VPATH=.libs to the makefile; this should help when building with one
4160 prefix path and installing with another. (Or so I'm told by someone who
4161 knows more about this stuff than I do.)
4162
4163 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4164 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4165 match everything, including newlines. This is not POSIX-compatible, but
4166 somebody wanted the feature. From pcretest it can be activated by using
4167 both the P and the s flags.
4168
4169 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4170
4171 19. libpcre.pc was being incorrectly installed as executable.
4172
4173 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4174 it now also looks for '\r' so that it will work unmodified on Windows.
4175
4176 21. Added Google's contributed C++ wrapper to the distribution.
4177
4178 22. Added some untidy missing memory free() calls in pcretest, to keep
4179 Electric Fence happy when testing.
4180
4181
4182
4183 Version 5.0 13-Sep-04
4184 ---------------------
4185
4186 1. Internal change: literal characters are no longer packed up into items
4187 containing multiple characters in a single byte-string. Each character
4188 is now matched using a separate opcode. However, there may be more than one
4189 byte in the character in UTF-8 mode.
4190
4191 2. The pcre_callout_block structure has two new fields: pattern_position and
4192 next_item_length. These contain the offset in the pattern to the next match
4193 item, and its length, respectively.
4194
4195 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4196 insertion of callouts before each pattern item. Added the /C option to
4197 pcretest to make use of this.
4198
4199 4. On the advice of a Windows user, the lines
4200
4201 #if defined(_WIN32) || defined(WIN32)
4202 _setmode( _fileno( stdout ), 0x8000 );
4203 #endif /* defined(_WIN32) || defined(WIN32) */
4204
4205 have been added to the source of pcretest. This apparently does useful
4206 magic in relation to line terminators.
4207
4208 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4209 for the benefit of those environments where the "b" makes a difference.
4210
4211 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4212 to know about it. I have put a hack into configure.in that adds in code
4213 to set GCC=yes if CC=icc. This seems to end up at a point in the
4214 generated configure script that is early enough to affect the setting of
4215 compiler options, which is what is needed, but I have no means of testing
4216 whether it really works. (The user who reported this had patched the
4217 generated configure script, which of course I cannot do.)
4218
4219 LATER: After change 22 below (new libtool files), the configure script
4220 seems to know about icc (and also ecc). Therefore, I have commented out
4221 this hack in configure.in.
4222
4223 7. Added support for pkg-config (2 patches were sent in).
4224
4225 8. Negated POSIX character classes that used a combination of internal tables
4226 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4227 [[:^ascii]]. Typically, they would match almost any characters. The other
4228 POSIX classes were not broken in this way.
4229
4230 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4231 to find the match, as PCRE was deluded into thinking that the match had to
4232 start at the start point or following a newline. The same bug applied to
4233 patterns with negative forward assertions or any backward assertions
4234 preceding ".*" at the start, unless the pattern required a fixed first
4235 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4236
4237 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4238 starting at the last subject character, bytes beyond the end of the subject
4239 string were read.
4240
4241 11. Renamed the variable "class" as "classbits" to make life easier for C++
4242 users. (Previously there was a macro definition, but it apparently wasn't
4243 enough.)
4244
4245 12. Added the new field "tables" to the extra data so that tables can be passed
4246 in at exec time, or the internal tables can be re-selected. This allows
4247 a compiled regex to be saved and re-used at a later time by a different
4248 program that might have everything at different addresses.
4249
4250 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4251 -R library as well as a -L library.
4252
4253 14. The debugging options of pcretest (-d on the command line or D on a
4254 pattern) showed incorrect output for anything following an extended class
4255 that contained multibyte characters and which was followed by a quantifier.
4256
4257 15. Added optional support for general category Unicode character properties
4258 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4259 support. It adds about 90K to the size of the library. The meanings of the
4260 inbuilt class escapes such as \d and \s have NOT been changed.
4261
4262 16. Updated pcredemo.c to include calls to free() to release the memory for the
4263 compiled pattern.
4264
4265 17. The generated file chartables.c was being created in the source directory
4266 instead of in the building directory. This caused the build to fail if the
4267 source directory was different from the building directory, and was
4268 read-only.
4269
4270 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4271 file. No doubt somebody will tell me if they don't make sense... Also added
4272 Dan Mooney's comments about building on OpenVMS.
4273
4274 19. Added support for partial matching via the PCRE_PARTIAL option for
4275 pcre_exec() and the \P data escape in pcretest.
4276
4277 20. Extended pcretest with 3 new pattern features:
4278
4279 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4280 write the compiled pattern to the file whose name is "rest-of-line".
4281 This is a straight binary dump of the data, with the saved pointer to
4282 the character tables forced to be NULL. The study data, if any, is
4283 written too. After writing, pcretest reads a new pattern.
4284
4285 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4286 compiled pattern from the given file. There must not be any
4287 occurrences of "<" in the file name (pretty unlikely); if there are,
4288 pcretest will instead treat the initial "<" as a pattern delimiter.
4289 After reading in the pattern, pcretest goes on to read data lines as
4290 usual.
4291
4292 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4293 and 16-bit fields in a compiled pattern, to simulate a pattern that
4294 was compiled on a host of opposite endianness.
4295
4296 21. The pcre-exec() function can now cope with patterns that were compiled on
4297 hosts of opposite endianness, with this restriction:
4298
4299 As for any compiled expression that is saved and used later, the tables
4300 pointer field cannot be preserved; the extra_data field in the arguments
4301 to pcre_exec() should be used to pass in a tables address if a value
4302 other than the default internal tables were used at compile time.
4303
4304 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4305 now diagnosed as an error. Previously, most of the time, a negative number
4306 would have been treated as zero, but if in addition "ovector" was passed as
4307 NULL, a crash could occur.
4308
4309 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4310 new versions from the libtool 1.5 distribution (the last one is a copy of
4311 a file called libtool.m4). This seems to have fixed the need to patch
4312 "configure" to support Darwin 1.3 (which I used to do). However, I still
4313 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4314 workstation).
4315
4316 24. Changed the PCRE licence to be the more standard "BSD" licence.
4317
4318
4319 Version 4.5 01-Dec-03
4320 ---------------------
4321
4322 1. There has been some re-arrangement of the code for the match() function so
4323 that it can be compiled in a version that does not call itself recursively.
4324 Instead, it keeps those local variables that need separate instances for
4325 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4326 needs to "recurse". Keeping track of where control must go is done by means
4327 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4328 hide most of the details from the main code, and operates only if
4329 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4330 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4331 operating.
4332
4333 To make it easier for callers to provide specially tailored get/free
4334 functions for this usage, two new functions, pcre_stack_malloc, and
4335 pcre_stack_free, are used. They are always called in strict stacking order,
4336 and the size of block requested is always the same.
4337
4338 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4339 PCRE has been compiled to use the stack or the heap for recursion. The
4340 -C option of pcretest uses this to show which version is compiled.
4341
4342 A new data escape \S, is added to pcretest; it causes the amounts of store
4343 obtained and freed by both kinds of malloc/free at match time to be added
4344 to the output.
4345
4346 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4347 what's available on my current Linux desktop machine.
4348
4349 3. When matching a UTF-8 string, the test for a valid string at the start has
4350 been extended. If start_offset is not zero, PCRE now checks that it points
4351 to a byte that is the start of a UTF-8 character. If not, it returns
4352 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4353 this is necessary because there may be backward assertions in the pattern.
4354 When matching the same subject several times, it may save resources to use
4355 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4356
4357 4. The code for checking the validity of UTF-8 strings has been tightened so
4358 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4359 containing "overlong sequences".
4360
4361 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4362 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4363 should have been "&", but it just so happened that all the cases this let
4364 through by mistake were picked up later in the function.
4365
4366 6. I had used a variable called "isblank" - this is a C99 function, causing
4367 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4368
4369 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4370 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4371 so the version is shown; (c) stop "make test" echoing "./RunTest".
4372
4373 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4374
4375 9. The prototype for memmove() for systems that don't have it was using
4376 size_t, but the inclusion of the header that defines size_t was later. I've
4377 moved the #includes for the C headers earlier to avoid this.
4378
4379 10. Added some adjustments to the code to make it easier to compiler on certain
4380 special systems:
4381
4382 (a) Some "const" qualifiers were missing.
4383 (b) Added the macro EXPORT before all exported functions; by default this
4384 is defined to be empty.
4385 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4386 that it reads its output file name as an argument instead of writing
4387 to the standard output and assuming this can be redirected.
4388
4389 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4390 class containing characters with values greater than 255, PCRE compilation
4391 went into a loop.
4392
4393 12. A recursive reference to a subpattern that was within another subpattern
4394 that had a minimum quantifier of zero caused PCRE to crash. For example,
4395 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4396 recursion. If the recursively-called subpattern itself had a zero repeat,
4397 that was OK.
4398
4399 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4400 buffer into which it was copied (for escape processing) was still set at
4401 1024, so long lines caused crashes.
4402
4403 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4404 "internal error: code overflow...". This applied to any character class
4405 that was followed by a possessive quantifier.
4406
4407 15. Modified the Makefile to add libpcre.la as a prerequisite for
4408 libpcreposix.la because I was told this is needed for a parallel build to
4409 work.
4410
4411 16. If a pattern that contained .* following optional items at the start was
4412 studied, the wrong optimizing data was generated, leading to matching
4413 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4414 matching string must start with a or b or c. The correct conclusion for
4415 this pattern is that a match can start with any character.
4416
4417
4418 Version 4.4 13-Aug-03
4419 ---------------------
4420
4421 1. In UTF-8 mode, a character class containing characters with values between
4422 127 and 255 was not handled correctly if the compiled pattern was studied.
4423 In fixing this, I have also improved the studying algorithm for such
4424 classes (slightly).
4425
4426 2. Three internal functions had redundant arguments passed to them. Removal
4427 might give a very teeny performance improvement.
4428
4429 3. Documentation bug: the value of the capture_top field in a callout is *one
4430 more than* the number of the hightest numbered captured substring.
4431
4432 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4433 in incorrectly linking with a previously installed version. They now link
4434 explicitly with libpcre.la.
4435
4436 5. configure.in no longer needs to recognize Cygwin specially.
4437
4438 6. A problem in pcre.in for Windows platforms is fixed.
4439
4440 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4441 pcretest, it used to include the size of the study block as part of its
4442 output. Unfortunately, the structure contains a field that has a different
4443 size on different hardware architectures. This meant that the tests that
4444 showed this size failed. As the block is currently always of a fixed size,
4445 this information isn't actually particularly useful in pcretest output, so
4446 I have just removed it.
4447
4448 8. Three pre-processor statements accidentally did not start in column 1.
4449 Sadly, there are *still* compilers around that complain, even though
4450 standard C has not required this for well over a decade. Sigh.
4451
4452 9. In pcretest, the code for checking callouts passed small integers in the
4453 callout_data field, which is a void * field. However, some picky compilers
4454 complained about the casts involved for this on 64-bit systems. Now
4455 pcretest passes the address of the small integer instead, which should get
4456 rid of the warnings.
4457
4458 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4459 both compile and run time, and gives an error if an invalid UTF-8 sequence
4460 is found. There is a option for disabling this check in cases where the
4461 string is known to be correct and/or the maximum performance is wanted.
4462
4463 11. In response to a bug report, I changed one line in Makefile.in from
4464
4465 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4466 to
4467 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4468
4469 to look similar to other lines, but I have no way of telling whether this
4470 is the right thing to do, as I do not use Windows. No doubt I'll get told
4471 if it's wrong...
4472
4473
4474 Version 4.3 21-May-03
4475 ---------------------
4476
4477 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4478 Makefile.
4479
4480 2. Some refactoring to improve the quality of the code:
4481
4482 (i) The utf8_table... variables are now declared "const".
4483
4484 (ii) The code for \cx, which used the "case flipping" table to upper case
4485 lower case letters, now just substracts 32. This is ASCII-specific,
4486 but the whole concept of \cx is ASCII-specific, so it seems
4487 reasonable.
4488
4489 (iii) PCRE was using its character types table to recognize decimal and
4490 hexadecimal digits in the pattern. This is silly, because it handles
4491 only 0-9, a-f, and A-F, but the character types table is locale-
4492 specific, which means strange things might happen. A private
4493 table is now used for this - though it costs 256 bytes, a table is
4494 much faster than multiple explicit tests. Of course, the standard
4495 character types table is still used for matching digits in subject
4496 strings against \d.
4497
4498 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4499 ending in _t are). So I've renamed it as ESC_tee.
4500
4501 3. The first argument for regexec() in the POSIX wrapper should have been
4502 defined as "const".
4503
4504 4. Changed pcretest to use malloc() for its buffers so that they can be
4505 Electric Fenced for debugging.
4506
4507 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4508 to read one or more bytes before the start of the subject string. Often this
4509 had no effect on PCRE's behaviour, but in some circumstances it could
4510 provoke a segmentation fault.
4511
4512 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4513 to try to read one or more bytes before the start of the subject string.
4514
4515 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4516 UTF-8 support could misbehave in various ways if the subject string
4517 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4518 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4519 back over UTF-8 characters.)
4520
4521
4522 Version 4.2 14-Apr-03
4523 ---------------------
4524
4525 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4526
4527 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4528 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4529 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4530 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4531 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4532 and BUILD_EXEEXT
4533 Note: automatic setting of the BUILD variables is not yet working
4534 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4535 compile-time but not at link-time
4536 [LINK]: use for linking executables only
4537 make different versions for Windows and non-Windows
4538 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4539 libraries
4540 [LINK_FOR_BUILD]: new variable
4541 [OBJEXT]: use throughout
4542 [EXEEXT]: use throughout
4543 <winshared>: new target
4544 <wininstall>: new target
4545 <dftables.o>: use native compiler
4546 <dftables>: use native linker
4547 <install>: handle Windows platform correctly
4548 <clean>: ditto
4549 <check>: ditto
4550 copy DLL to top builddir before testing
4551
4552 As part of these changes, -no-undefined was removed again. This was reported
4553 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4554 in any case.
4555
4556 3. Some tidies to get rid of compiler warnings:
4557
4558 . In the match_data structure, match_limit was an unsigned long int, whereas
4559 match_call_count was an int. I've made them both unsigned long ints.
4560
4561 . In pcretest the fact that a const uschar * doesn't automatically cast to
4562 a void * provoked a warning.
4563
4564 . Turning on some more compiler warnings threw up some "shadow" variables
4565 and a few more missing casts.
4566
4567 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4568 option, a class that contained a single character with a value between 128
4569 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4570
4571 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4572 option, a class that contained several characters, but with at least one
4573 whose value was between 128 and 255 caused PCRE to crash.
4574
4575
4576 Version 4.1 12-Mar-03
4577 ---------------------
4578
4579 1. Compiling with gcc -pedantic found a couple of places where casts were
4580 needed, and a string in dftables.c that was longer than standard compilers are
4581 required to support.
4582
4583 2. Compiling with Sun's compiler found a few more places where the code could
4584 be tidied up in order to avoid warnings.
4585
4586 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4587 first of these names is deprecated in the latest Autoconf in favour of the name
4588 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4589 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4590 analogy I have changed it to CFLAGS_FOR_BUILD.
4591
4592 4. Added -no-undefined to the linking command in the Makefile, because this is
4593 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4594 linking step for the pcreposix library.
4595
4596 5. PCRE was failing to diagnose the case of two named groups with the same
4597 name.
4598
4599 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4600 literal character that is needed in the subject for a match, and scans along to
4601 ensure that it is present before embarking on the full matching process. This
4602 saves time in cases of nested unlimited repeats that are never going to match.
4603 Problem: the scan can take a lot of time if the subject is very long (e.g.
4604 megabytes), thus penalizing straightforward matches. It is now done only if the
4605 amount of subject to be scanned is less than 1000 bytes.
4606
4607 7. A lesser problem with the same optimization is that it was recording the
4608 first character of an anchored pattern as "needed", thus provoking a search
4609 right along the subject, even when the first match of the pattern was going to
4610 fail. The "needed" character is now not set for anchored patterns, unless it
4611 follows something in the pattern that is of non-fixed length. Thus, it still
4612 fulfils its original purpose of finding quick non-matches in cases of nested
4613 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4614
4615
4616 Version 4.0 17-Feb-03
4617 ---------------------
4618
4619 1. If a comment in an extended regex that started immediately after a meta-item
4620 extended to the end of string, PCRE compiled incorrect data. This could lead to
4621 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4622
4623 2. Moved to autoconf 2.53 and libtool 1.4.2.
4624
4625 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4626 the special perltest8 script is no longer needed - all the tests can be run
4627 from a single perltest script.
4628
4629 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4630 by \s. It has now been removed in PCRE. This means it isn't recognized as
4631 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4632 class [:space:] *does* include VT, thereby creating a mess.
4633
4634 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4635 space and tab.
4636
4637 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4638 its new features into the main test script, reducing the number of scripts.
4639
4640 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4641 were backward compatible, and made the (?i) apply to the whole pattern, as if
4642 /i were given. Now it behaves more logically, and applies the option setting
4643 only to what follows. PCRE has been changed to follow suit. However, if it
4644 finds options settings right at the start of the pattern, it extracts them into
4645 the global options, as before. Thus, they show up in the info data.
4646
4647 8. Added support for the \Q...\E escape sequence. Characters in between are
4648 treated as literals. This is slightly different from Perl in that $ and @ are
4649 also handled as literals inside the quotes. In Perl, they will cause variable
4650 interpolation. Note the following examples:
4651
4652 Pattern PCRE matches Perl matches
4653
4654 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4655 \Qabc\$xyz\E abc\$xyz abc\$xyz
4656 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4657
4658 For compatibility with Perl, \Q...\E sequences are recognized inside character
4659 classes as well as outside them.
4660
4661 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4662 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4663 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4664 signed/unsigned warnings.
4665
4666 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4667 option for pcretest, so I've replaced it by a simple function that does just
4668 that job.
4669
4670 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4671 "pcregrep -".
4672
4673 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4674 Java package. This provides some syntactic sugar for simple cases of what my
4675 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4676 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4677 item, you can use this simplified notation. Note that only makes sense with
4678 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4679 greediness, whatever the setting of the PCRE_UNGREEDY option.
4680
4681 13. A change of greediness default within a pattern was not taking effect at
4682 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4683 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4684 was abstracted outside.
4685
4686 14. PCRE now supports the \G assertion. It is true when the current matching
4687 position is at the start point of the match. This differs from \A when the
4688 starting offset is non-zero. Used with the /g option of pcretest (or similar
4689 code), it works in the same way as it does for Perl's /g option. If all
4690 alternatives of a regex begin with \G, the expression is anchored to the start
4691 match position, and the "anchored" flag is set in the compiled expression.
4692
4693 15. Some bugs concerning the handling of certain option changes within patterns
4694 have been fixed. These applied to options other than (?ims). For example,
4695 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4696 been the other way round. Some of this was related to change 7 above.
4697
4698 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4699 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4700 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4701 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4702
4703 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4704 mode. Unlike ".", it always matches newline, whatever the setting of
4705 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4706 assertions. Perl allows it, but it doesn't (in general) work because it can't
4707 calculate the length of the lookbehind. At least, that's the case for Perl
4708 5.8.0 - I've been told they are going to document that it doesn't work in
4709 future.
4710
4711 18. Added an error diagnosis for escapes that PCRE does not support: these are
4712 \L, \l, \N, \P, \p, \U, \u, and \X.
4713
4714 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4715 reading past the end of the pattern in cases such as /[abcd/.
4716
4717 20. PCRE was getting more memory than necessary for patterns with classes that
4718 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4719
4720 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4721 compiling PCRE for use with Virtual Pascal.
4722
4723 22. Small fix to the Makefile to make it work properly if the build is done
4724 outside the source tree.
4725
4726 23. Added a new extension: a condition to go with recursion. If a conditional
4727 subpattern starts with (?(R) the "true" branch is used if recursion has
4728 happened, whereas the "false" branch is used only at the top level.
4729
4730 24. When there was a very long string of literal characters (over 255 bytes
4731 without UTF support, over 250 bytes with UTF support), the computation of how
4732 much memory was required could be incorrect, leading to segfaults or other
4733 strange effects.
4734
4735 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4736 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4737 there was a subsequent back reference to those brackets. This meant that, for
4738 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4739 possible to check for precisely this case. All we can do is abandon the
4740 optimization if .* occurs inside capturing brackets when there are any back
4741 references whatsoever. (See below for a better fix that came later.)
4742
4743 26. The handling of the optimization for finding the first character of a
4744 non-anchored pattern, and for finding a character that is required later in the
4745 match were failing in some cases. This didn't break the matching; it just
4746 failed to optimize when it could. The way this is done has been re-implemented.
4747
4748 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4749
4750 28. Added a new feature that provides some of the functionality that Perl
4751 provides with (?{...}). The facility is termed a "callout". The way it is done
4752 in PCRE is for the caller to provide an optional function, by setting
4753 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4754 global variable. By default it is unset, which disables all calling out. To get
4755 the function called, the regex must include (?C) at appropriate points. This
4756 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4757 This provides a means of identifying different callout points. When PCRE
4758 reaches such a point in the regex, if pcre_callout has been set, the external
4759 function is called. It is provided with data in a structure called
4760 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4761 matching continues; if it returns a non-zero value, the match at the current
4762 point fails. However, backtracking will occur if possible. [This was changed
4763 later and other features added - see item 49 below.]
4764
4765 29. pcretest is upgraded to test the callout functionality. It provides a
4766 callout function that displays information. By default, it shows the start of
4767 the match and the current position in the text. There are some new data escapes
4768 to vary what happens:
4769
4770 \C+ in addition, show current contents of captured substrings
4771 \C- do not supply a callout function
4772 \C!n return 1 when callout number n is reached
4773 \C!n!m return 1 when callout number n is reached for the mth time
4774
4775 30. If pcregrep was called with the -l option and just a single file name, it
4776 output "<stdin>" if a match was found, instead of the file name.
4777
4778 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4779 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4780 pcre_exec(). This saves a malloc/free per call. The default value of
4781 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4782 when configuring.
4783
4784 32. The default maximum size of a compiled pattern is 64K. There have been a
4785 few cases of people hitting this limit. The code now uses macros to handle the
4786 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4787 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4788 configuring. Tests 2 and 5 work only with 2-byte links because they output
4789 debugging information about compiled patterns.
4790
4791 33. Internal code re-arrangements:
4792
4793 (a) Moved the debugging function for printing out a compiled regex into
4794 its own source file (printint.c) and used #include to pull it into
4795 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4796 separate copies.
4797
4798 (b) Defined the list of op-code names for debugging as a macro in
4799 internal.h so that it is next to the definition of the opcodes.
4800
4801 (c) Defined a table of op-code lengths for simpler skipping along compiled
4802 code. This is again a macro in internal.h so that it is next to the
4803 definition of the opcodes.
4804
4805 34. Added support for recursive calls to individual subpatterns, along the
4806 lines of Robin Houston's patch (but implemented somewhat differently).
4807
4808 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4809 allow it to read and process whole directories in Win32. This code was
4810 contributed by Lionel Fourquaux; it has not been tested by me.
4811
4812 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4813 used to name a group. Names consist of alphanumerics and underscores, and must
4814 be unique. Back references use the syntax (?P=name) and recursive calls use
4815 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4816 numbers. The function pcre_fullinfo() can be used after compilation to extract
4817 a name/number map. There are three relevant calls:
4818
4819 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4820 PCRE_INFO_NAMECOUNT yields the number of entries
4821 PCRE_INFO_NAMETABLE yields a pointer to the map.
4822
4823 The map is a vector of fixed-size entries. The size of each entry depends on
4824 the length of the longest name used. The first two bytes of each entry are the
4825 group number, most significant byte first. There follows the corresponding
4826 name, zero terminated. The names are in alphabetical order.
4827
4828 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4829 case instead of 255. Making it the same both with and without UTF-8 support
4830 means that the same test output works with both.
4831
4832 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4833 calling malloc() with a zero argument.
4834
4835 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4836 optimization. I've improved things by keeping a bitmap of backreferences with
4837 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4838 fact referenced, the optimization can be applied. It is unlikely that a
4839 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4840 the match to follow \n) will appear inside brackets with a number greater than
4841 31, but if it does, any back reference > 31 suppresses the optimization.
4842
4843 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4844 of disabling numbered capturing parentheses. Any opening parenthesis that is
4845 not followed by ? behaves as if it were followed by ?: but named parentheses
4846 can still be used for capturing (and they will acquire numbers in the usual
4847 way).
4848
4849 41. Redesigned the return codes from the match() function into yes/no/error so
4850 that errors can be passed back from deep inside the nested calls. A malloc
4851 failure while inside a recursive subpattern call now causes the
4852 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4853
4854 42. It is now possible to set a limit on the number of times the match()
4855 function is called in a call to pcre_exec(). This facility makes it possible to
4856 limit the amount of recursion and backtracking, though not in a directly
4857 obvious way, because the match() function is used in a number of different
4858 circumstances. The count starts from zero for each position in the subject
4859 string (for non-anchored patterns). The default limit is, for compatibility, a
4860 large number, namely 10 000 000. You can change this in two ways:
4861
4862 (a) When configuring PCRE before making, you can use --with-match-limit=n
4863 to set a default value for the compiled library.
4864
4865 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4866 a different value is set. See 45 below.
4867
4868 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4869
4870 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4871 of things that can be changed at compile time. The first argument specifies
4872 what is wanted and the second points to where the information is to be placed.
4873 The current list of available information is:
4874
4875 PCRE_CONFIG_UTF8
4876
4877 The output is an integer that is set to one if UTF-8 support is available;
4878 otherwise it is set to zero.
4879
4880 PCRE_CONFIG_NEWLINE
4881
4882 The output is an integer that it set to the value of the code that is used for
4883 newline. It is either LF (10) or CR (13).
4884
4885 PCRE_CONFIG_LINK_SIZE
4886
4887 The output is an integer that contains the number of bytes used for internal
4888 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4889
4890 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4891
4892 The output is an integer that contains the threshold above which the POSIX
4893 interface uses malloc() for output vectors. See item 31 above.
4894
4895 PCRE_CONFIG_MATCH_LIMIT
4896
4897 The output is an unsigned integer that contains the default limit of the number
4898 of match() calls in a pcre_exec() execution. See 42 above.
4899
4900 44. pcretest has been upgraded by the addition of the -C option. This causes it
4901 to extract all the available output from the new pcre_config() function, and to
4902 output it. The program then exits immediately.
4903
4904 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4905 order to support additional features. One way would have been to define
4906 pcre_exec2() (for example) with extra arguments, but this would not have been
4907 extensible, and would also have required all calls to the original function to
4908 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4909 is used for passing in "extra" data from pcre_study().
4910
4911 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4912 contains the following fields:
4913
4914 flags a bitmap indicating which of the following fields are set
4915 study_data opaque data from pcre_study()
4916 match_limit a way of specifying a limit on match() calls for a specific
4917 call to pcre_exec()
4918 callout_data data for callouts (see 49 below)
4919
4920 The flag bits are also defined in pcre.h, and are
4921
4922 PCRE_EXTRA_STUDY_DATA
4923 PCRE_EXTRA_MATCH_LIMIT
4924 PCRE_EXTRA_CALLOUT_DATA
4925
4926 The pcre_study() function now returns one of these new pcre_extra blocks, with
4927 the actual study data pointed to by the study_data field, and the
4928 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4929 before. That is, this change is entirely upwards-compatible and requires no
4930 change to existing code.
4931
4932 If you want to pass in additional data to pcre_exec(), you can either place it
4933 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4934 block.
4935
4936 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4937 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4938 times with different match limits, until it finds the minimum value needed for
4939 pcre_exec() to complete. The value is then output. This can be instructive; for
4940 most simple matches the number is quite small, but for pathological cases it
4941 gets very large very quickly.
4942
4943 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4944 returns the size of the data block pointed to by the study_data field in a
4945 pcre_extra block, that is, the value that was passed as the argument to
4946 pcre_malloc() when PCRE was getting memory in which to place the information
4947 created by pcre_study(). The fourth argument should point to a size_t variable.
4948 pcretest has been extended so that this information is shown after a successful
4949 pcre_study() call when information about the compiled regex is being displayed.
4950
4951 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4952 because what follows is always an absolute path. (Later: it turns out that this
4953 is more than cosmetic for MinGW, because it doesn't like empty path
4954 components.)
4955
4956 49. Some changes have been made to the callout feature (see 28 above):
4957
4958 (i) A callout function now has three choices for what it returns:
4959
4960 0 => success, carry on matching
4961 > 0 => failure at this point, but backtrack if possible
4962 < 0 => serious error, return this value from pcre_exec()
4963
4964 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4965 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4966 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4967 use by callout functions. It will never be used by PCRE itself.
4968
4969 (ii) The pcre_extra structure (see 45 above) has a void * field called
4970 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4971 pcre_callout_block structure has a field of the same name. The contents of
4972 the field passed in the pcre_extra structure are passed to the callout
4973 function in the corresponding field in the callout block. This makes it
4974 easier to use the same callout-containing regex from multiple threads. For
4975 testing, the pcretest program has a new data escape
4976
4977 \C*n pass the number n (may be negative) as callout_data
4978
4979 If the callout function in pcretest receives a non-zero value as
4980 callout_data, it returns that value.
4981
4982 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4983 there were some redundant $(CFLAGS) in commands that are now specified as
4984 $(LINK), which already includes $(CFLAGS).
4985
4986 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4987 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4988 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4989 one-byte characters throughout. Note that case-insensitive matching applies
4990 only to characters whose values are less than 256. PCRE doesn't support the
4991 notion of cases for higher-valued characters.
4992
4993 (i) A character class whose characters are all within 0-255 is handled as
4994 a bit map, and the map is inverted for negative classes. Previously, a
4995 character > 255 always failed to match such a class; however it should
4996 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4997
4998 (ii) A negated character class with a single character < 255 is coded as
4999 "not this character" (OP_NOT). This wasn't working properly when the test
5000 character was multibyte, either singly or repeated.
5001
5002 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5003 mode, for example: \x{100}{2,3}.
5004
5005 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5006 singly or repeated) now correctly test multibyte characters. However,
5007 PCRE doesn't recognize any characters with values greater than 255 as
5008 digits, spaces, or word characters. Such characters always match \D, \S,
5009 and \W, and never match \d, \s, or \w.
5010
5011 (v) Classes may now contain characters and character ranges with values
5012 greater than 255. For example: [ab\x{100}-\x{400}].
5013
5014 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5015 PCRE in UTF-8 mode.
5016
5017 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5018 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5019 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5020 value.)
5021
5022 53. The single man page has become too large. I have therefore split it up into
5023 a number of separate man pages. These also give rise to individual HTML pages;
5024 these are now put in a separate directory, and there is an index.html page that
5025 lists them all. Some hyperlinking between the pages has been installed.
5026
5027 54. Added convenience functions for handling named capturing parentheses.
5028
5029 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5030 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5031 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5032 are faulted.
5033
5034 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5035 calling configure. These values are used when compiling the dftables.c program
5036 which is run to generate the source of the default character tables. They
5037 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5038 you will need to set these values.
5039
5040 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5041
5042
5043 Version 3.9 02-Jan-02
5044 ---------------------
5045
5046 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5047
5048 2. If --disable-static was given, the building process failed when trying to
5049 build pcretest and pcregrep. (For some reason it was using libtool to compile
5050 them, which is not right, as they aren't part of the library.)
5051
5052
5053 Version 3.8 18-Dec-01
5054 ---------------------
5055
5056 1. The experimental UTF-8 code was completely screwed up. It was packing the
5057 bytes in the wrong order. How dumb can you get?
5058
5059
5060 Version 3.7 29-Oct-01
5061 ---------------------
5062
5063 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5064 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5065 this didn't happen under Solaris 8, where I normally test things.
5066
5067 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5068 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5069 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5070 this fix an hour or so after the initial 3.7 release.)
5071
5072
5073 Version 3.6 23-Oct-01
5074 ---------------------
5075
5076 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5077 offsets passed as NULL with zero offset count.
5078
5079 2. The config.guess and config.sub files had not been updated when I moved to
5080 the latest autoconf.
5081
5082
5083 Version 3.5 15-Aug-01
5084 ---------------------
5085
5086 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5087 had been forgotten.
5088
5089 2. By using declared but undefined structures, we can avoid using "void"
5090 definitions in pcre.h while keeping the internal definitions of the structures
5091 private.
5092
5093 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5094 user point of view, this means that both static and shared libraries are built
5095 by default, but this can be individually controlled. More of the work of
5096 handling this static/shared cases is now inside libtool instead of PCRE's make
5097 file.
5098
5099 4. The pcretest utility is now installed along with pcregrep because it is
5100 useful for users (to test regexs) and by doing this, it automatically gets
5101 relinked by libtool. The documentation has been turned into a man page, so
5102 there are now .1, .txt, and .html versions in /doc.
5103
5104 5. Upgrades to pcregrep:
5105 (i) Added long-form option names like gnu grep.
5106 (ii) Added --help to list all options with an explanatory phrase.
5107 (iii) Added -r, --recursive to recurse into sub-directories.
5108 (iv) Added -f, --file to read patterns from a file.
5109
5110 6. pcre_exec() was referring to its "code" argument before testing that
5111 argument for NULL (and giving an error if it was NULL).
5112
5113 7. Upgraded Makefile.in to allow for compiling in a different directory from
5114 the source directory.
5115
5116 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5117 options bits, the pointer it was passed was to an int instead of to an unsigned
5118 long int. This mattered only on 64-bit systems.
5119
5120 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5121 generated) instead of pcre.in, which it its source. Also made the same change
5122 in several of the .c files.
5123
5124 10. A new release of gcc defines printf() as a macro, which broke pcretest
5125 because it had an ifdef in the middle of a string argument for printf(). Fixed
5126 by using separate calls to printf().
5127
5128 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5129 script, to force use of CR or LF instead of \n in the source. On non-Unix
5130 systems, the value can be set in config.h.
5131
5132 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5133 absolute limit. Changed the text of the error message to make this clear, and
5134 likewise updated the man page.
5135
5136 13. The limit of 99 on the number of capturing subpatterns has been removed.
5137 The new limit is 65535, which I hope will not be a "real" limit.
5138
5139
5140 Version 3.4 22-Aug-00
5141 ---------------------
5142
5143 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5144
5145 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5146
5147
5148 Version 3.3 01-Aug-00
5149 ---------------------
5150
5151 1. If an octal character was given, but the value was greater than \377, it
5152 was not getting masked to the least significant bits, as documented. This could
5153 lead to crashes in some systems.
5154
5155 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5156 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5157
5158 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5159 These just pass their arguments on to (pcre_free)(), but they are provided
5160 because some uses of PCRE bind it to non-C systems that can call its functions,
5161 but cannot call free() or pcre_free() directly.
5162
5163 4. Add "make test" as a synonym for "make check". Corrected some comments in
5164 the Makefile.
5165
5166 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5167 Makefile.
5168
5169 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5170 command called pgrep for grepping around the active processes.
5171
5172 7. Added the beginnings of support for UTF-8 character strings.
5173
5174 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5175 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5176 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5177 out for the ar command.)
5178
5179
5180 Version 3.2 12-May-00
5181 ---------------------
5182
5183 This is purely a bug fixing release.
5184
5185 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5186 of ZA. This was just one example of several cases that could provoke this bug,
5187 which was introduced by change 9 of version 2.00. The code for breaking
5188 infinite loops after an iteration that matches an empty string was't working
5189 correctly.
5190
5191 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5192 when matched against abbab (for example). After matching an empty string, it
5193 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5194 caused it to match further down the string than it should.
5195
5196 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5197 was there because it doesn't seem to be needed, and it causes trouble on some
5198 systems, as it is not a Standard C header. It has been removed.
5199
5200 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5201 were reported on the Macintosh. The changes were from
5202
5203 while ((c = *(++ptr)) != 0 && c != '\n');
5204 to
5205 while ((c = *(++ptr)) != 0 && c != '\n') ;
5206
5207 Totally extraordinary, but if that's what it takes...
5208
5209 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5210 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5211 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5212 assumes the way PCRE uses memmove() (always moving upwards).
5213
5214 6. PCRE is being used in one environment where strchr() is not available. There
5215 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5216 faster code anyway.
5217
5218
5219 Version 3.1 09-Feb-00
5220 ---------------------
5221
5222 The only change in this release is the fixing of some bugs in Makefile.in for
5223 the "install" target:
5224
5225 (1) It was failing to install pcreposix.h.
5226
5227 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5228
5229
5230 Version 3.0 01-Feb-00
5231 ---------------------
5232
5233 1. Add support for the /+ modifier to perltest (to output $` like it does in
5234 pcretest).
5235
5236 2. Add support for the /g modifier to perltest.
5237
5238 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5239 matches null strings.
5240
5241 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5242 pattern. Perl treats empty patterns specially - it reuses the most recent
5243 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5244 effect.
5245
5246 5. The POSIX interface was broken in that it was just handing over the POSIX
5247 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5248 required a bigger vector, with some working space on the end. This means that
5249 the POSIX wrapper now has to get and free some memory, and copy the results.
5250
5251 6. Added some simple autoconf support, placing the test data and the
5252 documentation in separate directories, re-organizing some of the
5253 information files, and making it build pcre-config (a GNU standard). Also added
5254 libtool support for building PCRE as a shared library, which is now the
5255 default.
5256
5257 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5258 09 are not valid octal constants. Single digits will be used for minor values
5259 less than 10.
5260
5261 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5262 existing programs that set these in the POSIX interface can use PCRE without
5263 modification.
5264
5265 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5266 return all that pcre_info() returns, plus additional data. The pcre_info()
5267 function is retained for compatibility, but is considered to be obsolete.
5268
5269 10. Added experimental recursion feature (?R) to handle one common case that
5270 Perl 5.6 will be able to do with (?p{...}).
5271
5272 11. Added support for POSIX character classes like [:alpha:], which Perl is
5273 adopting.
5274
5275
5276 Version 2.08 31-Aug-99
5277 ----------------------
5278
5279 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5280 trying to match at the startoffset position, but instead was moving forward to
5281 the next newline as if a previous match had failed.
5282
5283 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5284 and could get into a loop if a null string was matched other than at the start
5285 of the subject.
5286
5287 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5288 be distinguished at compile time, and for completeness also added PCRE_DATE.
5289
5290 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5291 in GnuWin32 environments.
5292
5293
5294 Version 2.07 29-Jul-99
5295 ----------------------
5296
5297 1. The documentation is now supplied in plain text form and HTML as well as in
5298 the form of man page sources.
5299
5300 2. C++ compilers don't like assigning (void *) values to other pointer types.
5301 In particular this affects malloc(). Although there is no problem in Standard
5302 C, I've put in casts to keep C++ compilers happy.
5303
5304 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5305 should be (const char *).
5306
5307 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5308 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5309 However, I haven't made this a standard facility. The documentation doesn't
5310 mention it, and the Makefile doesn't support it.
5311
5312 5. The Makefile now contains an "install" target, with editable destinations at
5313 the top of the file. The pcretest program is not installed.
5314
5315 6. pgrep -V now gives the PCRE version number and date.
5316
5317 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5318 causing the entire string to be ignored, instead of just the last character.
5319
5320 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5321 non-matching string, it can take a very, very long time, even for strings of
5322 quite modest length, because of the nested recursion. PCRE now does better in
5323 some of these cases. It does this by remembering the last required literal
5324 character in the pattern, and pre-searching the subject to ensure it is present
5325 before running the real match. In other words, it applies a heuristic to detect
5326 some types of certain failure quickly, and in the above example, if presented
5327 with a string that has no trailing " it gives "no match" very quickly.
5328
5329 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5330 other alternatives are tried instead.
5331
5332
5333 Version 2.06 09-Jun-99
5334 ----------------------
5335
5336 1. Change pcretest's output for amount of store used to show just the code
5337 space, because the remainder (the data block) varies in size between 32-bit and
5338 64-bit systems.
5339
5340 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5341 start matching at. This allows lookbehinds to work when searching for multiple
5342 occurrences in a string.
5343
5344 3. Added additional options to pcretest for testing multiple occurrences:
5345
5346 /+ outputs the rest of the string that follows a match
5347 /g loops for multiple occurrences, using the new startoffset argument
5348 /G loops for multiple occurrences by passing an incremented pointer
5349
5350 4. PCRE wasn't doing the "first character" optimization for patterns starting
5351 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5352 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5353 the letter 'x'. On long subject strings, this gives a significant speed-up.
5354
5355
5356 Version 2.05 21-Apr-99
5357 ----------------------
5358
5359 1. Changed the type of magic_number from int to long int so that it works
5360 properly on 16-bit systems.
5361
5362 2. Fixed a bug which caused patterns starting with .* not to work correctly
5363 when the subject string contained newline characters. PCRE was assuming
5364 anchoring for such patterns in all cases, which is not correct because .* will
5365 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5366 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5367 must be retried after every newline in the subject.
5368
5369
5370 Version 2.04 18-Feb-99
5371 ----------------------
5372
5373 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5374 computation of the store needed to hold the pattern was incorrect (too large).
5375 If such patterns were nested a few deep, this could multiply and become a real
5376 problem.
5377
5378 2. Added /M option to pcretest to show the memory requirement of a specific
5379 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5380
5381 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5382 compiled in such a way that the backtracking after subsequent failure was
5383 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5384 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5385
5386
5387 Version 2.03 02-Feb-99
5388 ----------------------
5389
5390 1. Fixed typo and small mistake in man page.
5391
5392 2. Added 4th condition (GPL supersedes if conflict) and created separate
5393 LICENCE file containing the conditions.
5394
5395 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5396 Perl, that is the internal \ allows the delimiter to be included in the
5397 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5398 the final delimiter, add \ to the end of the pattern (to test the error).
5399
5400 4. Added the convenience functions for extracting substrings after a successful
5401 match. Updated pcretest to make it able to test these functions.
5402
5403
5404 Version 2.02 14-Jan-99
5405 ----------------------
5406
5407 1. Initialized the working variables associated with each extraction so that
5408 their saving and restoring doesn't refer to uninitialized store.
5409
5410 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5411 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5412 fix the problem.
5413
5414 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5415 calls, and wasn't printing the correct value for compiling calls. Increased the
5416 default value of LOOPREPEAT, and the number of significant figures in the
5417 times.
5418
5419 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5420
5421 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5422 a building problem on Windows NT with a FAT file system.
5423
5424
5425 Version 2.01 21-Oct-98
5426 ----------------------
5427
5428 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5429 to character tables built by pcre_maketables() in the current locale. If NULL
5430 is passed, the default tables are used.
5431
5432
5433 Version 2.00 24-Sep-98
5434 ----------------------
5435
5436 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5437 it any more.
5438
5439 2. Allow quantification of (?>) groups, and make it work correctly.
5440
5441 3. The first character computation wasn't working for (?>) groups.
5442
5443 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5444 end of the subject) and add 5.005's \z, which really does match only at the
5445 very end of the subject.
5446
5447 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5448
5449 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5450 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5451 localized options. All options to pcre_study() were also removed.
5452
5453 7. Add other new features from 5.005:
5454
5455 $(?<= positive lookbehind
5456 $(?<! negative lookbehind
5457 (?imsx-imsx) added the unsetting capability
5458 such a setting is global if at outer level; local otherwise
5459 (?imsx-imsx:) non-capturing groups with option setting
5460 (?(cond)re|re) conditional pattern matching
5461
5462 A backreference to itself in a repeated group matches the previous
5463 captured string.
5464
5465 8. General tidying up of studying (both automatic and via "study")
5466 consequential on the addition of new assertions.
5467
5468 9. As in 5.005, unlimited repeated groups that could match an empty substring
5469 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5470 runtime if any iteration does actually match an empty substring.
5471
5472 10. Include the RunTest script in the distribution.
5473
5474 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5475 discrepancies, some of which were old and were also with respect to 5.004. They
5476 have now been fixed.
5477
5478
5479 Version 1.09 28-Apr-98
5480 ----------------------
5481
5482 1. A negated single character class followed by a quantifier with a minimum
5483 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5484 program crashes, or just wrong answers. This did not apply to negated classes
5485 containing more than one character, or to minima other than one.
5486
5487
5488 Version 1.08 27-Mar-98
5489 ----------------------
5490
5491 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5492
5493 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5494 latter must appear before anything that relies on it in the pattern.
5495
5496
5497 Version 1.07 16-Feb-98
5498 ----------------------
5499
5500 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5501 repeat of a potentially empty string).
5502
5503
5504 Version 1.06 23-Jan-98
5505 ----------------------
5506
5507 1. Added Markus Oberhumer's little patches for C++.
5508
5509 2. Literal strings longer than 255 characters were broken.
5510
5511
5512 Version 1.05 23-Dec-97
5513 ----------------------
5514
5515 1. Negated character classes containing more than one character were failing if
5516 PCRE_CASELESS was set at run time.
5517
5518
5519 Version 1.04 19-Dec-97
5520 ----------------------
5521
5522 1. Corrected the man page, where some "const" qualifiers had been omitted.
5523
5524 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5525 input syntax.
5526
5527 3. Fixed memory leak which occurred when a regex with back references was
5528 matched with an offsets vector that wasn't big enough. The temporary memory
5529 that is used in this case wasn't being freed if the match failed.
5530
5531 4. Tidied pcretest to ensure it frees memory that it gets.
5532
5533 5. Temporary memory was being obtained in the case where the passed offsets
5534 vector was exactly big enough.
5535
5536 6. Corrected definition of offsetof() from change 5 below.
5537
5538 7. I had screwed up change 6 below and broken the rules for the use of
5539 setjmp(). Now fixed.
5540
5541
5542 Version 1.03 18-Dec-97
5543 ----------------------
5544
5545 1. A erroneous regex with a missing opening parenthesis was correctly
5546 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5547 on some systems.
5548
5549 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5550 it was reported that one broken compiler failed on the former because "code" is
5551 also an independent variable.
5552
5553 3. The erroneous regex a[]b caused an array overrun reference.
5554
5555 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5556 fail on data ending with that character. (It was going on too far, and checking
5557 the next character, typically a binary zero.) This was specific to the
5558 optimized code for single-character negative classes.
5559
5560 5. Added a contributed patch from the TIN world which does the following:
5561
5562 + Add an undef for memmove, in case the the system defines a macro for it.
5563
5564 + Add a definition of offsetof(), in case there isn't one. (I don't know
5565 the reason behind this - offsetof() is part of the ANSI standard - but
5566 it does no harm).
5567
5568 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5569 most of the places where whitespace preceded '#'. I have given up and
5570 allowed the remaining 2 cases to be at the margin.
5571
5572 + Rename some variables in pcre to eliminate shadowing. This seems very
5573 pedantic, but does no harm, of course.
5574
5575 6. Moved the call to setjmp() into its own function, to get rid of warnings
5576 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5577
5578 7. Constructs such as \d{8,} were compiling into the equivalent of
5579 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5580 outcome, but in this particular case used more store than had been allocated,
5581 which caused the bug to be discovered because it threw up an internal error.
5582
5583 8. The debugging code in both pcre and pcretest for outputting the compiled
5584 form of a regex was going wrong in the case of back references followed by
5585 curly-bracketed repeats.
5586
5587
5588 Version 1.02 12-Dec-97
5589 ----------------------
5590
5591 1. Typos in pcre.3 and comments in the source fixed.
5592
5593 2. Applied a contributed patch to get rid of places where it used to remove
5594 'const' from variables, and fixed some signed/unsigned and uninitialized
5595 variable warnings.
5596
5597 3. Added the "runtest" target to Makefile.
5598
5599 4. Set default compiler flag to -O2 rather than just -O.
5600
5601
5602 Version 1.01 19-Nov-97
5603 ----------------------
5604
5605 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5606 like /([ab]*)*/, that is, for classes with more than one character in them.
5607
5608 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5609 as /((?>a*))*/ (a PCRE_EXTRA facility).
5610
5611
5612 Version 1.00 18-Nov-97
5613 ----------------------
5614
5615 1. Added compile-time macros to support systems such as SunOS4 which don't have
5616 memmove() or strerror() but have other things that can be used instead.
5617
5618 2. Arranged that "make clean" removes the executables.
5619
5620
5621 Version 0.99 27-Oct-97
5622 ----------------------
5623
5624 1. Fixed bug in code for optimizing classes with only one character. It was
5625 initializing a 32-byte map regardless, which could cause it to run off the end
5626 of the memory it had got.
5627
5628 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5629
5630
5631 Version 0.98 22-Oct-97
5632 ----------------------
5633
5634 1. Fixed bug in code for handling temporary memory usage when there are more
5635 back references than supplied space in the ovector. This could cause segfaults.
5636
5637
5638 Version 0.97 21-Oct-97
5639 ----------------------
5640
5641 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5642
5643 2. Optimized negated single characters not to use a bit map.
5644
5645 3. Brought error texts together as macro definitions; clarified some of them;
5646 fixed one that was wrong - it said "range out of order" when it meant "invalid
5647 escape sequence".
5648
5649 4. Changed some char * arguments to const char *.
5650
5651 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5652
5653 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5654 pcretest.
5655
5656
5657 Version 0.96 16-Oct-97
5658 ----------------------
5659
5660 1. Added a simple "pgrep" utility to the distribution.
5661
5662 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5663 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5664 where "ddd" means "one or more decimal digits".
5665
5666 3. Fixed serious bug. If a pattern had a back reference, but the call to
5667 pcre_exec() didn't supply a large enough ovector to record the related
5668 identifying subpattern, the match always failed. PCRE now remembers the number
5669 of the largest back reference, and gets some temporary memory in which to save
5670 the offsets during matching if necessary, in order to ensure that
5671 backreferences always work.
5672
5673 4. Increased the compatibility with Perl in a number of ways:
5674
5675 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5676 to request this handling. The option can be set at compile or exec time.
5677
5678 (b) $ matches before a terminating newline by default; an option
5679 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5680 mode). The option can be set at compile or exec time.
5681
5682 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5683 the same as Perl's. If the decimal number it represents is less than 10
5684 or there aren't that many previous left capturing parentheses, an octal
5685 escape is read. Inside a character class, it's always an octal escape,
5686 even if it is a single digit.
5687
5688 (d) An escaped but undefined alphabetic character is taken as a literal,
5689 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5690 escapes.
5691
5692 (e) {0} is now permitted. (The previous item is removed from the compiled
5693 pattern).
5694
5695 5. Changed all the names of code files so that the basic parts are no longer
5696 than 10 characters, and abolished the teeny "globals.c" file.
5697
5698 6. Changed the handling of character classes; they are now done with a 32-byte
5699 bit map always.
5700
5701 7. Added the -d and /D options to pcretest to make it possible to look at the
5702 internals of compilation without having to recompile pcre.
5703
5704
5705 Version 0.95 23-Sep-97
5706 ----------------------
5707
5708 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5709 \x20 at the start of a run of normal characters. These were being treated as
5710 real characters, instead of the source characters being re-checked.
5711
5712
5713 Version 0.94 18-Sep-97
5714 ----------------------
5715
5716 1. The functions are now thread-safe, with the caveat that the global variables
5717 containing pointers to malloc() and free() or alternative functions are the
5718 same for all threads.
5719
5720 2. Get pcre_study() to generate a bitmap of initial characters for non-
5721 anchored patterns when this is possible, and use it if passed to pcre_exec().
5722
5723
5724 Version 0.93 15-Sep-97
5725 ----------------------
5726
5727 1. /(b)|(:+)/ was computing an incorrect first character.
5728
5729 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5730 but not actually doing anything yet.
5731
5732 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5733 as Perl does (e.g. [-az] or [az-]).
5734
5735 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5736 all possible positions.
5737
5738 5. Split up into different modules to avoid including unneeded functions in a
5739 compiled binary. However, compile and exec are still in one module. The "study"
5740 function is split off.
5741
5742 6. The character tables are now in a separate module whose source is generated
5743 by an auxiliary program - but can then be edited by hand if required. There are
5744 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5745 toupper() in the code.
5746
5747 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5748 make them global. Abolish the function for setting them, as the caller can now
5749 set them directly.
5750
5751
5752 Version 0.92 11-Sep-97
5753 ----------------------
5754
5755 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5756 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5757
5758 2. Caseless matching was not working in character classes if the characters in
5759 the pattern were in upper case.
5760
5761 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5762
5763 4. Make PCRE_ANCHORED public and accept as a compile option.
5764
5765 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5766 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5767 pass them.
5768
5769 6. Give an error if bad option bits passed at compile or run time.
5770
5771 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5772 pcretest to cause it to pass that flag.
5773
5774 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5775 options, and the first character, if set.
5776
5777 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5778
5779
5780 Version 0.91 10-Sep-97
5781 ----------------------
5782
5783 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5784 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5785
5786 2. PCRE was looping on encountering an indefinitely repeated back reference to
5787 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5788 Perl does - treats the match as successful.
5789
5790 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5