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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1557 - (show annotations)
Fri May 8 16:39:40 2015 UTC (4 years, 5 months ago) by ph10
File size: 278142 byte(s)
Error occurred while calculating annotation data.
Fix buffer overflow for repeated conditional when referencing a duplicate name.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.38 xx-xxx-xxxx
5 ------------------------
6
7 1. If a group that contained a recursive back reference also contained a
8 forward reference subroutine call followed by a non-forward-reference
9 subroutine call, for example /.((?2)(?R)\1)()/, pcre2_compile() failed to
10 compile correct code, leading to undefined behaviour or an internally
11 detected error. This bug was discovered by the LLVM fuzzer.
12
13 2. Quantification of certain items (e.g. atomic back references) could cause
14 incorrect code to be compiled when recursive forward references were
15 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
16 This bug was discovered by the LLVM fuzzer.
17
18 3. A repeated conditional group whose condition was a reference by name caused
19 a buffer overflow if there was more than one group with the given name.
20 This bug was discovered by the LLVM fuzzer.
21
22
23 Version 8.37 28-April-2015
24 --------------------------
25
26 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
27 for those parentheses to be closed with whatever has been captured so far.
28 However, it was failing to mark any other groups between the hightest
29 capture so far and the currrent group as "unset". Thus, the ovector for
30 those groups contained whatever was previously there. An example is the
31 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
32
33 2. If an assertion condition was quantified with a minimum of zero (an odd
34 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
35
36 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
37 unrecognized modifier, a crash could occur.
38
39 4. An attempt to do global matching in pcretest with a zero-length ovector
40 caused a crash.
41
42 5. Fixed a memory leak during matching that could occur for a subpattern
43 subroutine call (recursive or otherwise) if the number of captured groups
44 that had to be saved was greater than ten.
45
46 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
47 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
48 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
49
50 7. A UTF pattern containing a "not" match of a non-ASCII character and a
51 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
52
53 8. When a pattern is compiled, it remembers the highest back reference so that
54 when matching, if the ovector is too small, extra memory can be obtained to
55 use instead. A conditional subpattern whose condition is a check on a
56 capture having happened, such as, for example in the pattern
57 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
58 setting the highest backreference number. This mattered only if pcre_exec()
59 was called with an ovector that was too small to hold the capture, and there
60 was no other kind of back reference (a situation which is probably quite
61 rare). The effect of the bug was that the condition was always treated as
62 FALSE when the capture could not be consulted, leading to a incorrect
63 behaviour by pcre_exec(). This bug has been fixed.
64
65 9. A reference to a duplicated named group (either a back reference or a test
66 for being set in a conditional) that occurred in a part of the pattern where
67 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
68 to be incorrectly calculated, leading to overwriting.
69
70 10. A mutually recursive set of back references such as (\2)(\1) caused a
71 segfault at study time (while trying to find the minimum matching length).
72 The infinite loop is now broken (with the minimum length unset, that is,
73 zero).
74
75 11. If an assertion that was used as a condition was quantified with a minimum
76 of zero, matching went wrong. In particular, if the whole group had
77 unlimited repetition and could match an empty string, a segfault was
78 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
79 assertions to be quantified, but not if they are being used as conditions,
80 so the above pattern is faulted by Perl. PCRE has now been changed so that
81 it also rejects such patterns.
82
83 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
84 failed to allow the zero-repeat case if pcre2_exec() was called with an
85 ovector too small to capture the group.
86
87 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
88 Red Hat Product Security:
89
90 (a) A crash if /K and /F were both set with the option to save the compiled
91 pattern.
92
93 (b) Another crash if the option to print captured substrings in a callout
94 was combined with setting a null ovector, for example \O\C+ as a subject
95 string.
96
97 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
98 forward reference repeated a large (but limited) number of times within a
99 repeated outer group that has a zero minimum quantifier, caused incorrect
100 code to be compiled, leading to the error "internal error:
101 previously-checked referenced subpattern not found" when an incorrect
102 memory address was read. This bug was reported as "heap overflow",
103 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
104 CVE-2015-2325.
105
106 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
107 call within a group that also contained a recursive back reference caused
108 incorrect code to be compiled. This bug was reported as "heap overflow",
109 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
110 number CVE-2015-2326.
111
112 24. Computing the size of the JIT read-only data in advance has been a source
113 of various issues, and new ones are still appear unfortunately. To fix
114 existing and future issues, size computation is eliminated from the code,
115 and replaced by on-demand memory allocation.
116
117 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
118 adjacent to the end of the range, and the range contained characters with
119 more than one other case, caused incorrect behaviour when compiled in UTF
120 mode. In that example, the range a-j was left out of the class.
121
122 26. Fix JIT compilation of conditional blocks, which assertion
123 is converted to (*FAIL). E.g: /(?(?!))/.
124
125 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
126 discovered by the LLVM fuzzer.
127
128 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
129 when this assertion was used as a condition, for example (?(?!)a|b). In
130 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
131 error about an unsupported item.
132
133 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
134 possessification code could take exponential time to complete. A recursion
135 depth limit of 1000 has been imposed to limit the resources used by this
136 optimization.
137
138 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
139 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
140 because \S ensures they are all in the class. The code for doing this was
141 interacting badly with the code for computing the amount of space needed to
142 compile the pattern, leading to a buffer overflow. This bug was discovered
143 by the LLVM fuzzer.
144
145 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
146 other kinds of group caused stack overflow at compile time. This bug was
147 discovered by the LLVM fuzzer.
148
149 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
150 between a subroutine call and its quantifier was incorrectly compiled,
151 leading to buffer overflow or other errors. This bug was discovered by the
152 LLVM fuzzer.
153
154 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
155 assertion after (?(. The code was failing to check the character after
156 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
157 was discovered by the LLVM fuzzer.
158
159 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
160 a fixed maximum following a group that contains a subroutine reference was
161 incorrectly compiled and could trigger buffer overflow. This bug was
162 discovered by the LLVM fuzzer.
163
164 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
165 caused a stack overflow instead of the diagnosis of a non-fixed length
166 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
167
168 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
169 (e.g. /(?<=\Ka)/) could make pcregrep loop.
170
171 37. There was a similar problem to 36 in pcretest for global matches.
172
173 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
174 and a subsequent item in the pattern caused a non-match, backtracking over
175 the repeated \X did not stop, but carried on past the start of the subject,
176 causing reference to random memory and/or a segfault. There were also some
177 other cases where backtracking after \C could crash. This set of bugs was
178 discovered by the LLVM fuzzer.
179
180 39. The function for finding the minimum length of a matching string could take
181 a very long time if mutual recursion was present many times in a pattern,
182 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
183 method has been implemented. This infelicity was discovered by the LLVM
184 fuzzer.
185
186 40. Static linking against the PCRE library using the pkg-config module was
187 failing on missing pthread symbols.
188
189
190 Version 8.36 26-September-2014
191 ------------------------------
192
193 1. Got rid of some compiler warnings in the C++ modules that were shown up by
194 -Wmissing-field-initializers and -Wunused-parameter.
195
196 2. The tests for quantifiers being too big (greater than 65535) were being
197 applied after reading the number, and stupidly assuming that integer
198 overflow would give a negative number. The tests are now applied as the
199 numbers are read.
200
201 3. Tidy code in pcre_exec.c where two branches that used to be different are
202 now the same.
203
204 4. The JIT compiler did not generate match limit checks for certain
205 bracketed expressions with quantifiers. This may lead to exponential
206 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
207 issue should be resolved now.
208
209 5. Fixed an issue, which occures when nested alternatives are optimized
210 with table jumps.
211
212 6. Inserted two casts and changed some ints to size_t in the light of some
213 reported 64-bit compiler warnings (Bugzilla 1477).
214
215 7. Fixed a bug concerned with zero-minimum possessive groups that could match
216 an empty string, which sometimes were behaving incorrectly in the
217 interpreter (though correctly in the JIT matcher). This pcretest input is
218 an example:
219
220 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
221 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
222
223 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
224 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
225 for an empty string was breaking the inner loop and carrying on at a lower
226 level, when possessive repeated groups should always return to a higher
227 level as they have no backtrack points in them. The empty string test now
228 occurs at the outer level.
229
230 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
231 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
232
233 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
234 doesn't).
235
236 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
237 to hold a pattern to be incorrectly computed (too small) when there were
238 named back references to duplicated names. This could cause "internal
239 error: code overflow" or "double free or corruption" or other memory
240 handling errors.
241
242 11. When named subpatterns had the same prefixes, back references could be
243 confused. For example, in this pattern:
244
245 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
246
247 the reference to 'Name' was incorrectly treated as a reference to a
248 duplicate name.
249
250 12. A pattern such as /^s?c/mi8 where the optional character has more than
251 one "other case" was incorrectly compiled such that it would only try to
252 match starting at "c".
253
254 13. When a pattern starting with \s was studied, VT was not included in the
255 list of possible starting characters; this should have been part of the
256 8.34/18 patch.
257
258 14. If a character class started [\Qx]... where x is any character, the class
259 was incorrectly terminated at the ].
260
261 15. If a pattern that started with a caseless match for a character with more
262 than one "other case" was studied, PCRE did not set up the starting code
263 unit bit map for the list of possible characters. Now it does. This is an
264 optimization improvement, not a bug fix.
265
266 16. The Unicode data tables have been updated to Unicode 7.0.0.
267
268 17. Fixed a number of memory leaks in pcregrep.
269
270 18. Avoid a compiler warning (from some compilers) for a function call with
271 a cast that removes "const" from an lvalue by using an intermediate
272 variable (to which the compiler does not object).
273
274 19. Incorrect code was compiled if a group that contained an internal recursive
275 back reference was optional (had quantifier with a minimum of zero). This
276 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
277 caused segmentation faults because of stack overflows at compile time.
278
279 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
280 group that is quantified with an indefinite repeat, caused a compile-time
281 loop which used up all the system stack and provoked a segmentation fault.
282 This was not the same bug as 19 above.
283
284 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
285 Patch by Mike Frysinger.
286
287
288 Version 8.35 04-April-2014
289 --------------------------
290
291 1. A new flag is set, when property checks are present in an XCLASS.
292 When this flag is not set, PCRE can perform certain optimizations
293 such as studying these XCLASS-es.
294
295 2. The auto-possessification of character sets were improved: a normal
296 and an extended character set can be compared now. Furthermore
297 the JIT compiler optimizes more character set checks.
298
299 3. Got rid of some compiler warnings for potentially uninitialized variables
300 that show up only when compiled with -O2.
301
302 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
303 of a match later then the end of the match. The pcretest program was not
304 handling the case sensibly - it was outputting from the start to the next
305 binary zero. It now reports this situation in a message, and outputs the
306 text from the end to the start.
307
308 5. Fast forward search is improved in JIT. Instead of the first three
309 characters, any three characters with fixed position can be searched.
310 Search order: first, last, middle.
311
312 6. Improve character range checks in JIT. Characters are read by an inprecise
313 function now, which returns with an unknown value if the character code is
314 above a certain threshold (e.g: 256). The only limitation is that the value
315 must be bigger than the threshold as well. This function is useful when
316 the characters above the threshold are handled in the same way.
317
318 7. The macros whose names start with RAWUCHAR are placeholders for a future
319 mode in which only the bottom 21 bits of 32-bit data items are used. To
320 make this more memorable for those maintaining the code, the names have
321 been changed to start with UCHAR21, and an extensive comment has been added
322 to their definition.
323
324 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
325 to the export list in Makefile.am (they were accidentally omitted from the
326 8.34 tarball).
327
328 9. The informational output from pcretest used the phrase "starting byte set"
329 which is inappropriate for the 16-bit and 32-bit libraries. As the output
330 for "first char" and "need char" really means "non-UTF-char", I've changed
331 "byte" to "char", and slightly reworded the output. The documentation about
332 these values has also been (I hope) clarified.
333
334 10. Another JIT related optimization: use table jumps for selecting the correct
335 backtracking path, when more than four alternatives are present inside a
336 bracket.
337
338 11. Empty match is not possible, when the minimum length is greater than zero,
339 and there is no \K in the pattern. JIT should avoid empty match checks in
340 such cases.
341
342 12. In a caseless character class with UCP support, when a character with more
343 than one alternative case was not the first character of a range, not all
344 the alternative cases were added to the class. For example, s and \x{17f}
345 are both alternative cases for S: the class [RST] was handled correctly,
346 but [R-T] was not.
347
348 13. The configure.ac file always checked for pthread support when JIT was
349 enabled. This is not used in Windows, so I have put this test inside a
350 check for the presence of windows.h (which was already tested for).
351
352 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
353 The algorithm provides a way to skip certain starting offsets, and usually
354 faster than linear prefix searches.
355
356 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
357 as for 'fr_FR' and 'french'. For some reason, however, it then used the
358 Windows-specific input and output files, which have 'french' screwed in.
359 So this could never have worked. One of the problems with locales is that
360 they aren't always the same. I have now updated RunTest so that it checks
361 the output of the locale test (test 3) against three different output
362 files, and it allows the test to pass if any one of them matches. With luck
363 this should make the test pass on some versions of Solaris where it was
364 failing. Because of the uncertainty, the script did not used to stop if
365 test 3 failed; it now does. If further versions of a French locale ever
366 come to light, they can now easily be added.
367
368 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
369 there was a message during ./configure, but it did not stop. This now
370 provokes an error. The invalid example in README has been corrected.
371 If a value less than the minimum is given, the minimum value has always
372 been used, but now a warning is given.
373
374 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
375 was a bug in the test system, which is now fixed. Also, the list of various
376 configurations that are tested for each release did not have one with both
377 16/32 bits and --enable-bar-anycrlf. It now does.
378
379 18. pcretest was missing "-C bsr" for displaying the \R default setting.
380
381 19. Little endian PowerPC systems are supported now by the JIT compiler.
382
383 20. The fast forward newline mechanism could enter to an infinite loop on
384 certain invalid UTF-8 input. Although we don't support these cases
385 this issue can be fixed by a performance optimization.
386
387 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
388 not take account if existing stack usage. There is now a new global
389 variable called pcre_stack_guard that can be set to point to an external
390 function to check stack availability. It is called at the start of
391 processing every parenthesized group.
392
393 22. A typo in the code meant that in ungreedy mode the max/min qualifier
394 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
395 not match "ab".
396
397 23. When UTF was disabled, the JIT program reported some incorrect compile
398 errors. These messages are silenced now.
399
400 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
401 compiler.
402
403 25. Change all the temporary files used in RunGrepTest to be different to those
404 used by RunTest so that the tests can be run simultaneously, for example by
405 "make -j check".
406
407
408 Version 8.34 15-December-2013
409 -----------------------------
410
411 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
412 executable memory. Patch inspired by Carsten Klein.
413
414 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
415 this macro is never tested and has no effect, because the work to support
416 coverage involves only compiling and linking options and special targets in
417 the Makefile. The comment in config.h implied that defining the macro would
418 enable coverage support, which is totally false. There was also support for
419 setting this macro in the CMake files (my fault, I just copied it from
420 configure). SUPPORT_GCOV has now been removed.
421
422 3. Make a small performance improvement in strlen16() and strlen32() in
423 pcretest.
424
425 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
426 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
427 these cases). There was also one in pcretest.c.
428
429 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
430
431 6. In UTF mode, the code for checking whether a group could match an empty
432 string (which is used for indefinitely repeated groups to allow for
433 breaking an infinite loop) was broken when the group contained a repeated
434 negated single-character class with a character that occupied more than one
435 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
436 UTF-8 mode). The effect was undefined: the group might or might not be
437 deemed as matching an empty string, or the program might have crashed.
438
439 7. The code for checking whether a group could match an empty string was not
440 recognizing that \h, \H, \v, \V, and \R must match a character.
441
442 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
443 an empty string. If it can, pcretest shows this in its information output.
444
445 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
446 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
447 matched by pcre_exec() without using JIT:
448
449 (a) If the rest of the pattern did not match after a maximal run of
450 grapheme clusters, the code for backing up to try with fewer of them
451 did not always back up over a full grapheme when characters that do not
452 have the modifier quality were involved, e.g. Hangul syllables.
453
454 (b) If the match point in a subject started with modifier character, and
455 there was no match, the code could incorrectly back up beyond the match
456 point, and potentially beyond the first character in the subject,
457 leading to a segfault or an incorrect match result.
458
459 10. A conditional group with an assertion condition could lead to PCRE
460 recording an incorrect first data item for a match if no other first data
461 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
462 first data item, and therefore matched "ca" after "c" instead of at the
463 start.
464
465 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
466 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
467
468 12. The source of pcregrep now includes z/OS-specific code so that it can be
469 compiled for z/OS as part of the special z/OS distribution.
470
471 13. Added the -T and -TM options to pcretest.
472
473 14. The code in pcre_compile.c for creating the table of named capturing groups
474 has been refactored. Instead of creating the table dynamically during the
475 actual compiling pass, the information is remembered during the pre-compile
476 pass (on the stack unless there are more than 20 named groups, in which
477 case malloc() is used) and the whole table is created before the actual
478 compile happens. This has simplified the code (it is now nearly 150 lines
479 shorter) and prepared the way for better handling of references to groups
480 with duplicate names.
481
482 15. A back reference to a named subpattern when there is more than one of the
483 same name now checks them in the order in which they appear in the pattern.
484 The first one that is set is used for the reference. Previously only the
485 first one was inspected. This change makes PCRE more compatible with Perl.
486
487 16. Unicode character properties were updated from Unicode 6.3.0.
488
489 17. The compile-time code for auto-possessification has been refactored, based
490 on a patch by Zoltan Herczeg. It now happens after instead of during
491 compilation. The code is cleaner, and more cases are handled. The option
492 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
493 options in pcretest are provided to set it. It can also be set by
494 (*NO_AUTO_POSSESS) at the start of a pattern.
495
496 18. The character VT has been added to the default ("C" locale) set of
497 characters that match \s and are generally treated as white space,
498 following this same change in Perl 5.18. There is now no difference between
499 "Perl space" and "POSIX space". Whether VT is treated as white space in
500 other locales depends on the locale.
501
502 19. The code for checking named groups as conditions, either for being set or
503 for being recursed, has been refactored (this is related to 14 and 15
504 above). Processing unduplicated named groups should now be as fast at
505 numerical groups, and processing duplicated groups should be faster than
506 before.
507
508 20. Two patches to the CMake build system, by Alexander Barkov:
509
510 (1) Replace the "source" command by "." in CMakeLists.txt because
511 "source" is a bash-ism.
512
513 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
514 without these the CMake build does not work on Solaris.
515
516 21. Perl has changed its handling of \8 and \9. If there is no previously
517 encountered capturing group of those numbers, they are treated as the
518 literal characters 8 and 9 instead of a binary zero followed by the
519 literals. PCRE now does the same.
520
521 22. Following Perl, added \o{} to specify codepoints in octal, making it
522 possible to specify values greater than 0777 and also making them
523 unambiguous.
524
525 23. Perl now gives an error for missing closing braces after \x{... instead of
526 treating the string as literal. PCRE now does the same.
527
528 24. RunTest used to grumble if an inappropriate test was selected explicitly,
529 but just skip it when running all tests. This make it awkward to run ranges
530 of tests when one of them was inappropriate. Now it just skips any
531 inappropriate tests, as it always did when running all tests.
532
533 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
534 character types such as \d or \w, too many callouts were inserted, and the
535 data that they returned was rubbish.
536
537 26. In UCP mode, \s was not matching two of the characters that Perl matches,
538 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
539 were matched by \h. The code has now been refactored so that the lists of
540 the horizontal and vertical whitespace characters used for \h and \v (which
541 are defined only in one place) are now also used for \s.
542
543 27. Add JIT support for the 64 bit TileGX architecture.
544 Patch by Jiong Wang (Tilera Corporation).
545
546 28. Possessive quantifiers for classes (both explicit and automatically
547 generated) now use special opcodes instead of wrapping in ONCE brackets.
548
549 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
550 (because it's meaningless), this was not happening when PCRE_CASELESS was
551 set. Not wrong, but inefficient.
552
553 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
554 properties for \w, \d, etc) is present in a test regex. Otherwise if the
555 test contains no characters greater than 255, Perl doesn't realise it
556 should be using Unicode semantics.
557
558 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
559 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
560 does in Unicode mode.
561
562 32. Added the "forbid" facility to pcretest so that putting tests into the
563 wrong test files can sometimes be quickly detected.
564
565 33. There is now a limit (default 250) on the depth of nesting of parentheses.
566 This limit is imposed to control the amount of system stack used at compile
567 time. It can be changed at build time by --with-parens-nest-limit=xxx or
568 the equivalent in CMake.
569
570 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
571 errors. Perl warns for these when in warning mode, but PCRE has no facility
572 for giving warnings.
573
574 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
575 However, this was not working for (?!) because it is optimized to (*FAIL),
576 for which PCRE does not allow quantifiers. The optimization is now disabled
577 when a quantifier follows (?!). I can't see any use for this, but it makes
578 things uniform.
579
580 36. Perl no longer allows group names to start with digits, so I have made this
581 change also in PCRE. It simplifies the code a bit.
582
583 37. In extended mode, Perl ignores spaces before a + that indicates a
584 possessive quantifier. PCRE allowed a space before the quantifier, but not
585 before the possessive +. It now does.
586
587 38. The use of \K (reset reported match start) within a repeated possessive
588 group such as (a\Kb)*+ was not working.
589
590 40. Document that the same character tables must be used at compile time and
591 run time, and that the facility to pass tables to pcre_exec() and
592 pcre_dfa_exec() is for use only with saved/restored patterns.
593
594 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
595 features for Builds with MSVC:
596
597 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
598 builds). The .rc files can be used to set FileDescription and many other
599 attributes.
600
601 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
602 This allows higher-level build scripts which want .pdb files to avoid
603 hard-coding the exact files needed."
604
605 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
606 mean "start of word" and "end of word", respectively, as a transition aid.
607
608 43. A minimizing repeat of a class containing codepoints greater than 255 in
609 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
610 compiled to use the heap for recursion.
611
612 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
613 is configured.
614
615
616 Version 8.33 28-May-2013
617 ------------------------
618
619 1. Added 'U' to some constants that are compared to unsigned integers, to
620 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
621 variables that are added to signed variables, to ensure the result is
622 signed and can be negated.
623
624 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
625 CMake config files.
626
627 3. Revise the creation of config.h.generic so that all boolean macros are
628 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
629 overriding via -D on the command line possible.
630
631 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
632 to unsigned int is reported to make a quite noticeable speed difference in
633 a specific Windows environment. Testing on Linux did also appear to show
634 some benefit (and it is clearly not harmful). Also fixed the definition of
635 Xop which should be unsigned.
636
637 5. Related to (4), changing the definition of the intermediate variable cc
638 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
639 improvements.
640
641 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
642 unnecessary spaces.
643
644 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
645 and later.
646
647 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
648
649 9. Optimizing fast_forward_start_bits in JIT.
650
651 10. Adding support for callouts in JIT, and fixing some issues revealed
652 during this work. Namely:
653
654 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
655
656 (b) Minimum length was not checked before the matching is started.
657
658 11. The value of capture_last that is passed to callouts was incorrect in some
659 cases when there was a capture on one path that was subsequently abandoned
660 after a backtrack. Also, the capture_last value is now reset after a
661 recursion, since all captures are also reset in this case.
662
663 12. The interpreter no longer returns the "too many substrings" error in the
664 case when an overflowing capture is in a branch that is subsequently
665 abandoned after a backtrack.
666
667 13. In the pathological case when an offset vector of size 2 is used, pcretest
668 now prints out the matched string after a yield of 0 or 1.
669
670 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
671 Only supported by the JIT compiler at the moment.
672
673 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
674
675 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
676 offset of the starting point of the matching process, provided the offsets
677 vector is large enough.
678
679 17. The \A escape now records a lookbehind value of 1, though its execution
680 does not actually inspect the previous character. This is to ensure that,
681 in partial multi-segment matching, at least one character from the old
682 segment is retained when a new segment is processed. Otherwise, if there
683 are no lookbehinds in the pattern, \A might match incorrectly at the start
684 of a new segment.
685
686 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
687
688 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
689 gives some modest performance improvement in 8-bit mode.
690
691 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
692 be expressed in certain programming languages using Universal Character
693 Names.
694
695 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
696 which points out that "non characters" are not "characters that may not
697 appear in Unicode strings" but rather "characters that are reserved for
698 internal use and have only local meaning".
699
700 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
701 there was a conditional group that depended on an assertion, if the
702 assertion was false, the callout that immediately followed the alternation
703 in the condition was skipped when pcre_exec() was used for matching.
704
705 23. Allow an explicit callout to be inserted before an assertion that is the
706 condition for a conditional group, for compatibility with automatic
707 callouts, which always insert a callout at this point.
708
709 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
710 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
711
712 25. (*PRUNE) is now supported by the JIT compiler.
713
714 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
715
716 27. Fix the case where there are two or more SKIPs with arguments that may be
717 ignored.
718
719 28. (*SKIP) is now supported by the JIT compiler.
720
721 29. (*THEN) is now supported by the JIT compiler.
722
723 30. Update RunTest with additional test selector options.
724
725 31. The way PCRE handles backtracking verbs has been changed in two ways.
726
727 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
728 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
729 backtracking. In some cases this makes it more Perl-compatible, but Perl's
730 rather obscure rules do not always do the same thing.
731
732 (2) Previously, backtracking verbs were confined within assertions. This is
733 no longer the case for positive assertions, except for (*ACCEPT). Again,
734 this sometimes improves Perl compatibility, and sometimes does not.
735
736 32. A number of tests that were in test 2 because Perl did things differently
737 have been moved to test 1, because either Perl or PCRE has changed, and
738 these tests are now compatible.
739
740 32. Backtracking control verbs are now handled in the same way in JIT and
741 interpreter.
742
743 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
744 contained a forward subroutine reference caused a compile error.
745
746 34. Auto-detect and optimize limited repetitions in JIT.
747
748 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
749 blocking (*UTF) etc.
750
751 36. In the interpreter, maximizing pattern repetitions for characters and
752 character types now use tail recursion, which reduces stack usage.
753
754 37. The value of the max lookbehind was not correctly preserved if a compiled
755 and saved regex was reloaded on a host of different endianness.
756
757 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
758 of the compiled pattern block, expand the flags field from 16 to 32 bits
759 because it was almost full.
760
761 39. Try madvise first before posix_madvise.
762
763 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
764 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
765 original change locked out all matches of empty strings. This has been
766 changed so that one match of an empty string per line is recognized.
767 Subsequent searches on the same line (for colouring or for --only-matching,
768 for example) do not recognize empty strings.
769
770 41. Applied a user patch to fix a number of spelling mistakes in comments.
771
772 42. Data lines longer than 65536 caused pcretest to crash.
773
774 43. Clarified the data type for length and startoffset arguments for pcre_exec
775 and pcre_dfa_exec in the function-specific man pages, where they were
776 explicitly stated to be in bytes, never having been updated. I also added
777 some clarification to the pcreapi man page.
778
779 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
780 a segmentation fault.
781
782
783 Version 8.32 30-November-2012
784 -----------------------------
785
786 1. Improved JIT compiler optimizations for first character search and single
787 character iterators.
788
789 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
790 Patch by Daniel Richard G.
791
792 3. Single character iterator optimizations in the JIT compiler.
793
794 4. Improved JIT compiler optimizations for character ranges.
795
796 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
797 Reported by Giuseppe D'Angelo.
798
799 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
800 of a line, was being set incorrectly in cases where .* appeared inside
801 atomic brackets at the start of a pattern, or where there was a subsequent
802 *PRUNE or *SKIP.
803
804 7. Improved instruction cache flush for POWER/PowerPC.
805 Patch by Daniel Richard G.
806
807 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
808 grep:
809
810 (a) There is now no limit to the number of patterns to be matched.
811
812 (b) An error is given if a pattern is too long.
813
814 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
815 are now supported.
816
817 (d) --exclude-from and --include-from (multiple use) have been added.
818
819 (e) Exclusions and inclusions now apply to all files and directories, not
820 just to those obtained from scanning a directory recursively.
821
822 (f) Multiple uses of -f and --file-list are now supported.
823
824 (g) In a Windows environment, the default for -d has been changed from
825 "read" (the GNU grep default) to "skip", because otherwise the presence
826 of a directory in the file list provokes an error.
827
828 (h) The documentation has been revised and clarified in places.
829
830 9. Improve the matching speed of capturing brackets.
831
832 10. Changed the meaning of \X so that it now matches a Unicode extended
833 grapheme cluster.
834
835 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
836 out POSIX threads when JIT support is configured.
837
838 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
839
840 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
841 structure could go wrong in environments where size_t is not the same size
842 as int.
843
844 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
845 set.
846
847 15. The EBCDIC support had decayed; later updates to the code had included
848 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
849 general tidy up of EBCDIC-related issues, and the documentation was also
850 not quite right. There is now a test that can be run on ASCII systems to
851 check some of the EBCDIC-related things (but is it not a full test).
852
853 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
854 in a small tidy to the code.
855
856 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
857
858 18. If the --only-matching (-o) option in pcregrep is specified multiple
859 times, each one causes appropriate output. For example, -o1 -o2 outputs the
860 substrings matched by the 1st and 2nd capturing parentheses. A separating
861 string can be specified by --om-separator (default empty).
862
863 19. Improving the first n character searches.
864
865 20. Turn case lists for horizontal and vertical white space into macros so that
866 they are defined only once.
867
868 21. This set of changes together give more compatible Unicode case-folding
869 behaviour for characters that have more than one other case when UCP
870 support is available.
871
872 (a) The Unicode property table now has offsets into a new table of sets of
873 three or more characters that are case-equivalent. The MultiStage2.py
874 script that generates these tables (the pcre_ucd.c file) now scans
875 CaseFolding.txt instead of UnicodeData.txt for character case
876 information.
877
878 (b) The code for adding characters or ranges of characters to a character
879 class has been abstracted into a generalized function that also handles
880 case-independence. In UTF-mode with UCP support, this uses the new data
881 to handle characters with more than one other case.
882
883 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
884 whose other case is greater than 256 are now correctly matched
885 caselessly. Previously, the high codepoint matched the low one, but not
886 vice versa.
887
888 (d) The processing of \h, \H, \v, and \ in character classes now makes use
889 of the new class addition function, using character lists defined as
890 macros alongside the case definitions of 20 above.
891
892 (e) Caseless back references now work with characters that have more than
893 one other case.
894
895 (f) General caseless matching of characters with more than one other case
896 is supported.
897
898 22. Unicode character properties were updated from Unicode 6.2.0
899
900 23. Improved CMake support under Windows. Patch by Daniel Richard G.
901
902 24. Add support for 32-bit character strings, and UTF-32
903
904 25. Major JIT compiler update (code refactoring and bugfixing).
905 Experimental Sparc 32 support is added.
906
907 26. Applied a modified version of Daniel Richard G's patch to create
908 pcre.h.generic and config.h.generic by "make" instead of in the
909 PrepareRelease script.
910
911 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
912 pcre_compile.c when checking for a zero character.
913
914 28. Introducing a native interface for JIT. Through this interface, the compiled
915 machine code can be directly executed. The purpose of this interface is to
916 provide fast pattern matching, so several sanity checks are not performed.
917 However, feature tests are still performed. The new interface provides
918 1.4x speedup compared to the old one.
919
920 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
921 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
922 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
923
924 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
925 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
926 the "old" RFC 2279). Instead, it ended up passing a negative length to
927 pcre_exec().
928
929 31. Add support for GCC's visibility feature to hide internal functions.
930
931 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
932 "unknown -C option" after outputting 0 or 1.
933
934 33. There is now support for generating a code coverage report for the test
935 suite in environments where gcc is the compiler and lcov is installed. This
936 is mainly for the benefit of the developers.
937
938 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
939 unaddressable using valgrind annotations, allowing valgrind to detect
940 invalid memory accesses. This is mainly for the benefit of the developers.
941
942 25. (*UTF) can now be used to start a pattern in any of the three libraries.
943
944 26. Give configure error if --enable-cpp but no C++ compiler found.
945
946
947 Version 8.31 06-July-2012
948 -------------------------
949
950 1. Fixing a wrong JIT test case and some compiler warnings.
951
952 2. Removed a bashism from the RunTest script.
953
954 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
955 to unsigned type, result still unsigned" that was given by an MS compiler
956 on encountering the code "-sizeof(xxx)".
957
958 4. Partial matching support is added to the JIT compiler.
959
960 5. Fixed several bugs concerned with partial matching of items that consist
961 of more than one character:
962
963 (a) /^(..)\1/ did not partially match "aba" because checking references was
964 done on an "all or nothing" basis. This also applied to repeated
965 references.
966
967 (b) \R did not give a hard partial match if \r was found at the end of the
968 subject.
969
970 (c) \X did not give a hard partial match after matching one or more
971 characters at the end of the subject.
972
973 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
974 a partial match for the string "\r".
975
976 (e) When newline was set to CRLF, the metacharacter "." did not recognize
977 a partial match for a CR character at the end of the subject string.
978
979 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
980 running pcretest, the text "(JIT)" added to the output whenever JIT is
981 actually used to run the match.
982
983 7. Individual JIT compile options can be set in pcretest by following -s+[+]
984 or /S+[+] with a digit between 1 and 7.
985
986 8. OP_NOT now supports any UTF character not just single-byte ones.
987
988 9. (*MARK) control verb is now supported by the JIT compiler.
989
990 10. The command "./RunTest list" lists the available tests without actually
991 running any of them. (Because I keep forgetting what they all are.)
992
993 11. Add PCRE_INFO_MAXLOOKBEHIND.
994
995 12. Applied a (slightly modified) user-supplied patch that improves performance
996 when the heap is used for recursion (compiled with --disable-stack-for-
997 recursion). Instead of malloc and free for each heap frame each time a
998 logical recursion happens, frames are retained on a chain and re-used where
999 possible. This sometimes gives as much as 30% improvement.
1000
1001 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1002 call.
1003
1004 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1005
1006 15. It is now possible to link pcretest with libedit as an alternative to
1007 libreadline.
1008
1009 16. (*COMMIT) control verb is now supported by the JIT compiler.
1010
1011 17. The Unicode data tables have been updated to Unicode 6.1.0.
1012
1013 18. Added --file-list option to pcregrep.
1014
1015 19. Added binary file support to pcregrep, including the -a, --binary-files,
1016 -I, and --text options.
1017
1018 20. The madvise function is renamed for posix_madvise for QNX compatibility
1019 reasons. Fixed by Giuseppe D'Angelo.
1020
1021 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1022
1023 22. Changed the option for creating symbolic links for 16-bit man pages from
1024 -s to -sf so that re-installing does not cause issues.
1025
1026 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1027
1028 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1029 matches in certain environments (the workspace was not being correctly
1030 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1031 some of the workspace data at the beginning of a restart.
1032
1033 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1034 was not doing so when it should - probably a typo introduced by SVN 528
1035 (change 8.10/14).
1036
1037 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1038 \w+ when the character tables indicated that \x{c4} was a word character.
1039 There were several related cases, all because the tests for doing a table
1040 lookup were testing for characters less than 127 instead of 255.
1041
1042 27. If a pattern contains capturing parentheses that are not used in a match,
1043 their slots in the ovector are set to -1. For those that are higher than
1044 any matched groups, this happens at the end of processing. In the case when
1045 there were back references that the ovector was too small to contain
1046 (causing temporary malloc'd memory to be used during matching), and the
1047 highest capturing number was not used, memory off the end of the ovector
1048 was incorrectly being set to -1. (It was using the size of the temporary
1049 memory instead of the true size.)
1050
1051 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1052 ovector size, it uses memory at the end of the block that it has got.
1053
1054 29. Check for an overlong MARK name and give an error at compile time. The
1055 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1056
1057 30. JIT compiler update.
1058
1059 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1060 Rill for the patch.
1061
1062 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1063
1064 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1065
1066 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1067 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1068 was enabled.
1069
1070 35. Improve JIT code generation for greedy plus quantifier.
1071
1072 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1073 1 to "aa" instead of to an empty string. The bug affected repeated groups
1074 that could potentially match an empty string.
1075
1076 37. Optimizing single character iterators in JIT.
1077
1078 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1079 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1080 codepoints that are too big for the mode are faulted, and in a UTF mode,
1081 disallowed codepoints are also faulted.
1082
1083 39. If PCRE was compiled with UTF support, in three places in the DFA
1084 matcher there was code that should only have been obeyed in UTF mode, but
1085 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1086 processing when bytes with values greater than 127 were present. In 16-bit
1087 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1088 both cases the values are those that cannot be the first data item in a UTF
1089 character. The three items that might have provoked this were recursions,
1090 possessively repeated groups, and atomic groups.
1091
1092 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1093 and pcregrep, because some OS require shared objects to be explicitly
1094 passed to ld, causing the link step to fail if they are not.
1095
1096 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1097 mode, patterns that started with \h* or \R* might be incorrectly matched.
1098
1099
1100 Version 8.30 04-February-2012
1101 -----------------------------
1102
1103 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1104 name is defined in ctype.h.
1105
1106 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1107 only in quite long subpatterns.
1108
1109 3. Removed the function pcre_info(), which has been obsolete and deprecated
1110 since it was replaced by pcre_fullinfo() in February 2000.
1111
1112 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1113 match a (*MARK), and the match failed at the start of the subject, a
1114 reference to memory before the start of the subject could occur. This bug
1115 was introduced by fix 17 of release 8.21.
1116
1117 5. A reference to an unset group with zero minimum repetition was giving
1118 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1119 /(another)?(\1?)test/ matched against "hello world test". This bug was
1120 introduced in release 8.13.
1121
1122 6. Add support for 16-bit character strings (a large amount of work involving
1123 many changes and refactorings).
1124
1125 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1126 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1127 from a file.
1128
1129 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1130 rounding is not applied in this particular case).
1131
1132 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1133 if they appear, or are escaped, in patterns.
1134
1135 10. Get rid of a number of -Wunused-but-set-variable warnings.
1136
1137 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1138 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1139 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1140 also returns the mark "x". This bug applied to capturing parentheses,
1141 non-capturing parentheses, and atomic parentheses. It also applied to some
1142 assertions.
1143
1144 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1145 information out of configure.ac instead of relying on pcre.h.generic, which
1146 is not stored in the repository.
1147
1148 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1149 -lreadline.
1150
1151 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1152
1153 15. Applied Graycode's patch to put the top-level frame on the stack rather
1154 than the heap when not using the stack for recursion. This gives a
1155 performance improvement in many cases when recursion is not deep.
1156
1157 16. Experimental code added to "pcretest -C" to output the stack frame size.
1158
1159
1160 Version 8.21 12-Dec-2011
1161 ------------------------
1162
1163 1. Updating the JIT compiler.
1164
1165 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1166 are added as well.
1167
1168 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1169 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1170 calling _pcre_jit_exec. Some extra comments are added.
1171
1172 4. (*MARK) settings inside atomic groups that do not contain any capturing
1173 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1174 was introduced by change 18 for 8.20.
1175
1176 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1177 ECMA-262 standard.
1178
1179 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1180 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1181 This bug was probably introduced by change 9 of 8.13.
1182
1183 7. While fixing 6 above, I noticed that a number of other items were being
1184 incorrectly rejected as "not fixed length". This arose partly because newer
1185 opcodes had not been added to the fixed-length checking code. I have (a)
1186 corrected the bug and added tests for these items, and (b) arranged for an
1187 error to occur if an unknown opcode is encountered while checking for fixed
1188 length instead of just assuming "not fixed length". The items that were
1189 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1190 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1191 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1192
1193 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1194 being incorrectly compiled and would have given unpredicatble results.
1195
1196 9. A possessively repeated subpattern with minimum repeat count greater than
1197 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1198 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1199 the first (A) could occur when it should not.
1200
1201 10. Add a cast and remove a redundant test from the code.
1202
1203 11. JIT should use pcre_malloc/pcre_free for allocation.
1204
1205 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1206 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1207 is anything other than /usr, -L is still shown).
1208
1209 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1210
1211 14. Perl does not support \N without a following name in a [] class; PCRE now
1212 also gives an error.
1213
1214 15. If a forward reference was repeated with an upper limit of around 2000,
1215 it caused the error "internal error: overran compiling workspace". The
1216 maximum number of forward references (including repeats) was limited by the
1217 internal workspace, and dependent on the LINK_SIZE. The code has been
1218 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1219 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1220 of around 200,000 forward references. While doing this, I also speeded up
1221 the filling in of repeated forward references.
1222
1223 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1224 incorrectly expecting the subject to contain another "a" after the start.
1225
1226 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1227 in the match, the SKIP should be ignored. This was not happening; instead
1228 the SKIP was being treated as NOMATCH. For patterns such as
1229 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1230 tested.
1231
1232 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1233 now much more compatible with Perl, in particular in cases where the result
1234 is a non-match for a non-anchored pattern. For example, if
1235 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1236 "m", where previously it did not return a name. A side effect of this
1237 change is that for partial matches, the last encountered mark name is
1238 returned, as for non matches. A number of tests that were previously not
1239 Perl-compatible have been moved into the Perl-compatible test files. The
1240 refactoring has had the pleasing side effect of removing one argument from
1241 the match() function, thus reducing its stack requirements.
1242
1243 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1244 subsequent uses of /S (without +) incorrectly behaved like /S+.
1245
1246 21. Retrieve executable code size support for the JIT compiler and fixing
1247 some warnings.
1248
1249 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1250 not work when the shorter character appeared right at the end of the
1251 subject string.
1252
1253 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1254 systems.
1255
1256 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1257 output it when the /M option is used in pcretest.
1258
1259 25. The CheckMan script was not being included in the distribution. Also, added
1260 an explicit "perl" to run Perl scripts from the PrepareRelease script
1261 because this is reportedly needed in Windows.
1262
1263 26. If study data was being save in a file and studying had not found a set of
1264 "starts with" bytes for the pattern, the data written to the file (though
1265 never used) was taken from uninitialized memory and so caused valgrind to
1266 complain.
1267
1268 27. Updated RunTest.bat as provided by Sheri Pierce.
1269
1270 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1271
1272 29. Computation of memory usage for the table of capturing group names was
1273 giving an unnecessarily large value.
1274
1275
1276 Version 8.20 21-Oct-2011
1277 ------------------------
1278
1279 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1280 a POSIX class. After further experiments with Perl, which convinced me that
1281 Perl has bugs and confusions, a closing square bracket is no longer allowed
1282 in a POSIX name. This bug also affected patterns with classes that started
1283 with full stops.
1284
1285 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1286 captured substring, but while checking the failing first alternative,
1287 substring 1 is temporarily captured. If the output vector supplied to
1288 pcre_exec() was not big enough for this capture, the yield of the function
1289 was still zero ("insufficient space for captured substrings"). This cannot
1290 be totally fixed without adding another stack variable, which seems a lot
1291 of expense for a edge case. However, I have improved the situation in cases
1292 such as /(a)(b)x|abc/ matched against "abc", where the return code
1293 indicates that fewer than the maximum number of slots in the ovector have
1294 been set.
1295
1296 3. Related to (2) above: when there are more back references in a pattern than
1297 slots in the output vector, pcre_exec() uses temporary memory during
1298 matching, and copies in the captures as far as possible afterwards. It was
1299 using the entire output vector, but this conflicts with the specification
1300 that only 2/3 is used for passing back captured substrings. Now it uses
1301 only the first 2/3, for compatibility. This is, of course, another edge
1302 case.
1303
1304 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1305 main code base, and can be used by building with --enable-jit. When this is
1306 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1307 runtime --no-jit option is given.
1308
1309 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1310 ovector, the return from the function was zero, implying that there were
1311 other matches that did not fit. The correct "exactly full" value is now
1312 returned.
1313
1314 6. If a subpattern that was called recursively or as a subroutine contained
1315 (*PRUNE) or any other control that caused it to give a non-standard return,
1316 invalid errors such as "Error -26 (nested recursion at the same subject
1317 position)" or even infinite loops could occur.
1318
1319 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1320 computing the minimum length on reaching *ACCEPT, and so ended up with the
1321 wrong value of 1 rather than 0. Further investigation indicates that
1322 computing a minimum subject length in the presence of *ACCEPT is difficult
1323 (think back references, subroutine calls), and so I have changed the code
1324 so that no minimum is registered for a pattern that contains *ACCEPT.
1325
1326 8. If (*THEN) was present in the first (true) branch of a conditional group,
1327 it was not handled as intended. [But see 16 below.]
1328
1329 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1330 Sheri Pierce.
1331
1332 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1333 the first byte in a match must be "a".
1334
1335 11. Change 17 for 8.13 increased the recursion depth for patterns like
1336 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1337 pattern contains any instances of (*THEN). If it does not, the old
1338 optimizations are restored. It would be nice to do this on a per-group
1339 basis, but at the moment that is not feasible.
1340
1341 12. In some environments, the output of pcretest -C is CRLF terminated. This
1342 broke RunTest's code that checks for the link size. A single white space
1343 character after the value is now allowed for.
1344
1345 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1346 For "fr", it uses the Windows-specific input and output files.
1347
1348 14. If (*THEN) appeared in a group that was called recursively or as a
1349 subroutine, it did not work as intended. [But see next item.]
1350
1351 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1352 pattern fragments (but not containing any | characters). If A and B are
1353 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1354 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1355 D. In other words, Perl considers parentheses that do not contain any |
1356 characters to be part of a surrounding alternative, whereas PCRE was
1357 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1358 differently. PCRE now behaves in the same way as Perl, except in the case
1359 of subroutine/recursion calls such as (?1) which have in any case always
1360 been different (but PCRE had them first :-).
1361
1362 16. Related to 15 above: Perl does not treat the | in a conditional group as
1363 creating alternatives. Such a group is treated in the same way as an
1364 ordinary group without any | characters when processing (*THEN). PCRE has
1365 been changed to match Perl's behaviour.
1366
1367 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1368 RunGrepTest script failed.
1369
1370 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1371 inevitable for groups that contain captures, but it can lead to a lot of
1372 stack use in large patterns. The old behaviour has been restored for atomic
1373 groups that do not contain any capturing parentheses.
1374
1375 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1376 suppress the check for a minimum subject length at run time. (If it was
1377 given to pcre_exec() or pcre_dfa_exec() it did work.)
1378
1379 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1380 fail to work when decoding hex characters in data strings in EBCDIC
1381 environments.
1382
1383 21. It appears that in at least one Mac OS environment, the isxdigit() function
1384 is implemented as a macro that evaluates to its argument more than once,
1385 contravening the C 90 Standard (I haven't checked a later standard). There
1386 was an instance in pcretest which caused it to go wrong when processing
1387 \x{...} escapes in subject strings. The has been rewritten to avoid using
1388 things like p++ in the argument of isxdigit().
1389
1390
1391 Version 8.13 16-Aug-2011
1392 ------------------------
1393
1394 1. The Unicode data tables have been updated to Unicode 6.0.0.
1395
1396 2. Two minor typos in pcre_internal.h have been fixed.
1397
1398 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1399 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1400 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1401
1402 4. There were a number of related bugs in the code for matching backrefences
1403 caselessly in UTF-8 mode when codes for the characters concerned were
1404 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1405 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1406 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1407 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1408 2-byte code at the end of the subject (it thought there wasn't enough data
1409 left).
1410
1411 5. Comprehensive information about what went wrong is now returned by
1412 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1413 as the output vector has at least 2 elements. The offset of the start of
1414 the failing character and a reason code are placed in the vector.
1415
1416 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1417 now returned is for the first byte of the failing character, instead of the
1418 last byte inspected. This is an incompatible change, but I hope it is small
1419 enough not to be a problem. It makes the returned offset consistent with
1420 pcre_exec() and pcre_dfa_exec().
1421
1422 7. pcretest now gives a text phrase as well as the error number when
1423 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1424 failure, the offset and reason code are output.
1425
1426 8. When \R was used with a maximizing quantifier it failed to skip backwards
1427 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1428 back over a single character (\n). This seems wrong (because it treated the
1429 two characters as a single entity when going forwards), conflicts with the
1430 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1431 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1432 has been changed.
1433
1434 9. Some internal refactoring has changed the processing so that the handling
1435 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1436 time (the PCRE_DOTALL option was changed this way some time ago: version
1437 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1438 which was always a bit of a fudge. It also means that there is one less
1439 argument for the match() function, which reduces its stack requirements
1440 slightly. This change also fixes an incompatibility with Perl: the pattern
1441 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1442
1443 10. More internal refactoring has drastically reduced the number of recursive
1444 calls to match() for possessively repeated groups such as (abc)++ when
1445 using pcre_exec().
1446
1447 11. While implementing 10, a number of bugs in the handling of groups were
1448 discovered and fixed:
1449
1450 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1451 (a|)*(?1) gave a compile-time internal error.
1452 ((a|)+)+ did not notice that the outer group could match an empty string.
1453 (^a|^)+ was not marked as anchored.
1454 (.*a|.*)+ was not marked as matching at start or after a newline.
1455
1456 12. Yet more internal refactoring has removed another argument from the match()
1457 function. Special calls to this function are now indicated by setting a
1458 value in a variable in the "match data" data block.
1459
1460 13. Be more explicit in pcre_study() instead of relying on "default" for
1461 opcodes that mean there is no starting character; this means that when new
1462 ones are added and accidentally left out of pcre_study(), testing should
1463 pick them up.
1464
1465 14. The -s option of pcretest has been documented for ages as being an old
1466 synonym of -m (show memory usage). I have changed it to mean "force study
1467 for every regex", that is, assume /S for every regex. This is similar to -i
1468 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1469 using it. It makes it easier to run collections of tests with and without
1470 study enabled, and thereby test pcre_study() more easily. All the standard
1471 tests are now run with and without -s (but some patterns can be marked as
1472 "never study" - see 20 below).
1473
1474 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1475 restoration of the capturing data to the outer values was not happening
1476 correctly.
1477
1478 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1479 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1480 pattern had matched an empty string, and so incorrectly returned a no
1481 match.
1482
1483 17. There was optimizing code for the last branch of non-capturing parentheses,
1484 and also for the obeyed branch of a conditional subexpression, which used
1485 tail recursion to cut down on stack usage. Unfortunately, now that there is
1486 the possibility of (*THEN) occurring in these branches, tail recursion is
1487 no longer possible because the return has to be checked for (*THEN). These
1488 two optimizations have therefore been removed. [But see 8.20/11 above.]
1489
1490 18. If a pattern containing \R was studied, it was assumed that \R always
1491 matched two bytes, thus causing the minimum subject length to be
1492 incorrectly computed because \R can also match just one byte.
1493
1494 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1495 was incorrectly computed.
1496
1497 20. If /S is present twice on a test pattern in pcretest input, it now
1498 *disables* studying, thereby overriding the use of -s on the command line
1499 (see 14 above). This is necessary for one or two tests to keep the output
1500 identical in both cases.
1501
1502 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1503 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1504
1505 22. When an atomic group that contained a capturing parenthesis was
1506 successfully matched, but the branch in which it appeared failed, the
1507 capturing was not being forgotten if a higher numbered group was later
1508 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1509 group 1 to "a", when in fact it should be unset. This applied to multi-
1510 branched capturing and non-capturing groups, repeated or not, and also to
1511 positive assertions (capturing in negative assertions does not happen
1512 in PCRE) and also to nested atomic groups.
1513
1514 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1515 subject after a captured substring, to make it easier to tell which of a
1516 number of identical substrings has been captured.
1517
1518 24. The way atomic groups are processed by pcre_exec() has been changed so that
1519 if they are repeated, backtracking one repetition now resets captured
1520 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1521 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1522 "aaa". Previously, it would have been "a". As part of this code
1523 refactoring, the way recursive calls are handled has also been changed.
1524
1525 25. If an assertion condition captured any substrings, they were not passed
1526 back unless some other capturing happened later. For example, if
1527 (?(?=(a))a) was matched against "a", no capturing was returned.
1528
1529 26. When studying a pattern that contained subroutine calls or assertions,
1530 the code for finding the minimum length of a possible match was handling
1531 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1532 group 1 called group 2 while simultaneously a separate group 2 called group
1533 1). A stack overflow occurred in this case. I have fixed this by limiting
1534 the recursion depth to 10.
1535
1536 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1537 Fortmann. This supports explicit test numbers on the command line, and has
1538 argument validation and error reporting.
1539
1540 28. An instance of \X with an unlimited repeat could fail if at any point the
1541 first character it looked at was a mark character.
1542
1543 29. Some minor code refactoring concerning Unicode properties and scripts
1544 should reduce the stack requirement of match() slightly.
1545
1546 30. Added the '=' option to pcretest to check the setting of unused capturing
1547 slots at the end of the pattern, which are documented as being -1, but are
1548 not included in the return count.
1549
1550 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1551 compiled something random. Now it gives a compile-time error (as does
1552 Perl).
1553
1554 32. A *MARK encountered during the processing of a positive assertion is now
1555 recorded and passed back (compatible with Perl).
1556
1557 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1558 had alternative anchored branches, the search for a second match in a line
1559 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1560 matched the line "0102" twice. The same bug affected patterns that started
1561 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1562 twice.
1563
1564 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1565 does, and because of capturing effects, quantifying parenthesized
1566 assertions may at times be useful. Quantifiers are now allowed for
1567 parenthesized assertions.
1568
1569 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1570
1571 36. \g was being checked for fancy things in a character class, when it should
1572 just be a literal "g".
1573
1574 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1575 appearance of a nested POSIX class supersedes an apparent external class.
1576 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1577 unescaped square brackets may also appear as part of class names. For
1578 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1579 more like Perl. (But see 8.20/1 above.)
1580
1581 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1582 was because it thought it was \N{name}, which is not supported).
1583
1584 39. Add minix to OS list not supporting the -S option in pcretest.
1585
1586 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1587 cannot analyze patterns in sufficient detail to catch mutual recursions
1588 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1589 subgroup is called recursively as a subpattern for a second time at the
1590 same position in the subject string. In previous releases this might have
1591 been caught by the recursion limit, or it might have run out of stack.
1592
1593 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1594 happen only once. PCRE was, however incorrectly giving a compile time error
1595 "recursive call could loop indefinitely" because it cannot analyze the
1596 pattern in sufficient detail. The compile time test no longer happens when
1597 PCRE is compiling a conditional subpattern, but actual runaway loops are
1598 now caught at runtime (see 40 above).
1599
1600 42. It seems that Perl allows any characters other than a closing parenthesis
1601 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1602 has been changed to be the same.
1603
1604 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1605 as not to get warnings when autogen.sh is called. Also changed
1606 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1607
1608 44. To help people who use pcregrep to scan files containing exceedingly long
1609 lines, the following changes have been made:
1610
1611 (a) The default value of the buffer size parameter has been increased from
1612 8K to 20K. (The actual buffer used is three times this size.)
1613
1614 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1615 PCRE is built.
1616
1617 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1618 to be set at run time.
1619
1620 (d) Numerical values in pcregrep options can be followed by K or M, for
1621 example --buffer-size=50K.
1622
1623 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1624 given and the return code is set to 2.
1625
1626 45. Add a pointer to the latest mark to the callout data block.
1627
1628 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1629 partial match of an empty string instead of no match. This was specific to
1630 the use of ".".
1631
1632 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1633 complete match instead of a partial match. This bug was dependent on both
1634 the PCRE_UTF8 and PCRE_DOTALL options being set.
1635
1636 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1637 starting byte set, because \b was not being ignored.
1638
1639
1640 Version 8.12 15-Jan-2011
1641 ------------------------
1642
1643 1. Fixed some typos in the markup of the man pages, and wrote a script that
1644 checks for such things as part of the documentation building process.
1645
1646 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1647 --match-limit and --recursion-limit options (added for 8.11). In
1648 particular, this made one of the standard tests fail. (The integer value
1649 went into the wrong half of a long int.)
1650
1651 3. If the --colour option was given to pcregrep with -v (invert match), it
1652 did strange things, either producing crazy output, or crashing. It should,
1653 of course, ignore a request for colour when reporting lines that do not
1654 match.
1655
1656 4. Another pcregrep bug caused similar problems if --colour was specified with
1657 -M (multiline) and the pattern match finished with a line ending.
1658
1659 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1660 matched in multiline mode, the following line was shown as part of the
1661 match. This seems wrong, so I have changed it.
1662
1663 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1664 the check for further matches in the same line (so they could be coloured)
1665 to overrun the end of the current line. If another match was found, it was
1666 incorrectly shown (and then shown again when found in the next line).
1667
1668 7. If pcregrep was compiled under Windows, there was a reference to the
1669 function pcregrep_exit() before it was defined. I am assuming this was
1670 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1671 reported by a user. I've moved the definition above the reference.
1672
1673
1674 Version 8.11 10-Dec-2010
1675 ------------------------
1676
1677 1. (*THEN) was not working properly if there were untried alternatives prior
1678 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1679 backtracked to try for "b" instead of moving to the next alternative branch
1680 at the same level (in this case, to look for "c"). The Perl documentation
1681 is clear that when (*THEN) is backtracked onto, it goes to the "next
1682 alternative in the innermost enclosing group".
1683
1684 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1685 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1686 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1687 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1688 (*THEN).
1689
1690 3. If \s appeared in a character class, it removed the VT character from
1691 the class, even if it had been included by some previous item, for example
1692 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1693 of \s, but is part of the POSIX "space" class.)
1694
1695 4. A partial match never returns an empty string (because you can always
1696 match an empty string at the end of the subject); however the checking for
1697 an empty string was starting at the "start of match" point. This has been
1698 changed to the "earliest inspected character" point, because the returned
1699 data for a partial match starts at this character. This means that, for
1700 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1701 (previously it gave "no match").
1702
1703 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1704 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1705 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1706 has an implication that the given string is incomplete (because a partial
1707 match is preferred over a full match). For this reason, these items now
1708 give a partial match in this situation. [Aside: previously, the one case
1709 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1710 match rather than a full match, which was wrong by the old rules, but is
1711 now correct.]
1712
1713 6. There was a bug in the handling of #-introduced comments, recognized when
1714 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1715 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1716 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1717 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1718 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1719 places in pcre_compile().
1720
1721 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1722 comments when looking ahead for named forward references to subpatterns,
1723 the only newline sequence it recognized was NL. It now handles newlines
1724 according to the set newline convention.
1725
1726 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1727 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1728 cater for a lack of strerror(). These oversights have been fixed.
1729
1730 9. Added --match-limit and --recursion-limit to pcregrep.
1731
1732 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1733
1734 11. When the -o option was used, pcregrep was setting a return code of 1, even
1735 when matches were found, and --line-buffered was not being honoured.
1736
1737 12. Added an optional parentheses number to the -o and --only-matching options
1738 of pcregrep.
1739
1740 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1741 can match an empty string. The code to do it in pcretest and pcredemo
1742 needed fixing:
1743
1744 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1745 only one byte after an empty string match just before CRLF (this case
1746 just got forgotten; "any" and "anycrlf" were OK).
1747
1748 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1749 mode when an empty string match preceded an ASCII character followed by
1750 a non-ASCII character. (The code for advancing by one character rather
1751 than one byte was nonsense.)
1752
1753 (c) The pcredemo.c sample program did not have any code at all to handle
1754 the cases when CRLF is a valid newline sequence.
1755
1756 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1757 as a starting offset was within the subject string. There is now a new
1758 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1759 negative or greater than the length of the string. In order to test this,
1760 pcretest is extended to allow the setting of negative starting offsets.
1761
1762 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1763 starting offset points to the beginning of a UTF-8 character was
1764 unnecessarily clumsy. I tidied it up.
1765
1766 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1767 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1768
1769 17. Nobody had reported that the --include_dir option, which was added in
1770 release 7.7 should have been called --include-dir (hyphen, not underscore)
1771 for compatibility with GNU grep. I have changed it to --include-dir, but
1772 left --include_dir as an undocumented synonym, and the same for
1773 --exclude-dir, though that is not available in GNU grep, at least as of
1774 release 2.5.4.
1775
1776 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1777 characters from a string of bytes) have been redefined so as not to use
1778 loops, in order to improve performance in some environments. At the same
1779 time, I abstracted some of the common code into auxiliary macros to save
1780 repetition (this should not affect the compiled code).
1781
1782 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1783 compile-time error is now given if \c is not followed by an ASCII
1784 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1785 different, and any byte value is allowed.)
1786
1787 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1788 START_OPTIMIZE option, which is now allowed at compile time - but just
1789 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1790 to pcregrep and other applications that have no direct access to PCRE
1791 options. The new /Y option in pcretest sets this option when calling
1792 pcre_compile().
1793
1794 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1795 back references. Groups containing recursive back references were forced to
1796 be atomic by that change, but in the case of named groups, the amount of
1797 memory required was incorrectly computed, leading to "Failed: internal
1798 error: code overflow". This has been fixed.
1799
1800 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1801 pcretest.c, to avoid build problems in some Borland environments.
1802
1803
1804 Version 8.10 25-Jun-2010
1805 ------------------------
1806
1807 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1808 THEN.
1809
1810 2. (*ACCEPT) was not working when inside an atomic group.
1811
1812 3. Inside a character class, \B is treated as a literal by default, but
1813 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1814 causes the error). The code is unchanged, but I tidied the documentation.
1815
1816 4. Inside a character class, PCRE always treated \R and \X as literals,
1817 whereas Perl faults them if its -w option is set. I have changed PCRE so
1818 that it faults them when PCRE_EXTRA is set.
1819
1820 5. Added support for \N, which always matches any character other than
1821 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1822
1823 6. When compiling pcregrep with newer versions of gcc which may have
1824 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1825 declared with attribute warn_unused_result" were given. Just casting the
1826 result to (void) does not stop the warnings; a more elaborate fudge is
1827 needed. I've used a macro to implement this.
1828
1829 7. Minor change to pcretest.c to avoid a compiler warning.
1830
1831 8. Added four artifical Unicode properties to help with an option to make
1832 \s etc use properties (see next item). The new properties are: Xan
1833 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1834
1835 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1836 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1837 this option. Modified pcretest to add /W to test this facility. Added
1838 REG_UCP to make it available via the POSIX interface.
1839
1840 10. Added --line-buffered to pcregrep.
1841
1842 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1843 studied, and the match started with a letter with a code point greater than
1844 127 whose first byte was different to the first byte of the other case of
1845 the letter, the other case of this starting letter was not recognized
1846 (#976).
1847
1848 12. If a pattern that was studied started with a repeated Unicode property
1849 test, for example, \p{Nd}+, there was the theoretical possibility of
1850 setting up an incorrect bitmap of starting bytes, but fortunately it could
1851 not have actually happened in practice until change 8 above was made (it
1852 added property types that matched character-matching opcodes).
1853
1854 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1855 possible starting bytes for non-anchored patterns.
1856
1857 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1858 \R, and also a number of cases that involve Unicode properties, both
1859 explicit and implicit when PCRE_UCP is set.
1860
1861 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1862 input, it could crash or give wrong results if characters with values
1863 greater than 0xc0 were present in the subject string. (Detail: it assumed
1864 UTF-8 input when processing these items.)
1865
1866 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1867 size_t is 64-bit (#991).
1868
1869 17. Added a check for running out of memory when PCRE is compiled with
1870 --disable-stack-for-recursion (#990).
1871
1872 18. If the last data line in a file for pcretest does not have a newline on
1873 the end, a newline was missing in the output.
1874
1875 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1876 less than 128) in its various bitmaps. However, there is a facility for
1877 generating tables according to the current locale when PCRE is compiled. It
1878 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1879 characters, are recognized by isspace() and therefore were getting set in
1880 these tables, and indeed these tables seem to approximate to ISO 8859. This
1881 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1882 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1883 which of course cannot start UTF-8 characters. I have changed the code so
1884 that only real ASCII characters (less than 128) and the correct starting
1885 bytes for UTF-8 encodings are set for characters greater than 127 when in
1886 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1887 altogether.)
1888
1889 20. Added the /T option to pcretest so as to be able to run tests with non-
1890 standard character tables, thus making it possible to include the tests
1891 used for 19 above in the standard set of tests.
1892
1893 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1894 reference to a subpattern the other side of a comment that contains an
1895 opening parenthesis caused either an internal compiling error, or a
1896 reference to the wrong subpattern.
1897
1898
1899 Version 8.02 19-Mar-2010
1900 ------------------------
1901
1902 1. The Unicode data tables have been updated to Unicode 5.2.0.
1903
1904 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1905 configured.
1906
1907 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1908 original author of that file, following a query about its status.
1909
1910 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1911 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1912
1913 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1914 quantifier applied to a forward-referencing subroutine call, could compile
1915 incorrect code or give the error "internal error: previously-checked
1916 referenced subpattern not found".
1917
1918 6. Both MS Visual Studio and Symbian OS have problems with initializing
1919 variables to point to external functions. For these systems, therefore,
1920 pcre_malloc etc. are now initialized to local functions that call the
1921 relevant global functions.
1922
1923 7. There were two entries missing in the vectors called coptable and poptable
1924 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1925 I've fixed the data, and added a kludgy way of testing at compile time that
1926 the lengths are correct (equal to the number of opcodes).
1927
1928 8. Following on from 7, I added a similar kludge to check the length of the
1929 eint vector in pcreposix.c.
1930
1931 9. Error texts for pcre_compile() are held as one long string to avoid too
1932 much relocation at load time. To find a text, the string is searched,
1933 counting zeros. There was no check for running off the end of the string,
1934 which could happen if a new error number was added without updating the
1935 string.
1936
1937 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1938
1939 11. \K was not working if it appeared in an atomic group or in a group that
1940 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1941 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1942 the assertion is positive, but not if it is negative.
1943
1944 12. Change 11 fortuitously reduced the size of the stack frame used in the
1945 "match()" function of pcre_exec.c by one pointer. Forthcoming
1946 implementation of support for (*MARK) will need an extra pointer on the
1947 stack; I have reserved it now, so that the stack frame size does not
1948 decrease.
1949
1950 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1951 item in branch that calls a recursion is a subroutine call - as in the
1952 second branch in the above example - was incorrectly given the compile-
1953 time error "recursive call could loop indefinitely" because pcre_compile()
1954 was not correctly checking the subroutine for matching a non-empty string.
1955
1956 14. The checks for overrunning compiling workspace could trigger after an
1957 overrun had occurred. This is a "should never occur" error, but it can be
1958 triggered by pathological patterns such as hundreds of nested parentheses.
1959 The checks now trigger 100 bytes before the end of the workspace.
1960
1961 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1962
1963
1964 Version 8.01 19-Jan-2010
1965 ------------------------
1966
1967 1. If a pattern contained a conditional subpattern with only one branch (in
1968 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1969 computed the wrong minimum data length (which is of course zero for such
1970 subpatterns). This could cause incorrect "no match" results.
1971
1972 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1973 the pattern is reset in the first branch, pcre_compile() failed with
1974 "internal error: code overflow at offset...". This happened only when
1975 the reset was to the original external option setting. (An optimization
1976 abstracts leading options settings into an external setting, which was the
1977 cause of this.)
1978
1979 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1980 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1981 assertion pattern did not match (meaning that the assertion was true), it
1982 was incorrectly treated as false if the SKIP had been reached during the
1983 matching. This also applied to assertions used as conditions.
1984
1985 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1986 assertion subpattern, including such a pattern used as a condition,
1987 unpredictable results occurred, instead of the error return
1988 PCRE_ERROR_DFA_UITEM.
1989
1990 5. The C++ GlobalReplace function was not working like Perl for the special
1991 situation when an empty string is matched. It now does the fancy magic
1992 stuff that is necessary.
1993
1994 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1995 removed. (These were left over from very, very early versions of PCRE.)
1996
1997 7. Some cosmetic changes to the code to make life easier when compiling it
1998 as part of something else:
1999
2000 (a) Change DEBUG to PCRE_DEBUG.
2001
2002 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2003 called "current" as "current_branch", to prevent a collision with the
2004 Linux macro when compiled as a kernel module.
2005
2006 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2007 prevent a collision with the Linux macro when compiled as a kernel
2008 module.
2009
2010 8. In pcre_compile() there are some checks for integer overflows that used to
2011 cast potentially large values to (double). This has been changed to that
2012 when building, a check for int64_t is made, and if it is found, it is used
2013 instead, thus avoiding the use of floating point arithmetic. (There is no
2014 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2015 double.
2016
2017 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2018 2005 (difference between two addresses compared to an unsigned value).
2019
2020 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2021 custom one, because of the following reported problem in Windows:
2022
2023 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2024 under Win32.
2025 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2026 therefore missing the function definition.
2027 - The compiler thus generates a "C" signature for the test function.
2028 - The linker fails to find the "C" function.
2029 - PCRE fails to configure if asked to do so against libbz2.
2030
2031 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2032 messages were output:
2033
2034 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2035 rerunning libtoolize, to keep the correct libtool macros in-tree.
2036 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2037
2038 I have done both of these things.
2039
2040 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2041 most of the time, it *can* run out if it is given a pattern that contains a
2042 runaway infinite recursion. I updated the discussion in the pcrestack man
2043 page.
2044
2045 13. Now that we have gone to the x.xx style of version numbers, the minor
2046 version may start with zero. Using 08 or 09 is a bad idea because users
2047 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2048 interpreted as invalid octal numbers. I've updated the previous comment in
2049 configure.ac, and also added a check that gives an error if 08 or 09 are
2050 used.
2051
2052 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2053 causing partial matching to fail when the end of the subject matched \W
2054 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2055
2056 15. There were some discrepancies between the declarations in pcre_internal.h
2057 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2058 their definitions. The declarations used "const uschar *" and the
2059 definitions used USPTR. Even though USPTR is normally defined as "const
2060 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2061 reported that: "This difference in casting confuses some C++ compilers, for
2062 example, SunCC recognizes above declarations as different functions and
2063 generates broken code for hbpcre." I have changed the declarations to use
2064 USPTR.
2065
2066 16. GNU libtool is named differently on some systems. The autogen.sh script now
2067 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2068 (FreeBSD).
2069
2070 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2071 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2072 comment: "Figure out how to create a longlong from a string: strtoll and
2073 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2074 instance, but it only takes 2 args instead of 3!"
2075
2076 18. A subtle bug concerned with back references has been fixed by a change of
2077 specification, with a corresponding code fix. A pattern such as
2078 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2079 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2080 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2081 same bug. Such groups have to be quantified to be useful, or contained
2082 inside another quantified group. (If there's no repetition, the reference
2083 can never match.) The problem arises because, having left the group and
2084 moved on to the rest of the pattern, a later failure that backtracks into
2085 the group uses the captured value from the final iteration of the group
2086 rather than the correct earlier one. I have fixed this in PCRE by forcing
2087 any group that contains a reference to itself to be an atomic group; that
2088 is, there cannot be any backtracking into it once it has completed. This is
2089 similar to recursive and subroutine calls.
2090
2091
2092 Version 8.00 19-Oct-09
2093 ----------------------
2094
2095 1. The table for translating pcre_compile() error codes into POSIX error codes
2096 was out-of-date, and there was no check on the pcre_compile() error code
2097 being within the table. This could lead to an OK return being given in
2098 error.
2099
2100 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2101 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2102 in a Windows environment.
2103
2104 3. The pcregrep --count option prints the count for each file even when it is
2105 zero, as does GNU grep. However, pcregrep was also printing all files when
2106 --files-with-matches was added. Now, when both options are given, it prints
2107 counts only for those files that have at least one match. (GNU grep just
2108 prints the file name in this circumstance, but including the count seems
2109 more useful - otherwise, why use --count?) Also ensured that the
2110 combination -clh just lists non-zero counts, with no names.
2111
2112 4. The long form of the pcregrep -F option was incorrectly implemented as
2113 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2114 but it seems right to fix it, and I didn't think it was worth preserving
2115 the old behaviour.
2116
2117 5. The command line items --regex=pattern and --regexp=pattern were not
2118 recognized by pcregrep, which required --regex pattern or --regexp pattern
2119 (with a space rather than an '='). The man page documented the '=' forms,
2120 which are compatible with GNU grep; these now work.
2121
2122 6. No libpcreposix.pc file was created for pkg-config; there was just
2123 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2124
2125 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2126 when UCP support is not needed, by modifying the Python script that
2127 generates it from Unicode data files. This should not matter if the module
2128 is correctly used as a library, but I received one complaint about 50K of
2129 unwanted data. My guess is that the person linked everything into his
2130 program rather than using a library. Anyway, it does no harm.
2131
2132 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2133 was a minimum greater than 1 for a wide character in a possessive
2134 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2135 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2136 character. Chaos in the form of incorrect output or a compiling loop could
2137 result.
2138
2139 9. The restrictions on what a pattern can contain when partial matching is
2140 requested for pcre_exec() have been removed. All patterns can now be
2141 partially matched by this function. In addition, if there are at least two
2142 slots in the offset vector, the offset of the earliest inspected character
2143 for the match and the offset of the end of the subject are set in them when
2144 PCRE_ERROR_PARTIAL is returned.
2145
2146 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2147 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2148 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2149 and may be more useful for multi-segment matching.
2150
2151 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2152 used to be given if ever the end of the subject was reached; now it is
2153 given only if matching could not proceed because another character was
2154 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2155 string "Z", which now yields "no match" instead of "partial match". In the
2156 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2157 final character ended with (*FAIL).
2158
2159 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2160 if the pattern had a "must contain" character that was already found in the
2161 earlier partial match, unless partial matching was again requested. For
2162 example, with the pattern /dog.(body)?/, the "must contain" character is
2163 "g". If the first part-match was for the string "dog", restarting with
2164 "sbody" failed. This bug has been fixed.
2165
2166 13. The string returned by pcre_dfa_exec() after a partial match has been
2167 changed so that it starts at the first inspected character rather than the
2168 first character of the match. This makes a difference only if the pattern
2169 starts with a lookbehind assertion or \b or \B (\K is not supported by
2170 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2171 matching functions compatible, and I think it's the right thing to do.
2172
2173 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2174 so that the demonstration program is easily available in environments where
2175 PCRE has not been installed from source.
2176
2177 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2178 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2179 library.
2180
2181 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2182 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2183 is not the first non-POSIX option to be added. Clearly some people find
2184 these options useful.
2185
2186 17. If a caller to the POSIX matching function regexec() passes a non-zero
2187 value for nmatch with a NULL value for pmatch, the value of
2188 nmatch is forced to zero.
2189
2190 18. RunGrepTest did not have a test for the availability of the -u option of
2191 the diff command, as RunTest does. It now checks in the same way as
2192 RunTest, and also checks for the -b option.
2193
2194 19. If an odd number of negated classes containing just a single character
2195 interposed, within parentheses, between a forward reference to a named
2196 subpattern and the definition of the subpattern, compilation crashed with
2197 an internal error, complaining that it could not find the referenced
2198 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2199 [The bug was that it was starting one character too far in when skipping
2200 over the character class, thus treating the ] as data rather than
2201 terminating the class. This meant it could skip too much.]
2202
2203 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2204 /g option in pcretest when the pattern contains \K, which makes it possible
2205 to have an empty string match not at the start, even when the pattern is
2206 anchored. Updated pcretest and pcredemo to use this option.
2207
2208 21. If the maximum number of capturing subpatterns in a recursion was greater
2209 than the maximum at the outer level, the higher number was returned, but
2210 with unset values at the outer level. The correct (outer level) value is
2211 now given.
2212
2213 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2214 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2215 make it do so. The string so far is captured, making this feature
2216 compatible with Perl.
2217
2218 23. The tests have been re-organized, adding tests 11 and 12, to make it
2219 possible to check the Perl 5.10 features against Perl 5.10.
2220
2221 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2222 pattern matches a fixed length string. PCRE did not allow this; now it
2223 does. Neither allows recursion.
2224
2225 25. I finally figured out how to implement a request to provide the minimum
2226 length of subject string that was needed in order to match a given pattern.
2227 (It was back references and recursion that I had previously got hung up
2228 on.) This code has now been added to pcre_study(); it finds a lower bound
2229 to the length of subject needed. It is not necessarily the greatest lower
2230 bound, but using it to avoid searching strings that are too short does give
2231 some useful speed-ups. The value is available to calling programs via
2232 pcre_fullinfo().
2233
2234 26. While implementing 25, I discovered to my embarrassment that pcretest had
2235 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2236 study optimizations had never been tested with that matching function.
2237 Oops. What is worse, even when it was passed study data, there was a bug in
2238 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2239 were also very few tests of studied patterns with pcre_dfa_exec().
2240
2241 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2242 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2243 on the other side of the coin, they are no longer allowed to have different
2244 names, because these cannot be distinguished in PCRE, and this has caused
2245 confusion. (This is a difference from Perl.)
2246
2247 28. When duplicate subpattern names are present (necessarily with different
2248 numbers, as required by 27 above), and a test is made by name in a
2249 conditional pattern, either for a subpattern having been matched, or for
2250 recursion in such a pattern, all the associated numbered subpatterns are
2251 tested, and the overall condition is true if the condition is true for any
2252 one of them. This is the way Perl works, and is also more like the way
2253 testing by number works.
2254
2255
2256 Version 7.9 11-Apr-09
2257 ---------------------
2258
2259 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2260 (pcretest), all targets were linked against these libraries. This included
2261 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2262 libraries. This caused unwanted dependencies to be created. This problem
2263 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2264 pcretest is linked with readline.
2265
2266 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2267 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2268 moved outside it again, because FALSE and TRUE are already defined in AIX,
2269 but BOOL is not.
2270
2271 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2272 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2273
2274 4. The pcregrep documentation said spaces were inserted as well as colons (or
2275 hyphens) following file names and line numbers when outputting matching
2276 lines. This is not true; no spaces are inserted. I have also clarified the
2277 wording for the --colour (or --color) option.
2278
2279 5. In pcregrep, when --colour was used with -o, the list of matching strings
2280 was not coloured; this is different to GNU grep, so I have changed it to be
2281 the same.
2282
2283 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2284 each matching line was coloured. Now it goes on to look for further matches
2285 of any of the test patterns, which is the same behaviour as GNU grep.
2286
2287 7. A pattern that could match an empty string could cause pcregrep to loop; it
2288 doesn't make sense to accept an empty string match in pcregrep, so I have
2289 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2290 seems to be how GNU grep behaves. [But see later change 40 for release
2291 8.33.]
2292
2293 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2294 start or after a newline", because the conditional assertion was not being
2295 correctly handled. The rule now is that both the assertion and what follows
2296 in the first alternative must satisfy the test.
2297
2298 9. If auto-callout was enabled in a pattern with a conditional group whose
2299 condition was an assertion, PCRE could crash during matching, both with
2300 pcre_exec() and pcre_dfa_exec().
2301
2302 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2303 used for matching.
2304
2305 11. Unicode property support in character classes was not working for
2306 characters (bytes) greater than 127 when not in UTF-8 mode.
2307
2308 12. Added the -M command line option to pcretest.
2309
2310 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2311
2312 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2313
2314 16. Added comments and documentation about mis-use of no_arg in the C++
2315 wrapper.
2316
2317 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2318 from Martin Jerabek that uses macro names for all relevant character and
2319 string constants.
2320
2321 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2322 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2323 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2324 these, but not everybody uses configure.
2325
2326 19. A conditional group that had only one branch was not being correctly
2327 recognized as an item that could match an empty string. This meant that an
2328 enclosing group might also not be so recognized, causing infinite looping
2329 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2330 with the subject "ab", where knowledge that the repeated group can match
2331 nothing is needed in order to break the loop.
2332
2333 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2334 exec(), but without supplying a callout function, matching went wrong.
2335
2336 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2337 leak if the size of the offset vector was greater than 30. When the vector
2338 is smaller, the saved offsets during recursion go onto a local stack
2339 vector, but for larger vectors malloc() is used. It was failing to free
2340 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2341 error, in fact).
2342
2343 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2344 heapframe that is used only when UTF-8 support is enabled. This caused no
2345 problem, but was untidy.
2346
2347 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2348 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2349 included within another project.
2350
2351 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2352 slightly modified by me:
2353
2354 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2355 not building pcregrep.
2356
2357 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2358 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2359
2360 25. Forward references, both numeric and by name, in patterns that made use of
2361 duplicate group numbers, could behave incorrectly or give incorrect errors,
2362 because when scanning forward to find the reference group, PCRE was not
2363 taking into account the duplicate group numbers. A pattern such as
2364 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2365
2366 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2367 the feature of a custom pointer more persuasive (as requested by a user).
2368
2369 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2370 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2371 pre-defined.
2372
2373 28. Added support for (*UTF8) at the start of a pattern.
2374
2375 29. Arrange for flags added by the "release type" setting in CMake to be shown
2376 in the configuration summary.
2377
2378
2379 Version 7.8 05-Sep-08
2380 ---------------------
2381
2382 1. Replaced UCP searching code with optimized version as implemented for Ad
2383 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2384 stage table and inline lookup instead of a function, giving speed ups of 2
2385 to 5 times on some simple patterns that I tested. Permission was given to
2386 distribute the MultiStage2.py script that generates the tables (it's not in
2387 the tarball, but is in the Subversion repository).
2388
2389 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2390 scripts.
2391
2392 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2393 a group with a zero qualifier. The result of the study could be incorrect,
2394 or the function might crash, depending on the pattern.
2395
2396 4. Caseless matching was not working for non-ASCII characters in back
2397 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2398 It now works when Unicode Property Support is available.
2399
2400 5. In pcretest, an escape such as \x{de} in the data was always generating
2401 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2402 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2403 truncation.
2404
2405 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2406
2407 7. Added two (int) casts to pcregrep when printing the difference of two
2408 pointers, in case they are 64-bit values.
2409
2410 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2411 test 2 if it fails.
2412
2413 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2414 and a #define of that name to empty if it is not externally set. This is to
2415 allow users of MSVC to set it if necessary.
2416
2417 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2418 the convenience functions in the pcre_get.c source file.
2419
2420 11. An option change at the start of a pattern that had top-level alternatives
2421 could cause overwriting and/or a crash. This command provoked a crash in
2422 some environments:
2423
2424 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2425
2426 This potential security problem was recorded as CVE-2008-2371.
2427
2428 12. For a pattern where the match had to start at the beginning or immediately
2429 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2430 pcre_dfa_exec() could read past the end of the passed subject if there was
2431 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2432 pcretest so that it places the subject at the end of its malloc-ed buffer.
2433
2434 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2435 exec() might read past the end of the data buffer in UTF-8 mode.
2436
2437 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2438 the data contained the byte 0x85 as part of a UTF-8 character within its
2439 first line. This applied both to normal and DFA matching.
2440
2441 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2442 /^[^d]*?$/8 failed to match "abc".
2443
2444 16. Added a missing copyright notice to pcrecpp_internal.h.
2445
2446 17. Make it more clear in the documentation that values returned from
2447 pcre_exec() in ovector are byte offsets, not character counts.
2448
2449 18. Tidied a few places to stop certain compilers from issuing warnings.
2450
2451 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2452 supplied by Stefan Weber. I made a further small update for 7.8 because
2453 there is a change of source arrangements: the pcre_searchfuncs.c module is
2454 replaced by pcre_ucd.c.
2455
2456
2457 Version 7.7 07-May-08
2458 ---------------------
2459
2460 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2461 a string to a long long, pretend we don't even have a long long." This is
2462 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2463
2464 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2465 pre-7.6 versions, which defined a global no_arg variable instead of putting
2466 it in the RE class. (See also #8 below.)
2467
2468 3. Remove a line of dead code, identified by coverity and reported by Nuno
2469 Lopes.
2470
2471 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2472
2473 (1) The include/exclude patterns were being applied to the whole pathnames
2474 of files, instead of just to the final components.
2475
2476 (2) If there was more than one level of directory, the subdirectories were
2477 skipped unless they satisfied the include/exclude conditions. This is
2478 inconsistent with GNU grep (and could even be seen as contrary to the
2479 pcregrep specification - which I improved to make it absolutely clear).
2480 The action now is always to scan all levels of directory, and just
2481 apply the include/exclude patterns to regular files.
2482
2483 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2484 --exclude_dir in the tests to avoid scanning .svn directories.
2485
2486 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2487 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2488 doesn't support NULs in patterns.
2489
2490 7. Added some missing "const"s to declarations of static tables in
2491 pcre_compile.c and pcre_dfa_exec.c.
2492
2493 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2494 caused by fix #2 above. (Subsequently also a second patch to fix the
2495 first patch. And a third patch - this was a messy problem.)
2496
2497 9. Applied Craig's patch to remove the use of push_back().
2498
2499 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2500 matching function regexec().
2501
2502 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2503 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2504 references. PCRE supports relative numbers with this syntax (I don't think
2505 Oniguruma does).
2506
2507 12. Previously, a group with a zero repeat such as (...){0} was completely
2508 omitted from the compiled regex. However, this means that if the group
2509 was called as a subroutine from elsewhere in the pattern, things went wrong
2510 (an internal error was given). Such groups are now left in the compiled
2511 pattern, with a new opcode that causes them to be skipped at execution
2512 time.
2513
2514 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2515 to the way PCRE behaves:
2516
2517 (a) A lone ] character is dis-allowed (Perl treats it as data).
2518
2519 (b) A back reference to an unmatched subpattern matches an empty string
2520 (Perl fails the current match path).
2521
2522 (c) A data ] in a character class must be notated as \] because if the
2523 first data character in a class is ], it defines an empty class. (In
2524 Perl it is not possible to have an empty class.) The empty class []
2525 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2526 The negative empty class [^] matches any one character, independently
2527 of the DOTALL setting.
2528
2529 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2530 non-existent subpattern following a character class starting with ']' and
2531 containing () gave an internal compiling error instead of "reference to
2532 non-existent subpattern". Fortunately, when the pattern did exist, the
2533 compiled code was correct. (When scanning forwards to check for the
2534 existence of the subpattern, it was treating the data ']' as terminating
2535 the class, so got the count wrong. When actually compiling, the reference
2536 was subsequently set up correctly.)
2537
2538 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2539 it was being rejected as not supported by pcre_dfa_exec(), even though
2540 other assertions are supported. I have made pcre_dfa_exec() support
2541 (*FAIL).
2542
2543 16. The implementation of 13c above involved the invention of a new opcode,
2544 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2545 cannot be changed at match time, I realized I could make a small
2546 improvement to matching performance by compiling OP_ALLANY instead of
2547 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2548 on the OP_ANY path.
2549
2550 17. Compiling pcretest on Windows with readline support failed without the
2551 following two fixes: (1) Make the unistd.h include conditional on
2552 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2553
2554 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2555 ncurses library to be included for pcretest when ReadLine support is
2556 requested, but also to allow for it to be overridden. This patch came from
2557 Daniel Bergström.
2558
2559 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2560 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2561 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2562 spotting this.
2563
2564
2565 Version 7.6 28-Jan-08
2566 ---------------------
2567
2568 1. A character class containing a very large number of characters with
2569 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2570 overflow.
2571
2572 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2573 HAVE_LONG_LONG is not defined.
2574
2575 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2576 bring them up to date and include new features. This patch includes:
2577
2578 - Fixed PH's badly added libz and libbz2 support.
2579 - Fixed a problem with static linking.
2580 - Added pcredemo. [But later removed - see 7 below.]
2581 - Fixed dftables problem and added an option.
2582 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2583 HAVE_LONG_LONG.
2584 - Added readline support for pcretest.
2585 - Added an listing of the option settings after cmake has run.
2586
2587 4. A user submitted a patch to Makefile that makes it easy to create
2588 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2589 Makefile.am that cause it to include this special target, without
2590 affecting anything else. Note that the same mingw target plus all
2591 the other distribution libraries and programs are now supported
2592 when configuring with CMake (see 6 below) instead of with
2593 Configure/Make.
2594
2595 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2596 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2597 exported in the Windows port". It has not yet been confirmed that the patch
2598 solves the problem, but it does no harm.
2599
2600 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2601 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2602 with CMake, and also correct the comment about stack recursion.
2603
2604 7. Remove the automatic building of pcredemo from the ./configure system and
2605 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2606 of a program that users should build themselves after PCRE is installed, so
2607 building it automatically is not really right. What is more, it gave
2608 trouble in some build environments.
2609
2610 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2611
2612
2613 Version 7.5 10-Jan-08
2614 ---------------------
2615
2616 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2617 values in parens when parsing an RE using the C++ wrapper."
2618
2619 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2620 Characters greater than 255 were excluded from the class instead of being
2621 included.
2622
2623 3. The same bug as (2) above applied to negated POSIX classes such as
2624 [:^space:].
2625
2626 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2627 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2628 I have changed it.
2629
2630 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2631 first named subpattern) and a construct such as (?&a) would reference the
2632 first named subpattern whose name started with "a" (in other words, the
2633 length check was missing). Both these problems are fixed. "Subpattern name
2634 expected" is now given for (?&) (a zero-length name), and this patch also
2635 makes it give the same error for \k'' (previously it complained that that
2636 was a reference to a non-existent subpattern).
2637
2638 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2639 this is right because (?- can be followed by option settings as well as by
2640 digits. I have, however, made the messages clearer.
2641
2642 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2643 than the number used in the conditional) now cause a compile-time error.
2644 This is actually not compatible with Perl, which accepts such patterns, but
2645 treats the conditional as always being FALSE (as PCRE used to), but it
2646 seems to me that giving a diagnostic is better.
2647
2648 8. Change "alphameric" to the more common word "alphanumeric" in comments
2649 and messages.
2650
2651 9. Fix two occurrences of "backslash" in comments that should have been
2652 "backspace".
2653
2654 10. Remove two redundant lines of code that can never be obeyed (their function
2655 was moved elsewhere).
2656
2657 11. The program that makes PCRE's Unicode character property table had a bug
2658 which caused it to generate incorrect table entries for sequences of
2659 characters that have the same character type, but are in different scripts.
2660 It amalgamated them into a single range, with the script of the first of
2661 them. In other words, some characters were in the wrong script. There were
2662 thirteen such cases, affecting characters in the following ranges:
2663
2664 U+002b0 - U+002c1
2665 U+0060c - U+0060d
2666 U+0061e - U+00612
2667 U+0064b - U+0065e
2668 U+0074d - U+0076d
2669 U+01800 - U+01805
2670 U+01d00 - U+01d77
2671 U+01d9b - U+01dbf
2672 U+0200b - U+0200f
2673 U+030fc - U+030fe
2674 U+03260 - U+0327f
2675 U+0fb46 - U+0fbb1
2676 U+10450 - U+1049d
2677
2678 12. The -o option (show only the matching part of a line) for pcregrep was not
2679 compatible with GNU grep in that, if there was more than one match in a
2680 line, it showed only the first of them. It now behaves in the same way as
2681 GNU grep.
2682
2683 13. If the -o and -v options were combined for pcregrep, it printed a blank
2684 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2685 does the same. The return code can be used to tell if there were any
2686 non-matching lines.
2687
2688 14. Added --file-offsets and --line-offsets to pcregrep.
2689
2690 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2691 infinitely looping recursion. The bug was that positive lookaheads were not
2692 being skipped when checking for a possible empty match (negative lookaheads
2693 and both kinds of lookbehind were skipped).
2694
2695 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2696 inclusion of <windows.h> to before rather than after the definition of
2697 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2698
2699 17. Specifying a possessive quantifier with a specific limit for a Unicode
2700 character property caused pcre_compile() to compile bad code, which led at
2701 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2702 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2703 caused the error; without that there was no problem.
2704
2705 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2706
2707 19. Added --enable-pcretest-libreadline.
2708
2709 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2710 RE::GlobalReplace(). As a result, the number of replacements returned was
2711 double what it should be. I removed one of the increments, but Craig sent a
2712 later patch that removed the other one (the right fix) and added unit tests
2713 that check the return values (which was not done before).
2714
2715 21. Several CMake things:
2716
2717 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2718 the names libpcre and libpcreposix, not just pcre and pcreposix.
2719
2720 (2) The above change means that pcretest and pcregrep are now correctly
2721 linked with the newly-built libraries, not previously installed ones.
2722
2723 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2724
2725 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2726 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2727 UTF-8 newline character). The key issue is that the pattern starts .*;
2728 this means that the match must be either at the beginning, or after a
2729 newline. The bug was in the code for advancing after a failed match and
2730 checking that the new position followed a newline. It was not taking
2731 account of UTF-8 characters correctly.
2732
2733 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2734 character classes. PCRE was not treating the sequence [:...:] as a
2735 character class unless the ... were all letters. Perl, however, seems to
2736 allow any characters between [: and :], though of course it rejects as
2737 unknown any "names" that contain non-letters, because all the known class
2738 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2739 for example, whereas PCRE did not - it did not recognize a POSIX character
2740 class. This seemed a bit dangerous, so the code has been changed to be
2741 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2742 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2743 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2744 Perl does, and where it didn't before.
2745
2746 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2747 Windows environments %n is disabled by default.
2748
2749
2750 Version 7.4 21-Sep-07
2751 ---------------------
2752
2753 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2754 means that a class such as [\s] counted as "explicit reference to CR or
2755 LF". That isn't really right - the whole point of the change was to try to
2756 help when there was an actual mention of one of the two characters. So now
2757 the change happens only if \r or \n (or a literal CR or LF) character is
2758 encountered.
2759
2760 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2761 of both had grown to the point where there were only 3 bits left.
2762 Fortunately, there was spare space in the data structure, and so I have
2763 moved the internal flags into a new 16-bit field to free up more option
2764 bits.
2765
2766 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2767 but did not set the internal JCHANGED flag - either of these is enough to
2768 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2769 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2770 start sets both bits.
2771
2772 4. Added options (at build time, compile time, exec time) to change \R from
2773 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2774
2775 5. doc/pcresyntax.html was missing from the distribution.
2776
2777 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2778 compatibility, even though it is no longer used.
2779
2780 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2781 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2782 windows.h file is present (where different names are used). [This was
2783 reversed later after testing - see 16 below.]
2784
2785 8. Changed all #include <config.h> to #include "config.h". There were also
2786 some further <pcre.h> cases that I changed to "pcre.h".
2787
2788 9. When pcregrep was used with the --colour option, it missed the line ending
2789 sequence off the lines that it output.
2790
2791 10. It was pointed out to me that arrays of string pointers cause lots of
2792 relocations when a shared library is dynamically loaded. A technique of
2793 using a single long string with a table of offsets can drastically reduce
2794 these. I have refactored PCRE in four places to do this. The result is
2795 dramatic:
2796
2797 Originally: 290
2798 After changing UCP table: 187
2799 After changing error message table: 43
2800 After changing table of "verbs" 36
2801 After changing table of Posix names 22
2802
2803 Thanks to the folks working on Gregex for glib for this insight.
2804
2805 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2806 unicode-properties was also set.
2807
2808 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2809
2810 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2811 checked only for CRLF.
2812
2813 14. Added casts to pcretest.c to avoid compiler warnings.
2814
2815 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2816
2817 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2818 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2819 entirely. This removes changes made in 7 above.
2820
2821 17. The CMake files have been updated, and there is now more information about
2822 building with CMake in the NON-UNIX-USE document.
2823
2824
2825 Version 7.3 28-Aug-07
2826 ---------------------
2827
2828 1. In the rejigging of the build system that eventually resulted in 7.1, the
2829 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2830 brackets there is not right, since it causes compilers to look for an
2831 installed pcre.h, not the version that is in the source that is being
2832 compiled (which of course may be different). I have changed it back to:
2833
2834 #include "pcre.h"
2835
2836 I have a vague recollection that the change was concerned with compiling in
2837 different directories, but in the new build system, that is taken care of
2838 by the VPATH setting the Makefile.
2839
2840 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2841 when the subject happened to end in the byte 0x85 (e.g. if the last
2842 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2843 characters but of course it shouldn't be taken as a newline when it is part
2844 of another character. The bug was that, for an unlimited repeat of . in
2845 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2846 characters when looking for a newline.
2847
2848 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2849
2850 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2851 in debug output.
2852
2853 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2854 long printing in the pcrecpp unittest when running under MinGW.
2855
2856 6. ESC_K was left out of the EBCDIC table.
2857
2858 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2859 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2860 limit also applies to "virtual nesting" when a pattern is recursive, and in
2861 this case 1000 isn't so big. I have been able to remove this limit at the
2862 expense of backing off one optimization in certain circumstances. Normally,
2863 when pcre_exec() would call its internal match() function recursively and
2864 immediately return the result unconditionally, it uses a "tail recursion"
2865 feature to save stack. However, when a subpattern that can match an empty
2866 string has an unlimited repetition quantifier, it no longer makes this
2867 optimization. That gives it a stack frame in which to save the data for
2868 checking that an empty string has been matched. Previously this was taken
2869 from the 1000-entry workspace that had been reserved. So now there is no
2870 explicit limit, but more stack is used.
2871
2872 8. Applied Daniel's patches to solve problems with the import/export magic
2873 syntax that is required for Windows, and which was going wrong for the
2874 pcreposix and pcrecpp parts of the library. These were overlooked when this
2875 problem was solved for the main library.
2876
2877 9. There were some crude static tests to avoid integer overflow when computing
2878 the size of patterns that contain repeated groups with explicit upper
2879 limits. As the maximum quantifier is 65535, the maximum group length was
2880 set at 30,000 so that the product of these two numbers did not overflow a
2881 32-bit integer. However, it turns out that people want to use groups that
2882 are longer than 30,000 bytes (though not repeat them that many times).
2883 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2884 made it possible to implement the integer overflow checks in a much more
2885 dynamic way, which I have now done. The artificial limitation on group
2886 length has been removed - we now have only the limit on the total length of
2887 the compiled pattern, which depends on the LINK_SIZE setting.
2888
2889 10. Fixed a bug in the documentation for get/copy named substring when
2890 duplicate names are permitted. If none of the named substrings are set, the
2891 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2892 empty string.
2893
2894 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2895 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2896 because the ] is interpreted as the first data character and the
2897 terminating ] is not found. PCRE has been made compatible with Perl in this
2898 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2899 cause memory overwriting.
2900
2901 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2902 string has been matched (to stop an infinite loop). It was not recognizing
2903 a conditional subpattern that could match an empty string if that
2904 subpattern was within another subpattern. For example, it looped when
2905 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2906 condition was not nested. This bug has been fixed.
2907
2908 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2909 past the start of the subject in the presence of bytes with the top bit
2910 set, for example "\x8aBCD".
2911
2912 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2913 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2914
2915 14. Optimized (?!) to (*FAIL).
2916
2917 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2918 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2919 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2920 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2921 does: it's just the validity check that is more restrictive.
2922
2923 16. Inserted checks for integer overflows during escape sequence (backslash)
2924 processing, and also fixed erroneous offset values for syntax errors during
2925 backslash processing.
2926
2927 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2928 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2929
2930 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2931 caused an overrun.
2932
2933 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2934 something other than just ASCII characters) inside a group that had an
2935 unlimited repeat caused a loop at compile time (while checking to see
2936 whether the group could match an empty string).
2937
2938 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2939 [\P{Any}] did so. (Error in the code for printing property names.)
2940
2941 21. An orphan \E inside a character class could cause a crash.
2942
2943 22. A repeated capturing bracket such as (A)? could cause a wild memory
2944 reference during compilation.
2945
2946 23. There are several functions in pcre_compile() that scan along a compiled
2947 expression for various reasons (e.g. to see if it's fixed length for look
2948 behind). There were bugs in these functions when a repeated \p or \P was
2949 present in the pattern. These operators have additional parameters compared
2950 with \d, etc, and these were not being taken into account when moving along
2951 the compiled data. Specifically:
2952
2953 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2954 length.
2955
2956 (b) An item such as \pL+ within a repeated group could cause crashes or
2957 loops.
2958
2959 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2960 "reference to non-existent subpattern" error.
2961
2962 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2963
2964 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2965 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2966
2967 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2968
2969 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2970 character were causing crashes (broken optimization).
2971
2972 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2973 \p or \P) caused a compile-time loop.
2974
2975 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2976 break. For example, the unstudied pattern [\r\n]A does not match the string
2977 "\r\nA" because change 7.0/46 below moves the current point on by two
2978 characters after failing to match at the start. However, the pattern \nA
2979 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2980 the same is true. There doesn't seem any very clean way out of this, but
2981 what I have chosen to do makes the common cases work: PCRE now takes note
2982 of whether there can be an explicit match for \r or \n anywhere in the
2983 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2984 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2985 pattern has explicit CR or LF references.
2986
2987 29. Added (*CR) etc for changing newline setting at start of pattern.
2988
2989
2990 Version 7.2 19-Jun-07
2991 ---------------------
2992
2993 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2994 which is apparently normally available under Windows.
2995
2996 2. Re-jig the pcregrep tests with different newline settings in an attempt
2997 to make them independent of the local environment's newline setting.
2998
2999 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3000
3001 4. Some of the "internals" tests were previously cut out when the link size
3002 was not 2, because the output contained actual offsets. The recent new
3003 "Z" feature of pcretest means that these can be cut out, making the tests
3004 usable with all link sizes.
3005
3006 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3007 stack recursion. This gives a massive performance boost under BSD, but just
3008 a small improvement under Linux. However, it saves one field in the frame
3009 in all cases.
3010
3011 6. Added more features from the forthcoming Perl 5.10:
3012
3013 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3014 recursion call. It refers to the nth most recently opened parentheses.
3015
3016 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3017 to be opened parentheses.
3018
3019 (c) Conditions that refer to capturing parentheses can be specified
3020 relatively, for example, (?(-2)... or (?(+3)...
3021
3022 (d) \K resets the start of the current match so that everything before
3023 is not part of it.
3024
3025 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3026
3027 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3028 reference syntax.
3029
3030 (g) (?| introduces a group in which the numbering of parentheses in each
3031 alternative starts with the same number.
3032
3033 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3034
3035 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3036 PCRE_INFO_JCHANGED.
3037
3038 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3039 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3040 for detecting groups that can match an empty string.
3041
3042 9. A pattern with a very large number of alternatives (more than several
3043 hundred) was running out of internal workspace during the pre-compile
3044 phase, where pcre_compile() figures out how much memory will be needed. A
3045 bit of new cunning has reduced the workspace needed for groups with
3046 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3047 workspace instead of running out of the 4096 that are available.
3048
3049 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3050
3051 11. Applied patch from Google to remove an optimization that didn't quite work.
3052 The report of the bug said:
3053
3054 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3055 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3056 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3057
3058 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3059 it matched the wrong number of bytes.
3060
3061
3062 Version 7.1 24-Apr-07
3063 ---------------------
3064
3065 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3066 that is more "standard", making use of automake and other Autotools. There
3067 is some re-arrangement of the files and adjustment of comments consequent
3068 on this.
3069
3070 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3071 for recursive directory scanning broke on some systems because the files
3072 are not scanned in any specific order and on different systems the order
3073 was different. A call to "sort" has been inserted into RunGrepTest for the
3074 approprate test as a short-term fix. In the longer term there may be an
3075 alternative.
3076
3077 3. I had an email from Eric Raymond about problems translating some of PCRE's
3078 man pages to HTML (despite the fact that I distribute HTML pages, some
3079 people do their own conversions for various reasons). The problems
3080 concerned the use of low-level troff macros .br and .in. I have therefore
3081 removed all such uses from the man pages (some were redundant, some could
3082 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3083 HTML has been updated to handle .nf/.fi and to complain if it encounters
3084 .br or .in.
3085
3086 4. Updated comments in configure.ac that get placed in config.h.in and also
3087 arranged for config.h to be included in the distribution, with the name
3088 config.h.generic, for the benefit of those who have to compile without
3089 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3090
3091 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3092 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3093 makevp.bat for the current PCRE, using the additional files
3094 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3095
3096 6. A Windows user reported a minor discrepancy with test 2, which turned out
3097 to be caused by a trailing space on an input line that had got lost in his
3098 copy. The trailing space was an accident, so I've just removed it.
3099
3100 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3101 that is needed.
3102
3103 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3104 as "const" (a) because they are and (b) because it helps the PHP
3105 maintainers who have recently made a script to detect big data structures
3106 in the php code that should be moved to the .rodata section. I remembered
3107 to update Builducptable as well, so it won't revert if ucptable.h is ever
3108 re-created.
3109
3110 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3111 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3112 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3113 support is not required. This saves 1.5-2K of code, which is important in
3114 some applications.
3115
3116 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3117 so as not to refer to the tables, even though these functions will never be
3118 called when UTF-8 support is disabled. Otherwise there are problems with a
3119 shared library.
3120
3121 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3122
3123 (a) It was defining its arguments as char * instead of void *.
3124
3125 (b) It was assuming that all moves were upwards in memory; this was true
3126 a long time ago when I wrote it, but is no longer the case.
3127
3128 The emulated memove() is provided for those environments that have neither
3129 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3130 is clearly not the case, as these two bugs were recently reported.
3131
3132 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3133 and Detrail to create the HTML documentation, the .txt form of the man
3134 pages, and it removes trailing spaces from listed files. It also creates
3135 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3136 case, it wraps all the #defines with #ifndefs. This script should be run
3137 before "make dist".
3138
3139 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3140 with Unicode property support.
3141
3142 (a) For a maximizing quantifier, if the two different cases of the
3143 character were of different lengths in their UTF-8 codings (there are
3144 some cases like this - I found 11), and the matching function had to
3145 back up over a mixture of the two cases, it incorrectly assumed they
3146 were both the same length.
3147
3148 (b) When PCRE was configured to use the heap rather than the stack for
3149 recursion during matching, it was not correctly preserving the data for
3150 the other case of a UTF-8 character when checking ahead for a match
3151 while processing a minimizing repeat. If the check also involved
3152 matching a wide character, but failed, corruption could cause an
3153 erroneous result when trying to check for a repeat of the original
3154 character.
3155
3156 13. Some tidying changes to the testing mechanism:
3157
3158 (a) The RunTest script now detects the internal link size and whether there
3159 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3160 values substituted by "configure". (The RunGrepTest script already did
3161 this for UTF-8.) The configure.ac script no longer substitutes the
3162 relevant variables.
3163
3164 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3165 with length and offset values. This means that the output is different
3166 for different internal link sizes. Test 2 is skipped for link sizes
3167 other than 2 because of this, bypassing the problem. Unfortunately,
3168 there was also a test in test 3 (the locale tests) that used /B and
3169 failed for link sizes other than 2. Rather than cut the whole test out,
3170 I have added a new /Z option to pcretest that replaces the length and
3171 offset values with spaces. This is now used to make test 3 independent
3172 of link size. (Test 2 will be tidied up later.)
3173
3174 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3175 segmentation fault instead of returning the appropriate error message.
3176
3177 15. In multiline mode when the newline sequence was set to "any", the pattern
3178 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3179 This doesn't seem right; it now treats the CRLF combination as the line
3180 ending, and so does not match in that case. It's only a pattern such as ^$
3181 that would hit this one: something like ^ABC$ would have failed after \r
3182 and then tried again after \r\n.
3183
3184 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3185 in an attempt to make files that differ only in their line terminators
3186 compare equal. This works on Linux.
3187
3188 17. Under certain error circumstances pcregrep might try to free random memory
3189 as it exited. This is now fixed, thanks to valgrind.
3190
3191 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3192 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3193 was because its rules for how to advance for /g after matching an empty
3194 string at the end of a line did not allow for this case. They now check for
3195 it specially.
3196
3197 20. pcretest is supposed to handle patterns and data of any length, by
3198 extending its buffers when necessary. It was getting this wrong when the
3199 buffer for a data line had to be extended.
3200
3201 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3202 CRLF as a newline sequence.
3203
3204 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3205 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3206 I have nevertheless tidied it up.
3207
3208 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3209
3210 24. Added a man page for pcre-config.
3211
3212
3213 Version 7.0 19-Dec-06
3214 ---------------------
3215
3216 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3217 moving to gcc 4.1.1.
3218
3219 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3220 sys/time.h, which is documented as needed for this function. It doesn't
3221 seem to matter on Linux, but it showed up on some releases of OS X.
3222
3223 3. It seems that there are systems where bytes whose values are greater than
3224 127 match isprint() in the "C" locale. The "C" locale should be the
3225 default when a C program starts up. In most systems, only ASCII printing
3226 characters match isprint(). This difference caused the output from pcretest
3227 to vary, making some of the tests fail. I have changed pcretest so that:
3228
3229 (a) When it is outputting text in the compiled version of a pattern, bytes
3230 other than 32-126 are always shown as hex escapes.
3231
3232 (b) When it is outputting text that is a matched part of a subject string,
3233 it does the same, unless a different locale has been set for the match
3234 (using the /L modifier). In this case, it uses isprint() to decide.
3235
3236 4. Fixed a major bug that caused incorrect computation of the amount of memory
3237 required for a compiled pattern when options that changed within the
3238 pattern affected the logic of the preliminary scan that determines the
3239 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3240 that the computed length was too small. The symptoms of this bug were
3241 either the PCRE error "internal error: code overflow" from pcre_compile(),
3242 or a glibc crash with a message such as "pcretest: free(): invalid next
3243 size (fast)". Examples of patterns that provoked this bug (shown in
3244 pcretest format) are:
3245
3246 /(?-x: )/x
3247 /(?x)(?-x: \s*#\s*)/
3248 /((?i)[\x{c0}])/8
3249 /(?i:[\x{c0}])/8
3250
3251 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3252 is now done differently.
3253
3254 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3255 wrapper classes; (b) implement a new function in the C++ scanner that is
3256 more efficient than the old way of doing things because it avoids levels of
3257 recursion in the regex matching; (c) add a paragraph to the documentation
3258 for the FullMatch() function.
3259
3260 6. The escape sequence \n was being treated as whatever was defined as
3261 "newline". Not only was this contrary to the documentation, which states
3262 that \n is character 10 (hex 0A), but it also went horribly wrong when
3263 "newline" was defined as CRLF. This has been fixed.
3264
3265 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3266 was being set to -1 for the "end of line" case (supposedly a value that no
3267 character can have). Though this value is never used (the check for end of
3268 line is "zero bytes in current character"), it caused compiler complaints.
3269 I've changed it to 0xffffffff.
3270
3271 8. In pcre_version.c, the version string was being built by a sequence of
3272 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3273 string (as it is for production releases) called a macro with an empty
3274 argument. The C standard says the result of this is undefined. The gcc
3275 compiler treats it as an empty string (which was what was wanted) but it is
3276 reported that Visual C gives an error. The source has been hacked around to
3277 avoid this problem.
3278
3279 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3280 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3281 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3282 of them did).
3283
3284 10. Originally, pcretest opened its input and output without "b"; then I was
3285 told that "b" was needed in some environments, so it was added for release
3286 5.0 to both the input and output. (It makes no difference on Unix-like
3287 systems.) Later I was told that it is wrong for the input on Windows. I've
3288 now abstracted the modes into two macros, to make it easier to fiddle with
3289 them, and removed "b" from the input mode under Windows.
3290
3291 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3292
3293 12. Added -help and --help to pcretest as an official way of being reminded
3294 of the options.
3295
3296 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3297 and pcrecpp.cc because they annoy compilers at high warning levels.
3298
3299 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3300
3301 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3302 scripts are not C programs :-) and which was not noticed because it works
3303 on Linux.
3304
3305 16. pcretest is supposed to handle any length of pattern and data line (as one
3306 line or as a continued sequence of lines) by extending its input buffer if
3307 necessary. This feature was broken for very long pattern lines, leading to
3308 a string of junk being passed to pcre_compile() if the pattern was longer
3309 than about 50K.
3310
3311 17. I have done a major re-factoring of the way pcre_compile() computes the
3312 amount of memory needed for a compiled pattern. Previously, there was code
3313 that made a preliminary scan of the pattern in order to do this. That was
3314 OK when PCRE was new, but as the facilities have expanded, it has become
3315 harder and harder to keep it in step with the real compile phase, and there
3316 have been a number of bugs (see for example, 4 above). I have now found a
3317 cunning way of running the real compile function in a "fake" mode that
3318 enables it to compute how much memory it would need, while actually only
3319 ever using a few hundred bytes of working memory and without too many
3320 tests of the mode. This should make future maintenance and development
3321 easier. A side effect of this work is that the limit of 200 on the nesting
3322 depth of parentheses has been removed (though this was never a serious
3323 limitation, I suspect). However, there is a downside: pcre_compile() now
3324 runs more slowly than before (30% or more, depending on the pattern). I
3325 hope this isn't a big issue. There is no effect on runtime performance.
3326
3327 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3328 newline (only possible for the last line of a file) and it was a
3329 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3330
3331 19. Added additional timing features to pcretest. (1) The -tm option now times
3332 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3333 separate command line item, by a number that specifies the number of
3334 repeats to use when timing. The default is 50000; this gives better
3335 precision, but takes uncomfortably long for very large patterns.
3336
3337 20. Extended pcre_study() to be more clever in cases where a branch of a
3338 subpattern has no definite first character. For example, (a*|b*)[cd] would
3339 previously give no result from pcre_study(). Now it recognizes that the
3340 first character must be a, b, c, or d.
3341
3342 21. There was an incorrect error "recursive call could loop indefinitely" if
3343 a subpattern (or the entire pattern) that was being tested for matching an
3344 empty string contained only one non-empty item after a nested subpattern.
3345 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3346 incorrectly, because the \d was being skipped in the check.
3347
3348 22. The pcretest program now has a new pattern option /B and a command line
3349 option -b, which is equivalent to adding /B to every pattern. This causes
3350 it to show the compiled bytecode, without the additional information that
3351 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3352 is the same as /B/I).
3353
3354 23. A new optimization is now able automatically to treat some sequences such
3355 as a*b as a*+b. More specifically, if something simple (such as a character
3356 or a simple class like \d) has an unlimited quantifier, and is followed by
3357 something that cannot possibly match the quantified thing, the quantifier
3358 is automatically "possessified".
3359
3360 24. A recursive reference to a subpattern whose number was greater than 39
3361 went wrong under certain circumstances in UTF-8 mode. This bug could also
3362 have affected the operation of pcre_study().
3363
3364 25. Realized that a little bit of performance could be had by replacing
3365 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3366
3367 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3368
3369 27. Possessive quantifiers such as a++ were previously implemented by turning
3370 them into atomic groups such as ($>a+). Now they have their own opcodes,
3371 which improves performance. This includes the automatically created ones
3372 from 23 above.
3373
3374 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3375 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3376 the first matched character to be a colon. This applied both to named and
3377 numbered groups.
3378
3379 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3380
3381 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3382 building PCRE on Windows easier, so I have included it in the distribution.
3383
3384 31. There is now a check in pcretest against a ridiculously large number being
3385 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3386 loop, the loop is abandoned.
3387
3388 32. Forward references to subpatterns in conditions such as (?(2)...) where
3389 subpattern 2 is defined later cause pcre_compile() to search forwards in
3390 the pattern for the relevant set of parentheses. This search went wrong
3391 when there were unescaped parentheses in a character class, parentheses
3392 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3393
3394 33. "Subroutine" calls and backreferences were previously restricted to
3395 referencing subpatterns earlier in the regex. This restriction has now
3396 been removed.
3397
3398 34. Added a number of extra features that are going to be in Perl 5.10. On the
3399 whole, these are just syntactic alternatives for features that PCRE had
3400 previously implemented using the Python syntax or my own invention. The
3401 other formats are all retained for compatibility.
3402
3403 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3404 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3405 also .NET compatible.
3406
3407 (b) A recursion or subroutine call to a named group can now be defined as
3408 (?&name) as well as (?P>name).
3409
3410 (c) A backreference to a named group can now be defined as \k<name> or
3411 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3412 5.10, are also .NET compatible.
3413
3414 (d) A conditional reference to a named group can now use the syntax
3415 (?(<name>) or (?('name') as well as (?(name).
3416
3417 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3418 groups (named and numbered) that are never evaluated inline, but can be
3419 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3420 is always false. There may be only one alternative in such a group.
3421
3422 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3423 as the simple (?(R). The condition is true only if the most recent
3424 recursion is that of the given number or name. It does not search out
3425 through the entire recursion stack.
3426
3427 (g) The escape \gN or \g{N} has been added, where N is a positive or
3428 negative number, specifying an absolute or relative reference.
3429
3430 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3431 some "unreachable code" warnings.
3432
3433 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3434 things, this adds five new scripts.
3435
3436 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3437 There were also incompatibilities regarding the handling of \Q..\E inside
3438 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3439 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3440
3441 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3442 matches an empty string, and forcibly breaks the loop. There were bugs in
3443 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3444 against aaaa the result was just "a" rather than "aaaa", for example. Two
3445 separate and independent bugs (that affected different cases) have been
3446 fixed.
3447
3448 39. Refactored the code to abolish the use of different opcodes for small
3449 capturing bracket numbers. This is a tidy that I avoided doing when I
3450 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3451 The new approach is not only tidier, it makes it possible to reduce the
3452 memory needed to fix the previous bug (38).
3453
3454 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3455 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3456 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3457 mode.
3458
3459 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3460 report.
3461
3462 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3463 copy construction and assignment in the C++ wrapper.
3464
3465 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3466 couple of bugs that could have given wrong results in the "--newline=crlf"
3467 case.
3468
3469 44. Added a number of casts and did some reorganization of signed/unsigned int
3470 variables following suggestions from Dair Grant. Also renamed the variable
3471 "this" as "item" because it is a C++ keyword.
3472
3473 45. Arranged for dftables to add
3474
3475 #include "pcre_internal.h"
3476
3477 to pcre_chartables.c because without it, gcc 4.x may remove the array
3478 definition from the final binary if PCRE is built into a static library and
3479 dead code stripping is activated.
3480
3481 46. For an unanchored pattern, if a match attempt fails at the start of a
3482 newline sequence, and the newline setting is CRLF or ANY, and the next two
3483 characters are CRLF, advance by two characters instead of one.
3484
3485
3486 Version 6.7 04-Jul-06
3487 ---------------------
3488
3489 1. In order to handle tests when input lines are enormously long, pcretest has
3490 been re-factored so that it automatically extends its buffers when
3491 necessary. The code is crude, but this _is_ just a test program. The
3492 default size has been increased from 32K to 50K.
3493
3494 2. The code in pcre_study() was using the value of the re argument before
3495 testing it for NULL. (Of course, in any sensible call of the function, it
3496 won't be NULL.)
3497
3498 3. The memmove() emulation function in pcre_internal.h, which is used on
3499 systems that lack both memmove() and bcopy() - that is, hardly ever -
3500 was missing a "static" storage class specifier.
3501
3502 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3503 containing an extended class (one that cannot be represented by a bitmap
3504 because it contains high-valued characters or Unicode property items, e.g.
3505 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3506 pattern, but PCRE should not loop if you do not (it no longer does).
3507 [Detail: two cases were found: (a) a repeated subpattern containing an
3508 extended class; (b) a recursive reference to a subpattern that followed a
3509 previous extended class. It wasn't skipping over the extended class
3510 correctly when UTF-8 mode was not set.]
3511
3512 5. A negated single-character class was not being recognized as fixed-length
3513 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3514 compile error "lookbehind assertion is not fixed length".
3515
3516 6. The RunPerlTest auxiliary script was showing an unexpected difference
3517 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3518 write a Perl script that can interpret lines of an input file either as
3519 byte characters or as UTF-8, which is what "perltest" was being required to
3520 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3521 can't do is switch easily at run time between having the "use utf8;" pragma
3522 or not. In the end, I fudged it by using the RunPerlTest script to insert
3523 "use utf8;" explicitly for the UTF-8 tests.
3524
3525 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3526 the end of the subject string, contrary to the documentation and to what
3527 Perl does. This was true of both matching functions. Now it matches only at
3528 the start of the subject and immediately after *internal* newlines.
3529
3530 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3531 a pointer to an int instead of a pointer to an unsigned long int. This
3532 caused problems on 64-bit systems.
3533
3534 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3535 instance of the 'standard' template library not being so standard".
3536
3537 10. There was no check on the number of named subpatterns nor the maximum
3538 length of a subpattern name. The product of these values is used to compute
3539 the size of the memory block for a compiled pattern. By supplying a very
3540 long subpattern name and a large number of named subpatterns, the size
3541 computation could be caused to overflow. This is now prevented by limiting
3542 the length of names to 32 characters, and the number of named subpatterns
3543 to 10,000.
3544
3545 11. Subpatterns that are repeated with specific counts have to be replicated in
3546 the compiled pattern. The size of memory for this was computed from the
3547 length of the subpattern and the repeat count. The latter is limited to
3548 65535, but there was no limit on the former, meaning that integer overflow
3549 could in principle occur. The compiled length of a repeated subpattern is
3550 now limited to 30,000 bytes in order to prevent this.
3551
3552 12. Added the optional facility to have named substrings with the same name.
3553
3554 13. Added the ability to use a named substring as a condition, using the
3555 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3556 are numbers (not recommended). Forward references are permitted.
3557
3558 14. Added forward references in named backreferences (if you see what I mean).
3559
3560 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3561 pattern could run off the end of the subject. For example, the pattern
3562 "(?s)(.{1,5})"8 did this with the subject "ab".
3563
3564 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3565 PCRE_CASELESS was set when matching characters that were quantified with ?
3566 or *.
3567
3568 17. A character class other than a single negated character that had a minimum
3569 but no maximum quantifier - for example [ab]{6,} - was not handled
3570 correctly by pce_dfa_exec(). It would match only one character.
3571
3572 18. A valid (though odd) pattern that looked like a POSIX character
3573 class but used an invalid character after [ (for example [[,abc,]]) caused
3574 pcre_compile() to give the error "Failed: internal error: code overflow" or
3575 in some cases to crash with a glibc free() error. This could even happen if
3576 the pattern terminated after [[ but there just happened to be a sequence of
3577 letters, a binary zero, and a closing ] in the memory that followed.
3578
3579 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3580 over the years. Originally (before any Unicode support), just the bottom 8
3581 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3582 output from "man perlunicode" includes this:
3583
3584 The regular expression compiler produces polymorphic opcodes. That
3585 is, the pattern adapts to the data and automatically switches to
3586 the Unicode character scheme when presented with Unicode data--or
3587 instead uses a traditional byte scheme when presented with byte
3588 data.
3589
3590 Sadly, a wide octal escape does not cause a switch, and in a string with
3591 no other multibyte characters, these octal escapes are treated as before.
3592 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3593 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3594 Unicode string.
3595
3596 I have not perpetrated such confusion in PCRE. Up till now, it took just
3597 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3598 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3599 translate to the appropriate multibyte character.
3600
3601 29. Applied some refactoring to reduce the number of warnings from Microsoft
3602 and Borland compilers. This has included removing the fudge introduced
3603 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3604 a warning about an unused variable.
3605
3606 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3607 characters since release 4.0, because Perl (from release 5.004) does not.
3608 [Or at least, is documented not to: some releases seem to be in conflict
3609 with the documentation.] However, when a pattern was studied with
3610 pcre_study() and all its branches started with \s, PCRE still included VT
3611 as a possible starting character. Of course, this did no harm; it just
3612 caused an unnecessary match attempt.
3613
3614 22. Removed a now-redundant internal flag bit that recorded the fact that case
3615 dependency changed within the pattern. This was once needed for "required
3616 byte" processing, but is no longer used. This recovers a now-scarce options
3617 bit. Also moved the least significant internal flag bit to the most-
3618 significant bit of the word, which was not previously used (hangover from
3619 the days when it was an int rather than a uint) to free up another bit for
3620 the future.
3621
3622 23. Added support for CRLF line endings as well as CR and LF. As well as the
3623 default being selectable at build time, it can now be changed at runtime
3624 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3625 specify that it is scanning data with non-default line endings.
3626
3627 24. Changed the definition of CXXLINK to make it agree with the definition of
3628 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3629
3630 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3631 recursions. This makes a big different to stack usage for some patterns.
3632
3633 26. If a subpattern containing a named recursion or subroutine reference such
3634 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3635 the space required for the compiled pattern went wrong and gave too small a
3636 value. Depending on the environment, this could lead to "Failed: internal
3637 error: code overflow at offset 49" or "glibc detected double free or
3638 corruption" errors.
3639
3640 27. Applied patches from Google (a) to support the new newline modes and (b) to
3641 advance over multibyte UTF-8 characters in GlobalReplace.
3642
3643 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3644 difference for some implementation of PCRE in some Windows version.
3645
3646 29. Added some extra testing facilities to pcretest:
3647
3648 \q<number> in a data line sets the "match limit" value
3649 \Q<number> in a data line sets the "match recursion limt" value
3650 -S <number> sets the stack size, where <number> is in megabytes
3651
3652 The -S option isn't available for Windows.
3653
3654
3655 Version 6.6 06-Feb-06
3656 ---------------------
3657
3658 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3659 in pcreposix.h. I have copied the definition from pcre.h.
3660
3661 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3662 because pcre.h is no longer a built file.
3663
3664 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3665 not normally included in the compiled code.
3666
3667
3668 Version 6.5 01-Feb-06
3669 ---------------------
3670
3671 1. When using the partial match feature with pcre_dfa_exec(), it was not
3672 anchoring the second and subsequent partial matches at the new starting
3673 point. This could lead to incorrect results. For example, with the pattern
3674 /1234/, partially matching against "123" and then "a4" gave a match.
3675
3676 2. Changes to pcregrep:
3677
3678 (a) All non-match returns from pcre_exec() were being treated as failures
3679 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3680 error message is output. Some extra information is given for the
3681 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3682 probably the only errors that are likely to be caused by users (by
3683 specifying a regex that has nested indefinite repeats, for instance).
3684 If there are more than 20 of these errors, pcregrep is abandoned.
3685
3686 (b) A binary zero was treated as data while matching, but terminated the
3687 output line if it was written out. This has been fixed: binary zeroes
3688 are now no different to any other data bytes.
3689
3690 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3691 used to set a locale for matching. The --locale=xxxx long option has
3692 been added (no short equivalent) to specify a locale explicitly on the
3693 pcregrep command, overriding the environment variables.
3694
3695 (d) When -B was used with -n, some line numbers in the output were one less
3696 than they should have been.
3697
3698 (e) Added the -o (--only-matching) option.
3699
3700 (f) If -A or -C was used with -c (count only), some lines of context were
3701 accidentally printed for the final match.
3702
3703 (g) Added the -H (--with-filename) option.
3704
3705 (h) The combination of options -rh failed to suppress file names for files
3706 that were found from directory arguments.
3707
3708 (i) Added the -D (--devices) and -d (--directories) options.
3709
3710 (j) Added the -F (--fixed-strings) option.
3711
3712 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3713
3714 (l) Added the --colo(u)r option.
3715
3716 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3717 is not present by default.
3718
3719 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3720 items such as (?R) or (?1), when the recursion could match a number of
3721 alternatives. If it matched one of the alternatives, but subsequently,
3722 outside the recursion, there was a failure, the code tried to back up into
3723 the recursion. However, because of the way PCRE is implemented, this is not
3724 possible, and the result was an incorrect result from the match.
3725
3726 In order to prevent this happening, the specification of recursion has
3727 been changed so that all such subpatterns are automatically treated as
3728 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3729
3730 4. I had overlooked the fact that, in some locales, there are characters for
3731 which isalpha() is true but neither isupper() nor islower() are true. In
3732 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3733 and ordfeminine) are like this. This affected the treatment of \w and \W
3734 when they appeared in character classes, but not when they appeared outside
3735 a character class. The bit map for "word" characters is now created
3736 separately from the results of isalnum() instead of just taking it from the
3737 upper, lower, and digit maps. (Plus the underscore character, of course.)
3738
3739 5. The above bug also affected the handling of POSIX character classes such as
3740 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3741 permanent tables. Instead, the bit maps for such a class were previously
3742 created as the appropriate unions of the upper, lower, and digit bitmaps.
3743 Now they are created by subtraction from the [[:word:]] class, which has
3744 its own bitmap.
3745
3746 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3747 It is created by subtracting the vertical space characters (\x09, \x0a,
3748 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3749 subtraction was done in the overall bitmap for a character class, meaning
3750 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3751 be recognized. This bug has been fixed.
3752
3753 7. Patches from the folks at Google:
3754
3755 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3756 real life, but is still worth protecting against".
3757
3758 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3759 regular expressions".
3760
3761 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3762 have it.
3763
3764 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3765 "configure" and the latter not, in order to fix a problem somebody had
3766 with compiling the Arg class on HP-UX.
3767
3768 (e) Improve the error-handling of the C++ wrapper a little bit.
3769
3770 (f) New tests for checking recursion limiting.
3771
3772 8. The pcre_memmove() function, which is used only if the environment does not
3773 have a standard memmove() function (and is therefore rarely compiled),
3774 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3775 returning a result (though PCRE never actually uses the result).
3776
3777 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3778 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3779 returned instead of calling malloc() with an overflowing number that would
3780 most likely cause subsequent chaos.
3781
3782 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3783
3784 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3785 with this option is matched, the nmatch and pmatch options of regexec() are
3786 ignored.
3787
3788 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3789 provided in case anyone wants to the the POSIX interface with UTF-8
3790 strings.
3791
3792 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3793 C++ linking (needed for some HP-UX environments).
3794
3795 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3796 (unused parameter) and in the pcre_printint() function (omitted "default"
3797 switch label when the default is to do nothing).
3798
3799 15. Added some code to make it possible, when PCRE is compiled as a C++
3800 library, to replace subject pointers for pcre_exec() with a smart pointer
3801 class, thus making it possible to process discontinuous strings.
3802
3803 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3804 much the same function. They were added by different people who were trying
3805 to make PCRE easy to compile on non-Unix systems. It has been suggested
3806 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3807 for compiling on Windows systems. I have therefore replaced it with
3808 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3809 defaults to "extern" for C or "extern C" for C++, which works fine on
3810 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3811 SCOPE with something explicit in config.h. In addition:
3812
3813 (a) pcreposix.h still had just "extern" instead of either of these macros;
3814 I have replaced it with PCRE_DATA_SCOPE.
3815
3816 (b) Functions such as _pcre_xclass(), which are internal to the library,
3817 but external in the C sense, all had PCRE_EXPORT in their definitions.
3818 This is apparently wrong for the Windows case, so I have removed it.
3819 (It makes no difference on Unix-like systems.)
3820
3821 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3822 of recursive calls to match(). This is different to MATCH_LIMIT because
3823 that limits the total number of calls to match(), not all of which increase
3824 the depth of recursion. Limiting the recursion depth limits the amount of
3825 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3826 when PCRE is compiled, and changed at run time. A patch from Google adds
3827 this functionality to the C++ interface.
3828
3829 18. Changes to the handling of Unicode character properties:
3830
3831 (a) Updated the table to Unicode 4.1.0.
3832
3833 (b) Recognize characters that are not in the table as "Cn" (undefined).
3834
3835 (c) I revised the way the table is implemented to a much improved format
3836 which includes recognition of ranges. It now supports the ranges that
3837 are defined in UnicodeData.txt, and it also amalgamates other
3838 characters into ranges. This has reduced the number of entries in the
3839 table from around 16,000 to around 3,000, thus reducing its size
3840 considerably. I realized I did not need to use a tree structure after
3841 all - a binary chop search is just as efficient. Having reduced the
3842 number of entries, I extended their size from 6 bytes to 8 bytes to
3843 allow for more data.
3844
3845 (d) Added support for Unicode script names via properties such as \p{Han}.
3846
3847 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3848 matching that character.
3849
3850 20. When matching a repeated Unicode property with a minimum greater than zero,
3851 (for example \pL{2,}), PCRE could look past the end of the subject if it
3852 reached it while seeking the minimum number of characters. This could
3853 happen only if some of the characters were more than one byte long, because
3854 there is a check for at least the minimum number of bytes.
3855
3856 21. Refactored the implementation of \p and \P so as to be more general, to
3857 allow for more different types of property in future. This has changed the
3858 compiled form incompatibly. Anybody with saved compiled patterns that use
3859 \p or \P will have to recompile them.
3860
3861 22. Added "Any" and "L&" to the supported property types.
3862
3863 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3864 but give a compile time error if the value is greater than 0xff.
3865
3866 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3867 accidentally not being installed or uninstalled.
3868
3869 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3870 made were to insert the current release number. This seemed silly, because
3871 it made things harder for people building PCRE on systems that don't run
3872 "configure". I have turned pcre.h into a distributed file, no longer built
3873 by "configure", with the version identification directly included. There is
3874 no longer a pcre.h.in file.
3875
3876 However, this change necessitated a change to the pcre-config script as
3877 well. It is built from pcre-config.in, and one of the substitutions was the
3878 release number. I have updated configure.ac so that ./configure now finds
3879 the release number by grepping pcre.h.
3880
3881 26. Added the ability to run the tests under valgrind.
3882
3883
3884 Version 6.4 05-Sep-05
3885 ---------------------
3886
3887 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3888 "--" to be printed when multiple files were scanned, even when none of the
3889 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3890 consider it to be a bug, and have restored the previous behaviour.
3891
3892 2. A couple of code tidies to get rid of compiler warnings.
3893
3894 3. The pcretest program used to cheat by referring to symbols in the library
3895 whose names begin with _pcre_. These are internal symbols that are not
3896 really supposed to be visible externally, and in some environments it is
3897 possible to suppress them. The cheating is now confined to including
3898 certain files from the library's source, which is a bit cleaner.
3899
3900 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3901 file's purpose clearer.
3902
3903 5. Reorganized pcre_ucp_findchar().
3904
3905
3906 Version 6.3 15-Aug-05
3907 ---------------------
3908
3909 1. The file libpcre.pc.in did not have general read permission in the tarball.
3910
3911 2. There were some problems when building without C++ support:
3912
3913 (a) If C++ support was not built, "make install" and "make test" still
3914 tried to test it.
3915
3916 (b) There were problems when the value of CXX was explicitly set. Some
3917 changes have been made to try to fix these, and ...
3918
3919 (c) --disable-cpp can now be used to explicitly disable C++ support.
3920
3921 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3922 backslash in a target when C++ was disabled. This confuses some
3923 versions of "make", apparently. Using an intermediate variable solves
3924 this. (Same for CPP_LOBJ.)
3925
3926 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3927 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3928 necessary on certain architectures.
3929
3930 4. Added a setting of -export-symbols-regex to the link command to remove
3931 those symbols that are exported in the C sense, but actually are local
3932 within the library, and not documented. Their names all begin with
3933 "_pcre_". This is not a perfect job, because (a) we have to except some
3934 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3935 available (and never for static libraries). I have made a note to try to
3936 find a way round (a) in the future.
3937
3938
3939 Version 6.2 01-Aug-05
3940 ---------------------
3941
3942 1. There was no test for integer overflow of quantifier values. A construction
3943 such as {1111111111111111} would give undefined results. What is worse, if
3944 a minimum quantifier for a parenthesized subpattern overflowed and became
3945 negative, the calculation of the memory size went wrong. This could have
3946 led to memory overwriting.
3947
3948 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3949
3950 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3951 operating environments where this matters.
3952
3953 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3954 PCRE options from within the C++ wrapper.
3955
3956 5. Named capturing subpatterns were not being correctly counted when a pattern
3957 was compiled. This caused two problems: (a) If there were more than 100
3958 such subpatterns, the calculation of the memory needed for the whole
3959 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3960 back references of the form \12, where the number was greater than 9, were
3961 not recognized as back references, even though there were sufficient
3962 previous subpatterns.
3963
3964 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3965 versions of gcc, e.g. 2.95.4.
3966
3967
3968 Version 6.1 21-Jun-05
3969 ---------------------
3970
3971 1. There was one reference to the variable "posix" in pcretest.c that was not
3972 surrounded by "#if !defined NOPOSIX".
3973
3974 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3975 the cross-check on the old pcre_info() function, for the benefit of the
3976 cut-down version of PCRE that is currently imported into Exim.
3977
3978 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3979 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3980 patterns that start (?i) but I don't think that matters. The use of (?i) is
3981 just an example; this all applies to the other options as well.
3982
3983 4. Since libtool seems to echo the compile commands it is issuing, the output
3984 from "make" can be reduced a bit by putting "@" in front of each libtool
3985 compile command.
3986
3987 5. Patch from the folks at Google for configure.in to be a bit more thorough
3988 in checking for a suitable C++ installation before trying to compile the
3989 C++ stuff. This should fix a reported problem when a compiler was present,
3990 but no suitable headers.
3991
3992 6. The man pages all had just "PCRE" as their title. I have changed them to
3993 be the relevant file name. I have also arranged that these names are
3994 retained in the file doc/pcre.txt, which is a concatenation in text format
3995 of all the man pages except the little individual ones for each function.
3996
3997 7. The NON-UNIX-USE file had not been updated for the different set of source
3998 files that come with release 6. I also added a few comments about the C++
3999 wrapper.
4000
4001
4002 Version 6.0 07-Jun-05
4003 ---------------------
4004
4005 1. Some minor internal re-organization to help with my DFA experiments.
4006
4007 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4008 didn't matter for the library itself when fully configured, but did matter
4009 when compiling without UCP support, or within Exim, where the ucp files are
4010 not imported.
4011
4012 3. Refactoring of the library code to split up the various functions into
4013 different source modules. The addition of the new DFA matching code (see
4014 below) to a single monolithic source would have made it really too
4015 unwieldy, quite apart from causing all the code to be include in a
4016 statically linked application, when only some functions are used. This is
4017 relevant even without the DFA addition now that patterns can be compiled in
4018 one application and matched in another.
4019
4020 The downside of splitting up is that there have to be some external
4021 functions and data tables that are used internally in different modules of
4022 the library but which are not part of the API. These have all had their
4023 names changed to start with "_pcre_" so that they are unlikely to clash
4024 with other external names.
4025
4026 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4027 a different (DFA) algorithm. Although it is slower than the original
4028 function, it does have some advantages for certain types of matching
4029 problem.
4030
4031 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4032 including restarting after a partial match.
4033
4034 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4035 defined when compiling for Windows was sent to me. I have put it into the
4036 code, though I have no means of testing or verifying it.
4037
4038 7. Added the pcre_refcount() auxiliary function.
4039
4040 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4041 match before or at the first newline in the subject string. In pcretest,
4042 the /f option on a pattern can be used to set this.
4043
4044 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4045 would behave wrongly. This has been present in PCRE since release 4.0.
4046
4047 10. A number of changes to the pcregrep command:
4048
4049 (a) Refactored how -x works; insert ^(...)$ instead of setting
4050 PCRE_ANCHORED and checking the length, in preparation for adding
4051 something similar for -w.
4052
4053 (b) Added the -w (match as a word) option.
4054
4055 (c) Refactored the way lines are read and buffered so as to have more
4056 than one at a time available.
4057
4058 (d) Implemented a pcregrep test script.
4059
4060 (e) Added the -M (multiline match) option. This allows patterns to match
4061 over several lines of the subject. The buffering ensures that at least
4062 8K, or the rest of the document (whichever is the shorter) is available
4063 for matching (and similarly the previous 8K for lookbehind assertions).
4064
4065 (f) Changed the --help output so that it now says
4066
4067 -w, --word-regex(p)
4068
4069 instead of two lines, one with "regex" and the other with "regexp"
4070 because that confused at least one person since the short forms are the
4071 same. (This required a bit of code, as the output is generated
4072 automatically from a table. It wasn't just a text change.)
4073
4074 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4075 option but starts with a hyphen. Could be a pattern or a path name
4076 starting with a hyphen, for instance.
4077
4078 (h) "-" can be given as a file name to represent stdin.
4079
4080 (i) When file names are being printed, "(standard input)" is used for
4081 the standard input, for compatibility with GNU grep. Previously
4082 "<stdin>" was used.
4083
4084 (j) The option --label=xxx can be used to supply a name to be used for
4085 stdin when file names are being printed. There is no short form.
4086
4087 (k) Re-factored the options decoding logic because we are going to add
4088 two more options that take data. Such options can now be given in four
4089 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4090
4091 (l) Added the -A, -B, and -C options for requesting that lines of context
4092 around matches be printed.
4093
4094 (m) Added the -L option to print the names of files that do not contain
4095 any matching lines, that is, the complement of -l.
4096
4097 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4098 continue to scan other files.
4099
4100 (o) The -s option was incorrectly implemented. For compatibility with other
4101 greps, it now suppresses the error message for a non-existent or non-
4102 accessible file (but not the return code). There is a new option called
4103 -q that suppresses the output of matching lines, which was what -s was
4104 previously doing.
4105
4106 (p) Added --include and --exclude options to specify files for inclusion
4107 and exclusion when recursing.
4108
4109 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4110 Hopefully, it now does.
4111
4112 12. Missing cast in pcre_study().
4113
4114 13. Added an "uninstall" target to the makefile.
4115
4116 14. Replaced "extern" in the function prototypes in Makefile.in with
4117 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4118 world, but is set differently for Windows.
4119
4120 15. Added a second compiling function called pcre_compile2(). The only
4121 difference is that it has an extra argument, which is a pointer to an
4122 integer error code. When there is a compile-time failure, this is set
4123 non-zero, in addition to the error test pointer being set to point to an
4124 error message. The new argument may be NULL if no error number is required
4125 (but then you may as well call pcre_compile(), which is now just a
4126 wrapper). This facility is provided because some applications need a
4127 numeric error indication, but it has also enabled me to tidy up the way
4128 compile-time errors are handled in the POSIX wrapper.
4129
4130 16. Added VPATH=.libs to the makefile; this should help when building with one
4131 prefix path and installing with another. (Or so I'm told by someone who
4132 knows more about this stuff than I do.)
4133
4134 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4135 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4136 match everything, including newlines. This is not POSIX-compatible, but
4137 somebody wanted the feature. From pcretest it can be activated by using
4138 both the P and the s flags.
4139
4140 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4141
4142 19. libpcre.pc was being incorrectly installed as executable.
4143
4144 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4145 it now also looks for '\r' so that it will work unmodified on Windows.
4146
4147 21. Added Google's contributed C++ wrapper to the distribution.
4148
4149 22. Added some untidy missing memory free() calls in pcretest, to keep
4150 Electric Fence happy when testing.
4151
4152
4153
4154 Version 5.0 13-Sep-04
4155 ---------------------
4156
4157 1. Internal change: literal characters are no longer packed up into items
4158 containing multiple characters in a single byte-string. Each character
4159 is now matched using a separate opcode. However, there may be more than one
4160 byte in the character in UTF-8 mode.
4161
4162 2. The pcre_callout_block structure has two new fields: pattern_position and
4163 next_item_length. These contain the offset in the pattern to the next match
4164 item, and its length, respectively.
4165
4166 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4167 insertion of callouts before each pattern item. Added the /C option to
4168 pcretest to make use of this.
4169
4170 4. On the advice of a Windows user, the lines
4171
4172 #if defined(_WIN32) || defined(WIN32)
4173 _setmode( _fileno( stdout ), 0x8000 );
4174 #endif /* defined(_WIN32) || defined(WIN32) */
4175
4176 have been added to the source of pcretest. This apparently does useful
4177 magic in relation to line terminators.
4178
4179 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4180 for the benefit of those environments where the "b" makes a difference.
4181
4182 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4183 to know about it. I have put a hack into configure.in that adds in code
4184 to set GCC=yes if CC=icc. This seems to end up at a point in the
4185 generated configure script that is early enough to affect the setting of
4186 compiler options, which is what is needed, but I have no means of testing
4187 whether it really works. (The user who reported this had patched the
4188 generated configure script, which of course I cannot do.)
4189
4190 LATER: After change 22 below (new libtool files), the configure script
4191 seems to know about icc (and also ecc). Therefore, I have commented out
4192 this hack in configure.in.
4193
4194 7. Added support for pkg-config (2 patches were sent in).
4195
4196 8. Negated POSIX character classes that used a combination of internal tables
4197 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4198 [[:^ascii]]. Typically, they would match almost any characters. The other
4199 POSIX classes were not broken in this way.
4200
4201 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4202 to find the match, as PCRE was deluded into thinking that the match had to
4203 start at the start point or following a newline. The same bug applied to
4204 patterns with negative forward assertions or any backward assertions
4205 preceding ".*" at the start, unless the pattern required a fixed first
4206 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4207
4208 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4209 starting at the last subject character, bytes beyond the end of the subject
4210 string were read.
4211
4212 11. Renamed the variable "class" as "classbits" to make life easier for C++
4213 users. (Previously there was a macro definition, but it apparently wasn't
4214 enough.)
4215
4216 12. Added the new field "tables" to the extra data so that tables can be passed
4217 in at exec time, or the internal tables can be re-selected. This allows
4218 a compiled regex to be saved and re-used at a later time by a different
4219 program that might have everything at different addresses.
4220
4221 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4222 -R library as well as a -L library.
4223
4224 14. The debugging options of pcretest (-d on the command line or D on a
4225 pattern) showed incorrect output for anything following an extended class
4226 that contained multibyte characters and which was followed by a quantifier.
4227
4228 15. Added optional support for general category Unicode character properties
4229 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4230 support. It adds about 90K to the size of the library. The meanings of the
4231 inbuilt class escapes such as \d and \s have NOT been changed.
4232
4233 16. Updated pcredemo.c to include calls to free() to release the memory for the
4234 compiled pattern.
4235
4236 17. The generated file chartables.c was being created in the source directory
4237 instead of in the building directory. This caused the build to fail if the
4238 source directory was different from the building directory, and was
4239 read-only.
4240
4241 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4242 file. No doubt somebody will tell me if they don't make sense... Also added
4243 Dan Mooney's comments about building on OpenVMS.
4244
4245 19. Added support for partial matching via the PCRE_PARTIAL option for
4246 pcre_exec() and the \P data escape in pcretest.
4247
4248 20. Extended pcretest with 3 new pattern features:
4249
4250 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4251 write the compiled pattern to the file whose name is "rest-of-line".
4252 This is a straight binary dump of the data, with the saved pointer to
4253 the character tables forced to be NULL. The study data, if any, is
4254 written too. After writing, pcretest reads a new pattern.
4255
4256 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4257 compiled pattern from the given file. There must not be any
4258 occurrences of "<" in the file name (pretty unlikely); if there are,
4259 pcretest will instead treat the initial "<" as a pattern delimiter.
4260 After reading in the pattern, pcretest goes on to read data lines as
4261 usual.
4262
4263 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4264 and 16-bit fields in a compiled pattern, to simulate a pattern that
4265 was compiled on a host of opposite endianness.
4266
4267 21. The pcre-exec() function can now cope with patterns that were compiled on
4268 hosts of opposite endianness, with this restriction:
4269
4270 As for any compiled expression that is saved and used later, the tables
4271 pointer field cannot be preserved; the extra_data field in the arguments
4272 to pcre_exec() should be used to pass in a tables address if a value
4273 other than the default internal tables were used at compile time.
4274
4275 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4276 now diagnosed as an error. Previously, most of the time, a negative number
4277 would have been treated as zero, but if in addition "ovector" was passed as
4278 NULL, a crash could occur.
4279
4280 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4281 new versions from the libtool 1.5 distribution (the last one is a copy of
4282 a file called libtool.m4). This seems to have fixed the need to patch
4283 "configure" to support Darwin 1.3 (which I used to do). However, I still
4284 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4285 workstation).
4286
4287 24. Changed the PCRE licence to be the more standard "BSD" licence.
4288
4289
4290 Version 4.5 01-Dec-03
4291 ---------------------
4292
4293 1. There has been some re-arrangement of the code for the match() function so
4294 that it can be compiled in a version that does not call itself recursively.
4295 Instead, it keeps those local variables that need separate instances for
4296 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4297 needs to "recurse". Keeping track of where control must go is done by means
4298 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4299 hide most of the details from the main code, and operates only if
4300 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4301 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4302 operating.
4303
4304 To make it easier for callers to provide specially tailored get/free
4305 functions for this usage, two new functions, pcre_stack_malloc, and
4306 pcre_stack_free, are used. They are always called in strict stacking order,
4307 and the size of block requested is always the same.
4308
4309 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4310 PCRE has been compiled to use the stack or the heap for recursion. The
4311 -C option of pcretest uses this to show which version is compiled.
4312
4313 A new data escape \S, is added to pcretest; it causes the amounts of store
4314 obtained and freed by both kinds of malloc/free at match time to be added
4315 to the output.
4316
4317 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4318 what's available on my current Linux desktop machine.
4319
4320 3. When matching a UTF-8 string, the test for a valid string at the start has
4321 been extended. If start_offset is not zero, PCRE now checks that it points
4322 to a byte that is the start of a UTF-8 character. If not, it returns
4323 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4324 this is necessary because there may be backward assertions in the pattern.
4325 When matching the same subject several times, it may save resources to use
4326 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4327
4328 4. The code for checking the validity of UTF-8 strings has been tightened so
4329 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4330 containing "overlong sequences".
4331
4332 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4333 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4334 should have been "&", but it just so happened that all the cases this let
4335 through by mistake were picked up later in the function.
4336
4337 6. I had used a variable called "isblank" - this is a C99 function, causing
4338 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4339
4340 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4341 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4342 so the version is shown; (c) stop "make test" echoing "./RunTest".
4343
4344 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4345
4346 9. The prototype for memmove() for systems that don't have it was using
4347 size_t, but the inclusion of the header that defines size_t was later. I've
4348 moved the #includes for the C headers earlier to avoid this.
4349
4350 10. Added some adjustments to the code to make it easier to compiler on certain
4351 special systems:
4352
4353 (a) Some "const" qualifiers were missing.
4354 (b) Added the macro EXPORT before all exported functions; by default this
4355 is defined to be empty.
4356 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4357 that it reads its output file name as an argument instead of writing
4358 to the standard output and assuming this can be redirected.
4359
4360 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4361 class containing characters with values greater than 255, PCRE compilation
4362 went into a loop.
4363
4364 12. A recursive reference to a subpattern that was within another subpattern
4365 that had a minimum quantifier of zero caused PCRE to crash. For example,
4366 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4367 recursion. If the recursively-called subpattern itself had a zero repeat,
4368 that was OK.
4369
4370 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4371 buffer into which it was copied (for escape processing) was still set at
4372 1024, so long lines caused crashes.
4373
4374 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4375 "internal error: code overflow...". This applied to any character class
4376 that was followed by a possessive quantifier.
4377
4378 15. Modified the Makefile to add libpcre.la as a prerequisite for
4379 libpcreposix.la because I was told this is needed for a parallel build to
4380 work.
4381
4382 16. If a pattern that contained .* following optional items at the start was
4383 studied, the wrong optimizing data was generated, leading to matching
4384 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4385 matching string must start with a or b or c. The correct conclusion for
4386 this pattern is that a match can start with any character.
4387
4388
4389 Version 4.4 13-Aug-03
4390 ---------------------
4391
4392 1. In UTF-8 mode, a character class containing characters with values between
4393 127 and 255 was not handled correctly if the compiled pattern was studied.
4394 In fixing this, I have also improved the studying algorithm for such
4395 classes (slightly).
4396
4397 2. Three internal functions had redundant arguments passed to them. Removal
4398 might give a very teeny performance improvement.
4399
4400 3. Documentation bug: the value of the capture_top field in a callout is *one
4401 more than* the number of the hightest numbered captured substring.
4402
4403 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4404 in incorrectly linking with a previously installed version. They now link
4405 explicitly with libpcre.la.
4406
4407 5. configure.in no longer needs to recognize Cygwin specially.
4408
4409 6. A problem in pcre.in for Windows platforms is fixed.
4410
4411 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4412 pcretest, it used to include the size of the study block as part of its
4413 output. Unfortunately, the structure contains a field that has a different
4414 size on different hardware architectures. This meant that the tests that
4415 showed this size failed. As the block is currently always of a fixed size,
4416 this information isn't actually particularly useful in pcretest output, so
4417 I have just removed it.
4418
4419 8. Three pre-processor statements accidentally did not start in column 1.
4420 Sadly, there are *still* compilers around that complain, even though
4421 standard C has not required this for well over a decade. Sigh.
4422
4423 9. In pcretest, the code for checking callouts passed small integers in the
4424 callout_data field, which is a void * field. However, some picky compilers
4425 complained about the casts involved for this on 64-bit systems. Now
4426 pcretest passes the address of the small integer instead, which should get
4427 rid of the warnings.
4428
4429 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4430 both compile and run time, and gives an error if an invalid UTF-8 sequence
4431 is found. There is a option for disabling this check in cases where the
4432 string is known to be correct and/or the maximum performance is wanted.
4433
4434 11. In response to a bug report, I changed one line in Makefile.in from
4435
4436 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4437 to
4438 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4439
4440 to look similar to other lines, but I have no way of telling whether this
4441 is the right thing to do, as I do not use Windows. No doubt I'll get told
4442 if it's wrong...
4443
4444
4445 Version 4.3 21-May-03
4446 ---------------------
4447
4448 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4449 Makefile.
4450
4451 2. Some refactoring to improve the quality of the code:
4452
4453 (i) The utf8_table... variables are now declared "const".
4454
4455 (ii) The code for \cx, which used the "case flipping" table to upper case
4456 lower case letters, now just substracts 32. This is ASCII-specific,
4457 but the whole concept of \cx is ASCII-specific, so it seems
4458 reasonable.
4459
4460 (iii) PCRE was using its character types table to recognize decimal and
4461 hexadecimal digits in the pattern. This is silly, because it handles
4462 only 0-9, a-f, and A-F, but the character types table is locale-
4463 specific, which means strange things might happen. A private
4464 table is now used for this - though it costs 256 bytes, a table is
4465 much faster than multiple explicit tests. Of course, the standard
4466 character types table is still used for matching digits in subject
4467 strings against \d.
4468
4469 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4470 ending in _t are). So I've renamed it as ESC_tee.
4471
4472 3. The first argument for regexec() in the POSIX wrapper should have been
4473 defined as "const".
4474
4475 4. Changed pcretest to use malloc() for its buffers so that they can be
4476 Electric Fenced for debugging.
4477
4478 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4479 to read one or more bytes before the start of the subject string. Often this
4480 had no effect on PCRE's behaviour, but in some circumstances it could
4481 provoke a segmentation fault.
4482
4483 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4484 to try to read one or more bytes before the start of the subject string.
4485
4486 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4487 UTF-8 support could misbehave in various ways if the subject string
4488 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4489 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4490 back over UTF-8 characters.)
4491
4492
4493 Version 4.2 14-Apr-03
4494 ---------------------
4495
4496 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4497
4498 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4499 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4500 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4501 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4502 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4503 and BUILD_EXEEXT
4504 Note: automatic setting of the BUILD variables is not yet working
4505 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4506 compile-time but not at link-time
4507 [LINK]: use for linking executables only
4508 make different versions for Windows and non-Windows
4509 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4510 libraries
4511 [LINK_FOR_BUILD]: new variable
4512 [OBJEXT]: use throughout
4513 [EXEEXT]: use throughout
4514 <winshared>: new target
4515 <wininstall>: new target
4516 <dftables.o>: use native compiler
4517 <dftables>: use native linker
4518 <install>: handle Windows platform correctly
4519 <clean>: ditto
4520 <check>: ditto
4521 copy DLL to top builddir before testing
4522
4523 As part of these changes, -no-undefined was removed again. This was reported
4524 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4525 in any case.
4526
4527 3. Some tidies to get rid of compiler warnings:
4528
4529 . In the match_data structure, match_limit was an unsigned long int, whereas
4530 match_call_count was an int. I've made them both unsigned long ints.
4531
4532 . In pcretest the fact that a const uschar * doesn't automatically cast to
4533 a void * provoked a warning.
4534
4535 . Turning on some more compiler warnings threw up some "shadow" variables
4536 and a few more missing casts.
4537
4538 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4539 option, a class that contained a single character with a value between 128
4540 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4541
4542 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4543 option, a class that contained several characters, but with at least one
4544 whose value was between 128 and 255 caused PCRE to crash.
4545
4546
4547 Version 4.1 12-Mar-03
4548 ---------------------
4549
4550 1. Compiling with gcc -pedantic found a couple of places where casts were
4551 needed, and a string in dftables.c that was longer than standard compilers are
4552 required to support.
4553
4554 2. Compiling with Sun's compiler found a few more places where the code could
4555 be tidied up in order to avoid warnings.
4556
4557 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4558 first of these names is deprecated in the latest Autoconf in favour of the name
4559 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4560 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4561 analogy I have changed it to CFLAGS_FOR_BUILD.
4562
4563 4. Added -no-undefined to the linking command in the Makefile, because this is
4564 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4565 linking step for the pcreposix library.
4566
4567 5. PCRE was failing to diagnose the case of two named groups with the same
4568 name.
4569
4570 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4571 literal character that is needed in the subject for a match, and scans along to
4572 ensure that it is present before embarking on the full matching process. This
4573 saves time in cases of nested unlimited repeats that are never going to match.
4574 Problem: the scan can take a lot of time if the subject is very long (e.g.
4575 megabytes), thus penalizing straightforward matches. It is now done only if the
4576 amount of subject to be scanned is less than 1000 bytes.
4577
4578 7. A lesser problem with the same optimization is that it was recording the
4579 first character of an anchored pattern as "needed", thus provoking a search
4580 right along the subject, even when the first match of the pattern was going to
4581 fail. The "needed" character is now not set for anchored patterns, unless it
4582 follows something in the pattern that is of non-fixed length. Thus, it still
4583 fulfils its original purpose of finding quick non-matches in cases of nested
4584 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4585
4586
4587 Version 4.0 17-Feb-03
4588 ---------------------
4589
4590 1. If a comment in an extended regex that started immediately after a meta-item
4591 extended to the end of string, PCRE compiled incorrect data. This could lead to
4592 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4593
4594 2. Moved to autoconf 2.53 and libtool 1.4.2.
4595
4596 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4597 the special perltest8 script is no longer needed - all the tests can be run
4598 from a single perltest script.
4599
4600 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4601 by \s. It has now been removed in PCRE. This means it isn't recognized as
4602 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4603 class [:space:] *does* include VT, thereby creating a mess.
4604
4605 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4606 space and tab.
4607
4608 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4609 its new features into the main test script, reducing the number of scripts.
4610
4611 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4612 were backward compatible, and made the (?i) apply to the whole pattern, as if
4613 /i were given. Now it behaves more logically, and applies the option setting
4614 only to what follows. PCRE has been changed to follow suit. However, if it
4615 finds options settings right at the start of the pattern, it extracts them into
4616 the global options, as before. Thus, they show up in the info data.
4617
4618 8. Added support for the \Q...\E escape sequence. Characters in between are
4619 treated as literals. This is slightly different from Perl in that $ and @ are
4620 also handled as literals inside the quotes. In Perl, they will cause variable
4621 interpolation. Note the following examples:
4622
4623 Pattern PCRE matches Perl matches
4624
4625 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4626 \Qabc\$xyz\E abc\$xyz abc\$xyz
4627 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4628
4629 For compatibility with Perl, \Q...\E sequences are recognized inside character
4630 classes as well as outside them.
4631
4632 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4633 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4634 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4635 signed/unsigned warnings.
4636
4637 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4638 option for pcretest, so I've replaced it by a simple function that does just
4639 that job.
4640
4641 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4642 "pcregrep -".
4643
4644 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4645 Java package. This provides some syntactic sugar for simple cases of what my
4646 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4647 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4648 item, you can use this simplified notation. Note that only makes sense with
4649 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4650 greediness, whatever the setting of the PCRE_UNGREEDY option.
4651
4652 13. A change of greediness default within a pattern was not taking effect at
4653 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4654 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4655 was abstracted outside.
4656
4657 14. PCRE now supports the \G assertion. It is true when the current matching
4658 position is at the start point of the match. This differs from \A when the
4659 starting offset is non-zero. Used with the /g option of pcretest (or similar
4660 code), it works in the same way as it does for Perl's /g option. If all
4661 alternatives of a regex begin with \G, the expression is anchored to the start
4662 match position, and the "anchored" flag is set in the compiled expression.
4663
4664 15. Some bugs concerning the handling of certain option changes within patterns
4665 have been fixed. These applied to options other than (?ims). For example,
4666 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4667 been the other way round. Some of this was related to change 7 above.
4668
4669 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4670 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4671 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4672 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4673
4674 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4675 mode. Unlike ".", it always matches newline, whatever the setting of
4676 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4677 assertions. Perl allows it, but it doesn't (in general) work because it can't
4678 calculate the length of the lookbehind. At least, that's the case for Perl
4679 5.8.0 - I've been told they are going to document that it doesn't work in
4680 future.
4681
4682 18. Added an error diagnosis for escapes that PCRE does not support: these are
4683 \L, \l, \N, \P, \p, \U, \u, and \X.
4684
4685 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4686 reading past the end of the pattern in cases such as /[abcd/.
4687
4688 20. PCRE was getting more memory than necessary for patterns with classes that
4689 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4690
4691 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4692 compiling PCRE for use with Virtual Pascal.
4693
4694 22. Small fix to the Makefile to make it work properly if the build is done
4695 outside the source tree.
4696
4697 23. Added a new extension: a condition to go with recursion. If a conditional
4698 subpattern starts with (?(R) the "true" branch is used if recursion has
4699 happened, whereas the "false" branch is used only at the top level.
4700
4701 24. When there was a very long string of literal characters (over 255 bytes
4702 without UTF support, over 250 bytes with UTF support), the computation of how
4703 much memory was required could be incorrect, leading to segfaults or other
4704 strange effects.
4705
4706 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4707 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4708 there was a subsequent back reference to those brackets. This meant that, for
4709 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4710 possible to check for precisely this case. All we can do is abandon the
4711 optimization if .* occurs inside capturing brackets when there are any back
4712 references whatsoever. (See below for a better fix that came later.)
4713
4714 26. The handling of the optimization for finding the first character of a
4715 non-anchored pattern, and for finding a character that is required later in the
4716 match were failing in some cases. This didn't break the matching; it just
4717 failed to optimize when it could. The way this is done has been re-implemented.
4718
4719 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4720
4721 28. Added a new feature that provides some of the functionality that Perl
4722 provides with (?{...}). The facility is termed a "callout". The way it is done
4723 in PCRE is for the caller to provide an optional function, by setting
4724 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4725 global variable. By default it is unset, which disables all calling out. To get
4726 the function called, the regex must include (?C) at appropriate points. This
4727 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4728 This provides a means of identifying different callout points. When PCRE
4729 reaches such a point in the regex, if pcre_callout has been set, the external
4730 function is called. It is provided with data in a structure called
4731 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4732 matching continues; if it returns a non-zero value, the match at the current
4733 point fails. However, backtracking will occur if possible. [This was changed
4734 later and other features added - see item 49 below.]
4735
4736 29. pcretest is upgraded to test the callout functionality. It provides a
4737 callout function that displays information. By default, it shows the start of
4738 the match and the current position in the text. There are some new data escapes
4739 to vary what happens:
4740
4741 \C+ in addition, show current contents of captured substrings
4742 \C- do not supply a callout function
4743 \C!n return 1 when callout number n is reached
4744 \C!n!m return 1 when callout number n is reached for the mth time
4745
4746 30. If pcregrep was called with the -l option and just a single file name, it
4747 output "<stdin>" if a match was found, instead of the file name.
4748
4749 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4750 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4751 pcre_exec(). This saves a malloc/free per call. The default value of
4752 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4753 when configuring.
4754
4755 32. The default maximum size of a compiled pattern is 64K. There have been a
4756 few cases of people hitting this limit. The code now uses macros to handle the
4757 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4758 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4759 configuring. Tests 2 and 5 work only with 2-byte links because they output
4760 debugging information about compiled patterns.
4761
4762 33. Internal code re-arrangements:
4763
4764 (a) Moved the debugging function for printing out a compiled regex into
4765 its own source file (printint.c) and used #include to pull it into
4766 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4767 separate copies.
4768
4769 (b) Defined the list of op-code names for debugging as a macro in
4770 internal.h so that it is next to the definition of the opcodes.
4771
4772 (c) Defined a table of op-code lengths for simpler skipping along compiled
4773 code. This is again a macro in internal.h so that it is next to the
4774 definition of the opcodes.
4775
4776 34. Added support for recursive calls to individual subpatterns, along the
4777 lines of Robin Houston's patch (but implemented somewhat differently).
4778
4779 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4780 allow it to read and process whole directories in Win32. This code was
4781 contributed by Lionel Fourquaux; it has not been tested by me.
4782
4783 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4784 used to name a group. Names consist of alphanumerics and underscores, and must
4785 be unique. Back references use the syntax (?P=name) and recursive calls use
4786 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4787 numbers. The function pcre_fullinfo() can be used after compilation to extract
4788 a name/number map. There are three relevant calls:
4789
4790 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4791 PCRE_INFO_NAMECOUNT yields the number of entries
4792 PCRE_INFO_NAMETABLE yields a pointer to the map.
4793
4794 The map is a vector of fixed-size entries. The size of each entry depends on
4795 the length of the longest name used. The first two bytes of each entry are the
4796 group number, most significant byte first. There follows the corresponding
4797 name, zero terminated. The names are in alphabetical order.
4798
4799 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4800 case instead of 255. Making it the same both with and without UTF-8 support
4801 means that the same test output works with both.
4802
4803 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4804 calling malloc() with a zero argument.
4805
4806 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4807 optimization. I've improved things by keeping a bitmap of backreferences with
4808 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4809 fact referenced, the optimization can be applied. It is unlikely that a
4810 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4811 the match to follow \n) will appear inside brackets with a number greater than
4812 31, but if it does, any back reference > 31 suppresses the optimization.
4813
4814 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4815 of disabling numbered capturing parentheses. Any opening parenthesis that is
4816 not followed by ? behaves as if it were followed by ?: but named parentheses
4817 can still be used for capturing (and they will acquire numbers in the usual
4818 way).
4819
4820 41. Redesigned the return codes from the match() function into yes/no/error so
4821 that errors can be passed back from deep inside the nested calls. A malloc
4822 failure while inside a recursive subpattern call now causes the
4823 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4824
4825 42. It is now possible to set a limit on the number of times the match()
4826 function is called in a call to pcre_exec(). This facility makes it possible to
4827 limit the amount of recursion and backtracking, though not in a directly
4828 obvious way, because the match() function is used in a number of different
4829 circumstances. The count starts from zero for each position in the subject
4830 string (for non-anchored patterns). The default limit is, for compatibility, a
4831 large number, namely 10 000 000. You can change this in two ways:
4832
4833 (a) When configuring PCRE before making, you can use --with-match-limit=n
4834 to set a default value for the compiled library.
4835
4836 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4837 a different value is set. See 45 below.
4838
4839 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4840
4841 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4842 of things that can be changed at compile time. The first argument specifies
4843 what is wanted and the second points to where the information is to be placed.
4844 The current list of available information is:
4845
4846 PCRE_CONFIG_UTF8
4847
4848 The output is an integer that is set to one if UTF-8 support is available;
4849 otherwise it is set to zero.
4850
4851 PCRE_CONFIG_NEWLINE
4852
4853 The output is an integer that it set to the value of the code that is used for
4854 newline. It is either LF (10) or CR (13).
4855
4856 PCRE_CONFIG_LINK_SIZE
4857
4858 The output is an integer that contains the number of bytes used for internal
4859 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4860
4861 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4862
4863 The output is an integer that contains the threshold above which the POSIX
4864 interface uses malloc() for output vectors. See item 31 above.
4865
4866 PCRE_CONFIG_MATCH_LIMIT
4867
4868 The output is an unsigned integer that contains the default limit of the number
4869 of match() calls in a pcre_exec() execution. See 42 above.
4870
4871 44. pcretest has been upgraded by the addition of the -C option. This causes it
4872 to extract all the available output from the new pcre_config() function, and to
4873 output it. The program then exits immediately.
4874
4875 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4876 order to support additional features. One way would have been to define
4877 pcre_exec2() (for example) with extra arguments, but this would not have been
4878 extensible, and would also have required all calls to the original function to
4879 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4880 is used for passing in "extra" data from pcre_study().
4881
4882 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4883 contains the following fields:
4884
4885 flags a bitmap indicating which of the following fields are set
4886 study_data opaque data from pcre_study()
4887 match_limit a way of specifying a limit on match() calls for a specific
4888 call to pcre_exec()
4889 callout_data data for callouts (see 49 below)
4890
4891 The flag bits are also defined in pcre.h, and are
4892
4893 PCRE_EXTRA_STUDY_DATA
4894 PCRE_EXTRA_MATCH_LIMIT
4895 PCRE_EXTRA_CALLOUT_DATA
4896
4897 The pcre_study() function now returns one of these new pcre_extra blocks, with
4898 the actual study data pointed to by the study_data field, and the
4899 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4900 before. That is, this change is entirely upwards-compatible and requires no
4901 change to existing code.
4902
4903 If you want to pass in additional data to pcre_exec(), you can either place it
4904 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4905 block.
4906
4907 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4908 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4909 times with different match limits, until it finds the minimum value needed for
4910 pcre_exec() to complete. The value is then output. This can be instructive; for
4911 most simple matches the number is quite small, but for pathological cases it
4912 gets very large very quickly.
4913
4914 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4915 returns the size of the data block pointed to by the study_data field in a
4916 pcre_extra block, that is, the value that was passed as the argument to
4917 pcre_malloc() when PCRE was getting memory in which to place the information
4918 created by pcre_study(). The fourth argument should point to a size_t variable.
4919 pcretest has been extended so that this information is shown after a successful
4920 pcre_study() call when information about the compiled regex is being displayed.
4921
4922 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4923 because what follows is always an absolute path. (Later: it turns out that this
4924 is more than cosmetic for MinGW, because it doesn't like empty path
4925 components.)
4926
4927 49. Some changes have been made to the callout feature (see 28 above):
4928
4929 (i) A callout function now has three choices for what it returns:
4930
4931 0 => success, carry on matching
4932 > 0 => failure at this point, but backtrack if possible
4933 < 0 => serious error, return this value from pcre_exec()
4934
4935 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4936 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4937 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4938 use by callout functions. It will never be used by PCRE itself.
4939
4940 (ii) The pcre_extra structure (see 45 above) has a void * field called
4941 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4942 pcre_callout_block structure has a field of the same name. The contents of
4943 the field passed in the pcre_extra structure are passed to the callout
4944 function in the corresponding field in the callout block. This makes it
4945 easier to use the same callout-containing regex from multiple threads. For
4946 testing, the pcretest program has a new data escape
4947
4948 \C*n pass the number n (may be negative) as callout_data
4949
4950 If the callout function in pcretest receives a non-zero value as
4951 callout_data, it returns that value.
4952
4953 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4954 there were some redundant $(CFLAGS) in commands that are now specified as
4955 $(LINK), which already includes $(CFLAGS).
4956
4957 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4958 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4959 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4960 one-byte characters throughout. Note that case-insensitive matching applies
4961 only to characters whose values are less than 256. PCRE doesn't support the
4962 notion of cases for higher-valued characters.
4963
4964 (i) A character class whose characters are all within 0-255 is handled as
4965 a bit map, and the map is inverted for negative classes. Previously, a
4966 character > 255 always failed to match such a class; however it should
4967 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4968
4969 (ii) A negated character class with a single character < 255 is coded as
4970 "not this character" (OP_NOT). This wasn't working properly when the test
4971 character was multibyte, either singly or repeated.
4972
4973 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4974 mode, for example: \x{100}{2,3}.
4975
4976 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4977 singly or repeated) now correctly test multibyte characters. However,
4978 PCRE doesn't recognize any characters with values greater than 255 as
4979 digits, spaces, or word characters. Such characters always match \D, \S,
4980 and \W, and never match \d, \s, or \w.
4981
4982 (v) Classes may now contain characters and character ranges with values
4983 greater than 255. For example: [ab\x{100}-\x{400}].
4984
4985 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4986 PCRE in UTF-8 mode.
4987
4988 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4989 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4990 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4991 value.)
4992
4993 53. The single man page has become too large. I have therefore split it up into
4994 a number of separate man pages. These also give rise to individual HTML pages;
4995 these are now put in a separate directory, and there is an index.html page that
4996 lists them all. Some hyperlinking between the pages has been installed.
4997
4998 54. Added convenience functions for handling named capturing parentheses.
4999
5000 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5001 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5002 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5003 are faulted.
5004
5005 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5006 calling configure. These values are used when compiling the dftables.c program
5007 which is run to generate the source of the default character tables. They
5008 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5009 you will need to set these values.
5010
5011 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5012
5013
5014 Version 3.9 02-Jan-02
5015 ---------------------
5016
5017 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5018
5019 2. If --disable-static was given, the building process failed when trying to
5020 build pcretest and pcregrep. (For some reason it was using libtool to compile
5021 them, which is not right, as they aren't part of the library.)
5022
5023
5024 Version 3.8 18-Dec-01
5025 ---------------------
5026
5027 1. The experimental UTF-8 code was completely screwed up. It was packing the
5028 bytes in the wrong order. How dumb can you get?
5029
5030
5031 Version 3.7 29-Oct-01
5032 ---------------------
5033
5034 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5035 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5036 this didn't happen under Solaris 8, where I normally test things.
5037
5038 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5039 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5040 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5041 this fix an hour or so after the initial 3.7 release.)
5042
5043
5044 Version 3.6 23-Oct-01
5045 ---------------------
5046
5047 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5048 offsets passed as NULL with zero offset count.
5049
5050 2. The config.guess and config.sub files had not been updated when I moved to
5051 the latest autoconf.
5052
5053
5054 Version 3.5 15-Aug-01
5055 ---------------------
5056
5057 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5058 had been forgotten.
5059
5060 2. By using declared but undefined structures, we can avoid using "void"
5061 definitions in pcre.h while keeping the internal definitions of the structures
5062 private.
5063
5064 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5065 user point of view, this means that both static and shared libraries are built
5066 by default, but this can be individually controlled. More of the work of
5067 handling this static/shared cases is now inside libtool instead of PCRE's make
5068 file.
5069
5070 4. The pcretest utility is now installed along with pcregrep because it is
5071 useful for users (to test regexs) and by doing this, it automatically gets
5072 relinked by libtool. The documentation has been turned into a man page, so
5073 there are now .1, .txt, and .html versions in /doc.
5074
5075 5. Upgrades to pcregrep:
5076 (i) Added long-form option names like gnu grep.
5077 (ii) Added --help to list all options with an explanatory phrase.
5078 (iii) Added -r, --recursive to recurse into sub-directories.
5079 (iv) Added -f, --file to read patterns from a file.
5080
5081 6. pcre_exec() was referring to its "code" argument before testing that
5082 argument for NULL (and giving an error if it was NULL).
5083
5084 7. Upgraded Makefile.in to allow for compiling in a different directory from
5085 the source directory.
5086
5087 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5088 options bits, the pointer it was passed was to an int instead of to an unsigned
5089 long int. This mattered only on 64-bit systems.
5090
5091 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5092 generated) instead of pcre.in, which it its source. Also made the same change
5093 in several of the .c files.
5094
5095 10. A new release of gcc defines printf() as a macro, which broke pcretest
5096 because it had an ifdef in the middle of a string argument for printf(). Fixed
5097 by using separate calls to printf().
5098
5099 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5100 script, to force use of CR or LF instead of \n in the source. On non-Unix
5101 systems, the value can be set in config.h.
5102
5103 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5104 absolute limit. Changed the text of the error message to make this clear, and
5105 likewise updated the man page.
5106
5107 13. The limit of 99 on the number of capturing subpatterns has been removed.
5108 The new limit is 65535, which I hope will not be a "real" limit.
5109
5110
5111 Version 3.4 22-Aug-00
5112 ---------------------
5113
5114 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5115
5116 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5117
5118
5119 Version 3.3 01-Aug-00
5120 ---------------------
5121
5122 1. If an octal character was given, but the value was greater than \377, it
5123 was not getting masked to the least significant bits, as documented. This could
5124 lead to crashes in some systems.
5125
5126 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5127 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5128
5129 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5130 These just pass their arguments on to (pcre_free)(), but they are provided
5131 because some uses of PCRE bind it to non-C systems that can call its functions,
5132 but cannot call free() or pcre_free() directly.
5133
5134 4. Add "make test" as a synonym for "make check". Corrected some comments in
5135 the Makefile.
5136
5137 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5138 Makefile.
5139
5140 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5141 command called pgrep for grepping around the active processes.
5142
5143 7. Added the beginnings of support for UTF-8 character strings.
5144
5145 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5146 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5147 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5148 out for the ar command.)
5149
5150
5151 Version 3.2 12-May-00
5152 ---------------------
5153
5154 This is purely a bug fixing release.
5155
5156 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5157 of ZA. This was just one example of several cases that could provoke this bug,
5158 which was introduced by change 9 of version 2.00. The code for breaking
5159 infinite loops after an iteration that matches an empty string was't working
5160 correctly.
5161
5162 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5163 when matched against abbab (for example). After matching an empty string, it
5164 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5165 caused it to match further down the string than it should.
5166
5167 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5168 was there because it doesn't seem to be needed, and it causes trouble on some
5169 systems, as it is not a Standard C header. It has been removed.
5170
5171 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5172 were reported on the Macintosh. The changes were from
5173
5174 while ((c = *(++ptr)) != 0 && c != '\n');
5175 to
5176 while ((c = *(++ptr)) != 0 && c != '\n') ;
5177
5178 Totally extraordinary, but if that's what it takes...
5179
5180 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5181 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5182 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5183 assumes the way PCRE uses memmove() (always moving upwards).
5184
5185 6. PCRE is being used in one environment where strchr() is not available. There
5186 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5187 faster code anyway.
5188
5189
5190 Version 3.1 09-Feb-00
5191 ---------------------
5192
5193 The only change in this release is the fixing of some bugs in Makefile.in for
5194 the "install" target:
5195
5196 (1) It was failing to install pcreposix.h.
5197
5198 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5199
5200
5201 Version 3.0 01-Feb-00
5202 ---------------------
5203
5204 1. Add support for the /+ modifier to perltest (to output $` like it does in
5205 pcretest).
5206
5207 2. Add support for the /g modifier to perltest.
5208
5209 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5210 matches null strings.
5211
5212 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5213 pattern. Perl treats empty patterns specially - it reuses the most recent
5214 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5215 effect.
5216
5217 5. The POSIX interface was broken in that it was just handing over the POSIX
5218 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5219 required a bigger vector, with some working space on the end. This means that
5220 the POSIX wrapper now has to get and free some memory, and copy the results.
5221
5222 6. Added some simple autoconf support, placing the test data and the
5223 documentation in separate directories, re-organizing some of the
5224 information files, and making it build pcre-config (a GNU standard). Also added
5225 libtool support for building PCRE as a shared library, which is now the
5226 default.
5227
5228 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5229 09 are not valid octal constants. Single digits will be used for minor values
5230 less than 10.
5231
5232 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5233 existing programs that set these in the POSIX interface can use PCRE without
5234 modification.
5235
5236 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5237 return all that pcre_info() returns, plus additional data. The pcre_info()
5238 function is retained for compatibility, but is considered to be obsolete.
5239
5240 10. Added experimental recursion feature (?R) to handle one common case that
5241 Perl 5.6 will be able to do with (?p{...}).
5242
5243 11. Added support for POSIX character classes like [:alpha:], which Perl is
5244 adopting.
5245
5246
5247 Version 2.08 31-Aug-99
5248 ----------------------
5249
5250 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5251 trying to match at the startoffset position, but instead was moving forward to
5252 the next newline as if a previous match had failed.
5253
5254 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5255 and could get into a loop if a null string was matched other than at the start
5256 of the subject.
5257
5258 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5259 be distinguished at compile time, and for completeness also added PCRE_DATE.
5260
5261 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5262 in GnuWin32 environments.
5263
5264
5265 Version 2.07 29-Jul-99
5266 ----------------------
5267
5268 1. The documentation is now supplied in plain text form and HTML as well as in
5269 the form of man page sources.
5270
5271 2. C++ compilers don't like assigning (void *) values to other pointer types.
5272 In particular this affects malloc(). Although there is no problem in Standard
5273 C, I've put in casts to keep C++ compilers happy.
5274
5275 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5276 should be (const char *).
5277
5278 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5279 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5280 However, I haven't made this a standard facility. The documentation doesn't
5281 mention it, and the Makefile doesn't support it.
5282
5283 5. The Makefile now contains an "install" target, with editable destinations at
5284 the top of the file. The pcretest program is not installed.
5285
5286 6. pgrep -V now gives the PCRE version number and date.
5287
5288 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5289 causing the entire string to be ignored, instead of just the last character.
5290
5291 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5292 non-matching string, it can take a very, very long time, even for strings of
5293 quite modest length, because of the nested recursion. PCRE now does better in
5294 some of these cases. It does this by remembering the last required literal
5295 character in the pattern, and pre-searching the subject to ensure it is present
5296 before running the real match. In other words, it applies a heuristic to detect
5297 some types of certain failure quickly, and in the above example, if presented
5298 with a string that has no trailing " it gives "no match" very quickly.
5299
5300 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5301 other alternatives are tried instead.
5302
5303
5304 Version 2.06 09-Jun-99
5305 ----------------------
5306
5307 1. Change pcretest's output for amount of store used to show just the code
5308 space, because the remainder (the data block) varies in size between 32-bit and
5309 64-bit systems.
5310
5311 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5312 start matching at. This allows lookbehinds to work when searching for multiple
5313 occurrences in a string.
5314
5315 3. Added additional options to pcretest for testing multiple occurrences:
5316
5317 /+ outputs the rest of the string that follows a match
5318 /g loops for multiple occurrences, using the new startoffset argument
5319 /G loops for multiple occurrences by passing an incremented pointer
5320
5321 4. PCRE wasn't doing the "first character" optimization for patterns starting
5322 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5323 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5324 the letter 'x'. On long subject strings, this gives a significant speed-up.
5325
5326
5327 Version 2.05 21-Apr-99
5328 ----------------------
5329
5330 1. Changed the type of magic_number from int to long int so that it works
5331 properly on 16-bit systems.
5332
5333 2. Fixed a bug which caused patterns starting with .* not to work correctly
5334 when the subject string contained newline characters. PCRE was assuming
5335 anchoring for such patterns in all cases, which is not correct because .* will
5336 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5337 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5338 must be retried after every newline in the subject.
5339
5340
5341 Version 2.04 18-Feb-99
5342 ----------------------
5343
5344 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5345 computation of the store needed to hold the pattern was incorrect (too large).
5346 If such patterns were nested a few deep, this could multiply and become a real
5347 problem.
5348
5349 2. Added /M option to pcretest to show the memory requirement of a specific
5350 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5351
5352 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5353 compiled in such a way that the backtracking after subsequent failure was
5354 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5355 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5356
5357
5358 Version 2.03 02-Feb-99
5359 ----------------------
5360
5361 1. Fixed typo and small mistake in man page.
5362
5363 2. Added 4th condition (GPL supersedes if conflict) and created separate
5364 LICENCE file containing the conditions.
5365
5366 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5367 Perl, that is the internal \ allows the delimiter to be included in the
5368 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5369 the final delimiter, add \ to the end of the pattern (to test the error).
5370
5371 4. Added the convenience functions for extracting substrings after a successful
5372 match. Updated pcretest to make it able to test these functions.
5373
5374
5375 Version 2.02 14-Jan-99
5376 ----------------------
5377
5378 1. Initialized the working variables associated with each extraction so that
5379 their saving and restoring doesn't refer to uninitialized store.
5380
5381 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5382 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5383 fix the problem.
5384
5385 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5386 calls, and wasn't printing the correct value for compiling calls. Increased the
5387 default value of LOOPREPEAT, and the number of significant figures in the
5388 times.
5389
5390 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5391
5392 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5393 a building problem on Windows NT with a FAT file system.
5394
5395
5396 Version 2.01 21-Oct-98
5397 ----------------------
5398
5399 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5400 to character tables built by pcre_maketables() in the current locale. If NULL
5401 is passed, the default tables are used.
5402
5403
5404 Version 2.00 24-Sep-98
5405 ----------------------
5406
5407 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5408 it any more.
5409
5410 2. Allow quantification of (?>) groups, and make it work correctly.
5411
5412 3. The first character computation wasn't working for (?>) groups.
5413
5414 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5415 end of the subject) and add 5.005's \z, which really does match only at the
5416 very end of the subject.
5417
5418 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5419
5420 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5421 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5422 localized options. All options to pcre_study() were also removed.
5423
5424 7. Add other new features from 5.005:
5425
5426 $(?<= positive lookbehind
5427 $(?<! negative lookbehind
5428 (?imsx-imsx) added the unsetting capability
5429 such a setting is global if at outer level; local otherwise
5430 (?imsx-imsx:) non-capturing groups with option setting
5431 (?(cond)re|re) conditional pattern matching
5432
5433 A backreference to itself in a repeated group matches the previous
5434 captured string.
5435
5436 8. General tidying up of studying (both automatic and via "study")
5437 consequential on the addition of new assertions.
5438
5439 9. As in 5.005, unlimited repeated groups that could match an empty substring
5440 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5441 runtime if any iteration does actually match an empty substring.
5442
5443 10. Include the RunTest script in the distribution.
5444
5445 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5446 discrepancies, some of which were old and were also with respect to 5.004. They
5447 have now been fixed.
5448
5449
5450 Version 1.09 28-Apr-98
5451 ----------------------
5452
5453 1. A negated single character class followed by a quantifier with a minimum
5454 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5455 program crashes, or just wrong answers. This did not apply to negated classes
5456 containing more than one character, or to minima other than one.
5457
5458
5459 Version 1.08 27-Mar-98
5460 ----------------------
5461
5462 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5463
5464 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5465 latter must appear before anything that relies on it in the pattern.
5466
5467
5468 Version 1.07 16-Feb-98
5469 ----------------------
5470
5471 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5472 repeat of a potentially empty string).
5473
5474
5475 Version 1.06 23-Jan-98
5476 ----------------------
5477
5478 1. Added Markus Oberhumer's little patches for C++.
5479
5480 2. Literal strings longer than 255 characters were broken.
5481
5482
5483 Version 1.05 23-Dec-97
5484 ----------------------
5485
5486 1. Negated character classes containing more than one character were failing if
5487 PCRE_CASELESS was set at run time.
5488
5489
5490 Version 1.04 19-Dec-97
5491 ----------------------
5492
5493 1. Corrected the man page, where some "const" qualifiers had been omitted.
5494
5495 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5496 input syntax.
5497
5498 3. Fixed memory leak which occurred when a regex with back references was
5499 matched with an offsets vector that wasn't big enough. The temporary memory
5500 that is used in this case wasn't being freed if the match failed.
5501
5502 4. Tidied pcretest to ensure it frees memory that it gets.
5503
5504 5. Temporary memory was being obtained in the case where the passed offsets
5505 vector was exactly big enough.
5506
5507 6. Corrected definition of offsetof() from change 5 below.
5508
5509 7. I had screwed up change 6 below and broken the rules for the use of
5510 setjmp(). Now fixed.
5511
5512
5513 Version 1.03 18-Dec-97
5514 ----------------------
5515
5516 1. A erroneous regex with a missing opening parenthesis was correctly
5517 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5518 on some systems.
5519
5520 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5521 it was reported that one broken compiler failed on the former because "code" is
5522 also an independent variable.
5523
5524 3. The erroneous regex a[]b caused an array overrun reference.
5525
5526 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5527 fail on data ending with that character. (It was going on too far, and checking
5528 the next character, typically a binary zero.) This was specific to the
5529 optimized code for single-character negative classes.
5530
5531 5. Added a contributed patch from the TIN world which does the following:
5532
5533 + Add an undef for memmove, in case the the system defines a macro for it.
5534
5535 + Add a definition of offsetof(), in case there isn't one. (I don't know
5536 the reason behind this - offsetof() is part of the ANSI standard - but
5537 it does no harm).
5538
5539 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5540 most of the places where whitespace preceded '#'. I have given up and
5541 allowed the remaining 2 cases to be at the margin.
5542
5543 + Rename some variables in pcre to eliminate shadowing. This seems very
5544 pedantic, but does no harm, of course.
5545
5546 6. Moved the call to setjmp() into its own function, to get rid of warnings
5547 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5548
5549 7. Constructs such as \d{8,} were compiling into the equivalent of
5550 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5551 outcome, but in this particular case used more store than had been allocated,
5552 which caused the bug to be discovered because it threw up an internal error.
5553
5554 8. The debugging code in both pcre and pcretest for outputting the compiled
5555 form of a regex was going wrong in the case of back references followed by
5556 curly-bracketed repeats.
5557
5558
5559 Version 1.02 12-Dec-97
5560 ----------------------
5561
5562 1. Typos in pcre.3 and comments in the source fixed.
5563
5564 2. Applied a contributed patch to get rid of places where it used to remove
5565 'const' from variables, and fixed some signed/unsigned and uninitialized
5566 variable warnings.
5567
5568 3. Added the "runtest" target to Makefile.
5569
5570 4. Set default compiler flag to -O2 rather than just -O.
5571
5572
5573 Version 1.01 19-Nov-97
5574 ----------------------
5575
5576 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5577 like /([ab]*)*/, that is, for classes with more than one character in them.
5578
5579 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5580 as /((?>a*))*/ (a PCRE_EXTRA facility).
5581
5582
5583 Version 1.00 18-Nov-97
5584 ----------------------
5585
5586 1. Added compile-time macros to support systems such as SunOS4 which don't have
5587 memmove() or strerror() but have other things that can be used instead.
5588
5589 2. Arranged that "make clean" removes the executables.
5590
5591
5592 Version 0.99 27-Oct-97
5593 ----------------------
5594
5595 1. Fixed bug in code for optimizing classes with only one character. It was
5596 initializing a 32-byte map regardless, which could cause it to run off the end
5597 of the memory it had got.
5598
5599 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5600
5601
5602 Version 0.98 22-Oct-97
5603 ----------------------
5604
5605 1. Fixed bug in code for handling temporary memory usage when there are more
5606 back references than supplied space in the ovector. This could cause segfaults.
5607
5608
5609 Version 0.97 21-Oct-97
5610 ----------------------
5611
5612 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5613
5614 2. Optimized negated single characters not to use a bit map.
5615
5616 3. Brought error texts together as macro definitions; clarified some of them;
5617 fixed one that was wrong - it said "range out of order" when it meant "invalid
5618 escape sequence".
5619
5620 4. Changed some char * arguments to const char *.
5621
5622 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5623
5624 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5625 pcretest.
5626
5627
5628 Version 0.96 16-Oct-97
5629 ----------------------
5630
5631 1. Added a simple "pgrep" utility to the distribution.
5632
5633 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5634 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5635 where "ddd" means "one or more decimal digits".
5636
5637 3. Fixed serious bug. If a pattern had a back reference, but the call to
5638 pcre_exec() didn't supply a large enough ovector to record the related
5639 identifying subpattern, the match always failed. PCRE now remembers the number
5640 of the largest back reference, and gets some temporary memory in which to save
5641 the offsets during matching if necessary, in order to ensure that
5642 backreferences always work.
5643
5644 4. Increased the compatibility with Perl in a number of ways:
5645
5646 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5647 to request this handling. The option can be set at compile or exec time.
5648
5649 (b) $ matches before a terminating newline by default; an option
5650 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5651 mode). The option can be set at compile or exec time.
5652
5653 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5654 the same as Perl's. If the decimal number it represents is less than 10
5655 or there aren't that many previous left capturing parentheses, an octal
5656 escape is read. Inside a character class, it's always an octal escape,
5657 even if it is a single digit.
5658
5659 (d) An escaped but undefined alphabetic character is taken as a literal,
5660 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5661 escapes.
5662
5663 (e) {0} is now permitted. (The previous item is removed from the compiled
5664 pattern).
5665
5666 5. Changed all the names of code files so that the basic parts are no longer
5667 than 10 characters, and abolished the teeny "globals.c" file.
5668
5669 6. Changed the handling of character classes; they are now done with a 32-byte
5670 bit map always.
5671
5672 7. Added the -d and /D options to pcretest to make it possible to look at the
5673 internals of compilation without having to recompile pcre.
5674
5675
5676 Version 0.95 23-Sep-97
5677 ----------------------
5678
5679 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5680 \x20 at the start of a run of normal characters. These were being treated as
5681 real characters, instead of the source characters being re-checked.
5682
5683
5684 Version 0.94 18-Sep-97
5685 ----------------------
5686
5687 1. The functions are now thread-safe, with the caveat that the global variables
5688 containing pointers to malloc() and free() or alternative functions are the
5689 same for all threads.
5690
5691 2. Get pcre_study() to generate a bitmap of initial characters for non-
5692 anchored patterns when this is possible, and use it if passed to pcre_exec().
5693
5694
5695 Version 0.93 15-Sep-97
5696 ----------------------
5697
5698 1. /(b)|(:+)/ was computing an incorrect first character.
5699
5700 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5701 but not actually doing anything yet.
5702
5703 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5704 as Perl does (e.g. [-az] or [az-]).
5705
5706 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5707 all possible positions.
5708
5709 5. Split up into different modules to avoid including unneeded functions in a
5710 compiled binary. However, compile and exec are still in one module. The "study"
5711 function is split off.
5712
5713 6. The character tables are now in a separate module whose source is generated
5714 by an auxiliary program - but can then be edited by hand if required. There are
5715 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5716 toupper() in the code.
5717
5718 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5719 make them global. Abolish the function for setting them, as the caller can now
5720 set them directly.
5721
5722
5723 Version 0.92 11-Sep-97
5724 ----------------------
5725
5726 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5727 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5728
5729 2. Caseless matching was not working in character classes if the characters in
5730 the pattern were in upper case.
5731
5732 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5733
5734 4. Make PCRE_ANCHORED public and accept as a compile option.
5735
5736 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5737 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5738 pass them.
5739
5740 6. Give an error if bad option bits passed at compile or run time.
5741
5742 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5743 pcretest to cause it to pass that flag.
5744
5745 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5746 options, and the first character, if set.
5747
5748 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5749
5750
5751 Version 0.91 10-Sep-97
5752 ----------------------
5753
5754 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5755 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5756
5757 2. PCRE was looping on encountering an indefinitely repeated back reference to
5758 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5759 Perl does - treats the match as successful.
5760
5761 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5