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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1556 - (show annotations)
Thu May 7 16:31:58 2015 UTC (4 years, 5 months ago) by ph10
File size: 277932 byte(s)
Fix forward referencing bugs.
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
19 Version 8.37 28-April-2015
20 --------------------------
21
22 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
23 for those parentheses to be closed with whatever has been captured so far.
24 However, it was failing to mark any other groups between the hightest
25 capture so far and the currrent group as "unset". Thus, the ovector for
26 those groups contained whatever was previously there. An example is the
27 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
28
29 2. If an assertion condition was quantified with a minimum of zero (an odd
30 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
31
32 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
33 unrecognized modifier, a crash could occur.
34
35 4. An attempt to do global matching in pcretest with a zero-length ovector
36 caused a crash.
37
38 5. Fixed a memory leak during matching that could occur for a subpattern
39 subroutine call (recursive or otherwise) if the number of captured groups
40 that had to be saved was greater than ten.
41
42 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
43 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
44 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
45
46 7. A UTF pattern containing a "not" match of a non-ASCII character and a
47 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
48
49 8. When a pattern is compiled, it remembers the highest back reference so that
50 when matching, if the ovector is too small, extra memory can be obtained to
51 use instead. A conditional subpattern whose condition is a check on a
52 capture having happened, such as, for example in the pattern
53 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
54 setting the highest backreference number. This mattered only if pcre_exec()
55 was called with an ovector that was too small to hold the capture, and there
56 was no other kind of back reference (a situation which is probably quite
57 rare). The effect of the bug was that the condition was always treated as
58 FALSE when the capture could not be consulted, leading to a incorrect
59 behaviour by pcre_exec(). This bug has been fixed.
60
61 9. A reference to a duplicated named group (either a back reference or a test
62 for being set in a conditional) that occurred in a part of the pattern where
63 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
64 to be incorrectly calculated, leading to overwriting.
65
66 10. A mutually recursive set of back references such as (\2)(\1) caused a
67 segfault at study time (while trying to find the minimum matching length).
68 The infinite loop is now broken (with the minimum length unset, that is,
69 zero).
70
71 11. If an assertion that was used as a condition was quantified with a minimum
72 of zero, matching went wrong. In particular, if the whole group had
73 unlimited repetition and could match an empty string, a segfault was
74 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
75 assertions to be quantified, but not if they are being used as conditions,
76 so the above pattern is faulted by Perl. PCRE has now been changed so that
77 it also rejects such patterns.
78
79 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
80 failed to allow the zero-repeat case if pcre2_exec() was called with an
81 ovector too small to capture the group.
82
83 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
84 Red Hat Product Security:
85
86 (a) A crash if /K and /F were both set with the option to save the compiled
87 pattern.
88
89 (b) Another crash if the option to print captured substrings in a callout
90 was combined with setting a null ovector, for example \O\C+ as a subject
91 string.
92
93 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
94 forward reference repeated a large (but limited) number of times within a
95 repeated outer group that has a zero minimum quantifier, caused incorrect
96 code to be compiled, leading to the error "internal error:
97 previously-checked referenced subpattern not found" when an incorrect
98 memory address was read. This bug was reported as "heap overflow",
99 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
100 CVE-2015-2325.
101
102 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
103 call within a group that also contained a recursive back reference caused
104 incorrect code to be compiled. This bug was reported as "heap overflow",
105 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
106 number CVE-2015-2326.
107
108 24. Computing the size of the JIT read-only data in advance has been a source
109 of various issues, and new ones are still appear unfortunately. To fix
110 existing and future issues, size computation is eliminated from the code,
111 and replaced by on-demand memory allocation.
112
113 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
114 adjacent to the end of the range, and the range contained characters with
115 more than one other case, caused incorrect behaviour when compiled in UTF
116 mode. In that example, the range a-j was left out of the class.
117
118 26. Fix JIT compilation of conditional blocks, which assertion
119 is converted to (*FAIL). E.g: /(?(?!))/.
120
121 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
122 discovered by the LLVM fuzzer.
123
124 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
125 when this assertion was used as a condition, for example (?(?!)a|b). In
126 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
127 error about an unsupported item.
128
129 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
130 possessification code could take exponential time to complete. A recursion
131 depth limit of 1000 has been imposed to limit the resources used by this
132 optimization.
133
134 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
135 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
136 because \S ensures they are all in the class. The code for doing this was
137 interacting badly with the code for computing the amount of space needed to
138 compile the pattern, leading to a buffer overflow. This bug was discovered
139 by the LLVM fuzzer.
140
141 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
142 other kinds of group caused stack overflow at compile time. This bug was
143 discovered by the LLVM fuzzer.
144
145 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
146 between a subroutine call and its quantifier was incorrectly compiled,
147 leading to buffer overflow or other errors. This bug was discovered by the
148 LLVM fuzzer.
149
150 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
151 assertion after (?(. The code was failing to check the character after
152 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
153 was discovered by the LLVM fuzzer.
154
155 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
156 a fixed maximum following a group that contains a subroutine reference was
157 incorrectly compiled and could trigger buffer overflow. This bug was
158 discovered by the LLVM fuzzer.
159
160 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
161 caused a stack overflow instead of the diagnosis of a non-fixed length
162 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
163
164 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
165 (e.g. /(?<=\Ka)/) could make pcregrep loop.
166
167 37. There was a similar problem to 36 in pcretest for global matches.
168
169 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
170 and a subsequent item in the pattern caused a non-match, backtracking over
171 the repeated \X did not stop, but carried on past the start of the subject,
172 causing reference to random memory and/or a segfault. There were also some
173 other cases where backtracking after \C could crash. This set of bugs was
174 discovered by the LLVM fuzzer.
175
176 39. The function for finding the minimum length of a matching string could take
177 a very long time if mutual recursion was present many times in a pattern,
178 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
179 method has been implemented. This infelicity was discovered by the LLVM
180 fuzzer.
181
182 40. Static linking against the PCRE library using the pkg-config module was
183 failing on missing pthread symbols.
184
185
186 Version 8.36 26-September-2014
187 ------------------------------
188
189 1. Got rid of some compiler warnings in the C++ modules that were shown up by
190 -Wmissing-field-initializers and -Wunused-parameter.
191
192 2. The tests for quantifiers being too big (greater than 65535) were being
193 applied after reading the number, and stupidly assuming that integer
194 overflow would give a negative number. The tests are now applied as the
195 numbers are read.
196
197 3. Tidy code in pcre_exec.c where two branches that used to be different are
198 now the same.
199
200 4. The JIT compiler did not generate match limit checks for certain
201 bracketed expressions with quantifiers. This may lead to exponential
202 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
203 issue should be resolved now.
204
205 5. Fixed an issue, which occures when nested alternatives are optimized
206 with table jumps.
207
208 6. Inserted two casts and changed some ints to size_t in the light of some
209 reported 64-bit compiler warnings (Bugzilla 1477).
210
211 7. Fixed a bug concerned with zero-minimum possessive groups that could match
212 an empty string, which sometimes were behaving incorrectly in the
213 interpreter (though correctly in the JIT matcher). This pcretest input is
214 an example:
215
216 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
217 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
218
219 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
220 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
221 for an empty string was breaking the inner loop and carrying on at a lower
222 level, when possessive repeated groups should always return to a higher
223 level as they have no backtrack points in them. The empty string test now
224 occurs at the outer level.
225
226 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
227 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
228
229 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
230 doesn't).
231
232 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
233 to hold a pattern to be incorrectly computed (too small) when there were
234 named back references to duplicated names. This could cause "internal
235 error: code overflow" or "double free or corruption" or other memory
236 handling errors.
237
238 11. When named subpatterns had the same prefixes, back references could be
239 confused. For example, in this pattern:
240
241 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
242
243 the reference to 'Name' was incorrectly treated as a reference to a
244 duplicate name.
245
246 12. A pattern such as /^s?c/mi8 where the optional character has more than
247 one "other case" was incorrectly compiled such that it would only try to
248 match starting at "c".
249
250 13. When a pattern starting with \s was studied, VT was not included in the
251 list of possible starting characters; this should have been part of the
252 8.34/18 patch.
253
254 14. If a character class started [\Qx]... where x is any character, the class
255 was incorrectly terminated at the ].
256
257 15. If a pattern that started with a caseless match for a character with more
258 than one "other case" was studied, PCRE did not set up the starting code
259 unit bit map for the list of possible characters. Now it does. This is an
260 optimization improvement, not a bug fix.
261
262 16. The Unicode data tables have been updated to Unicode 7.0.0.
263
264 17. Fixed a number of memory leaks in pcregrep.
265
266 18. Avoid a compiler warning (from some compilers) for a function call with
267 a cast that removes "const" from an lvalue by using an intermediate
268 variable (to which the compiler does not object).
269
270 19. Incorrect code was compiled if a group that contained an internal recursive
271 back reference was optional (had quantifier with a minimum of zero). This
272 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
273 caused segmentation faults because of stack overflows at compile time.
274
275 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
276 group that is quantified with an indefinite repeat, caused a compile-time
277 loop which used up all the system stack and provoked a segmentation fault.
278 This was not the same bug as 19 above.
279
280 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
281 Patch by Mike Frysinger.
282
283
284 Version 8.35 04-April-2014
285 --------------------------
286
287 1. A new flag is set, when property checks are present in an XCLASS.
288 When this flag is not set, PCRE can perform certain optimizations
289 such as studying these XCLASS-es.
290
291 2. The auto-possessification of character sets were improved: a normal
292 and an extended character set can be compared now. Furthermore
293 the JIT compiler optimizes more character set checks.
294
295 3. Got rid of some compiler warnings for potentially uninitialized variables
296 that show up only when compiled with -O2.
297
298 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
299 of a match later then the end of the match. The pcretest program was not
300 handling the case sensibly - it was outputting from the start to the next
301 binary zero. It now reports this situation in a message, and outputs the
302 text from the end to the start.
303
304 5. Fast forward search is improved in JIT. Instead of the first three
305 characters, any three characters with fixed position can be searched.
306 Search order: first, last, middle.
307
308 6. Improve character range checks in JIT. Characters are read by an inprecise
309 function now, which returns with an unknown value if the character code is
310 above a certain threshold (e.g: 256). The only limitation is that the value
311 must be bigger than the threshold as well. This function is useful when
312 the characters above the threshold are handled in the same way.
313
314 7. The macros whose names start with RAWUCHAR are placeholders for a future
315 mode in which only the bottom 21 bits of 32-bit data items are used. To
316 make this more memorable for those maintaining the code, the names have
317 been changed to start with UCHAR21, and an extensive comment has been added
318 to their definition.
319
320 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
321 to the export list in Makefile.am (they were accidentally omitted from the
322 8.34 tarball).
323
324 9. The informational output from pcretest used the phrase "starting byte set"
325 which is inappropriate for the 16-bit and 32-bit libraries. As the output
326 for "first char" and "need char" really means "non-UTF-char", I've changed
327 "byte" to "char", and slightly reworded the output. The documentation about
328 these values has also been (I hope) clarified.
329
330 10. Another JIT related optimization: use table jumps for selecting the correct
331 backtracking path, when more than four alternatives are present inside a
332 bracket.
333
334 11. Empty match is not possible, when the minimum length is greater than zero,
335 and there is no \K in the pattern. JIT should avoid empty match checks in
336 such cases.
337
338 12. In a caseless character class with UCP support, when a character with more
339 than one alternative case was not the first character of a range, not all
340 the alternative cases were added to the class. For example, s and \x{17f}
341 are both alternative cases for S: the class [RST] was handled correctly,
342 but [R-T] was not.
343
344 13. The configure.ac file always checked for pthread support when JIT was
345 enabled. This is not used in Windows, so I have put this test inside a
346 check for the presence of windows.h (which was already tested for).
347
348 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
349 The algorithm provides a way to skip certain starting offsets, and usually
350 faster than linear prefix searches.
351
352 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
353 as for 'fr_FR' and 'french'. For some reason, however, it then used the
354 Windows-specific input and output files, which have 'french' screwed in.
355 So this could never have worked. One of the problems with locales is that
356 they aren't always the same. I have now updated RunTest so that it checks
357 the output of the locale test (test 3) against three different output
358 files, and it allows the test to pass if any one of them matches. With luck
359 this should make the test pass on some versions of Solaris where it was
360 failing. Because of the uncertainty, the script did not used to stop if
361 test 3 failed; it now does. If further versions of a French locale ever
362 come to light, they can now easily be added.
363
364 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
365 there was a message during ./configure, but it did not stop. This now
366 provokes an error. The invalid example in README has been corrected.
367 If a value less than the minimum is given, the minimum value has always
368 been used, but now a warning is given.
369
370 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
371 was a bug in the test system, which is now fixed. Also, the list of various
372 configurations that are tested for each release did not have one with both
373 16/32 bits and --enable-bar-anycrlf. It now does.
374
375 18. pcretest was missing "-C bsr" for displaying the \R default setting.
376
377 19. Little endian PowerPC systems are supported now by the JIT compiler.
378
379 20. The fast forward newline mechanism could enter to an infinite loop on
380 certain invalid UTF-8 input. Although we don't support these cases
381 this issue can be fixed by a performance optimization.
382
383 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
384 not take account if existing stack usage. There is now a new global
385 variable called pcre_stack_guard that can be set to point to an external
386 function to check stack availability. It is called at the start of
387 processing every parenthesized group.
388
389 22. A typo in the code meant that in ungreedy mode the max/min qualifier
390 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
391 not match "ab".
392
393 23. When UTF was disabled, the JIT program reported some incorrect compile
394 errors. These messages are silenced now.
395
396 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
397 compiler.
398
399 25. Change all the temporary files used in RunGrepTest to be different to those
400 used by RunTest so that the tests can be run simultaneously, for example by
401 "make -j check".
402
403
404 Version 8.34 15-December-2013
405 -----------------------------
406
407 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
408 executable memory. Patch inspired by Carsten Klein.
409
410 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
411 this macro is never tested and has no effect, because the work to support
412 coverage involves only compiling and linking options and special targets in
413 the Makefile. The comment in config.h implied that defining the macro would
414 enable coverage support, which is totally false. There was also support for
415 setting this macro in the CMake files (my fault, I just copied it from
416 configure). SUPPORT_GCOV has now been removed.
417
418 3. Make a small performance improvement in strlen16() and strlen32() in
419 pcretest.
420
421 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
422 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
423 these cases). There was also one in pcretest.c.
424
425 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
426
427 6. In UTF mode, the code for checking whether a group could match an empty
428 string (which is used for indefinitely repeated groups to allow for
429 breaking an infinite loop) was broken when the group contained a repeated
430 negated single-character class with a character that occupied more than one
431 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
432 UTF-8 mode). The effect was undefined: the group might or might not be
433 deemed as matching an empty string, or the program might have crashed.
434
435 7. The code for checking whether a group could match an empty string was not
436 recognizing that \h, \H, \v, \V, and \R must match a character.
437
438 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
439 an empty string. If it can, pcretest shows this in its information output.
440
441 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
442 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
443 matched by pcre_exec() without using JIT:
444
445 (a) If the rest of the pattern did not match after a maximal run of
446 grapheme clusters, the code for backing up to try with fewer of them
447 did not always back up over a full grapheme when characters that do not
448 have the modifier quality were involved, e.g. Hangul syllables.
449
450 (b) If the match point in a subject started with modifier character, and
451 there was no match, the code could incorrectly back up beyond the match
452 point, and potentially beyond the first character in the subject,
453 leading to a segfault or an incorrect match result.
454
455 10. A conditional group with an assertion condition could lead to PCRE
456 recording an incorrect first data item for a match if no other first data
457 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
458 first data item, and therefore matched "ca" after "c" instead of at the
459 start.
460
461 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
462 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
463
464 12. The source of pcregrep now includes z/OS-specific code so that it can be
465 compiled for z/OS as part of the special z/OS distribution.
466
467 13. Added the -T and -TM options to pcretest.
468
469 14. The code in pcre_compile.c for creating the table of named capturing groups
470 has been refactored. Instead of creating the table dynamically during the
471 actual compiling pass, the information is remembered during the pre-compile
472 pass (on the stack unless there are more than 20 named groups, in which
473 case malloc() is used) and the whole table is created before the actual
474 compile happens. This has simplified the code (it is now nearly 150 lines
475 shorter) and prepared the way for better handling of references to groups
476 with duplicate names.
477
478 15. A back reference to a named subpattern when there is more than one of the
479 same name now checks them in the order in which they appear in the pattern.
480 The first one that is set is used for the reference. Previously only the
481 first one was inspected. This change makes PCRE more compatible with Perl.
482
483 16. Unicode character properties were updated from Unicode 6.3.0.
484
485 17. The compile-time code for auto-possessification has been refactored, based
486 on a patch by Zoltan Herczeg. It now happens after instead of during
487 compilation. The code is cleaner, and more cases are handled. The option
488 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
489 options in pcretest are provided to set it. It can also be set by
490 (*NO_AUTO_POSSESS) at the start of a pattern.
491
492 18. The character VT has been added to the default ("C" locale) set of
493 characters that match \s and are generally treated as white space,
494 following this same change in Perl 5.18. There is now no difference between
495 "Perl space" and "POSIX space". Whether VT is treated as white space in
496 other locales depends on the locale.
497
498 19. The code for checking named groups as conditions, either for being set or
499 for being recursed, has been refactored (this is related to 14 and 15
500 above). Processing unduplicated named groups should now be as fast at
501 numerical groups, and processing duplicated groups should be faster than
502 before.
503
504 20. Two patches to the CMake build system, by Alexander Barkov:
505
506 (1) Replace the "source" command by "." in CMakeLists.txt because
507 "source" is a bash-ism.
508
509 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
510 without these the CMake build does not work on Solaris.
511
512 21. Perl has changed its handling of \8 and \9. If there is no previously
513 encountered capturing group of those numbers, they are treated as the
514 literal characters 8 and 9 instead of a binary zero followed by the
515 literals. PCRE now does the same.
516
517 22. Following Perl, added \o{} to specify codepoints in octal, making it
518 possible to specify values greater than 0777 and also making them
519 unambiguous.
520
521 23. Perl now gives an error for missing closing braces after \x{... instead of
522 treating the string as literal. PCRE now does the same.
523
524 24. RunTest used to grumble if an inappropriate test was selected explicitly,
525 but just skip it when running all tests. This make it awkward to run ranges
526 of tests when one of them was inappropriate. Now it just skips any
527 inappropriate tests, as it always did when running all tests.
528
529 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
530 character types such as \d or \w, too many callouts were inserted, and the
531 data that they returned was rubbish.
532
533 26. In UCP mode, \s was not matching two of the characters that Perl matches,
534 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
535 were matched by \h. The code has now been refactored so that the lists of
536 the horizontal and vertical whitespace characters used for \h and \v (which
537 are defined only in one place) are now also used for \s.
538
539 27. Add JIT support for the 64 bit TileGX architecture.
540 Patch by Jiong Wang (Tilera Corporation).
541
542 28. Possessive quantifiers for classes (both explicit and automatically
543 generated) now use special opcodes instead of wrapping in ONCE brackets.
544
545 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
546 (because it's meaningless), this was not happening when PCRE_CASELESS was
547 set. Not wrong, but inefficient.
548
549 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
550 properties for \w, \d, etc) is present in a test regex. Otherwise if the
551 test contains no characters greater than 255, Perl doesn't realise it
552 should be using Unicode semantics.
553
554 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
555 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
556 does in Unicode mode.
557
558 32. Added the "forbid" facility to pcretest so that putting tests into the
559 wrong test files can sometimes be quickly detected.
560
561 33. There is now a limit (default 250) on the depth of nesting of parentheses.
562 This limit is imposed to control the amount of system stack used at compile
563 time. It can be changed at build time by --with-parens-nest-limit=xxx or
564 the equivalent in CMake.
565
566 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
567 errors. Perl warns for these when in warning mode, but PCRE has no facility
568 for giving warnings.
569
570 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
571 However, this was not working for (?!) because it is optimized to (*FAIL),
572 for which PCRE does not allow quantifiers. The optimization is now disabled
573 when a quantifier follows (?!). I can't see any use for this, but it makes
574 things uniform.
575
576 36. Perl no longer allows group names to start with digits, so I have made this
577 change also in PCRE. It simplifies the code a bit.
578
579 37. In extended mode, Perl ignores spaces before a + that indicates a
580 possessive quantifier. PCRE allowed a space before the quantifier, but not
581 before the possessive +. It now does.
582
583 38. The use of \K (reset reported match start) within a repeated possessive
584 group such as (a\Kb)*+ was not working.
585
586 40. Document that the same character tables must be used at compile time and
587 run time, and that the facility to pass tables to pcre_exec() and
588 pcre_dfa_exec() is for use only with saved/restored patterns.
589
590 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
591 features for Builds with MSVC:
592
593 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
594 builds). The .rc files can be used to set FileDescription and many other
595 attributes.
596
597 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
598 This allows higher-level build scripts which want .pdb files to avoid
599 hard-coding the exact files needed."
600
601 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
602 mean "start of word" and "end of word", respectively, as a transition aid.
603
604 43. A minimizing repeat of a class containing codepoints greater than 255 in
605 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
606 compiled to use the heap for recursion.
607
608 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
609 is configured.
610
611
612 Version 8.33 28-May-2013
613 ------------------------
614
615 1. Added 'U' to some constants that are compared to unsigned integers, to
616 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
617 variables that are added to signed variables, to ensure the result is
618 signed and can be negated.
619
620 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
621 CMake config files.
622
623 3. Revise the creation of config.h.generic so that all boolean macros are
624 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
625 overriding via -D on the command line possible.
626
627 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
628 to unsigned int is reported to make a quite noticeable speed difference in
629 a specific Windows environment. Testing on Linux did also appear to show
630 some benefit (and it is clearly not harmful). Also fixed the definition of
631 Xop which should be unsigned.
632
633 5. Related to (4), changing the definition of the intermediate variable cc
634 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
635 improvements.
636
637 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
638 unnecessary spaces.
639
640 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
641 and later.
642
643 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
644
645 9. Optimizing fast_forward_start_bits in JIT.
646
647 10. Adding support for callouts in JIT, and fixing some issues revealed
648 during this work. Namely:
649
650 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
651
652 (b) Minimum length was not checked before the matching is started.
653
654 11. The value of capture_last that is passed to callouts was incorrect in some
655 cases when there was a capture on one path that was subsequently abandoned
656 after a backtrack. Also, the capture_last value is now reset after a
657 recursion, since all captures are also reset in this case.
658
659 12. The interpreter no longer returns the "too many substrings" error in the
660 case when an overflowing capture is in a branch that is subsequently
661 abandoned after a backtrack.
662
663 13. In the pathological case when an offset vector of size 2 is used, pcretest
664 now prints out the matched string after a yield of 0 or 1.
665
666 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
667 Only supported by the JIT compiler at the moment.
668
669 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
670
671 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
672 offset of the starting point of the matching process, provided the offsets
673 vector is large enough.
674
675 17. The \A escape now records a lookbehind value of 1, though its execution
676 does not actually inspect the previous character. This is to ensure that,
677 in partial multi-segment matching, at least one character from the old
678 segment is retained when a new segment is processed. Otherwise, if there
679 are no lookbehinds in the pattern, \A might match incorrectly at the start
680 of a new segment.
681
682 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
683
684 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
685 gives some modest performance improvement in 8-bit mode.
686
687 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
688 be expressed in certain programming languages using Universal Character
689 Names.
690
691 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
692 which points out that "non characters" are not "characters that may not
693 appear in Unicode strings" but rather "characters that are reserved for
694 internal use and have only local meaning".
695
696 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
697 there was a conditional group that depended on an assertion, if the
698 assertion was false, the callout that immediately followed the alternation
699 in the condition was skipped when pcre_exec() was used for matching.
700
701 23. Allow an explicit callout to be inserted before an assertion that is the
702 condition for a conditional group, for compatibility with automatic
703 callouts, which always insert a callout at this point.
704
705 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
706 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
707
708 25. (*PRUNE) is now supported by the JIT compiler.
709
710 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
711
712 27. Fix the case where there are two or more SKIPs with arguments that may be
713 ignored.
714
715 28. (*SKIP) is now supported by the JIT compiler.
716
717 29. (*THEN) is now supported by the JIT compiler.
718
719 30. Update RunTest with additional test selector options.
720
721 31. The way PCRE handles backtracking verbs has been changed in two ways.
722
723 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
724 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
725 backtracking. In some cases this makes it more Perl-compatible, but Perl's
726 rather obscure rules do not always do the same thing.
727
728 (2) Previously, backtracking verbs were confined within assertions. This is
729 no longer the case for positive assertions, except for (*ACCEPT). Again,
730 this sometimes improves Perl compatibility, and sometimes does not.
731
732 32. A number of tests that were in test 2 because Perl did things differently
733 have been moved to test 1, because either Perl or PCRE has changed, and
734 these tests are now compatible.
735
736 32. Backtracking control verbs are now handled in the same way in JIT and
737 interpreter.
738
739 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
740 contained a forward subroutine reference caused a compile error.
741
742 34. Auto-detect and optimize limited repetitions in JIT.
743
744 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
745 blocking (*UTF) etc.
746
747 36. In the interpreter, maximizing pattern repetitions for characters and
748 character types now use tail recursion, which reduces stack usage.
749
750 37. The value of the max lookbehind was not correctly preserved if a compiled
751 and saved regex was reloaded on a host of different endianness.
752
753 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
754 of the compiled pattern block, expand the flags field from 16 to 32 bits
755 because it was almost full.
756
757 39. Try madvise first before posix_madvise.
758
759 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
760 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
761 original change locked out all matches of empty strings. This has been
762 changed so that one match of an empty string per line is recognized.
763 Subsequent searches on the same line (for colouring or for --only-matching,
764 for example) do not recognize empty strings.
765
766 41. Applied a user patch to fix a number of spelling mistakes in comments.
767
768 42. Data lines longer than 65536 caused pcretest to crash.
769
770 43. Clarified the data type for length and startoffset arguments for pcre_exec
771 and pcre_dfa_exec in the function-specific man pages, where they were
772 explicitly stated to be in bytes, never having been updated. I also added
773 some clarification to the pcreapi man page.
774
775 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
776 a segmentation fault.
777
778
779 Version 8.32 30-November-2012
780 -----------------------------
781
782 1. Improved JIT compiler optimizations for first character search and single
783 character iterators.
784
785 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
786 Patch by Daniel Richard G.
787
788 3. Single character iterator optimizations in the JIT compiler.
789
790 4. Improved JIT compiler optimizations for character ranges.
791
792 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
793 Reported by Giuseppe D'Angelo.
794
795 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
796 of a line, was being set incorrectly in cases where .* appeared inside
797 atomic brackets at the start of a pattern, or where there was a subsequent
798 *PRUNE or *SKIP.
799
800 7. Improved instruction cache flush for POWER/PowerPC.
801 Patch by Daniel Richard G.
802
803 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
804 grep:
805
806 (a) There is now no limit to the number of patterns to be matched.
807
808 (b) An error is given if a pattern is too long.
809
810 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
811 are now supported.
812
813 (d) --exclude-from and --include-from (multiple use) have been added.
814
815 (e) Exclusions and inclusions now apply to all files and directories, not
816 just to those obtained from scanning a directory recursively.
817
818 (f) Multiple uses of -f and --file-list are now supported.
819
820 (g) In a Windows environment, the default for -d has been changed from
821 "read" (the GNU grep default) to "skip", because otherwise the presence
822 of a directory in the file list provokes an error.
823
824 (h) The documentation has been revised and clarified in places.
825
826 9. Improve the matching speed of capturing brackets.
827
828 10. Changed the meaning of \X so that it now matches a Unicode extended
829 grapheme cluster.
830
831 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
832 out POSIX threads when JIT support is configured.
833
834 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
835
836 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
837 structure could go wrong in environments where size_t is not the same size
838 as int.
839
840 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
841 set.
842
843 15. The EBCDIC support had decayed; later updates to the code had included
844 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
845 general tidy up of EBCDIC-related issues, and the documentation was also
846 not quite right. There is now a test that can be run on ASCII systems to
847 check some of the EBCDIC-related things (but is it not a full test).
848
849 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
850 in a small tidy to the code.
851
852 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
853
854 18. If the --only-matching (-o) option in pcregrep is specified multiple
855 times, each one causes appropriate output. For example, -o1 -o2 outputs the
856 substrings matched by the 1st and 2nd capturing parentheses. A separating
857 string can be specified by --om-separator (default empty).
858
859 19. Improving the first n character searches.
860
861 20. Turn case lists for horizontal and vertical white space into macros so that
862 they are defined only once.
863
864 21. This set of changes together give more compatible Unicode case-folding
865 behaviour for characters that have more than one other case when UCP
866 support is available.
867
868 (a) The Unicode property table now has offsets into a new table of sets of
869 three or more characters that are case-equivalent. The MultiStage2.py
870 script that generates these tables (the pcre_ucd.c file) now scans
871 CaseFolding.txt instead of UnicodeData.txt for character case
872 information.
873
874 (b) The code for adding characters or ranges of characters to a character
875 class has been abstracted into a generalized function that also handles
876 case-independence. In UTF-mode with UCP support, this uses the new data
877 to handle characters with more than one other case.
878
879 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
880 whose other case is greater than 256 are now correctly matched
881 caselessly. Previously, the high codepoint matched the low one, but not
882 vice versa.
883
884 (d) The processing of \h, \H, \v, and \ in character classes now makes use
885 of the new class addition function, using character lists defined as
886 macros alongside the case definitions of 20 above.
887
888 (e) Caseless back references now work with characters that have more than
889 one other case.
890
891 (f) General caseless matching of characters with more than one other case
892 is supported.
893
894 22. Unicode character properties were updated from Unicode 6.2.0
895
896 23. Improved CMake support under Windows. Patch by Daniel Richard G.
897
898 24. Add support for 32-bit character strings, and UTF-32
899
900 25. Major JIT compiler update (code refactoring and bugfixing).
901 Experimental Sparc 32 support is added.
902
903 26. Applied a modified version of Daniel Richard G's patch to create
904 pcre.h.generic and config.h.generic by "make" instead of in the
905 PrepareRelease script.
906
907 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
908 pcre_compile.c when checking for a zero character.
909
910 28. Introducing a native interface for JIT. Through this interface, the compiled
911 machine code can be directly executed. The purpose of this interface is to
912 provide fast pattern matching, so several sanity checks are not performed.
913 However, feature tests are still performed. The new interface provides
914 1.4x speedup compared to the old one.
915
916 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
917 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
918 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
919
920 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
921 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
922 the "old" RFC 2279). Instead, it ended up passing a negative length to
923 pcre_exec().
924
925 31. Add support for GCC's visibility feature to hide internal functions.
926
927 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
928 "unknown -C option" after outputting 0 or 1.
929
930 33. There is now support for generating a code coverage report for the test
931 suite in environments where gcc is the compiler and lcov is installed. This
932 is mainly for the benefit of the developers.
933
934 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
935 unaddressable using valgrind annotations, allowing valgrind to detect
936 invalid memory accesses. This is mainly for the benefit of the developers.
937
938 25. (*UTF) can now be used to start a pattern in any of the three libraries.
939
940 26. Give configure error if --enable-cpp but no C++ compiler found.
941
942
943 Version 8.31 06-July-2012
944 -------------------------
945
946 1. Fixing a wrong JIT test case and some compiler warnings.
947
948 2. Removed a bashism from the RunTest script.
949
950 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
951 to unsigned type, result still unsigned" that was given by an MS compiler
952 on encountering the code "-sizeof(xxx)".
953
954 4. Partial matching support is added to the JIT compiler.
955
956 5. Fixed several bugs concerned with partial matching of items that consist
957 of more than one character:
958
959 (a) /^(..)\1/ did not partially match "aba" because checking references was
960 done on an "all or nothing" basis. This also applied to repeated
961 references.
962
963 (b) \R did not give a hard partial match if \r was found at the end of the
964 subject.
965
966 (c) \X did not give a hard partial match after matching one or more
967 characters at the end of the subject.
968
969 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
970 a partial match for the string "\r".
971
972 (e) When newline was set to CRLF, the metacharacter "." did not recognize
973 a partial match for a CR character at the end of the subject string.
974
975 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
976 running pcretest, the text "(JIT)" added to the output whenever JIT is
977 actually used to run the match.
978
979 7. Individual JIT compile options can be set in pcretest by following -s+[+]
980 or /S+[+] with a digit between 1 and 7.
981
982 8. OP_NOT now supports any UTF character not just single-byte ones.
983
984 9. (*MARK) control verb is now supported by the JIT compiler.
985
986 10. The command "./RunTest list" lists the available tests without actually
987 running any of them. (Because I keep forgetting what they all are.)
988
989 11. Add PCRE_INFO_MAXLOOKBEHIND.
990
991 12. Applied a (slightly modified) user-supplied patch that improves performance
992 when the heap is used for recursion (compiled with --disable-stack-for-
993 recursion). Instead of malloc and free for each heap frame each time a
994 logical recursion happens, frames are retained on a chain and re-used where
995 possible. This sometimes gives as much as 30% improvement.
996
997 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
998 call.
999
1000 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1001
1002 15. It is now possible to link pcretest with libedit as an alternative to
1003 libreadline.
1004
1005 16. (*COMMIT) control verb is now supported by the JIT compiler.
1006
1007 17. The Unicode data tables have been updated to Unicode 6.1.0.
1008
1009 18. Added --file-list option to pcregrep.
1010
1011 19. Added binary file support to pcregrep, including the -a, --binary-files,
1012 -I, and --text options.
1013
1014 20. The madvise function is renamed for posix_madvise for QNX compatibility
1015 reasons. Fixed by Giuseppe D'Angelo.
1016
1017 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1018
1019 22. Changed the option for creating symbolic links for 16-bit man pages from
1020 -s to -sf so that re-installing does not cause issues.
1021
1022 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1023
1024 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1025 matches in certain environments (the workspace was not being correctly
1026 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1027 some of the workspace data at the beginning of a restart.
1028
1029 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1030 was not doing so when it should - probably a typo introduced by SVN 528
1031 (change 8.10/14).
1032
1033 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1034 \w+ when the character tables indicated that \x{c4} was a word character.
1035 There were several related cases, all because the tests for doing a table
1036 lookup were testing for characters less than 127 instead of 255.
1037
1038 27. If a pattern contains capturing parentheses that are not used in a match,
1039 their slots in the ovector are set to -1. For those that are higher than
1040 any matched groups, this happens at the end of processing. In the case when
1041 there were back references that the ovector was too small to contain
1042 (causing temporary malloc'd memory to be used during matching), and the
1043 highest capturing number was not used, memory off the end of the ovector
1044 was incorrectly being set to -1. (It was using the size of the temporary
1045 memory instead of the true size.)
1046
1047 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1048 ovector size, it uses memory at the end of the block that it has got.
1049
1050 29. Check for an overlong MARK name and give an error at compile time. The
1051 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1052
1053 30. JIT compiler update.
1054
1055 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1056 Rill for the patch.
1057
1058 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1059
1060 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1061
1062 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1063 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1064 was enabled.
1065
1066 35. Improve JIT code generation for greedy plus quantifier.
1067
1068 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1069 1 to "aa" instead of to an empty string. The bug affected repeated groups
1070 that could potentially match an empty string.
1071
1072 37. Optimizing single character iterators in JIT.
1073
1074 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1075 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1076 codepoints that are too big for the mode are faulted, and in a UTF mode,
1077 disallowed codepoints are also faulted.
1078
1079 39. If PCRE was compiled with UTF support, in three places in the DFA
1080 matcher there was code that should only have been obeyed in UTF mode, but
1081 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1082 processing when bytes with values greater than 127 were present. In 16-bit
1083 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1084 both cases the values are those that cannot be the first data item in a UTF
1085 character. The three items that might have provoked this were recursions,
1086 possessively repeated groups, and atomic groups.
1087
1088 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1089 and pcregrep, because some OS require shared objects to be explicitly
1090 passed to ld, causing the link step to fail if they are not.
1091
1092 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1093 mode, patterns that started with \h* or \R* might be incorrectly matched.
1094
1095
1096 Version 8.30 04-February-2012
1097 -----------------------------
1098
1099 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1100 name is defined in ctype.h.
1101
1102 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1103 only in quite long subpatterns.
1104
1105 3. Removed the function pcre_info(), which has been obsolete and deprecated
1106 since it was replaced by pcre_fullinfo() in February 2000.
1107
1108 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1109 match a (*MARK), and the match failed at the start of the subject, a
1110 reference to memory before the start of the subject could occur. This bug
1111 was introduced by fix 17 of release 8.21.
1112
1113 5. A reference to an unset group with zero minimum repetition was giving
1114 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1115 /(another)?(\1?)test/ matched against "hello world test". This bug was
1116 introduced in release 8.13.
1117
1118 6. Add support for 16-bit character strings (a large amount of work involving
1119 many changes and refactorings).
1120
1121 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1122 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1123 from a file.
1124
1125 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1126 rounding is not applied in this particular case).
1127
1128 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1129 if they appear, or are escaped, in patterns.
1130
1131 10. Get rid of a number of -Wunused-but-set-variable warnings.
1132
1133 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1134 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1135 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1136 also returns the mark "x". This bug applied to capturing parentheses,
1137 non-capturing parentheses, and atomic parentheses. It also applied to some
1138 assertions.
1139
1140 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1141 information out of configure.ac instead of relying on pcre.h.generic, which
1142 is not stored in the repository.
1143
1144 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1145 -lreadline.
1146
1147 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1148
1149 15. Applied Graycode's patch to put the top-level frame on the stack rather
1150 than the heap when not using the stack for recursion. This gives a
1151 performance improvement in many cases when recursion is not deep.
1152
1153 16. Experimental code added to "pcretest -C" to output the stack frame size.
1154
1155
1156 Version 8.21 12-Dec-2011
1157 ------------------------
1158
1159 1. Updating the JIT compiler.
1160
1161 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1162 are added as well.
1163
1164 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1165 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1166 calling _pcre_jit_exec. Some extra comments are added.
1167
1168 4. (*MARK) settings inside atomic groups that do not contain any capturing
1169 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1170 was introduced by change 18 for 8.20.
1171
1172 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1173 ECMA-262 standard.
1174
1175 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1176 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1177 This bug was probably introduced by change 9 of 8.13.
1178
1179 7. While fixing 6 above, I noticed that a number of other items were being
1180 incorrectly rejected as "not fixed length". This arose partly because newer
1181 opcodes had not been added to the fixed-length checking code. I have (a)
1182 corrected the bug and added tests for these items, and (b) arranged for an
1183 error to occur if an unknown opcode is encountered while checking for fixed
1184 length instead of just assuming "not fixed length". The items that were
1185 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1186 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1187 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1188
1189 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1190 being incorrectly compiled and would have given unpredicatble results.
1191
1192 9. A possessively repeated subpattern with minimum repeat count greater than
1193 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1194 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1195 the first (A) could occur when it should not.
1196
1197 10. Add a cast and remove a redundant test from the code.
1198
1199 11. JIT should use pcre_malloc/pcre_free for allocation.
1200
1201 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1202 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1203 is anything other than /usr, -L is still shown).
1204
1205 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1206
1207 14. Perl does not support \N without a following name in a [] class; PCRE now
1208 also gives an error.
1209
1210 15. If a forward reference was repeated with an upper limit of around 2000,
1211 it caused the error "internal error: overran compiling workspace". The
1212 maximum number of forward references (including repeats) was limited by the
1213 internal workspace, and dependent on the LINK_SIZE. The code has been
1214 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1215 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1216 of around 200,000 forward references. While doing this, I also speeded up
1217 the filling in of repeated forward references.
1218
1219 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1220 incorrectly expecting the subject to contain another "a" after the start.
1221
1222 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1223 in the match, the SKIP should be ignored. This was not happening; instead
1224 the SKIP was being treated as NOMATCH. For patterns such as
1225 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1226 tested.
1227
1228 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1229 now much more compatible with Perl, in particular in cases where the result
1230 is a non-match for a non-anchored pattern. For example, if
1231 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1232 "m", where previously it did not return a name. A side effect of this
1233 change is that for partial matches, the last encountered mark name is
1234 returned, as for non matches. A number of tests that were previously not
1235 Perl-compatible have been moved into the Perl-compatible test files. The
1236 refactoring has had the pleasing side effect of removing one argument from
1237 the match() function, thus reducing its stack requirements.
1238
1239 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1240 subsequent uses of /S (without +) incorrectly behaved like /S+.
1241
1242 21. Retrieve executable code size support for the JIT compiler and fixing
1243 some warnings.
1244
1245 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1246 not work when the shorter character appeared right at the end of the
1247 subject string.
1248
1249 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1250 systems.
1251
1252 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1253 output it when the /M option is used in pcretest.
1254
1255 25. The CheckMan script was not being included in the distribution. Also, added
1256 an explicit "perl" to run Perl scripts from the PrepareRelease script
1257 because this is reportedly needed in Windows.
1258
1259 26. If study data was being save in a file and studying had not found a set of
1260 "starts with" bytes for the pattern, the data written to the file (though
1261 never used) was taken from uninitialized memory and so caused valgrind to
1262 complain.
1263
1264 27. Updated RunTest.bat as provided by Sheri Pierce.
1265
1266 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1267
1268 29. Computation of memory usage for the table of capturing group names was
1269 giving an unnecessarily large value.
1270
1271
1272 Version 8.20 21-Oct-2011
1273 ------------------------
1274
1275 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1276 a POSIX class. After further experiments with Perl, which convinced me that
1277 Perl has bugs and confusions, a closing square bracket is no longer allowed
1278 in a POSIX name. This bug also affected patterns with classes that started
1279 with full stops.
1280
1281 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1282 captured substring, but while checking the failing first alternative,
1283 substring 1 is temporarily captured. If the output vector supplied to
1284 pcre_exec() was not big enough for this capture, the yield of the function
1285 was still zero ("insufficient space for captured substrings"). This cannot
1286 be totally fixed without adding another stack variable, which seems a lot
1287 of expense for a edge case. However, I have improved the situation in cases
1288 such as /(a)(b)x|abc/ matched against "abc", where the return code
1289 indicates that fewer than the maximum number of slots in the ovector have
1290 been set.
1291
1292 3. Related to (2) above: when there are more back references in a pattern than
1293 slots in the output vector, pcre_exec() uses temporary memory during
1294 matching, and copies in the captures as far as possible afterwards. It was
1295 using the entire output vector, but this conflicts with the specification
1296 that only 2/3 is used for passing back captured substrings. Now it uses
1297 only the first 2/3, for compatibility. This is, of course, another edge
1298 case.
1299
1300 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1301 main code base, and can be used by building with --enable-jit. When this is
1302 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1303 runtime --no-jit option is given.
1304
1305 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1306 ovector, the return from the function was zero, implying that there were
1307 other matches that did not fit. The correct "exactly full" value is now
1308 returned.
1309
1310 6. If a subpattern that was called recursively or as a subroutine contained
1311 (*PRUNE) or any other control that caused it to give a non-standard return,
1312 invalid errors such as "Error -26 (nested recursion at the same subject
1313 position)" or even infinite loops could occur.
1314
1315 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1316 computing the minimum length on reaching *ACCEPT, and so ended up with the
1317 wrong value of 1 rather than 0. Further investigation indicates that
1318 computing a minimum subject length in the presence of *ACCEPT is difficult
1319 (think back references, subroutine calls), and so I have changed the code
1320 so that no minimum is registered for a pattern that contains *ACCEPT.
1321
1322 8. If (*THEN) was present in the first (true) branch of a conditional group,
1323 it was not handled as intended. [But see 16 below.]
1324
1325 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1326 Sheri Pierce.
1327
1328 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1329 the first byte in a match must be "a".
1330
1331 11. Change 17 for 8.13 increased the recursion depth for patterns like
1332 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1333 pattern contains any instances of (*THEN). If it does not, the old
1334 optimizations are restored. It would be nice to do this on a per-group
1335 basis, but at the moment that is not feasible.
1336
1337 12. In some environments, the output of pcretest -C is CRLF terminated. This
1338 broke RunTest's code that checks for the link size. A single white space
1339 character after the value is now allowed for.
1340
1341 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1342 For "fr", it uses the Windows-specific input and output files.
1343
1344 14. If (*THEN) appeared in a group that was called recursively or as a
1345 subroutine, it did not work as intended. [But see next item.]
1346
1347 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1348 pattern fragments (but not containing any | characters). If A and B are
1349 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1350 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1351 D. In other words, Perl considers parentheses that do not contain any |
1352 characters to be part of a surrounding alternative, whereas PCRE was
1353 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1354 differently. PCRE now behaves in the same way as Perl, except in the case
1355 of subroutine/recursion calls such as (?1) which have in any case always
1356 been different (but PCRE had them first :-).
1357
1358 16. Related to 15 above: Perl does not treat the | in a conditional group as
1359 creating alternatives. Such a group is treated in the same way as an
1360 ordinary group without any | characters when processing (*THEN). PCRE has
1361 been changed to match Perl's behaviour.
1362
1363 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1364 RunGrepTest script failed.
1365
1366 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1367 inevitable for groups that contain captures, but it can lead to a lot of
1368 stack use in large patterns. The old behaviour has been restored for atomic
1369 groups that do not contain any capturing parentheses.
1370
1371 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1372 suppress the check for a minimum subject length at run time. (If it was
1373 given to pcre_exec() or pcre_dfa_exec() it did work.)
1374
1375 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1376 fail to work when decoding hex characters in data strings in EBCDIC
1377 environments.
1378
1379 21. It appears that in at least one Mac OS environment, the isxdigit() function
1380 is implemented as a macro that evaluates to its argument more than once,
1381 contravening the C 90 Standard (I haven't checked a later standard). There
1382 was an instance in pcretest which caused it to go wrong when processing
1383 \x{...} escapes in subject strings. The has been rewritten to avoid using
1384 things like p++ in the argument of isxdigit().
1385
1386
1387 Version 8.13 16-Aug-2011
1388 ------------------------
1389
1390 1. The Unicode data tables have been updated to Unicode 6.0.0.
1391
1392 2. Two minor typos in pcre_internal.h have been fixed.
1393
1394 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1395 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1396 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1397
1398 4. There were a number of related bugs in the code for matching backrefences
1399 caselessly in UTF-8 mode when codes for the characters concerned were
1400 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1401 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1402 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1403 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1404 2-byte code at the end of the subject (it thought there wasn't enough data
1405 left).
1406
1407 5. Comprehensive information about what went wrong is now returned by
1408 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1409 as the output vector has at least 2 elements. The offset of the start of
1410 the failing character and a reason code are placed in the vector.
1411
1412 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1413 now returned is for the first byte of the failing character, instead of the
1414 last byte inspected. This is an incompatible change, but I hope it is small
1415 enough not to be a problem. It makes the returned offset consistent with
1416 pcre_exec() and pcre_dfa_exec().
1417
1418 7. pcretest now gives a text phrase as well as the error number when
1419 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1420 failure, the offset and reason code are output.
1421
1422 8. When \R was used with a maximizing quantifier it failed to skip backwards
1423 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1424 back over a single character (\n). This seems wrong (because it treated the
1425 two characters as a single entity when going forwards), conflicts with the
1426 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1427 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1428 has been changed.
1429
1430 9. Some internal refactoring has changed the processing so that the handling
1431 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1432 time (the PCRE_DOTALL option was changed this way some time ago: version
1433 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1434 which was always a bit of a fudge. It also means that there is one less
1435 argument for the match() function, which reduces its stack requirements
1436 slightly. This change also fixes an incompatibility with Perl: the pattern
1437 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1438
1439 10. More internal refactoring has drastically reduced the number of recursive
1440 calls to match() for possessively repeated groups such as (abc)++ when
1441 using pcre_exec().
1442
1443 11. While implementing 10, a number of bugs in the handling of groups were
1444 discovered and fixed:
1445
1446 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1447 (a|)*(?1) gave a compile-time internal error.
1448 ((a|)+)+ did not notice that the outer group could match an empty string.
1449 (^a|^)+ was not marked as anchored.
1450 (.*a|.*)+ was not marked as matching at start or after a newline.
1451
1452 12. Yet more internal refactoring has removed another argument from the match()
1453 function. Special calls to this function are now indicated by setting a
1454 value in a variable in the "match data" data block.
1455
1456 13. Be more explicit in pcre_study() instead of relying on "default" for
1457 opcodes that mean there is no starting character; this means that when new
1458 ones are added and accidentally left out of pcre_study(), testing should
1459 pick them up.
1460
1461 14. The -s option of pcretest has been documented for ages as being an old
1462 synonym of -m (show memory usage). I have changed it to mean "force study
1463 for every regex", that is, assume /S for every regex. This is similar to -i
1464 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1465 using it. It makes it easier to run collections of tests with and without
1466 study enabled, and thereby test pcre_study() more easily. All the standard
1467 tests are now run with and without -s (but some patterns can be marked as
1468 "never study" - see 20 below).
1469
1470 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1471 restoration of the capturing data to the outer values was not happening
1472 correctly.
1473
1474 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1475 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1476 pattern had matched an empty string, and so incorrectly returned a no
1477 match.
1478
1479 17. There was optimizing code for the last branch of non-capturing parentheses,
1480 and also for the obeyed branch of a conditional subexpression, which used
1481 tail recursion to cut down on stack usage. Unfortunately, now that there is
1482 the possibility of (*THEN) occurring in these branches, tail recursion is
1483 no longer possible because the return has to be checked for (*THEN). These
1484 two optimizations have therefore been removed. [But see 8.20/11 above.]
1485
1486 18. If a pattern containing \R was studied, it was assumed that \R always
1487 matched two bytes, thus causing the minimum subject length to be
1488 incorrectly computed because \R can also match just one byte.
1489
1490 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1491 was incorrectly computed.
1492
1493 20. If /S is present twice on a test pattern in pcretest input, it now
1494 *disables* studying, thereby overriding the use of -s on the command line
1495 (see 14 above). This is necessary for one or two tests to keep the output
1496 identical in both cases.
1497
1498 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1499 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1500
1501 22. When an atomic group that contained a capturing parenthesis was
1502 successfully matched, but the branch in which it appeared failed, the
1503 capturing was not being forgotten if a higher numbered group was later
1504 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1505 group 1 to "a", when in fact it should be unset. This applied to multi-
1506 branched capturing and non-capturing groups, repeated or not, and also to
1507 positive assertions (capturing in negative assertions does not happen
1508 in PCRE) and also to nested atomic groups.
1509
1510 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1511 subject after a captured substring, to make it easier to tell which of a
1512 number of identical substrings has been captured.
1513
1514 24. The way atomic groups are processed by pcre_exec() has been changed so that
1515 if they are repeated, backtracking one repetition now resets captured
1516 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1517 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1518 "aaa". Previously, it would have been "a". As part of this code
1519 refactoring, the way recursive calls are handled has also been changed.
1520
1521 25. If an assertion condition captured any substrings, they were not passed
1522 back unless some other capturing happened later. For example, if
1523 (?(?=(a))a) was matched against "a", no capturing was returned.
1524
1525 26. When studying a pattern that contained subroutine calls or assertions,
1526 the code for finding the minimum length of a possible match was handling
1527 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1528 group 1 called group 2 while simultaneously a separate group 2 called group
1529 1). A stack overflow occurred in this case. I have fixed this by limiting
1530 the recursion depth to 10.
1531
1532 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1533 Fortmann. This supports explicit test numbers on the command line, and has
1534 argument validation and error reporting.
1535
1536 28. An instance of \X with an unlimited repeat could fail if at any point the
1537 first character it looked at was a mark character.
1538
1539 29. Some minor code refactoring concerning Unicode properties and scripts
1540 should reduce the stack requirement of match() slightly.
1541
1542 30. Added the '=' option to pcretest to check the setting of unused capturing
1543 slots at the end of the pattern, which are documented as being -1, but are
1544 not included in the return count.
1545
1546 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1547 compiled something random. Now it gives a compile-time error (as does
1548 Perl).
1549
1550 32. A *MARK encountered during the processing of a positive assertion is now
1551 recorded and passed back (compatible with Perl).
1552
1553 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1554 had alternative anchored branches, the search for a second match in a line
1555 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1556 matched the line "0102" twice. The same bug affected patterns that started
1557 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1558 twice.
1559
1560 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1561 does, and because of capturing effects, quantifying parenthesized
1562 assertions may at times be useful. Quantifiers are now allowed for
1563 parenthesized assertions.
1564
1565 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1566
1567 36. \g was being checked for fancy things in a character class, when it should
1568 just be a literal "g".
1569
1570 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1571 appearance of a nested POSIX class supersedes an apparent external class.
1572 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1573 unescaped square brackets may also appear as part of class names. For
1574 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1575 more like Perl. (But see 8.20/1 above.)
1576
1577 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1578 was because it thought it was \N{name}, which is not supported).
1579
1580 39. Add minix to OS list not supporting the -S option in pcretest.
1581
1582 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1583 cannot analyze patterns in sufficient detail to catch mutual recursions
1584 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1585 subgroup is called recursively as a subpattern for a second time at the
1586 same position in the subject string. In previous releases this might have
1587 been caught by the recursion limit, or it might have run out of stack.
1588
1589 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1590 happen only once. PCRE was, however incorrectly giving a compile time error
1591 "recursive call could loop indefinitely" because it cannot analyze the
1592 pattern in sufficient detail. The compile time test no longer happens when
1593 PCRE is compiling a conditional subpattern, but actual runaway loops are
1594 now caught at runtime (see 40 above).
1595
1596 42. It seems that Perl allows any characters other than a closing parenthesis
1597 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1598 has been changed to be the same.
1599
1600 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1601 as not to get warnings when autogen.sh is called. Also changed
1602 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1603
1604 44. To help people who use pcregrep to scan files containing exceedingly long
1605 lines, the following changes have been made:
1606
1607 (a) The default value of the buffer size parameter has been increased from
1608 8K to 20K. (The actual buffer used is three times this size.)
1609
1610 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1611 PCRE is built.
1612
1613 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1614 to be set at run time.
1615
1616 (d) Numerical values in pcregrep options can be followed by K or M, for
1617 example --buffer-size=50K.
1618
1619 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1620 given and the return code is set to 2.
1621
1622 45. Add a pointer to the latest mark to the callout data block.
1623
1624 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1625 partial match of an empty string instead of no match. This was specific to
1626 the use of ".".
1627
1628 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1629 complete match instead of a partial match. This bug was dependent on both
1630 the PCRE_UTF8 and PCRE_DOTALL options being set.
1631
1632 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1633 starting byte set, because \b was not being ignored.
1634
1635
1636 Version 8.12 15-Jan-2011
1637 ------------------------
1638
1639 1. Fixed some typos in the markup of the man pages, and wrote a script that
1640 checks for such things as part of the documentation building process.
1641
1642 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1643 --match-limit and --recursion-limit options (added for 8.11). In
1644 particular, this made one of the standard tests fail. (The integer value
1645 went into the wrong half of a long int.)
1646
1647 3. If the --colour option was given to pcregrep with -v (invert match), it
1648 did strange things, either producing crazy output, or crashing. It should,
1649 of course, ignore a request for colour when reporting lines that do not
1650 match.
1651
1652 4. Another pcregrep bug caused similar problems if --colour was specified with
1653 -M (multiline) and the pattern match finished with a line ending.
1654
1655 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1656 matched in multiline mode, the following line was shown as part of the
1657 match. This seems wrong, so I have changed it.
1658
1659 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1660 the check for further matches in the same line (so they could be coloured)
1661 to overrun the end of the current line. If another match was found, it was
1662 incorrectly shown (and then shown again when found in the next line).
1663
1664 7. If pcregrep was compiled under Windows, there was a reference to the
1665 function pcregrep_exit() before it was defined. I am assuming this was
1666 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1667 reported by a user. I've moved the definition above the reference.
1668
1669
1670 Version 8.11 10-Dec-2010
1671 ------------------------
1672
1673 1. (*THEN) was not working properly if there were untried alternatives prior
1674 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1675 backtracked to try for "b" instead of moving to the next alternative branch
1676 at the same level (in this case, to look for "c"). The Perl documentation
1677 is clear that when (*THEN) is backtracked onto, it goes to the "next
1678 alternative in the innermost enclosing group".
1679
1680 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1681 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1682 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1683 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1684 (*THEN).
1685
1686 3. If \s appeared in a character class, it removed the VT character from
1687 the class, even if it had been included by some previous item, for example
1688 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1689 of \s, but is part of the POSIX "space" class.)
1690
1691 4. A partial match never returns an empty string (because you can always
1692 match an empty string at the end of the subject); however the checking for
1693 an empty string was starting at the "start of match" point. This has been
1694 changed to the "earliest inspected character" point, because the returned
1695 data for a partial match starts at this character. This means that, for
1696 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1697 (previously it gave "no match").
1698
1699 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1700 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1701 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1702 has an implication that the given string is incomplete (because a partial
1703 match is preferred over a full match). For this reason, these items now
1704 give a partial match in this situation. [Aside: previously, the one case
1705 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1706 match rather than a full match, which was wrong by the old rules, but is
1707 now correct.]
1708
1709 6. There was a bug in the handling of #-introduced comments, recognized when
1710 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1711 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1712 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1713 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1714 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1715 places in pcre_compile().
1716
1717 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1718 comments when looking ahead for named forward references to subpatterns,
1719 the only newline sequence it recognized was NL. It now handles newlines
1720 according to the set newline convention.
1721
1722 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1723 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1724 cater for a lack of strerror(). These oversights have been fixed.
1725
1726 9. Added --match-limit and --recursion-limit to pcregrep.
1727
1728 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1729
1730 11. When the -o option was used, pcregrep was setting a return code of 1, even
1731 when matches were found, and --line-buffered was not being honoured.
1732
1733 12. Added an optional parentheses number to the -o and --only-matching options
1734 of pcregrep.
1735
1736 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1737 can match an empty string. The code to do it in pcretest and pcredemo
1738 needed fixing:
1739
1740 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1741 only one byte after an empty string match just before CRLF (this case
1742 just got forgotten; "any" and "anycrlf" were OK).
1743
1744 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1745 mode when an empty string match preceded an ASCII character followed by
1746 a non-ASCII character. (The code for advancing by one character rather
1747 than one byte was nonsense.)
1748
1749 (c) The pcredemo.c sample program did not have any code at all to handle
1750 the cases when CRLF is a valid newline sequence.
1751
1752 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1753 as a starting offset was within the subject string. There is now a new
1754 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1755 negative or greater than the length of the string. In order to test this,
1756 pcretest is extended to allow the setting of negative starting offsets.
1757
1758 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1759 starting offset points to the beginning of a UTF-8 character was
1760 unnecessarily clumsy. I tidied it up.
1761
1762 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1763 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1764
1765 17. Nobody had reported that the --include_dir option, which was added in
1766 release 7.7 should have been called --include-dir (hyphen, not underscore)
1767 for compatibility with GNU grep. I have changed it to --include-dir, but
1768 left --include_dir as an undocumented synonym, and the same for
1769 --exclude-dir, though that is not available in GNU grep, at least as of
1770 release 2.5.4.
1771
1772 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1773 characters from a string of bytes) have been redefined so as not to use
1774 loops, in order to improve performance in some environments. At the same
1775 time, I abstracted some of the common code into auxiliary macros to save
1776 repetition (this should not affect the compiled code).
1777
1778 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1779 compile-time error is now given if \c is not followed by an ASCII
1780 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1781 different, and any byte value is allowed.)
1782
1783 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1784 START_OPTIMIZE option, which is now allowed at compile time - but just
1785 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1786 to pcregrep and other applications that have no direct access to PCRE
1787 options. The new /Y option in pcretest sets this option when calling
1788 pcre_compile().
1789
1790 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1791 back references. Groups containing recursive back references were forced to
1792 be atomic by that change, but in the case of named groups, the amount of
1793 memory required was incorrectly computed, leading to "Failed: internal
1794 error: code overflow". This has been fixed.
1795
1796 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1797 pcretest.c, to avoid build problems in some Borland environments.
1798
1799
1800 Version 8.10 25-Jun-2010
1801 ------------------------
1802
1803 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1804 THEN.
1805
1806 2. (*ACCEPT) was not working when inside an atomic group.
1807
1808 3. Inside a character class, \B is treated as a literal by default, but
1809 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1810 causes the error). The code is unchanged, but I tidied the documentation.
1811
1812 4. Inside a character class, PCRE always treated \R and \X as literals,
1813 whereas Perl faults them if its -w option is set. I have changed PCRE so
1814 that it faults them when PCRE_EXTRA is set.
1815
1816 5. Added support for \N, which always matches any character other than
1817 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1818
1819 6. When compiling pcregrep with newer versions of gcc which may have
1820 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1821 declared with attribute warn_unused_result" were given. Just casting the
1822 result to (void) does not stop the warnings; a more elaborate fudge is
1823 needed. I've used a macro to implement this.
1824
1825 7. Minor change to pcretest.c to avoid a compiler warning.
1826
1827 8. Added four artifical Unicode properties to help with an option to make
1828 \s etc use properties (see next item). The new properties are: Xan
1829 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1830
1831 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1832 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1833 this option. Modified pcretest to add /W to test this facility. Added
1834 REG_UCP to make it available via the POSIX interface.
1835
1836 10. Added --line-buffered to pcregrep.
1837
1838 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1839 studied, and the match started with a letter with a code point greater than
1840 127 whose first byte was different to the first byte of the other case of
1841 the letter, the other case of this starting letter was not recognized
1842 (#976).
1843
1844 12. If a pattern that was studied started with a repeated Unicode property
1845 test, for example, \p{Nd}+, there was the theoretical possibility of
1846 setting up an incorrect bitmap of starting bytes, but fortunately it could
1847 not have actually happened in practice until change 8 above was made (it
1848 added property types that matched character-matching opcodes).
1849
1850 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1851 possible starting bytes for non-anchored patterns.
1852
1853 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1854 \R, and also a number of cases that involve Unicode properties, both
1855 explicit and implicit when PCRE_UCP is set.
1856
1857 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1858 input, it could crash or give wrong results if characters with values
1859 greater than 0xc0 were present in the subject string. (Detail: it assumed
1860 UTF-8 input when processing these items.)
1861
1862 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1863 size_t is 64-bit (#991).
1864
1865 17. Added a check for running out of memory when PCRE is compiled with
1866 --disable-stack-for-recursion (#990).
1867
1868 18. If the last data line in a file for pcretest does not have a newline on
1869 the end, a newline was missing in the output.
1870
1871 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1872 less than 128) in its various bitmaps. However, there is a facility for
1873 generating tables according to the current locale when PCRE is compiled. It
1874 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1875 characters, are recognized by isspace() and therefore were getting set in
1876 these tables, and indeed these tables seem to approximate to ISO 8859. This
1877 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1878 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1879 which of course cannot start UTF-8 characters. I have changed the code so
1880 that only real ASCII characters (less than 128) and the correct starting
1881 bytes for UTF-8 encodings are set for characters greater than 127 when in
1882 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1883 altogether.)
1884
1885 20. Added the /T option to pcretest so as to be able to run tests with non-
1886 standard character tables, thus making it possible to include the tests
1887 used for 19 above in the standard set of tests.
1888
1889 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1890 reference to a subpattern the other side of a comment that contains an
1891 opening parenthesis caused either an internal compiling error, or a
1892 reference to the wrong subpattern.
1893
1894
1895 Version 8.02 19-Mar-2010
1896 ------------------------
1897
1898 1. The Unicode data tables have been updated to Unicode 5.2.0.
1899
1900 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1901 configured.
1902
1903 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1904 original author of that file, following a query about its status.
1905
1906 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1907 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1908
1909 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1910 quantifier applied to a forward-referencing subroutine call, could compile
1911 incorrect code or give the error "internal error: previously-checked
1912 referenced subpattern not found".
1913
1914 6. Both MS Visual Studio and Symbian OS have problems with initializing
1915 variables to point to external functions. For these systems, therefore,
1916 pcre_malloc etc. are now initialized to local functions that call the
1917 relevant global functions.
1918
1919 7. There were two entries missing in the vectors called coptable and poptable
1920 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1921 I've fixed the data, and added a kludgy way of testing at compile time that
1922 the lengths are correct (equal to the number of opcodes).
1923
1924 8. Following on from 7, I added a similar kludge to check the length of the
1925 eint vector in pcreposix.c.
1926
1927 9. Error texts for pcre_compile() are held as one long string to avoid too
1928 much relocation at load time. To find a text, the string is searched,
1929 counting zeros. There was no check for running off the end of the string,
1930 which could happen if a new error number was added without updating the
1931 string.
1932
1933 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1934
1935 11. \K was not working if it appeared in an atomic group or in a group that
1936 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1937 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1938 the assertion is positive, but not if it is negative.
1939
1940 12. Change 11 fortuitously reduced the size of the stack frame used in the
1941 "match()" function of pcre_exec.c by one pointer. Forthcoming
1942 implementation of support for (*MARK) will need an extra pointer on the
1943 stack; I have reserved it now, so that the stack frame size does not
1944 decrease.
1945
1946 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1947 item in branch that calls a recursion is a subroutine call - as in the
1948 second branch in the above example - was incorrectly given the compile-
1949 time error "recursive call could loop indefinitely" because pcre_compile()
1950 was not correctly checking the subroutine for matching a non-empty string.
1951
1952 14. The checks for overrunning compiling workspace could trigger after an
1953 overrun had occurred. This is a "should never occur" error, but it can be
1954 triggered by pathological patterns such as hundreds of nested parentheses.
1955 The checks now trigger 100 bytes before the end of the workspace.
1956
1957 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1958
1959
1960 Version 8.01 19-Jan-2010
1961 ------------------------
1962
1963 1. If a pattern contained a conditional subpattern with only one branch (in
1964 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1965 computed the wrong minimum data length (which is of course zero for such
1966 subpatterns). This could cause incorrect "no match" results.
1967
1968 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1969 the pattern is reset in the first branch, pcre_compile() failed with
1970 "internal error: code overflow at offset...". This happened only when
1971 the reset was to the original external option setting. (An optimization
1972 abstracts leading options settings into an external setting, which was the
1973 cause of this.)
1974
1975 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1976 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1977 assertion pattern did not match (meaning that the assertion was true), it
1978 was incorrectly treated as false if the SKIP had been reached during the
1979 matching. This also applied to assertions used as conditions.
1980
1981 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1982 assertion subpattern, including such a pattern used as a condition,
1983 unpredictable results occurred, instead of the error return
1984 PCRE_ERROR_DFA_UITEM.
1985
1986 5. The C++ GlobalReplace function was not working like Perl for the special
1987 situation when an empty string is matched. It now does the fancy magic
1988 stuff that is necessary.
1989
1990 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1991 removed. (These were left over from very, very early versions of PCRE.)
1992
1993 7. Some cosmetic changes to the code to make life easier when compiling it
1994 as part of something else:
1995
1996 (a) Change DEBUG to PCRE_DEBUG.
1997
1998 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1999 called "current" as "current_branch", to prevent a collision with the
2000 Linux macro when compiled as a kernel module.
2001
2002 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2003 prevent a collision with the Linux macro when compiled as a kernel
2004 module.
2005
2006 8. In pcre_compile() there are some checks for integer overflows that used to
2007 cast potentially large values to (double). This has been changed to that
2008 when building, a check for int64_t is made, and if it is found, it is used
2009 instead, thus avoiding the use of floating point arithmetic. (There is no
2010 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2011 double.
2012
2013 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2014 2005 (difference between two addresses compared to an unsigned value).
2015
2016 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2017 custom one, because of the following reported problem in Windows:
2018
2019 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2020 under Win32.
2021 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2022 therefore missing the function definition.
2023 - The compiler thus generates a "C" signature for the test function.
2024 - The linker fails to find the "C" function.
2025 - PCRE fails to configure if asked to do so against libbz2.
2026
2027 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2028 messages were output:
2029
2030 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2031 rerunning libtoolize, to keep the correct libtool macros in-tree.
2032 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2033
2034 I have done both of these things.
2035
2036 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2037 most of the time, it *can* run out if it is given a pattern that contains a
2038 runaway infinite recursion. I updated the discussion in the pcrestack man
2039 page.
2040
2041 13. Now that we have gone to the x.xx style of version numbers, the minor
2042 version may start with zero. Using 08 or 09 is a bad idea because users
2043 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2044 interpreted as invalid octal numbers. I've updated the previous comment in
2045 configure.ac, and also added a check that gives an error if 08 or 09 are
2046 used.
2047
2048 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2049 causing partial matching to fail when the end of the subject matched \W
2050 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2051
2052 15. There were some discrepancies between the declarations in pcre_internal.h
2053 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2054 their definitions. The declarations used "const uschar *" and the
2055 definitions used USPTR. Even though USPTR is normally defined as "const
2056 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2057 reported that: "This difference in casting confuses some C++ compilers, for
2058 example, SunCC recognizes above declarations as different functions and
2059 generates broken code for hbpcre." I have changed the declarations to use
2060 USPTR.
2061
2062 16. GNU libtool is named differently on some systems. The autogen.sh script now
2063 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2064 (FreeBSD).
2065
2066 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2067 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2068 comment: "Figure out how to create a longlong from a string: strtoll and
2069 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2070 instance, but it only takes 2 args instead of 3!"
2071
2072 18. A subtle bug concerned with back references has been fixed by a change of
2073 specification, with a corresponding code fix. A pattern such as
2074 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2075 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2076 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2077 same bug. Such groups have to be quantified to be useful, or contained
2078 inside another quantified group. (If there's no repetition, the reference
2079 can never match.) The problem arises because, having left the group and
2080 moved on to the rest of the pattern, a later failure that backtracks into
2081 the group uses the captured value from the final iteration of the group
2082 rather than the correct earlier one. I have fixed this in PCRE by forcing
2083 any group that contains a reference to itself to be an atomic group; that
2084 is, there cannot be any backtracking into it once it has completed. This is
2085 similar to recursive and subroutine calls.
2086
2087
2088 Version 8.00 19-Oct-09
2089 ----------------------
2090
2091 1. The table for translating pcre_compile() error codes into POSIX error codes
2092 was out-of-date, and there was no check on the pcre_compile() error code
2093 being within the table. This could lead to an OK return being given in
2094 error.
2095
2096 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2097 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2098 in a Windows environment.
2099
2100 3. The pcregrep --count option prints the count for each file even when it is
2101 zero, as does GNU grep. However, pcregrep was also printing all files when
2102 --files-with-matches was added. Now, when both options are given, it prints
2103 counts only for those files that have at least one match. (GNU grep just
2104 prints the file name in this circumstance, but including the count seems
2105 more useful - otherwise, why use --count?) Also ensured that the
2106 combination -clh just lists non-zero counts, with no names.
2107
2108 4. The long form of the pcregrep -F option was incorrectly implemented as
2109 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2110 but it seems right to fix it, and I didn't think it was worth preserving
2111 the old behaviour.
2112
2113 5. The command line items --regex=pattern and --regexp=pattern were not
2114 recognized by pcregrep, which required --regex pattern or --regexp pattern
2115 (with a space rather than an '='). The man page documented the '=' forms,
2116 which are compatible with GNU grep; these now work.
2117
2118 6. No libpcreposix.pc file was created for pkg-config; there was just
2119 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2120
2121 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2122 when UCP support is not needed, by modifying the Python script that
2123 generates it from Unicode data files. This should not matter if the module
2124 is correctly used as a library, but I received one complaint about 50K of
2125 unwanted data. My guess is that the person linked everything into his
2126 program rather than using a library. Anyway, it does no harm.
2127
2128 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2129 was a minimum greater than 1 for a wide character in a possessive
2130 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2131 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2132 character. Chaos in the form of incorrect output or a compiling loop could
2133 result.
2134
2135 9. The restrictions on what a pattern can contain when partial matching is
2136 requested for pcre_exec() have been removed. All patterns can now be
2137 partially matched by this function. In addition, if there are at least two
2138 slots in the offset vector, the offset of the earliest inspected character
2139 for the match and the offset of the end of the subject are set in them when
2140 PCRE_ERROR_PARTIAL is returned.
2141
2142 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2143 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2144 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2145 and may be more useful for multi-segment matching.
2146
2147 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2148 used to be given if ever the end of the subject was reached; now it is
2149 given only if matching could not proceed because another character was
2150 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2151 string "Z", which now yields "no match" instead of "partial match". In the
2152 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2153 final character ended with (*FAIL).
2154
2155 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2156 if the pattern had a "must contain" character that was already found in the
2157 earlier partial match, unless partial matching was again requested. For
2158 example, with the pattern /dog.(body)?/, the "must contain" character is
2159 "g". If the first part-match was for the string "dog", restarting with
2160 "sbody" failed. This bug has been fixed.
2161
2162 13. The string returned by pcre_dfa_exec() after a partial match has been
2163 changed so that it starts at the first inspected character rather than the
2164 first character of the match. This makes a difference only if the pattern
2165 starts with a lookbehind assertion or \b or \B (\K is not supported by
2166 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2167 matching functions compatible, and I think it's the right thing to do.
2168
2169 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2170 so that the demonstration program is easily available in environments where
2171 PCRE has not been installed from source.
2172
2173 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2174 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2175 library.
2176
2177 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2178 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2179 is not the first non-POSIX option to be added. Clearly some people find
2180 these options useful.
2181
2182 17. If a caller to the POSIX matching function regexec() passes a non-zero
2183 value for nmatch with a NULL value for pmatch, the value of
2184 nmatch is forced to zero.
2185
2186 18. RunGrepTest did not have a test for the availability of the -u option of
2187 the diff command, as RunTest does. It now checks in the same way as
2188 RunTest, and also checks for the -b option.
2189
2190 19. If an odd number of negated classes containing just a single character
2191 interposed, within parentheses, between a forward reference to a named
2192 subpattern and the definition of the subpattern, compilation crashed with
2193 an internal error, complaining that it could not find the referenced
2194 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2195 [The bug was that it was starting one character too far in when skipping
2196 over the character class, thus treating the ] as data rather than
2197 terminating the class. This meant it could skip too much.]
2198
2199 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2200 /g option in pcretest when the pattern contains \K, which makes it possible
2201 to have an empty string match not at the start, even when the pattern is
2202 anchored. Updated pcretest and pcredemo to use this option.
2203
2204 21. If the maximum number of capturing subpatterns in a recursion was greater
2205 than the maximum at the outer level, the higher number was returned, but
2206 with unset values at the outer level. The correct (outer level) value is
2207 now given.
2208
2209 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2210 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2211 make it do so. The string so far is captured, making this feature
2212 compatible with Perl.
2213
2214 23. The tests have been re-organized, adding tests 11 and 12, to make it
2215 possible to check the Perl 5.10 features against Perl 5.10.
2216
2217 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2218 pattern matches a fixed length string. PCRE did not allow this; now it
2219 does. Neither allows recursion.
2220
2221 25. I finally figured out how to implement a request to provide the minimum
2222 length of subject string that was needed in order to match a given pattern.
2223 (It was back references and recursion that I had previously got hung up
2224 on.) This code has now been added to pcre_study(); it finds a lower bound
2225 to the length of subject needed. It is not necessarily the greatest lower
2226 bound, but using it to avoid searching strings that are too short does give
2227 some useful speed-ups. The value is available to calling programs via
2228 pcre_fullinfo().
2229
2230 26. While implementing 25, I discovered to my embarrassment that pcretest had
2231 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2232 study optimizations had never been tested with that matching function.
2233 Oops. What is worse, even when it was passed study data, there was a bug in
2234 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2235 were also very few tests of studied patterns with pcre_dfa_exec().
2236
2237 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2238 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2239 on the other side of the coin, they are no longer allowed to have different
2240 names, because these cannot be distinguished in PCRE, and this has caused
2241 confusion. (This is a difference from Perl.)
2242
2243 28. When duplicate subpattern names are present (necessarily with different
2244 numbers, as required by 27 above), and a test is made by name in a
2245 conditional pattern, either for a subpattern having been matched, or for
2246 recursion in such a pattern, all the associated numbered subpatterns are
2247 tested, and the overall condition is true if the condition is true for any
2248 one of them. This is the way Perl works, and is also more like the way
2249 testing by number works.
2250
2251
2252 Version 7.9 11-Apr-09
2253 ---------------------
2254
2255 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2256 (pcretest), all targets were linked against these libraries. This included
2257 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2258 libraries. This caused unwanted dependencies to be created. This problem
2259 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2260 pcretest is linked with readline.
2261
2262 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2263 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2264 moved outside it again, because FALSE and TRUE are already defined in AIX,
2265 but BOOL is not.
2266
2267 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2268 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2269
2270 4. The pcregrep documentation said spaces were inserted as well as colons (or
2271 hyphens) following file names and line numbers when outputting matching
2272 lines. This is not true; no spaces are inserted. I have also clarified the
2273 wording for the --colour (or --color) option.
2274
2275 5. In pcregrep, when --colour was used with -o, the list of matching strings
2276 was not coloured; this is different to GNU grep, so I have changed it to be
2277 the same.
2278
2279 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2280 each matching line was coloured. Now it goes on to look for further matches
2281 of any of the test patterns, which is the same behaviour as GNU grep.
2282
2283 7. A pattern that could match an empty string could cause pcregrep to loop; it
2284 doesn't make sense to accept an empty string match in pcregrep, so I have
2285 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2286 seems to be how GNU grep behaves. [But see later change 40 for release
2287 8.33.]
2288
2289 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2290 start or after a newline", because the conditional assertion was not being
2291 correctly handled. The rule now is that both the assertion and what follows
2292 in the first alternative must satisfy the test.
2293
2294 9. If auto-callout was enabled in a pattern with a conditional group whose
2295 condition was an assertion, PCRE could crash during matching, both with
2296 pcre_exec() and pcre_dfa_exec().
2297
2298 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2299 used for matching.
2300
2301 11. Unicode property support in character classes was not working for
2302 characters (bytes) greater than 127 when not in UTF-8 mode.
2303
2304 12. Added the -M command line option to pcretest.
2305
2306 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2307
2308 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2309
2310 16. Added comments and documentation about mis-use of no_arg in the C++
2311 wrapper.
2312
2313 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2314 from Martin Jerabek that uses macro names for all relevant character and
2315 string constants.
2316
2317 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2318 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2319 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2320 these, but not everybody uses configure.
2321
2322 19. A conditional group that had only one branch was not being correctly
2323 recognized as an item that could match an empty string. This meant that an
2324 enclosing group might also not be so recognized, causing infinite looping
2325 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2326 with the subject "ab", where knowledge that the repeated group can match
2327 nothing is needed in order to break the loop.
2328
2329 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2330 exec(), but without supplying a callout function, matching went wrong.
2331
2332 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2333 leak if the size of the offset vector was greater than 30. When the vector
2334 is smaller, the saved offsets during recursion go onto a local stack
2335 vector, but for larger vectors malloc() is used. It was failing to free
2336 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2337 error, in fact).
2338
2339 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2340 heapframe that is used only when UTF-8 support is enabled. This caused no
2341 problem, but was untidy.
2342
2343 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2344 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2345 included within another project.
2346
2347 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2348 slightly modified by me:
2349
2350 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2351 not building pcregrep.
2352
2353 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2354 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2355
2356 25. Forward references, both numeric and by name, in patterns that made use of
2357 duplicate group numbers, could behave incorrectly or give incorrect errors,
2358 because when scanning forward to find the reference group, PCRE was not
2359 taking into account the duplicate group numbers. A pattern such as
2360 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2361
2362 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2363 the feature of a custom pointer more persuasive (as requested by a user).
2364
2365 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2366 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2367 pre-defined.
2368
2369 28. Added support for (*UTF8) at the start of a pattern.
2370
2371 29. Arrange for flags added by the "release type" setting in CMake to be shown
2372 in the configuration summary.
2373
2374
2375 Version 7.8 05-Sep-08
2376 ---------------------
2377
2378 1. Replaced UCP searching code with optimized version as implemented for Ad
2379 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2380 stage table and inline lookup instead of a function, giving speed ups of 2
2381 to 5 times on some simple patterns that I tested. Permission was given to
2382 distribute the MultiStage2.py script that generates the tables (it's not in
2383 the tarball, but is in the Subversion repository).
2384
2385 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2386 scripts.
2387
2388 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2389 a group with a zero qualifier. The result of the study could be incorrect,
2390 or the function might crash, depending on the pattern.
2391
2392 4. Caseless matching was not working for non-ASCII characters in back
2393 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2394 It now works when Unicode Property Support is available.
2395
2396 5. In pcretest, an escape such as \x{de} in the data was always generating
2397 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2398 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2399 truncation.
2400
2401 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2402
2403 7. Added two (int) casts to pcregrep when printing the difference of two
2404 pointers, in case they are 64-bit values.
2405
2406 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2407 test 2 if it fails.
2408
2409 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2410 and a #define of that name to empty if it is not externally set. This is to
2411 allow users of MSVC to set it if necessary.
2412
2413 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2414 the convenience functions in the pcre_get.c source file.
2415
2416 11. An option change at the start of a pattern that had top-level alternatives
2417 could cause overwriting and/or a crash. This command provoked a crash in
2418 some environments:
2419
2420 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2421
2422 This potential security problem was recorded as CVE-2008-2371.
2423
2424 12. For a pattern where the match had to start at the beginning or immediately
2425 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2426 pcre_dfa_exec() could read past the end of the passed subject if there was
2427 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2428 pcretest so that it places the subject at the end of its malloc-ed buffer.
2429
2430 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2431 exec() might read past the end of the data buffer in UTF-8 mode.
2432
2433 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2434 the data contained the byte 0x85 as part of a UTF-8 character within its
2435 first line. This applied both to normal and DFA matching.
2436
2437 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2438 /^[^d]*?$/8 failed to match "abc".
2439
2440 16. Added a missing copyright notice to pcrecpp_internal.h.
2441
2442 17. Make it more clear in the documentation that values returned from
2443 pcre_exec() in ovector are byte offsets, not character counts.
2444
2445 18. Tidied a few places to stop certain compilers from issuing warnings.
2446
2447 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2448 supplied by Stefan Weber. I made a further small update for 7.8 because
2449 there is a change of source arrangements: the pcre_searchfuncs.c module is
2450 replaced by pcre_ucd.c.
2451
2452
2453 Version 7.7 07-May-08
2454 ---------------------
2455
2456 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2457 a string to a long long, pretend we don't even have a long long." This is
2458 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2459
2460 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2461 pre-7.6 versions, which defined a global no_arg variable instead of putting
2462 it in the RE class. (See also #8 below.)
2463
2464 3. Remove a line of dead code, identified by coverity and reported by Nuno
2465 Lopes.
2466
2467 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2468
2469 (1) The include/exclude patterns were being applied to the whole pathnames
2470 of files, instead of just to the final components.
2471
2472 (2) If there was more than one level of directory, the subdirectories were
2473 skipped unless they satisfied the include/exclude conditions. This is
2474 inconsistent with GNU grep (and could even be seen as contrary to the
2475 pcregrep specification - which I improved to make it absolutely clear).
2476 The action now is always to scan all levels of directory, and just
2477 apply the include/exclude patterns to regular files.
2478
2479 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2480 --exclude_dir in the tests to avoid scanning .svn directories.
2481
2482 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2483 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2484 doesn't support NULs in patterns.
2485
2486 7. Added some missing "const"s to declarations of static tables in
2487 pcre_compile.c and pcre_dfa_exec.c.
2488
2489 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2490 caused by fix #2 above. (Subsequently also a second patch to fix the
2491 first patch. And a third patch - this was a messy problem.)
2492
2493 9. Applied Craig's patch to remove the use of push_back().
2494
2495 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2496 matching function regexec().
2497
2498 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2499 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2500 references. PCRE supports relative numbers with this syntax (I don't think
2501 Oniguruma does).
2502
2503 12. Previously, a group with a zero repeat such as (...){0} was completely
2504 omitted from the compiled regex. However, this means that if the group
2505 was called as a subroutine from elsewhere in the pattern, things went wrong
2506 (an internal error was given). Such groups are now left in the compiled
2507 pattern, with a new opcode that causes them to be skipped at execution
2508 time.
2509
2510 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2511 to the way PCRE behaves:
2512
2513 (a) A lone ] character is dis-allowed (Perl treats it as data).
2514
2515 (b) A back reference to an unmatched subpattern matches an empty string
2516 (Perl fails the current match path).
2517
2518 (c) A data ] in a character class must be notated as \] because if the
2519 first data character in a class is ], it defines an empty class. (In
2520 Perl it is not possible to have an empty class.) The empty class []
2521 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2522 The negative empty class [^] matches any one character, independently
2523 of the DOTALL setting.
2524
2525 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2526 non-existent subpattern following a character class starting with ']' and
2527 containing () gave an internal compiling error instead of "reference to
2528 non-existent subpattern". Fortunately, when the pattern did exist, the
2529 compiled code was correct. (When scanning forwards to check for the
2530 existence of the subpattern, it was treating the data ']' as terminating
2531 the class, so got the count wrong. When actually compiling, the reference
2532 was subsequently set up correctly.)
2533
2534 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2535 it was being rejected as not supported by pcre_dfa_exec(), even though
2536 other assertions are supported. I have made pcre_dfa_exec() support
2537 (*FAIL).
2538
2539 16. The implementation of 13c above involved the invention of a new opcode,
2540 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2541 cannot be changed at match time, I realized I could make a small
2542 improvement to matching performance by compiling OP_ALLANY instead of
2543 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2544 on the OP_ANY path.
2545
2546 17. Compiling pcretest on Windows with readline support failed without the
2547 following two fixes: (1) Make the unistd.h include conditional on
2548 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2549
2550 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2551 ncurses library to be included for pcretest when ReadLine support is
2552 requested, but also to allow for it to be overridden. This patch came from
2553 Daniel Bergström.
2554
2555 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2556 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2557 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2558 spotting this.
2559
2560
2561 Version 7.6 28-Jan-08
2562 ---------------------
2563
2564 1. A character class containing a very large number of characters with
2565 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2566 overflow.
2567
2568 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2569 HAVE_LONG_LONG is not defined.
2570
2571 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2572 bring them up to date and include new features. This patch includes:
2573
2574 - Fixed PH's badly added libz and libbz2 support.
2575 - Fixed a problem with static linking.
2576 - Added pcredemo. [But later removed - see 7 below.]
2577 - Fixed dftables problem and added an option.
2578 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2579 HAVE_LONG_LONG.
2580 - Added readline support for pcretest.
2581 - Added an listing of the option settings after cmake has run.
2582
2583 4. A user submitted a patch to Makefile that makes it easy to create
2584 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2585 Makefile.am that cause it to include this special target, without
2586 affecting anything else. Note that the same mingw target plus all
2587 the other distribution libraries and programs are now supported
2588 when configuring with CMake (see 6 below) instead of with
2589 Configure/Make.
2590
2591 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2592 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2593 exported in the Windows port". It has not yet been confirmed that the patch
2594 solves the problem, but it does no harm.
2595
2596 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2597 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2598 with CMake, and also correct the comment about stack recursion.
2599
2600 7. Remove the automatic building of pcredemo from the ./configure system and
2601 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2602 of a program that users should build themselves after PCRE is installed, so
2603 building it automatically is not really right. What is more, it gave
2604 trouble in some build environments.
2605
2606 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2607
2608
2609 Version 7.5 10-Jan-08
2610 ---------------------
2611
2612 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2613 values in parens when parsing an RE using the C++ wrapper."
2614
2615 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2616 Characters greater than 255 were excluded from the class instead of being
2617 included.
2618
2619 3. The same bug as (2) above applied to negated POSIX classes such as
2620 [:^space:].
2621
2622 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2623 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2624 I have changed it.
2625
2626 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2627 first named subpattern) and a construct such as (?&a) would reference the
2628 first named subpattern whose name started with "a" (in other words, the
2629 length check was missing). Both these problems are fixed. "Subpattern name
2630 expected" is now given for (?&) (a zero-length name), and this patch also
2631 makes it give the same error for \k'' (previously it complained that that
2632 was a reference to a non-existent subpattern).
2633
2634 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2635 this is right because (?- can be followed by option settings as well as by
2636 digits. I have, however, made the messages clearer.
2637
2638 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2639 than the number used in the conditional) now cause a compile-time error.
2640 This is actually not compatible with Perl, which accepts such patterns, but
2641 treats the conditional as always being FALSE (as PCRE used to), but it
2642 seems to me that giving a diagnostic is better.
2643
2644 8. Change "alphameric" to the more common word "alphanumeric" in comments
2645 and messages.
2646
2647 9. Fix two occurrences of "backslash" in comments that should have been
2648 "backspace".
2649
2650 10. Remove two redundant lines of code that can never be obeyed (their function
2651 was moved elsewhere).
2652
2653 11. The program that makes PCRE's Unicode character property table had a bug
2654 which caused it to generate incorrect table entries for sequences of
2655 characters that have the same character type, but are in different scripts.
2656 It amalgamated them into a single range, with the script of the first of
2657 them. In other words, some characters were in the wrong script. There were
2658 thirteen such cases, affecting characters in the following ranges:
2659
2660 U+002b0 - U+002c1
2661 U+0060c - U+0060d
2662 U+0061e - U+00612
2663 U+0064b - U+0065e
2664 U+0074d - U+0076d
2665 U+01800 - U+01805
2666 U+01d00 - U+01d77
2667 U+01d9b - U+01dbf
2668 U+0200b - U+0200f
2669 U+030fc - U+030fe
2670 U+03260 - U+0327f
2671 U+0fb46 - U+0fbb1
2672 U+10450 - U+1049d
2673
2674 12. The -o option (show only the matching part of a line) for pcregrep was not
2675 compatible with GNU grep in that, if there was more than one match in a
2676 line, it showed only the first of them. It now behaves in the same way as
2677 GNU grep.
2678
2679 13. If the -o and -v options were combined for pcregrep, it printed a blank
2680 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2681 does the same. The return code can be used to tell if there were any
2682 non-matching lines.
2683
2684 14. Added --file-offsets and --line-offsets to pcregrep.
2685
2686 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2687 infinitely looping recursion. The bug was that positive lookaheads were not
2688 being skipped when checking for a possible empty match (negative lookaheads
2689 and both kinds of lookbehind were skipped).
2690
2691 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2692 inclusion of <windows.h> to before rather than after the definition of
2693 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2694
2695 17. Specifying a possessive quantifier with a specific limit for a Unicode
2696 character property caused pcre_compile() to compile bad code, which led at
2697 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2698 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2699 caused the error; without that there was no problem.
2700
2701 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2702
2703 19. Added --enable-pcretest-libreadline.
2704
2705 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2706 RE::GlobalReplace(). As a result, the number of replacements returned was
2707 double what it should be. I removed one of the increments, but Craig sent a
2708 later patch that removed the other one (the right fix) and added unit tests
2709 that check the return values (which was not done before).
2710
2711 21. Several CMake things:
2712
2713 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2714 the names libpcre and libpcreposix, not just pcre and pcreposix.
2715
2716 (2) The above change means that pcretest and pcregrep are now correctly
2717 linked with the newly-built libraries, not previously installed ones.
2718
2719 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2720
2721 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2722 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2723 UTF-8 newline character). The key issue is that the pattern starts .*;
2724 this means that the match must be either at the beginning, or after a
2725 newline. The bug was in the code for advancing after a failed match and
2726 checking that the new position followed a newline. It was not taking
2727 account of UTF-8 characters correctly.
2728
2729 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2730 character classes. PCRE was not treating the sequence [:...:] as a
2731 character class unless the ... were all letters. Perl, however, seems to
2732 allow any characters between [: and :], though of course it rejects as
2733 unknown any "names" that contain non-letters, because all the known class
2734 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2735 for example, whereas PCRE did not - it did not recognize a POSIX character
2736 class. This seemed a bit dangerous, so the code has been changed to be
2737 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2738 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2739 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2740 Perl does, and where it didn't before.
2741
2742 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2743 Windows environments %n is disabled by default.
2744
2745
2746 Version 7.4 21-Sep-07
2747 ---------------------
2748
2749 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2750 means that a class such as [\s] counted as "explicit reference to CR or
2751 LF". That isn't really right - the whole point of the change was to try to
2752 help when there was an actual mention of one of the two characters. So now
2753 the change happens only if \r or \n (or a literal CR or LF) character is
2754 encountered.
2755
2756 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2757 of both had grown to the point where there were only 3 bits left.
2758 Fortunately, there was spare space in the data structure, and so I have
2759 moved the internal flags into a new 16-bit field to free up more option
2760 bits.
2761
2762 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2763 but did not set the internal JCHANGED flag - either of these is enough to
2764 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2765 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2766 start sets both bits.
2767
2768 4. Added options (at build time, compile time, exec time) to change \R from
2769 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2770
2771 5. doc/pcresyntax.html was missing from the distribution.
2772
2773 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2774 compatibility, even though it is no longer used.
2775
2776 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2777 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2778 windows.h file is present (where different names are used). [This was
2779 reversed later after testing - see 16 below.]
2780
2781 8. Changed all #include <config.h> to #include "config.h". There were also
2782 some further <pcre.h> cases that I changed to "pcre.h".
2783
2784 9. When pcregrep was used with the --colour option, it missed the line ending
2785 sequence off the lines that it output.
2786
2787 10. It was pointed out to me that arrays of string pointers cause lots of
2788 relocations when a shared library is dynamically loaded. A technique of
2789 using a single long string with a table of offsets can drastically reduce
2790 these. I have refactored PCRE in four places to do this. The result is
2791 dramatic:
2792
2793 Originally: 290
2794 After changing UCP table: 187
2795 After changing error message table: 43
2796 After changing table of "verbs" 36
2797 After changing table of Posix names 22
2798
2799 Thanks to the folks working on Gregex for glib for this insight.
2800
2801 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2802 unicode-properties was also set.
2803
2804 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2805
2806 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2807 checked only for CRLF.
2808
2809 14. Added casts to pcretest.c to avoid compiler warnings.
2810
2811 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2812
2813 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2814 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2815 entirely. This removes changes made in 7 above.
2816
2817 17. The CMake files have been updated, and there is now more information about
2818 building with CMake in the NON-UNIX-USE document.
2819
2820
2821 Version 7.3 28-Aug-07
2822 ---------------------
2823
2824 1. In the rejigging of the build system that eventually resulted in 7.1, the
2825 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2826 brackets there is not right, since it causes compilers to look for an
2827 installed pcre.h, not the version that is in the source that is being
2828 compiled (which of course may be different). I have changed it back to:
2829
2830 #include "pcre.h"
2831
2832 I have a vague recollection that the change was concerned with compiling in
2833 different directories, but in the new build system, that is taken care of
2834 by the VPATH setting the Makefile.
2835
2836 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2837 when the subject happened to end in the byte 0x85 (e.g. if the last
2838 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2839 characters but of course it shouldn't be taken as a newline when it is part
2840 of another character. The bug was that, for an unlimited repeat of . in
2841 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2842 characters when looking for a newline.
2843
2844 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2845
2846 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2847 in debug output.
2848
2849 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2850 long printing in the pcrecpp unittest when running under MinGW.
2851
2852 6. ESC_K was left out of the EBCDIC table.
2853
2854 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2855 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2856 limit also applies to "virtual nesting" when a pattern is recursive, and in
2857 this case 1000 isn't so big. I have been able to remove this limit at the
2858 expense of backing off one optimization in certain circumstances. Normally,
2859 when pcre_exec() would call its internal match() function recursively and
2860 immediately return the result unconditionally, it uses a "tail recursion"
2861 feature to save stack. However, when a subpattern that can match an empty
2862 string has an unlimited repetition quantifier, it no longer makes this
2863 optimization. That gives it a stack frame in which to save the data for
2864 checking that an empty string has been matched. Previously this was taken
2865 from the 1000-entry workspace that had been reserved. So now there is no
2866 explicit limit, but more stack is used.
2867
2868 8. Applied Daniel's patches to solve problems with the import/export magic
2869 syntax that is required for Windows, and which was going wrong for the
2870 pcreposix and pcrecpp parts of the library. These were overlooked when this
2871 problem was solved for the main library.
2872
2873 9. There were some crude static tests to avoid integer overflow when computing
2874 the size of patterns that contain repeated groups with explicit upper
2875 limits. As the maximum quantifier is 65535, the maximum group length was
2876 set at 30,000 so that the product of these two numbers did not overflow a
2877 32-bit integer. However, it turns out that people want to use groups that
2878 are longer than 30,000 bytes (though not repeat them that many times).
2879 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2880 made it possible to implement the integer overflow checks in a much more
2881 dynamic way, which I have now done. The artificial limitation on group
2882 length has been removed - we now have only the limit on the total length of
2883 the compiled pattern, which depends on the LINK_SIZE setting.
2884
2885 10. Fixed a bug in the documentation for get/copy named substring when
2886 duplicate names are permitted. If none of the named substrings are set, the
2887 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2888 empty string.
2889
2890 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2891 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2892 because the ] is interpreted as the first data character and the
2893 terminating ] is not found. PCRE has been made compatible with Perl in this
2894 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2895 cause memory overwriting.
2896
2897 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2898 string has been matched (to stop an infinite loop). It was not recognizing
2899 a conditional subpattern that could match an empty string if that
2900 subpattern was within another subpattern. For example, it looped when
2901 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2902 condition was not nested. This bug has been fixed.
2903
2904 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2905 past the start of the subject in the presence of bytes with the top bit
2906 set, for example "\x8aBCD".
2907
2908 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2909 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2910
2911 14. Optimized (?!) to (*FAIL).
2912
2913 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2914 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2915 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2916 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2917 does: it's just the validity check that is more restrictive.
2918
2919 16. Inserted checks for integer overflows during escape sequence (backslash)
2920 processing, and also fixed erroneous offset values for syntax errors during
2921 backslash processing.
2922
2923 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2924 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2925
2926 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2927 caused an overrun.
2928
2929 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2930 something other than just ASCII characters) inside a group that had an
2931 unlimited repeat caused a loop at compile time (while checking to see
2932 whether the group could match an empty string).
2933
2934 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2935 [\P{Any}] did so. (Error in the code for printing property names.)
2936
2937 21. An orphan \E inside a character class could cause a crash.
2938
2939 22. A repeated capturing bracket such as (A)? could cause a wild memory
2940 reference during compilation.
2941
2942 23. There are several functions in pcre_compile() that scan along a compiled
2943 expression for various reasons (e.g. to see if it's fixed length for look
2944 behind). There were bugs in these functions when a repeated \p or \P was
2945 present in the pattern. These operators have additional parameters compared
2946 with \d, etc, and these were not being taken into account when moving along
2947 the compiled data. Specifically:
2948
2949 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2950 length.
2951
2952 (b) An item such as \pL+ within a repeated group could cause crashes or
2953 loops.
2954
2955 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2956 "reference to non-existent subpattern" error.
2957
2958 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2959
2960 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2961 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2962
2963 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2964
2965 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2966 character were causing crashes (broken optimization).
2967
2968 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2969 \p or \P) caused a compile-time loop.
2970
2971 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2972 break. For example, the unstudied pattern [\r\n]A does not match the string
2973 "\r\nA" because change 7.0/46 below moves the current point on by two
2974 characters after failing to match at the start. However, the pattern \nA
2975 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2976 the same is true. There doesn't seem any very clean way out of this, but
2977 what I have chosen to do makes the common cases work: PCRE now takes note
2978 of whether there can be an explicit match for \r or \n anywhere in the
2979 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2980 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2981 pattern has explicit CR or LF references.
2982
2983 29. Added (*CR) etc for changing newline setting at start of pattern.
2984
2985
2986 Version 7.2 19-Jun-07
2987 ---------------------
2988
2989 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2990 which is apparently normally available under Windows.
2991
2992 2. Re-jig the pcregrep tests with different newline settings in an attempt
2993 to make them independent of the local environment's newline setting.
2994
2995 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2996
2997 4. Some of the "internals" tests were previously cut out when the link size
2998 was not 2, because the output contained actual offsets. The recent new
2999 "Z" feature of pcretest means that these can be cut out, making the tests
3000 usable with all link sizes.
3001
3002 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3003 stack recursion. This gives a massive performance boost under BSD, but just
3004 a small improvement under Linux. However, it saves one field in the frame
3005 in all cases.
3006
3007 6. Added more features from the forthcoming Perl 5.10:
3008
3009 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3010 recursion call. It refers to the nth most recently opened parentheses.
3011
3012 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3013 to be opened parentheses.
3014
3015 (c) Conditions that refer to capturing parentheses can be specified
3016 relatively, for example, (?(-2)... or (?(+3)...
3017
3018 (d) \K resets the start of the current match so that everything before
3019 is not part of it.
3020
3021 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3022
3023 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3024 reference syntax.
3025
3026 (g) (?| introduces a group in which the numbering of parentheses in each
3027 alternative starts with the same number.
3028
3029 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3030
3031 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3032 PCRE_INFO_JCHANGED.
3033
3034 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3035 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3036 for detecting groups that can match an empty string.
3037
3038 9. A pattern with a very large number of alternatives (more than several
3039 hundred) was running out of internal workspace during the pre-compile
3040 phase, where pcre_compile() figures out how much memory will be needed. A
3041 bit of new cunning has reduced the workspace needed for groups with
3042 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3043 workspace instead of running out of the 4096 that are available.
3044
3045 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3046
3047 11. Applied patch from Google to remove an optimization that didn't quite work.
3048 The report of the bug said:
3049
3050 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3051 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3052 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3053
3054 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3055 it matched the wrong number of bytes.
3056
3057
3058 Version 7.1 24-Apr-07
3059 ---------------------
3060
3061 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3062 that is more "standard", making use of automake and other Autotools. There
3063 is some re-arrangement of the files and adjustment of comments consequent
3064 on this.
3065
3066 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3067 for recursive directory scanning broke on some systems because the files
3068 are not scanned in any specific order and on different systems the order
3069 was different. A call to "sort" has been inserted into RunGrepTest for the
3070 approprate test as a short-term fix. In the longer term there may be an
3071 alternative.
3072
3073 3. I had an email from Eric Raymond about problems translating some of PCRE's
3074 man pages to HTML (despite the fact that I distribute HTML pages, some
3075 people do their own conversions for various reasons). The problems
3076 concerned the use of low-level troff macros .br and .in. I have therefore
3077 removed all such uses from the man pages (some were redundant, some could
3078 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3079 HTML has been updated to handle .nf/.fi and to complain if it encounters
3080 .br or .in.
3081
3082 4. Updated comments in configure.ac that get placed in config.h.in and also
3083 arranged for config.h to be included in the distribution, with the name
3084 config.h.generic, for the benefit of those who have to compile without
3085 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3086
3087 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3088 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3089 makevp.bat for the current PCRE, using the additional files
3090 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3091
3092 6. A Windows user reported a minor discrepancy with test 2, which turned out
3093 to be caused by a trailing space on an input line that had got lost in his
3094 copy. The trailing space was an accident, so I've just removed it.
3095
3096 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3097 that is needed.
3098
3099 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3100 as "const" (a) because they are and (b) because it helps the PHP
3101 maintainers who have recently made a script to detect big data structures
3102 in the php code that should be moved to the .rodata section. I remembered
3103 to update Builducptable as well, so it won't revert if ucptable.h is ever
3104 re-created.
3105
3106 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3107 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3108 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3109 support is not required. This saves 1.5-2K of code, which is important in
3110 some applications.
3111
3112 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3113 so as not to refer to the tables, even though these functions will never be
3114 called when UTF-8 support is disabled. Otherwise there are problems with a
3115 shared library.
3116
3117 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3118
3119 (a) It was defining its arguments as char * instead of void *.
3120
3121 (b) It was assuming that all moves were upwards in memory; this was true
3122 a long time ago when I wrote it, but is no longer the case.
3123
3124 The emulated memove() is provided for those environments that have neither
3125 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3126 is clearly not the case, as these two bugs were recently reported.
3127
3128 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3129 and Detrail to create the HTML documentation, the .txt form of the man
3130 pages, and it removes trailing spaces from listed files. It also creates
3131 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3132 case, it wraps all the #defines with #ifndefs. This script should be run
3133 before "make dist".
3134
3135 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3136 with Unicode property support.
3137
3138 (a) For a maximizing quantifier, if the two different cases of the
3139 character were of different lengths in their UTF-8 codings (there are
3140 some cases like this - I found 11), and the matching function had to
3141 back up over a mixture of the two cases, it incorrectly assumed they
3142 were both the same length.
3143
3144 (b) When PCRE was configured to use the heap rather than the stack for
3145 recursion during matching, it was not correctly preserving the data for
3146 the other case of a UTF-8 character when checking ahead for a match
3147 while processing a minimizing repeat. If the check also involved
3148 matching a wide character, but failed, corruption could cause an
3149 erroneous result when trying to check for a repeat of the original
3150 character.
3151
3152 13. Some tidying changes to the testing mechanism:
3153
3154 (a) The RunTest script now detects the internal link size and whether there
3155 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3156 values substituted by "configure". (The RunGrepTest script already did
3157 this for UTF-8.) The configure.ac script no longer substitutes the
3158 relevant variables.
3159
3160 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3161 with length and offset values. This means that the output is different
3162 for different internal link sizes. Test 2 is skipped for link sizes
3163 other than 2 because of this, bypassing the problem. Unfortunately,
3164 there was also a test in test 3 (the locale tests) that used /B and
3165 failed for link sizes other than 2. Rather than cut the whole test out,
3166 I have added a new /Z option to pcretest that replaces the length and
3167 offset values with spaces. This is now used to make test 3 independent
3168 of link size. (Test 2 will be tidied up later.)
3169
3170 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3171 segmentation fault instead of returning the appropriate error message.
3172
3173 15. In multiline mode when the newline sequence was set to "any", the pattern
3174 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3175 This doesn't seem right; it now treats the CRLF combination as the line
3176 ending, and so does not match in that case. It's only a pattern such as ^$
3177 that would hit this one: something like ^ABC$ would have failed after \r
3178 and then tried again after \r\n.
3179
3180 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3181 in an attempt to make files that differ only in their line terminators
3182 compare equal. This works on Linux.
3183
3184 17. Under certain error circumstances pcregrep might try to free random memory
3185 as it exited. This is now fixed, thanks to valgrind.
3186
3187 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3188 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3189 was because its rules for how to advance for /g after matching an empty
3190 string at the end of a line did not allow for this case. They now check for
3191 it specially.
3192
3193 20. pcretest is supposed to handle patterns and data of any length, by
3194 extending its buffers when necessary. It was getting this wrong when the
3195 buffer for a data line had to be extended.
3196
3197 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3198 CRLF as a newline sequence.
3199
3200 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3201 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3202 I have nevertheless tidied it up.
3203
3204 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3205
3206 24. Added a man page for pcre-config.
3207
3208
3209 Version 7.0 19-Dec-06
3210 ---------------------
3211
3212 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3213 moving to gcc 4.1.1.
3214
3215 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3216 sys/time.h, which is documented as needed for this function. It doesn't
3217 seem to matter on Linux, but it showed up on some releases of OS X.
3218
3219 3. It seems that there are systems where bytes whose values are greater than
3220 127 match isprint() in the "C" locale. The "C" locale should be the
3221 default when a C program starts up. In most systems, only ASCII printing
3222 characters match isprint(). This difference caused the output from pcretest
3223 to vary, making some of the tests fail. I have changed pcretest so that:
3224
3225 (a) When it is outputting text in the compiled version of a pattern, bytes
3226 other than 32-126 are always shown as hex escapes.
3227
3228 (b) When it is outputting text that is a matched part of a subject string,
3229 it does the same, unless a different locale has been set for the match
3230 (using the /L modifier). In this case, it uses isprint() to decide.
3231
3232 4. Fixed a major bug that caused incorrect computation of the amount of memory
3233 required for a compiled pattern when options that changed within the
3234 pattern affected the logic of the preliminary scan that determines the
3235 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3236 that the computed length was too small. The symptoms of this bug were
3237 either the PCRE error "internal error: code overflow" from pcre_compile(),
3238 or a glibc crash with a message such as "pcretest: free(): invalid next
3239 size (fast)". Examples of patterns that provoked this bug (shown in
3240 pcretest format) are:
3241
3242 /(?-x: )/x
3243 /(?x)(?-x: \s*#\s*)/
3244 /((?i)[\x{c0}])/8
3245 /(?i:[\x{c0}])/8
3246
3247 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3248 is now done differently.
3249
3250 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3251 wrapper classes; (b) implement a new function in the C++ scanner that is
3252 more efficient than the old way of doing things because it avoids levels of
3253 recursion in the regex matching; (c) add a paragraph to the documentation
3254 for the FullMatch() function.
3255
3256 6. The escape sequence \n was being treated as whatever was defined as
3257 "newline". Not only was this contrary to the documentation, which states
3258 that \n is character 10 (hex 0A), but it also went horribly wrong when
3259 "newline" was defined as CRLF. This has been fixed.
3260
3261 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3262 was being set to -1 for the "end of line" case (supposedly a value that no
3263 character can have). Though this value is never used (the check for end of
3264 line is "zero bytes in current character"), it caused compiler complaints.
3265 I've changed it to 0xffffffff.
3266
3267 8. In pcre_version.c, the version string was being built by a sequence of
3268 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3269 string (as it is for production releases) called a macro with an empty
3270 argument. The C standard says the result of this is undefined. The gcc
3271 compiler treats it as an empty string (which was what was wanted) but it is
3272 reported that Visual C gives an error. The source has been hacked around to
3273 avoid this problem.
3274
3275 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3276 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3277 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3278 of them did).
3279
3280 10. Originally, pcretest opened its input and output without "b"; then I was
3281 told that "b" was needed in some environments, so it was added for release
3282 5.0 to both the input and output. (It makes no difference on Unix-like
3283 systems.) Later I was told that it is wrong for the input on Windows. I've
3284 now abstracted the modes into two macros, to make it easier to fiddle with
3285 them, and removed "b" from the input mode under Windows.
3286
3287 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3288
3289 12. Added -help and --help to pcretest as an official way of being reminded
3290 of the options.
3291
3292 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3293 and pcrecpp.cc because they annoy compilers at high warning levels.
3294
3295 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3296
3297 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3298 scripts are not C programs :-) and which was not noticed because it works
3299 on Linux.
3300
3301 16. pcretest is supposed to handle any length of pattern and data line (as one
3302 line or as a continued sequence of lines) by extending its input buffer if
3303 necessary. This feature was broken for very long pattern lines, leading to
3304 a string of junk being passed to pcre_compile() if the pattern was longer
3305 than about 50K.
3306
3307 17. I have done a major re-factoring of the way pcre_compile() computes the
3308 amount of memory needed for a compiled pattern. Previously, there was code
3309 that made a preliminary scan of the pattern in order to do this. That was
3310 OK when PCRE was new, but as the facilities have expanded, it has become
3311 harder and harder to keep it in step with the real compile phase, and there
3312 have been a number of bugs (see for example, 4 above). I have now found a
3313 cunning way of running the real compile function in a "fake" mode that
3314 enables it to compute how much memory it would need, while actually only
3315 ever using a few hundred bytes of working memory and without too many
3316 tests of the mode. This should make future maintenance and development
3317 easier. A side effect of this work is that the limit of 200 on the nesting
3318 depth of parentheses has been removed (though this was never a serious
3319 limitation, I suspect). However, there is a downside: pcre_compile() now
3320 runs more slowly than before (30% or more, depending on the pattern). I
3321 hope this isn't a big issue. There is no effect on runtime performance.
3322
3323 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3324 newline (only possible for the last line of a file) and it was a
3325 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3326
3327 19. Added additional timing features to pcretest. (1) The -tm option now times
3328 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3329 separate command line item, by a number that specifies the number of
3330 repeats to use when timing. The default is 50000; this gives better
3331 precision, but takes uncomfortably long for very large patterns.
3332
3333 20. Extended pcre_study() to be more clever in cases where a branch of a
3334 subpattern has no definite first character. For example, (a*|b*)[cd] would
3335 previously give no result from pcre_study(). Now it recognizes that the
3336 first character must be a, b, c, or d.
3337
3338 21. There was an incorrect error "recursive call could loop indefinitely" if
3339 a subpattern (or the entire pattern) that was being tested for matching an
3340 empty string contained only one non-empty item after a nested subpattern.
3341 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3342 incorrectly, because the \d was being skipped in the check.
3343
3344 22. The pcretest program now has a new pattern option /B and a command line
3345 option -b, which is equivalent to adding /B to every pattern. This causes
3346 it to show the compiled bytecode, without the additional information that
3347 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3348 is the same as /B/I).
3349
3350 23. A new optimization is now able automatically to treat some sequences such
3351 as a*b as a*+b. More specifically, if something simple (such as a character
3352 or a simple class like \d) has an unlimited quantifier, and is followed by
3353 something that cannot possibly match the quantified thing, the quantifier
3354 is automatically "possessified".
3355
3356 24. A recursive reference to a subpattern whose number was greater than 39
3357 went wrong under certain circumstances in UTF-8 mode. This bug could also
3358 have affected the operation of pcre_study().
3359
3360 25. Realized that a little bit of performance could be had by replacing
3361 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3362
3363 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3364
3365 27. Possessive quantifiers such as a++ were previously implemented by turning
3366 them into atomic groups such as ($>a+). Now they have their own opcodes,
3367 which improves performance. This includes the automatically created ones
3368 from 23 above.
3369
3370 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3371 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3372 the first matched character to be a colon. This applied both to named and
3373 numbered groups.
3374
3375 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3376
3377 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3378 building PCRE on Windows easier, so I have included it in the distribution.
3379
3380 31. There is now a check in pcretest against a ridiculously large number being
3381 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3382 loop, the loop is abandoned.
3383
3384 32. Forward references to subpatterns in conditions such as (?(2)...) where
3385 subpattern 2 is defined later cause pcre_compile() to search forwards in
3386 the pattern for the relevant set of parentheses. This search went wrong
3387 when there were unescaped parentheses in a character class, parentheses
3388 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3389
3390 33. "Subroutine" calls and backreferences were previously restricted to
3391 referencing subpatterns earlier in the regex. This restriction has now
3392 been removed.
3393
3394 34. Added a number of extra features that are going to be in Perl 5.10. On the
3395 whole, these are just syntactic alternatives for features that PCRE had
3396 previously implemented using the Python syntax or my own invention. The
3397 other formats are all retained for compatibility.
3398
3399 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3400 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3401 also .NET compatible.
3402
3403 (b) A recursion or subroutine call to a named group can now be defined as
3404 (?&name) as well as (?P>name).
3405
3406 (c) A backreference to a named group can now be defined as \k<name> or
3407 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3408 5.10, are also .NET compatible.
3409
3410 (d) A conditional reference to a named group can now use the syntax
3411 (?(<name>) or (?('name') as well as (?(name).
3412
3413 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3414 groups (named and numbered) that are never evaluated inline, but can be
3415 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3416 is always false. There may be only one alternative in such a group.
3417
3418 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3419 as the simple (?(R). The condition is true only if the most recent
3420 recursion is that of the given number or name. It does not search out
3421 through the entire recursion stack.
3422
3423 (g) The escape \gN or \g{N} has been added, where N is a positive or
3424 negative number, specifying an absolute or relative reference.
3425
3426 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3427 some "unreachable code" warnings.
3428
3429 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3430 things, this adds five new scripts.
3431
3432 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3433 There were also incompatibilities regarding the handling of \Q..\E inside
3434 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3435 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3436
3437 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3438 matches an empty string, and forcibly breaks the loop. There were bugs in
3439 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3440 against aaaa the result was just "a" rather than "aaaa", for example. Two
3441 separate and independent bugs (that affected different cases) have been
3442 fixed.
3443
3444 39. Refactored the code to abolish the use of different opcodes for small
3445 capturing bracket numbers. This is a tidy that I avoided doing when I
3446 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3447 The new approach is not only tidier, it makes it possible to reduce the
3448 memory needed to fix the previous bug (38).
3449
3450 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3451 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3452 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3453 mode.
3454
3455 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3456 report.
3457
3458 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3459 copy construction and assignment in the C++ wrapper.
3460
3461 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3462 couple of bugs that could have given wrong results in the "--newline=crlf"
3463 case.
3464
3465 44. Added a number of casts and did some reorganization of signed/unsigned int
3466 variables following suggestions from Dair Grant. Also renamed the variable
3467 "this" as "item" because it is a C++ keyword.
3468
3469 45. Arranged for dftables to add
3470
3471 #include "pcre_internal.h"
3472
3473 to pcre_chartables.c because without it, gcc 4.x may remove the array
3474 definition from the final binary if PCRE is built into a static library and
3475 dead code stripping is activated.
3476
3477 46. For an unanchored pattern, if a match attempt fails at the start of a
3478 newline sequence, and the newline setting is CRLF or ANY, and the next two
3479 characters are CRLF, advance by two characters instead of one.
3480
3481
3482 Version 6.7 04-Jul-06
3483 ---------------------
3484
3485 1. In order to handle tests when input lines are enormously long, pcretest has
3486 been re-factored so that it automatically extends its buffers when
3487 necessary. The code is crude, but this _is_ just a test program. The
3488 default size has been increased from 32K to 50K.
3489
3490 2. The code in pcre_study() was using the value of the re argument before
3491 testing it for NULL. (Of course, in any sensible call of the function, it
3492 won't be NULL.)
3493
3494 3. The memmove() emulation function in pcre_internal.h, which is used on
3495 systems that lack both memmove() and bcopy() - that is, hardly ever -
3496 was missing a "static" storage class specifier.
3497
3498 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3499 containing an extended class (one that cannot be represented by a bitmap
3500 because it contains high-valued characters or Unicode property items, e.g.
3501 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3502 pattern, but PCRE should not loop if you do not (it no longer does).
3503 [Detail: two cases were found: (a) a repeated subpattern containing an
3504 extended class; (b) a recursive reference to a subpattern that followed a
3505 previous extended class. It wasn't skipping over the extended class
3506 correctly when UTF-8 mode was not set.]
3507
3508 5. A negated single-character class was not being recognized as fixed-length
3509 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3510 compile error "lookbehind assertion is not fixed length".
3511
3512 6. The RunPerlTest auxiliary script was showing an unexpected difference
3513 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3514 write a Perl script that can interpret lines of an input file either as
3515 byte characters or as UTF-8, which is what "perltest" was being required to
3516 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3517 can't do is switch easily at run time between having the "use utf8;" pragma
3518 or not. In the end, I fudged it by using the RunPerlTest script to insert
3519 "use utf8;" explicitly for the UTF-8 tests.
3520
3521 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3522 the end of the subject string, contrary to the documentation and to what
3523 Perl does. This was true of both matching functions. Now it matches only at
3524 the start of the subject and immediately after *internal* newlines.
3525
3526 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3527 a pointer to an int instead of a pointer to an unsigned long int. This
3528 caused problems on 64-bit systems.
3529
3530 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3531 instance of the 'standard' template library not being so standard".
3532
3533 10. There was no check on the number of named subpatterns nor the maximum
3534 length of a subpattern name. The product of these values is used to compute
3535 the size of the memory block for a compiled pattern. By supplying a very
3536 long subpattern name and a large number of named subpatterns, the size
3537 computation could be caused to overflow. This is now prevented by limiting
3538 the length of names to 32 characters, and the number of named subpatterns
3539 to 10,000.
3540
3541 11. Subpatterns that are repeated with specific counts have to be replicated in
3542 the compiled pattern. The size of memory for this was computed from the
3543 length of the subpattern and the repeat count. The latter is limited to
3544 65535, but there was no limit on the former, meaning that integer overflow
3545 could in principle occur. The compiled length of a repeated subpattern is
3546 now limited to 30,000 bytes in order to prevent this.
3547
3548 12. Added the optional facility to have named substrings with the same name.
3549
3550 13. Added the ability to use a named substring as a condition, using the
3551 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3552 are numbers (not recommended). Forward references are permitted.
3553
3554 14. Added forward references in named backreferences (if you see what I mean).
3555
3556 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3557 pattern could run off the end of the subject. For example, the pattern
3558 "(?s)(.{1,5})"8 did this with the subject "ab".
3559
3560 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3561 PCRE_CASELESS was set when matching characters that were quantified with ?
3562 or *.
3563
3564 17. A character class other than a single negated character that had a minimum
3565 but no maximum quantifier - for example [ab]{6,} - was not handled
3566 correctly by pce_dfa_exec(). It would match only one character.
3567
3568 18. A valid (though odd) pattern that looked like a POSIX character
3569 class but used an invalid character after [ (for example [[,abc,]]) caused
3570 pcre_compile() to give the error "Failed: internal error: code overflow" or
3571 in some cases to crash with a glibc free() error. This could even happen if
3572 the pattern terminated after [[ but there just happened to be a sequence of
3573 letters, a binary zero, and a closing ] in the memory that followed.
3574
3575 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3576 over the years. Originally (before any Unicode support), just the bottom 8
3577 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3578 output from "man perlunicode" includes this:
3579
3580 The regular expression compiler produces polymorphic opcodes. That
3581 is, the pattern adapts to the data and automatically switches to
3582 the Unicode character scheme when presented with Unicode data--or
3583 instead uses a traditional byte scheme when presented with byte
3584 data.
3585
3586 Sadly, a wide octal escape does not cause a switch, and in a string with
3587 no other multibyte characters, these octal escapes are treated as before.
3588 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3589 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3590 Unicode string.
3591
3592 I have not perpetrated such confusion in PCRE. Up till now, it took just
3593 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3594 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3595 translate to the appropriate multibyte character.
3596
3597 29. Applied some refactoring to reduce the number of warnings from Microsoft
3598 and Borland compilers. This has included removing the fudge introduced
3599 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3600 a warning about an unused variable.
3601
3602 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3603 characters since release 4.0, because Perl (from release 5.004) does not.
3604 [Or at least, is documented not to: some releases seem to be in conflict
3605 with the documentation.] However, when a pattern was studied with
3606 pcre_study() and all its branches started with \s, PCRE still included VT
3607 as a possible starting character. Of course, this did no harm; it just
3608 caused an unnecessary match attempt.
3609
3610 22. Removed a now-redundant internal flag bit that recorded the fact that case
3611 dependency changed within the pattern. This was once needed for "required
3612 byte" processing, but is no longer used. This recovers a now-scarce options
3613 bit. Also moved the least significant internal flag bit to the most-
3614 significant bit of the word, which was not previously used (hangover from
3615 the days when it was an int rather than a uint) to free up another bit for
3616 the future.
3617
3618 23. Added support for CRLF line endings as well as CR and LF. As well as the
3619 default being selectable at build time, it can now be changed at runtime
3620 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3621 specify that it is scanning data with non-default line endings.
3622
3623 24. Changed the definition of CXXLINK to make it agree with the definition of
3624 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3625
3626 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3627 recursions. This makes a big different to stack usage for some patterns.
3628
3629 26. If a subpattern containing a named recursion or subroutine reference such
3630 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3631 the space required for the compiled pattern went wrong and gave too small a
3632 value. Depending on the environment, this could lead to "Failed: internal
3633 error: code overflow at offset 49" or "glibc detected double free or
3634 corruption" errors.
3635
3636 27. Applied patches from Google (a) to support the new newline modes and (b) to
3637 advance over multibyte UTF-8 characters in GlobalReplace.
3638
3639 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3640 difference for some implementation of PCRE in some Windows version.
3641
3642 29. Added some extra testing facilities to pcretest:
3643
3644 \q<number> in a data line sets the "match limit" value
3645 \Q<number> in a data line sets the "match recursion limt" value
3646 -S <number> sets the stack size, where <number> is in megabytes
3647
3648 The -S option isn't available for Windows.
3649
3650
3651 Version 6.6 06-Feb-06
3652 ---------------------
3653
3654 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3655 in pcreposix.h. I have copied the definition from pcre.h.
3656
3657 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3658 because pcre.h is no longer a built file.
3659
3660 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3661 not normally included in the compiled code.
3662
3663
3664 Version 6.5 01-Feb-06
3665 ---------------------
3666
3667 1. When using the partial match feature with pcre_dfa_exec(), it was not
3668 anchoring the second and subsequent partial matches at the new starting
3669 point. This could lead to incorrect results. For example, with the pattern
3670 /1234/, partially matching against "123" and then "a4" gave a match.
3671
3672 2. Changes to pcregrep:
3673
3674 (a) All non-match returns from pcre_exec() were being treated as failures
3675 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3676 error message is output. Some extra information is given for the
3677 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3678 probably the only errors that are likely to be caused by users (by
3679 specifying a regex that has nested indefinite repeats, for instance).
3680 If there are more than 20 of these errors, pcregrep is abandoned.
3681
3682 (b) A binary zero was treated as data while matching, but terminated the
3683 output line if it was written out. This has been fixed: binary zeroes
3684 are now no different to any other data bytes.
3685
3686 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3687 used to set a locale for matching. The --locale=xxxx long option has
3688 been added (no short equivalent) to specify a locale explicitly on the
3689 pcregrep command, overriding the environment variables.
3690
3691 (d) When -B was used with -n, some line numbers in the output were one less
3692 than they should have been.
3693
3694 (e) Added the -o (--only-matching) option.
3695
3696 (f) If -A or -C was used with -c (count only), some lines of context were
3697 accidentally printed for the final match.
3698
3699 (g) Added the -H (--with-filename) option.
3700
3701 (h) The combination of options -rh failed to suppress file names for files
3702 that were found from directory arguments.
3703
3704 (i) Added the -D (--devices) and -d (--directories) options.
3705
3706 (j) Added the -F (--fixed-strings) option.
3707
3708 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3709
3710 (l) Added the --colo(u)r option.
3711
3712 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3713 is not present by default.
3714
3715 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3716 items such as (?R) or (?1), when the recursion could match a number of
3717 alternatives. If it matched one of the alternatives, but subsequently,
3718 outside the recursion, there was a failure, the code tried to back up into
3719 the recursion. However, because of the way PCRE is implemented, this is not
3720 possible, and the result was an incorrect result from the match.
3721
3722 In order to prevent this happening, the specification of recursion has
3723 been changed so that all such subpatterns are automatically treated as
3724 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3725
3726 4. I had overlooked the fact that, in some locales, there are characters for
3727 which isalpha() is true but neither isupper() nor islower() are true. In
3728 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3729 and ordfeminine) are like this. This affected the treatment of \w and \W
3730 when they appeared in character classes, but not when they appeared outside
3731 a character class. The bit map for "word" characters is now created
3732 separately from the results of isalnum() instead of just taking it from the
3733 upper, lower, and digit maps. (Plus the underscore character, of course.)
3734
3735 5. The above bug also affected the handling of POSIX character classes such as
3736 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3737 permanent tables. Instead, the bit maps for such a class were previously
3738 created as the appropriate unions of the upper, lower, and digit bitmaps.
3739 Now they are created by subtraction from the [[:word:]] class, which has
3740 its own bitmap.
3741
3742 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3743 It is created by subtracting the vertical space characters (\x09, \x0a,
3744 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3745 subtraction was done in the overall bitmap for a character class, meaning
3746 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3747 be recognized. This bug has been fixed.
3748
3749 7. Patches from the folks at Google:
3750
3751 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3752 real life, but is still worth protecting against".
3753
3754 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3755 regular expressions".
3756
3757 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3758 have it.
3759
3760 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3761 "configure" and the latter not, in order to fix a problem somebody had
3762 with compiling the Arg class on HP-UX.
3763
3764 (e) Improve the error-handling of the C++ wrapper a little bit.
3765
3766 (f) New tests for checking recursion limiting.
3767
3768 8. The pcre_memmove() function, which is used only if the environment does not
3769 have a standard memmove() function (and is therefore rarely compiled),
3770 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3771 returning a result (though PCRE never actually uses the result).
3772
3773 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3774 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3775 returned instead of calling malloc() with an overflowing number that would
3776 most likely cause subsequent chaos.
3777
3778 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3779
3780 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3781 with this option is matched, the nmatch and pmatch options of regexec() are
3782 ignored.
3783
3784 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3785 provided in case anyone wants to the the POSIX interface with UTF-8
3786 strings.
3787
3788 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3789 C++ linking (needed for some HP-UX environments).
3790
3791 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3792 (unused parameter) and in the pcre_printint() function (omitted "default"
3793 switch label when the default is to do nothing).
3794
3795 15. Added some code to make it possible, when PCRE is compiled as a C++
3796 library, to replace subject pointers for pcre_exec() with a smart pointer
3797 class, thus making it possible to process discontinuous strings.
3798
3799 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3800 much the same function. They were added by different people who were trying
3801 to make PCRE easy to compile on non-Unix systems. It has been suggested
3802 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3803 for compiling on Windows systems. I have therefore replaced it with
3804 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3805 defaults to "extern" for C or "extern C" for C++, which works fine on
3806 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3807 SCOPE with something explicit in config.h. In addition:
3808
3809 (a) pcreposix.h still had just "extern" instead of either of these macros;
3810 I have replaced it with PCRE_DATA_SCOPE.
3811
3812 (b) Functions such as _pcre_xclass(), which are internal to the library,
3813 but external in the C sense, all had PCRE_EXPORT in their definitions.
3814 This is apparently wrong for the Windows case, so I have removed it.
3815 (It makes no difference on Unix-like systems.)
3816
3817 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3818 of recursive calls to match(). This is different to MATCH_LIMIT because
3819 that limits the total number of calls to match(), not all of which increase
3820 the depth of recursion. Limiting the recursion depth limits the amount of
3821 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3822 when PCRE is compiled, and changed at run time. A patch from Google adds
3823 this functionality to the C++ interface.
3824
3825 18. Changes to the handling of Unicode character properties:
3826
3827 (a) Updated the table to Unicode 4.1.0.
3828
3829 (b) Recognize characters that are not in the table as "Cn" (undefined).
3830
3831 (c) I revised the way the table is implemented to a much improved format
3832 which includes recognition of ranges. It now supports the ranges that
3833 are defined in UnicodeData.txt, and it also amalgamates other
3834 characters into ranges. This has reduced the number of entries in the
3835 table from around 16,000 to around 3,000, thus reducing its size
3836 considerably. I realized I did not need to use a tree structure after
3837 all - a binary chop search is just as efficient. Having reduced the
3838 number of entries, I extended their size from 6 bytes to 8 bytes to
3839 allow for more data.
3840
3841 (d) Added support for Unicode script names via properties such as \p{Han}.
3842
3843 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3844 matching that character.
3845
3846 20. When matching a repeated Unicode property with a minimum greater than zero,
3847 (for example \pL{2,}), PCRE could look past the end of the subject if it
3848 reached it while seeking the minimum number of characters. This could
3849 happen only if some of the characters were more than one byte long, because
3850 there is a check for at least the minimum number of bytes.
3851
3852 21. Refactored the implementation of \p and \P so as to be more general, to
3853 allow for more different types of property in future. This has changed the
3854 compiled form incompatibly. Anybody with saved compiled patterns that use
3855 \p or \P will have to recompile them.
3856
3857 22. Added "Any" and "L&" to the supported property types.
3858
3859 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3860 but give a compile time error if the value is greater than 0xff.
3861
3862 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3863 accidentally not being installed or uninstalled.
3864
3865 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3866 made were to insert the current release number. This seemed silly, because
3867 it made things harder for people building PCRE on systems that don't run
3868 "configure". I have turned pcre.h into a distributed file, no longer built
3869 by "configure", with the version identification directly included. There is
3870 no longer a pcre.h.in file.
3871
3872 However, this change necessitated a change to the pcre-config script as
3873 well. It is built from pcre-config.in, and one of the substitutions was the
3874 release number. I have updated configure.ac so that ./configure now finds
3875 the release number by grepping pcre.h.
3876
3877 26. Added the ability to run the tests under valgrind.
3878
3879
3880 Version 6.4 05-Sep-05
3881 ---------------------
3882
3883 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3884 "--" to be printed when multiple files were scanned, even when none of the
3885 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3886 consider it to be a bug, and have restored the previous behaviour.
3887
3888 2. A couple of code tidies to get rid of compiler warnings.
3889
3890 3. The pcretest program used to cheat by referring to symbols in the library
3891 whose names begin with _pcre_. These are internal symbols that are not
3892 really supposed to be visible externally, and in some environments it is
3893 possible to suppress them. The cheating is now confined to including
3894 certain files from the library's source, which is a bit cleaner.
3895
3896 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3897 file's purpose clearer.
3898
3899 5. Reorganized pcre_ucp_findchar().
3900
3901
3902 Version 6.3 15-Aug-05
3903 ---------------------
3904
3905 1. The file libpcre.pc.in did not have general read permission in the tarball.
3906
3907 2. There were some problems when building without C++ support:
3908
3909 (a) If C++ support was not built, "make install" and "make test" still
3910 tried to test it.
3911
3912 (b) There were problems when the value of CXX was explicitly set. Some
3913 changes have been made to try to fix these, and ...
3914
3915 (c) --disable-cpp can now be used to explicitly disable C++ support.
3916
3917 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3918 backslash in a target when C++ was disabled. This confuses some
3919 versions of "make", apparently. Using an intermediate variable solves
3920 this. (Same for CPP_LOBJ.)
3921
3922 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3923 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3924 necessary on certain architectures.
3925
3926 4. Added a setting of -export-symbols-regex to the link command to remove
3927 those symbols that are exported in the C sense, but actually are local
3928 within the library, and not documented. Their names all begin with
3929 "_pcre_". This is not a perfect job, because (a) we have to except some
3930 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3931 available (and never for static libraries). I have made a note to try to
3932 find a way round (a) in the future.
3933
3934
3935 Version 6.2 01-Aug-05
3936 ---------------------
3937
3938 1. There was no test for integer overflow of quantifier values. A construction
3939 such as {1111111111111111} would give undefined results. What is worse, if
3940 a minimum quantifier for a parenthesized subpattern overflowed and became
3941 negative, the calculation of the memory size went wrong. This could have
3942 led to memory overwriting.
3943
3944 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3945
3946 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3947 operating environments where this matters.
3948
3949 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3950 PCRE options from within the C++ wrapper.
3951
3952 5. Named capturing subpatterns were not being correctly counted when a pattern
3953 was compiled. This caused two problems: (a) If there were more than 100
3954 such subpatterns, the calculation of the memory needed for the whole
3955 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3956 back references of the form \12, where the number was greater than 9, were
3957 not recognized as back references, even though there were sufficient
3958 previous subpatterns.
3959
3960 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3961 versions of gcc, e.g. 2.95.4.
3962
3963
3964 Version 6.1 21-Jun-05
3965 ---------------------
3966
3967 1. There was one reference to the variable "posix" in pcretest.c that was not
3968 surrounded by "#if !defined NOPOSIX".
3969
3970 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3971 the cross-check on the old pcre_info() function, for the benefit of the
3972 cut-down version of PCRE that is currently imported into Exim.
3973
3974 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3975 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3976 patterns that start (?i) but I don't think that matters. The use of (?i) is
3977 just an example; this all applies to the other options as well.
3978
3979 4. Since libtool seems to echo the compile commands it is issuing, the output
3980 from "make" can be reduced a bit by putting "@" in front of each libtool
3981 compile command.
3982
3983 5. Patch from the folks at Google for configure.in to be a bit more thorough
3984 in checking for a suitable C++ installation before trying to compile the
3985 C++ stuff. This should fix a reported problem when a compiler was present,
3986 but no suitable headers.
3987
3988 6. The man pages all had just "PCRE" as their title. I have changed them to
3989 be the relevant file name. I have also arranged that these names are
3990 retained in the file doc/pcre.txt, which is a concatenation in text format
3991 of all the man pages except the little individual ones for each function.
3992
3993 7. The NON-UNIX-USE file had not been updated for the different set of source
3994 files that come with release 6. I also added a few comments about the C++
3995 wrapper.
3996
3997
3998 Version 6.0 07-Jun-05
3999 ---------------------
4000
4001 1. Some minor internal re-organization to help with my DFA experiments.
4002
4003 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4004 didn't matter for the library itself when fully configured, but did matter
4005 when compiling without UCP support, or within Exim, where the ucp files are
4006 not imported.
4007
4008 3. Refactoring of the library code to split up the various functions into
4009 different source modules. The addition of the new DFA matching code (see
4010 below) to a single monolithic source would have made it really too
4011 unwieldy, quite apart from causing all the code to be include in a
4012 statically linked application, when only some functions are used. This is
4013 relevant even without the DFA addition now that patterns can be compiled in
4014 one application and matched in another.
4015
4016 The downside of splitting up is that there have to be some external
4017 functions and data tables that are used internally in different modules of
4018 the library but which are not part of the API. These have all had their
4019 names changed to start with "_pcre_" so that they are unlikely to clash
4020 with other external names.
4021
4022 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4023 a different (DFA) algorithm. Although it is slower than the original
4024 function, it does have some advantages for certain types of matching
4025 problem.
4026
4027 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4028 including restarting after a partial match.
4029
4030 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4031 defined when compiling for Windows was sent to me. I have put it into the
4032 code, though I have no means of testing or verifying it.
4033
4034 7. Added the pcre_refcount() auxiliary function.
4035
4036 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4037 match before or at the first newline in the subject string. In pcretest,
4038 the /f option on a pattern can be used to set this.
4039
4040 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4041 would behave wrongly. This has been present in PCRE since release 4.0.
4042
4043 10. A number of changes to the pcregrep command:
4044
4045 (a) Refactored how -x works; insert ^(...)$ instead of setting
4046 PCRE_ANCHORED and checking the length, in preparation for adding
4047 something similar for -w.
4048
4049 (b) Added the -w (match as a word) option.
4050
4051 (c) Refactored the way lines are read and buffered so as to have more
4052 than one at a time available.
4053
4054 (d) Implemented a pcregrep test script.
4055
4056 (e) Added the -M (multiline match) option. This allows patterns to match
4057 over several lines of the subject. The buffering ensures that at least
4058 8K, or the rest of the document (whichever is the shorter) is available
4059 for matching (and similarly the previous 8K for lookbehind assertions).
4060
4061 (f) Changed the --help output so that it now says
4062
4063 -w, --word-regex(p)
4064
4065 instead of two lines, one with "regex" and the other with "regexp"
4066 because that confused at least one person since the short forms are the
4067 same. (This required a bit of code, as the output is generated
4068 automatically from a table. It wasn't just a text change.)
4069
4070 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4071 option but starts with a hyphen. Could be a pattern or a path name
4072 starting with a hyphen, for instance.
4073
4074 (h) "-" can be given as a file name to represent stdin.
4075
4076 (i) When file names are being printed, "(standard input)" is used for
4077 the standard input, for compatibility with GNU grep. Previously
4078 "<stdin>" was used.
4079
4080 (j) The option --label=xxx can be used to supply a name to be used for
4081 stdin when file names are being printed. There is no short form.
4082
4083 (k) Re-factored the options decoding logic because we are going to add
4084 two more options that take data. Such options can now be given in four
4085 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4086
4087 (l) Added the -A, -B, and -C options for requesting that lines of context
4088 around matches be printed.
4089
4090 (m) Added the -L option to print the names of files that do not contain
4091 any matching lines, that is, the complement of -l.
4092
4093 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4094 continue to scan other files.
4095
4096 (o) The -s option was incorrectly implemented. For compatibility with other
4097 greps, it now suppresses the error message for a non-existent or non-
4098 accessible file (but not the return code). There is a new option called
4099 -q that suppresses the output of matching lines, which was what -s was
4100 previously doing.
4101
4102 (p) Added --include and --exclude options to specify files for inclusion
4103 and exclusion when recursing.
4104
4105 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4106 Hopefully, it now does.
4107
4108 12. Missing cast in pcre_study().
4109
4110 13. Added an "uninstall" target to the makefile.
4111
4112 14. Replaced "extern" in the function prototypes in Makefile.in with
4113 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4114 world, but is set differently for Windows.
4115
4116 15. Added a second compiling function called pcre_compile2(). The only
4117 difference is that it has an extra argument, which is a pointer to an
4118 integer error code. When there is a compile-time failure, this is set
4119 non-zero, in addition to the error test pointer being set to point to an
4120 error message. The new argument may be NULL if no error number is required
4121 (but then you may as well call pcre_compile(), which is now just a
4122 wrapper). This facility is provided because some applications need a
4123 numeric error indication, but it has also enabled me to tidy up the way
4124 compile-time errors are handled in the POSIX wrapper.
4125
4126 16. Added VPATH=.libs to the makefile; this should help when building with one
4127 prefix path and installing with another. (Or so I'm told by someone who
4128 knows more about this stuff than I do.)
4129
4130 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4131 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4132 match everything, including newlines. This is not POSIX-compatible, but
4133 somebody wanted the feature. From pcretest it can be activated by using
4134 both the P and the s flags.
4135
4136 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4137
4138 19. libpcre.pc was being incorrectly installed as executable.
4139
4140 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4141 it now also looks for '\r' so that it will work unmodified on Windows.
4142
4143 21. Added Google's contributed C++ wrapper to the distribution.
4144
4145 22. Added some untidy missing memory free() calls in pcretest, to keep
4146 Electric Fence happy when testing.
4147
4148
4149
4150 Version 5.0 13-Sep-04
4151 ---------------------
4152
4153 1. Internal change: literal characters are no longer packed up into items
4154 containing multiple characters in a single byte-string. Each character
4155 is now matched using a separate opcode. However, there may be more than one
4156 byte in the character in UTF-8 mode.
4157
4158 2. The pcre_callout_block structure has two new fields: pattern_position and
4159 next_item_length. These contain the offset in the pattern to the next match
4160 item, and its length, respectively.
4161
4162 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4163 insertion of callouts before each pattern item. Added the /C option to
4164 pcretest to make use of this.
4165
4166 4. On the advice of a Windows user, the lines
4167
4168 #if defined(_WIN32) || defined(WIN32)
4169 _setmode( _fileno( stdout ), 0x8000 );
4170 #endif /* defined(_WIN32) || defined(WIN32) */
4171
4172 have been added to the source of pcretest. This apparently does useful
4173 magic in relation to line terminators.
4174
4175 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4176 for the benefit of those environments where the "b" makes a difference.
4177
4178 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4179 to know about it. I have put a hack into configure.in that adds in code
4180 to set GCC=yes if CC=icc. This seems to end up at a point in the
4181 generated configure script that is early enough to affect the setting of
4182 compiler options, which is what is needed, but I have no means of testing
4183 whether it really works. (The user who reported this had patched the
4184 generated configure script, which of course I cannot do.)
4185
4186 LATER: After change 22 below (new libtool files), the configure script
4187 seems to know about icc (and also ecc). Therefore, I have commented out
4188 this hack in configure.in.
4189
4190 7. Added support for pkg-config (2 patches were sent in).
4191
4192 8. Negated POSIX character classes that used a combination of internal tables
4193 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4194 [[:^ascii]]. Typically, they would match almost any characters. The other
4195 POSIX classes were not broken in this way.
4196
4197 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4198 to find the match, as PCRE was deluded into thinking that the match had to
4199 start at the start point or following a newline. The same bug applied to
4200 patterns with negative forward assertions or any backward assertions
4201 preceding ".*" at the start, unless the pattern required a fixed first
4202 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4203
4204 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4205 starting at the last subject character, bytes beyond the end of the subject
4206 string were read.
4207
4208 11. Renamed the variable "class" as "classbits" to make life easier for C++
4209 users. (Previously there was a macro definition, but it apparently wasn't
4210 enough.)
4211
4212 12. Added the new field "tables" to the extra data so that tables can be passed
4213 in at exec time, or the internal tables can be re-selected. This allows
4214 a compiled regex to be saved and re-used at a later time by a different
4215 program that might have everything at different addresses.
4216
4217 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4218 -R library as well as a -L library.
4219
4220 14. The debugging options of pcretest (-d on the command line or D on a
4221 pattern) showed incorrect output for anything following an extended class
4222 that contained multibyte characters and which was followed by a quantifier.
4223
4224 15. Added optional support for general category Unicode character properties
4225 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4226 support. It adds about 90K to the size of the library. The meanings of the
4227 inbuilt class escapes such as \d and \s have NOT been changed.
4228
4229 16. Updated pcredemo.c to include calls to free() to release the memory for the
4230 compiled pattern.
4231
4232 17. The generated file chartables.c was being created in the source directory
4233 instead of in the building directory. This caused the build to fail if the
4234 source directory was different from the building directory, and was
4235 read-only.
4236
4237 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4238 file. No doubt somebody will tell me if they don't make sense... Also added
4239 Dan Mooney's comments about building on OpenVMS.
4240
4241 19. Added support for partial matching via the PCRE_PARTIAL option for
4242 pcre_exec() and the \P data escape in pcretest.
4243
4244 20. Extended pcretest with 3 new pattern features:
4245
4246 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4247 write the compiled pattern to the file whose name is "rest-of-line".
4248 This is a straight binary dump of the data, with the saved pointer to
4249 the character tables forced to be NULL. The study data, if any, is
4250 written too. After writing, pcretest reads a new pattern.
4251
4252 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4253 compiled pattern from the given file. There must not be any
4254 occurrences of "<" in the file name (pretty unlikely); if there are,
4255 pcretest will instead treat the initial "<" as a pattern delimiter.
4256 After reading in the pattern, pcretest goes on to read data lines as
4257 usual.
4258
4259 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4260 and 16-bit fields in a compiled pattern, to simulate a pattern that
4261 was compiled on a host of opposite endianness.
4262
4263 21. The pcre-exec() function can now cope with patterns that were compiled on
4264 hosts of opposite endianness, with this restriction:
4265
4266 As for any compiled expression that is saved and used later, the tables
4267 pointer field cannot be preserved; the extra_data field in the arguments
4268 to pcre_exec() should be used to pass in a tables address if a value
4269 other than the default internal tables were used at compile time.
4270
4271 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4272 now diagnosed as an error. Previously, most of the time, a negative number
4273 would have been treated as zero, but if in addition "ovector" was passed as
4274 NULL, a crash could occur.
4275
4276 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4277 new versions from the libtool 1.5 distribution (the last one is a copy of
4278 a file called libtool.m4). This seems to have fixed the need to patch
4279 "configure" to support Darwin 1.3 (which I used to do). However, I still
4280 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4281 workstation).
4282
4283 24. Changed the PCRE licence to be the more standard "BSD" licence.
4284
4285
4286 Version 4.5 01-Dec-03
4287 ---------------------
4288
4289 1. There has been some re-arrangement of the code for the match() function so
4290 that it can be compiled in a version that does not call itself recursively.
4291 Instead, it keeps those local variables that need separate instances for
4292 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4293 needs to "recurse". Keeping track of where control must go is done by means
4294 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4295 hide most of the details from the main code, and operates only if
4296 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4297 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4298 operating.
4299
4300 To make it easier for callers to provide specially tailored get/free
4301 functions for this usage, two new functions, pcre_stack_malloc, and
4302 pcre_stack_free, are used. They are always called in strict stacking order,
4303 and the size of block requested is always the same.
4304
4305 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4306 PCRE has been compiled to use the stack or the heap for recursion. The
4307 -C option of pcretest uses this to show which version is compiled.
4308
4309 A new data escape \S, is added to pcretest; it causes the amounts of store
4310 obtained and freed by both kinds of malloc/free at match time to be added
4311 to the output.
4312
4313 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4314 what's available on my current Linux desktop machine.
4315
4316 3. When matching a UTF-8 string, the test for a valid string at the start has
4317 been extended. If start_offset is not zero, PCRE now checks that it points
4318 to a byte that is the start of a UTF-8 character. If not, it returns
4319 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4320 this is necessary because there may be backward assertions in the pattern.
4321 When matching the same subject several times, it may save resources to use
4322 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4323
4324 4. The code for checking the validity of UTF-8 strings has been tightened so
4325 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4326 containing "overlong sequences".
4327
4328 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4329 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4330 should have been "&", but it just so happened that all the cases this let
4331 through by mistake were picked up later in the function.
4332
4333 6. I had used a variable called "isblank" - this is a C99 function, causing
4334 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4335
4336 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4337 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4338 so the version is shown; (c) stop "make test" echoing "./RunTest".
4339
4340 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4341
4342 9. The prototype for memmove() for systems that don't have it was using
4343 size_t, but the inclusion of the header that defines size_t was later. I've
4344 moved the #includes for the C headers earlier to avoid this.
4345
4346 10. Added some adjustments to the code to make it easier to compiler on certain
4347 special systems:
4348
4349 (a) Some "const" qualifiers were missing.
4350 (b) Added the macro EXPORT before all exported functions; by default this
4351 is defined to be empty.
4352 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4353 that it reads its output file name as an argument instead of writing
4354 to the standard output and assuming this can be redirected.
4355
4356 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4357 class containing characters with values greater than 255, PCRE compilation
4358 went into a loop.
4359
4360 12. A recursive reference to a subpattern that was within another subpattern
4361 that had a minimum quantifier of zero caused PCRE to crash. For example,
4362 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4363 recursion. If the recursively-called subpattern itself had a zero repeat,
4364 that was OK.
4365
4366 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4367 buffer into which it was copied (for escape processing) was still set at
4368 1024, so long lines caused crashes.
4369
4370 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4371 "internal error: code overflow...". This applied to any character class
4372 that was followed by a possessive quantifier.
4373
4374 15. Modified the Makefile to add libpcre.la as a prerequisite for
4375 libpcreposix.la because I was told this is needed for a parallel build to
4376 work.
4377
4378 16. If a pattern that contained .* following optional items at the start was
4379 studied, the wrong optimizing data was generated, leading to matching
4380 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4381 matching string must start with a or b or c. The correct conclusion for
4382 this pattern is that a match can start with any character.
4383
4384
4385 Version 4.4 13-Aug-03
4386 ---------------------
4387
4388 1. In UTF-8 mode, a character class containing characters with values between
4389 127 and 255 was not handled correctly if the compiled pattern was studied.
4390 In fixing this, I have also improved the studying algorithm for such
4391 classes (slightly).
4392
4393 2. Three internal functions had redundant arguments passed to them. Removal
4394 might give a very teeny performance improvement.
4395
4396 3. Documentation bug: the value of the capture_top field in a callout is *one
4397 more than* the number of the hightest numbered captured substring.
4398
4399 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4400 in incorrectly linking with a previously installed version. They now link
4401 explicitly with libpcre.la.
4402
4403 5. configure.in no longer needs to recognize Cygwin specially.
4404
4405 6. A problem in pcre.in for Windows platforms is fixed.
4406
4407 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4408 pcretest, it used to include the size of the study block as part of its
4409 output. Unfortunately, the structure contains a field that has a different
4410 size on different hardware architectures. This meant that the tests that
4411 showed this size failed. As the block is currently always of a fixed size,
4412 this information isn't actually particularly useful in pcretest output, so
4413 I have just removed it.
4414
4415 8. Three pre-processor statements accidentally did not start in column 1.
4416 Sadly, there are *still* compilers around that complain, even though
4417 standard C has not required this for well over a decade. Sigh.
4418
4419 9. In pcretest, the code for checking callouts passed small integers in the
4420 callout_data field, which is a void * field. However, some picky compilers
4421 complained about the casts involved for this on 64-bit systems. Now
4422 pcretest passes the address of the small integer instead, which should get
4423 rid of the warnings.
4424
4425 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4426 both compile and run time, and gives an error if an invalid UTF-8 sequence
4427 is found. There is a option for disabling this check in cases where the
4428 string is known to be correct and/or the maximum performance is wanted.
4429
4430 11. In response to a bug report, I changed one line in Makefile.in from
4431
4432 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4433 to
4434 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4435
4436 to look similar to other lines, but I have no way of telling whether this
4437 is the right thing to do, as I do not use Windows. No doubt I'll get told
4438 if it's wrong...
4439
4440
4441 Version 4.3 21-May-03
4442 ---------------------
4443
4444 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4445 Makefile.
4446
4447 2. Some refactoring to improve the quality of the code:
4448
4449 (i) The utf8_table... variables are now declared "const".
4450
4451 (ii) The code for \cx, which used the "case flipping" table to upper case
4452 lower case letters, now just substracts 32. This is ASCII-specific,
4453 but the whole concept of \cx is ASCII-specific, so it seems
4454 reasonable.
4455
4456 (iii) PCRE was using its character types table to recognize decimal and
4457 hexadecimal digits in the pattern. This is silly, because it handles
4458 only 0-9, a-f, and A-F, but the character types table is locale-
4459 specific, which means strange things might happen. A private
4460 table is now used for this - though it costs 256 bytes, a table is
4461 much faster than multiple explicit tests. Of course, the standard
4462 character types table is still used for matching digits in subject
4463 strings against \d.
4464
4465 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4466 ending in _t are). So I've renamed it as ESC_tee.
4467
4468 3. The first argument for regexec() in the POSIX wrapper should have been
4469 defined as "const".
4470
4471 4. Changed pcretest to use malloc() for its buffers so that they can be
4472 Electric Fenced for debugging.
4473
4474 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4475 to read one or more bytes before the start of the subject string. Often this
4476 had no effect on PCRE's behaviour, but in some circumstances it could
4477 provoke a segmentation fault.
4478
4479 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4480 to try to read one or more bytes before the start of the subject string.
4481
4482 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4483 UTF-8 support could misbehave in various ways if the subject string
4484 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4485 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4486 back over UTF-8 characters.)
4487
4488
4489 Version 4.2 14-Apr-03
4490 ---------------------
4491
4492 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4493
4494 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4495 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4496 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4497 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4498 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4499 and BUILD_EXEEXT
4500 Note: automatic setting of the BUILD variables is not yet working
4501 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4502 compile-time but not at link-time
4503 [LINK]: use for linking executables only
4504 make different versions for Windows and non-Windows
4505 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4506 libraries
4507 [LINK_FOR_BUILD]: new variable
4508 [OBJEXT]: use throughout
4509 [EXEEXT]: use throughout
4510 <winshared>: new target
4511 <wininstall>: new target
4512 <dftables.o>: use native compiler
4513 <dftables>: use native linker
4514 <install>: handle Windows platform correctly
4515 <clean>: ditto
4516 <check>: ditto
4517 copy DLL to top builddir before testing
4518
4519 As part of these changes, -no-undefined was removed again. This was reported
4520 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4521 in any case.
4522
4523 3. Some tidies to get rid of compiler warnings:
4524
4525 . In the match_data structure, match_limit was an unsigned long int, whereas
4526 match_call_count was an int. I've made them both unsigned long ints.
4527
4528 . In pcretest the fact that a const uschar * doesn't automatically cast to
4529 a void * provoked a warning.
4530
4531 . Turning on some more compiler warnings threw up some "shadow" variables
4532 and a few more missing casts.
4533
4534 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4535 option, a class that contained a single character with a value between 128
4536 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4537
4538 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4539 option, a class that contained several characters, but with at least one
4540 whose value was between 128 and 255 caused PCRE to crash.
4541
4542
4543 Version 4.1 12-Mar-03
4544 ---------------------
4545
4546 1. Compiling with gcc -pedantic found a couple of places where casts were
4547 needed, and a string in dftables.c that was longer than standard compilers are
4548 required to support.
4549
4550 2. Compiling with Sun's compiler found a few more places where the code could
4551 be tidied up in order to avoid warnings.
4552
4553 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4554 first of these names is deprecated in the latest Autoconf in favour of the name
4555 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4556 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4557 analogy I have changed it to CFLAGS_FOR_BUILD.
4558
4559 4. Added -no-undefined to the linking command in the Makefile, because this is
4560 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4561 linking step for the pcreposix library.
4562
4563 5. PCRE was failing to diagnose the case of two named groups with the same
4564 name.
4565
4566 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4567 literal character that is needed in the subject for a match, and scans along to
4568 ensure that it is present before embarking on the full matching process. This
4569 saves time in cases of nested unlimited repeats that are never going to match.
4570 Problem: the scan can take a lot of time if the subject is very long (e.g.
4571 megabytes), thus penalizing straightforward matches. It is now done only if the
4572 amount of subject to be scanned is less than 1000 bytes.
4573
4574 7. A lesser problem with the same optimization is that it was recording the
4575 first character of an anchored pattern as "needed", thus provoking a search
4576 right along the subject, even when the first match of the pattern was going to
4577 fail. The "needed" character is now not set for anchored patterns, unless it
4578 follows something in the pattern that is of non-fixed length. Thus, it still
4579 fulfils its original purpose of finding quick non-matches in cases of nested
4580 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4581
4582
4583 Version 4.0 17-Feb-03
4584 ---------------------
4585
4586 1. If a comment in an extended regex that started immediately after a meta-item
4587 extended to the end of string, PCRE compiled incorrect data. This could lead to
4588 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4589
4590 2. Moved to autoconf 2.53 and libtool 1.4.2.
4591
4592 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4593 the special perltest8 script is no longer needed - all the tests can be run
4594 from a single perltest script.
4595
4596 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4597 by \s. It has now been removed in PCRE. This means it isn't recognized as
4598 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4599 class [:space:] *does* include VT, thereby creating a mess.
4600
4601 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4602 space and tab.
4603
4604 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4605 its new features into the main test script, reducing the number of scripts.
4606
4607 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4608 were backward compatible, and made the (?i) apply to the whole pattern, as if
4609 /i were given. Now it behaves more logically, and applies the option setting
4610 only to what follows. PCRE has been changed to follow suit. However, if it
4611 finds options settings right at the start of the pattern, it extracts them into
4612 the global options, as before. Thus, they show up in the info data.
4613
4614 8. Added support for the \Q...\E escape sequence. Characters in between are
4615 treated as literals. This is slightly different from Perl in that $ and @ are
4616 also handled as literals inside the quotes. In Perl, they will cause variable
4617 interpolation. Note the following examples:
4618
4619 Pattern PCRE matches Perl matches
4620
4621 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4622 \Qabc\$xyz\E abc\$xyz abc\$xyz
4623 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4624
4625 For compatibility with Perl, \Q...\E sequences are recognized inside character
4626 classes as well as outside them.
4627
4628 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4629 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4630 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4631 signed/unsigned warnings.
4632
4633 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4634 option for pcretest, so I've replaced it by a simple function that does just
4635 that job.
4636
4637 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4638 "pcregrep -".
4639
4640 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4641 Java package. This provides some syntactic sugar for simple cases of what my
4642 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4643 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4644 item, you can use this simplified notation. Note that only makes sense with
4645 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4646 greediness, whatever the setting of the PCRE_UNGREEDY option.
4647
4648 13. A change of greediness default within a pattern was not taking effect at
4649 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4650 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4651 was abstracted outside.
4652
4653 14. PCRE now supports the \G assertion. It is true when the current matching
4654 position is at the start point of the match. This differs from \A when the
4655 starting offset is non-zero. Used with the /g option of pcretest (or similar
4656 code), it works in the same way as it does for Perl's /g option. If all
4657 alternatives of a regex begin with \G, the expression is anchored to the start
4658 match position, and the "anchored" flag is set in the compiled expression.
4659
4660 15. Some bugs concerning the handling of certain option changes within patterns
4661 have been fixed. These applied to options other than (?ims). For example,
4662 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4663 been the other way round. Some of this was related to change 7 above.
4664
4665 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4666 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4667 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4668 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4669
4670 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4671 mode. Unlike ".", it always matches newline, whatever the setting of
4672 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4673 assertions. Perl allows it, but it doesn't (in general) work because it can't
4674 calculate the length of the lookbehind. At least, that's the case for Perl
4675 5.8.0 - I've been told they are going to document that it doesn't work in
4676 future.
4677
4678 18. Added an error diagnosis for escapes that PCRE does not support: these are
4679 \L, \l, \N, \P, \p, \U, \u, and \X.
4680
4681 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4682 reading past the end of the pattern in cases such as /[abcd/.
4683
4684 20. PCRE was getting more memory than necessary for patterns with classes that
4685 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4686
4687 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4688 compiling PCRE for use with Virtual Pascal.
4689
4690 22. Small fix to the Makefile to make it work properly if the build is done
4691 outside the source tree.
4692
4693 23. Added a new extension: a condition to go with recursion. If a conditional
4694 subpattern starts with (?(R) the "true" branch is used if recursion has
4695 happened, whereas the "false" branch is used only at the top level.
4696
4697 24. When there was a very long string of literal characters (over 255 bytes
4698 without UTF support, over 250 bytes with UTF support), the computation of how
4699 much memory was required could be incorrect, leading to segfaults or other
4700 strange effects.
4701
4702 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4703 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4704 there was a subsequent back reference to those brackets. This meant that, for
4705 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4706 possible to check for precisely this case. All we can do is abandon the
4707 optimization if .* occurs inside capturing brackets when there are any back
4708 references whatsoever. (See below for a better fix that came later.)
4709
4710 26. The handling of the optimization for finding the first character of a
4711 non-anchored pattern, and for finding a character that is required later in the
4712 match were failing in some cases. This didn't break the matching; it just
4713 failed to optimize when it could. The way this is done has been re-implemented.
4714
4715 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4716
4717 28. Added a new feature that provides some of the functionality that Perl
4718 provides with (?{...}). The facility is termed a "callout". The way it is done
4719 in PCRE is for the caller to provide an optional function, by setting
4720 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4721 global variable. By default it is unset, which disables all calling out. To get
4722 the function called, the regex must include (?C) at appropriate points. This
4723 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4724 This provides a means of identifying different callout points. When PCRE
4725 reaches such a point in the regex, if pcre_callout has been set, the external
4726 function is called. It is provided with data in a structure called
4727 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4728 matching continues; if it returns a non-zero value, the match at the current
4729 point fails. However, backtracking will occur if possible. [This was changed
4730 later and other features added - see item 49 below.]
4731
4732 29. pcretest is upgraded to test the callout functionality. It provides a
4733 callout function that displays information. By default, it shows the start of
4734 the match and the current position in the text. There are some new data escapes
4735 to vary what happens:
4736
4737 \C+ in addition, show current contents of captured substrings
4738 \C- do not supply a callout function
4739 \C!n return 1 when callout number n is reached
4740 \C!n!m return 1 when callout number n is reached for the mth time
4741
4742 30. If pcregrep was called with the -l option and just a single file name, it
4743 output "<stdin>" if a match was found, instead of the file name.
4744
4745 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4746 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4747 pcre_exec(). This saves a malloc/free per call. The default value of
4748 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4749 when configuring.
4750
4751 32. The default maximum size of a compiled pattern is 64K. There have been a
4752 few cases of people hitting this limit. The code now uses macros to handle the
4753 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4754 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4755 configuring. Tests 2 and 5 work only with 2-byte links because they output
4756 debugging information about compiled patterns.
4757
4758 33. Internal code re-arrangements:
4759
4760 (a) Moved the debugging function for printing out a compiled regex into
4761 its own source file (printint.c) and used #include to pull it into
4762 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4763 separate copies.
4764
4765 (b) Defined the list of op-code names for debugging as a macro in
4766 internal.h so that it is next to the definition of the opcodes.
4767
4768 (c) Defined a table of op-code lengths for simpler skipping along compiled
4769 code. This is again a macro in internal.h so that it is next to the
4770 definition of the opcodes.
4771
4772 34. Added support for recursive calls to individual subpatterns, along the
4773 lines of Robin Houston's patch (but implemented somewhat differently).
4774
4775 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4776 allow it to read and process whole directories in Win32. This code was
4777 contributed by Lionel Fourquaux; it has not been tested by me.
4778
4779 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4780 used to name a group. Names consist of alphanumerics and underscores, and must
4781 be unique. Back references use the syntax (?P=name) and recursive calls use
4782 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4783 numbers. The function pcre_fullinfo() can be used after compilation to extract
4784 a name/number map. There are three relevant calls:
4785
4786 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4787 PCRE_INFO_NAMECOUNT yields the number of entries
4788 PCRE_INFO_NAMETABLE yields a pointer to the map.
4789
4790 The map is a vector of fixed-size entries. The size of each entry depends on
4791 the length of the longest name used. The first two bytes of each entry are the
4792 group number, most significant byte first. There follows the corresponding
4793 name, zero terminated. The names are in alphabetical order.
4794
4795 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4796 case instead of 255. Making it the same both with and without UTF-8 support
4797 means that the same test output works with both.
4798
4799 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4800 calling malloc() with a zero argument.
4801
4802 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4803 optimization. I've improved things by keeping a bitmap of backreferences with
4804 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4805 fact referenced, the optimization can be applied. It is unlikely that a
4806 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4807 the match to follow \n) will appear inside brackets with a number greater than
4808 31, but if it does, any back reference > 31 suppresses the optimization.
4809
4810 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4811 of disabling numbered capturing parentheses. Any opening parenthesis that is
4812 not followed by ? behaves as if it were followed by ?: but named parentheses
4813 can still be used for capturing (and they will acquire numbers in the usual
4814 way).
4815
4816 41. Redesigned the return codes from the match() function into yes/no/error so
4817 that errors can be passed back from deep inside the nested calls. A malloc
4818 failure while inside a recursive subpattern call now causes the
4819 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4820
4821 42. It is now possible to set a limit on the number of times the match()
4822 function is called in a call to pcre_exec(). This facility makes it possible to
4823 limit the amount of recursion and backtracking, though not in a directly
4824 obvious way, because the match() function is used in a number of different
4825 circumstances. The count starts from zero for each position in the subject
4826 string (for non-anchored patterns). The default limit is, for compatibility, a
4827 large number, namely 10 000 000. You can change this in two ways:
4828
4829 (a) When configuring PCRE before making, you can use --with-match-limit=n
4830 to set a default value for the compiled library.
4831
4832 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4833 a different value is set. See 45 below.
4834
4835 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4836
4837 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4838 of things that can be changed at compile time. The first argument specifies
4839 what is wanted and the second points to where the information is to be placed.
4840 The current list of available information is:
4841
4842 PCRE_CONFIG_UTF8
4843
4844 The output is an integer that is set to one if UTF-8 support is available;
4845 otherwise it is set to zero.
4846
4847 PCRE_CONFIG_NEWLINE
4848
4849 The output is an integer that it set to the value of the code that is used for
4850 newline. It is either LF (10) or CR (13).
4851
4852 PCRE_CONFIG_LINK_SIZE
4853
4854 The output is an integer that contains the number of bytes used for internal
4855 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4856
4857 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4858
4859 The output is an integer that contains the threshold above which the POSIX
4860 interface uses malloc() for output vectors. See item 31 above.
4861
4862 PCRE_CONFIG_MATCH_LIMIT
4863
4864 The output is an unsigned integer that contains the default limit of the number
4865 of match() calls in a pcre_exec() execution. See 42 above.
4866
4867 44. pcretest has been upgraded by the addition of the -C option. This causes it
4868 to extract all the available output from the new pcre_config() function, and to
4869 output it. The program then exits immediately.
4870
4871 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4872 order to support additional features. One way would have been to define
4873 pcre_exec2() (for example) with extra arguments, but this would not have been
4874 extensible, and would also have required all calls to the original function to
4875 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4876 is used for passing in "extra" data from pcre_study().
4877
4878 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4879 contains the following fields:
4880
4881 flags a bitmap indicating which of the following fields are set
4882 study_data opaque data from pcre_study()
4883 match_limit a way of specifying a limit on match() calls for a specific
4884 call to pcre_exec()
4885 callout_data data for callouts (see 49 below)
4886
4887 The flag bits are also defined in pcre.h, and are
4888
4889 PCRE_EXTRA_STUDY_DATA
4890 PCRE_EXTRA_MATCH_LIMIT
4891 PCRE_EXTRA_CALLOUT_DATA
4892
4893 The pcre_study() function now returns one of these new pcre_extra blocks, with
4894 the actual study data pointed to by the study_data field, and the
4895 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4896 before. That is, this change is entirely upwards-compatible and requires no
4897 change to existing code.
4898
4899 If you want to pass in additional data to pcre_exec(), you can either place it
4900 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4901 block.
4902
4903 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4904 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4905 times with different match limits, until it finds the minimum value needed for
4906 pcre_exec() to complete. The value is then output. This can be instructive; for
4907 most simple matches the number is quite small, but for pathological cases it
4908 gets very large very quickly.
4909
4910 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4911 returns the size of the data block pointed to by the study_data field in a
4912 pcre_extra block, that is, the value that was passed as the argument to
4913 pcre_malloc() when PCRE was getting memory in which to place the information
4914 created by pcre_study(). The fourth argument should point to a size_t variable.
4915 pcretest has been extended so that this information is shown after a successful
4916 pcre_study() call when information about the compiled regex is being displayed.
4917
4918 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4919 because what follows is always an absolute path. (Later: it turns out that this
4920 is more than cosmetic for MinGW, because it doesn't like empty path
4921 components.)
4922
4923 49. Some changes have been made to the callout feature (see 28 above):
4924
4925 (i) A callout function now has three choices for what it returns:
4926
4927 0 => success, carry on matching
4928 > 0 => failure at this point, but backtrack if possible
4929 < 0 => serious error, return this value from pcre_exec()
4930
4931 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4932 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4933 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4934 use by callout functions. It will never be used by PCRE itself.
4935
4936 (ii) The pcre_extra structure (see 45 above) has a void * field called
4937 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4938 pcre_callout_block structure has a field of the same name. The contents of
4939 the field passed in the pcre_extra structure are passed to the callout
4940 function in the corresponding field in the callout block. This makes it
4941 easier to use the same callout-containing regex from multiple threads. For
4942 testing, the pcretest program has a new data escape
4943
4944 \C*n pass the number n (may be negative) as callout_data
4945
4946 If the callout function in pcretest receives a non-zero value as
4947 callout_data, it returns that value.
4948
4949 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4950 there were some redundant $(CFLAGS) in commands that are now specified as
4951 $(LINK), which already includes $(CFLAGS).
4952
4953 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4954 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4955 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4956 one-byte characters throughout. Note that case-insensitive matching applies
4957 only to characters whose values are less than 256. PCRE doesn't support the
4958 notion of cases for higher-valued characters.
4959
4960 (i) A character class whose characters are all within 0-255 is handled as
4961 a bit map, and the map is inverted for negative classes. Previously, a
4962 character > 255 always failed to match such a class; however it should
4963 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4964
4965 (ii) A negated character class with a single character < 255 is coded as
4966 "not this character" (OP_NOT). This wasn't working properly when the test
4967 character was multibyte, either singly or repeated.
4968
4969 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4970 mode, for example: \x{100}{2,3}.
4971
4972 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4973 singly or repeated) now correctly test multibyte characters. However,
4974 PCRE doesn't recognize any characters with values greater than 255 as
4975 digits, spaces, or word characters. Such characters always match \D, \S,
4976 and \W, and never match \d, \s, or \w.
4977
4978 (v) Classes may now contain characters and character ranges with values
4979 greater than 255. For example: [ab\x{100}-\x{400}].
4980
4981 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4982 PCRE in UTF-8 mode.
4983
4984 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4985 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4986 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4987 value.)
4988
4989 53. The single man page has become too large. I have therefore split it up into
4990 a number of separate man pages. These also give rise to individual HTML pages;
4991 these are now put in a separate directory, and there is an index.html page that
4992 lists them all. Some hyperlinking between the pages has been installed.
4993
4994 54. Added convenience functions for handling named capturing parentheses.
4995
4996 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4997 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4998 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4999 are faulted.
5000
5001 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5002 calling configure. These values are used when compiling the dftables.c program
5003 which is run to generate the source of the default character tables. They
5004 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5005 you will need to set these values.
5006
5007 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5008
5009
5010 Version 3.9 02-Jan-02
5011 ---------------------
5012
5013 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5014
5015 2. If --disable-static was given, the building process failed when trying to
5016 build pcretest and pcregrep. (For some reason it was using libtool to compile
5017 them, which is not right, as they aren't part of the library.)
5018
5019
5020 Version 3.8 18-Dec-01
5021 ---------------------
5022
5023 1. The experimental UTF-8 code was completely screwed up. It was packing the
5024 bytes in the wrong order. How dumb can you get?
5025
5026
5027 Version 3.7 29-Oct-01
5028 ---------------------
5029
5030 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5031 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5032 this didn't happen under Solaris 8, where I normally test things.
5033
5034 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5035 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5036 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5037 this fix an hour or so after the initial 3.7 release.)
5038
5039
5040 Version 3.6 23-Oct-01
5041 ---------------------
5042
5043 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5044 offsets passed as NULL with zero offset count.
5045
5046 2. The config.guess and config.sub files had not been updated when I moved to
5047 the latest autoconf.
5048
5049
5050 Version 3.5 15-Aug-01
5051 ---------------------
5052
5053 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5054 had been forgotten.
5055
5056 2. By using declared but undefined structures, we can avoid using "void"
5057 definitions in pcre.h while keeping the internal definitions of the structures
5058 private.
5059
5060 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5061 user point of view, this means that both static and shared libraries are built
5062 by default, but this can be individually controlled. More of the work of
5063 handling this static/shared cases is now inside libtool instead of PCRE's make
5064 file.
5065
5066 4. The pcretest utility is now installed along with pcregrep because it is
5067 useful for users (to test regexs) and by doing this, it automatically gets
5068 relinked by libtool. The documentation has been turned into a man page, so
5069 there are now .1, .txt, and .html versions in /doc.
5070
5071 5. Upgrades to pcregrep:
5072 (i) Added long-form option names like gnu grep.
5073 (ii) Added --help to list all options with an explanatory phrase.
5074 (iii) Added -r, --recursive to recurse into sub-directories.
5075 (iv) Added -f, --file to read patterns from a file.
5076
5077 6. pcre_exec() was referring to its "code" argument before testing that
5078 argument for NULL (and giving an error if it was NULL).
5079
5080 7. Upgraded Makefile.in to allow for compiling in a different directory from
5081 the source directory.
5082
5083 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5084 options bits, the pointer it was passed was to an int instead of to an unsigned
5085 long int. This mattered only on 64-bit systems.
5086
5087 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5088 generated) instead of pcre.in, which it its source. Also made the same change
5089 in several of the .c files.
5090
5091 10. A new release of gcc defines printf() as a macro, which broke pcretest
5092 because it had an ifdef in the middle of a string argument for printf(). Fixed
5093 by using separate calls to printf().
5094
5095 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5096 script, to force use of CR or LF instead of \n in the source. On non-Unix
5097 systems, the value can be set in config.h.
5098
5099 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5100 absolute limit. Changed the text of the error message to make this clear, and
5101 likewise updated the man page.
5102
5103 13. The limit of 99 on the number of capturing subpatterns has been removed.
5104 The new limit is 65535, which I hope will not be a "real" limit.
5105
5106
5107 Version 3.4 22-Aug-00
5108 ---------------------
5109
5110 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5111
5112 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5113
5114
5115 Version 3.3 01-Aug-00
5116 ---------------------
5117
5118 1. If an octal character was given, but the value was greater than \377, it
5119 was not getting masked to the least significant bits, as documented. This could
5120 lead to crashes in some systems.
5121
5122 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5123 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5124
5125 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5126 These just pass their arguments on to (pcre_free)(), but they are provided
5127 because some uses of PCRE bind it to non-C systems that can call its functions,
5128 but cannot call free() or pcre_free() directly.
5129
5130 4. Add "make test" as a synonym for "make check". Corrected some comments in
5131 the Makefile.
5132
5133 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5134 Makefile.
5135
5136 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5137 command called pgrep for grepping around the active processes.
5138
5139 7. Added the beginnings of support for UTF-8 character strings.
5140
5141 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5142 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5143 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5144 out for the ar command.)
5145
5146
5147 Version 3.2 12-May-00
5148 ---------------------
5149
5150 This is purely a bug fixing release.
5151
5152 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5153 of ZA. This was just one example of several cases that could provoke this bug,
5154 which was introduced by change 9 of version 2.00. The code for breaking
5155 infinite loops after an iteration that matches an empty string was't working
5156 correctly.
5157
5158 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5159 when matched against abbab (for example). After matching an empty string, it
5160 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5161 caused it to match further down the string than it should.
5162
5163 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5164 was there because it doesn't seem to be needed, and it causes trouble on some
5165 systems, as it is not a Standard C header. It has been removed.
5166
5167 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5168 were reported on the Macintosh. The changes were from
5169
5170 while ((c = *(++ptr)) != 0 && c != '\n');
5171 to
5172 while ((c = *(++ptr)) != 0 && c != '\n') ;
5173
5174 Totally extraordinary, but if that's what it takes...
5175
5176 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5177 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5178 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5179 assumes the way PCRE uses memmove() (always moving upwards).
5180
5181 6. PCRE is being used in one environment where strchr() is not available. There
5182 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5183 faster code anyway.
5184
5185
5186 Version 3.1 09-Feb-00
5187 ---------------------
5188
5189 The only change in this release is the fixing of some bugs in Makefile.in for
5190 the "install" target:
5191
5192 (1) It was failing to install pcreposix.h.
5193
5194 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5195
5196
5197 Version 3.0 01-Feb-00
5198 ---------------------
5199
5200 1. Add support for the /+ modifier to perltest (to output $` like it does in
5201 pcretest).
5202
5203 2. Add support for the /g modifier to perltest.
5204
5205 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5206 matches null strings.
5207
5208 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5209 pattern. Perl treats empty patterns specially - it reuses the most recent
5210 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5211 effect.
5212
5213 5. The POSIX interface was broken in that it was just handing over the POSIX
5214 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5215 required a bigger vector, with some working space on the end. This means that
5216 the POSIX wrapper now has to get and free some memory, and copy the results.
5217
5218 6. Added some simple autoconf support, placing the test data and the
5219 documentation in separate directories, re-organizing some of the
5220 information files, and making it build pcre-config (a GNU standard). Also added
5221 libtool support for building PCRE as a shared library, which is now the
5222 default.
5223
5224 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5225 09 are not valid octal constants. Single digits will be used for minor values
5226 less than 10.
5227
5228 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5229 existing programs that set these in the POSIX interface can use PCRE without
5230 modification.
5231
5232 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5233 return all that pcre_info() returns, plus additional data. The pcre_info()
5234 function is retained for compatibility, but is considered to be obsolete.
5235
5236 10. Added experimental recursion feature (?R) to handle one common case that
5237 Perl 5.6 will be able to do with (?p{...}).
5238
5239 11. Added support for POSIX character classes like [:alpha:], which Perl is
5240 adopting.
5241
5242
5243 Version 2.08 31-Aug-99
5244 ----------------------
5245
5246 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5247 trying to match at the startoffset position, but instead was moving forward to
5248 the next newline as if a previous match had failed.
5249
5250 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5251 and could get into a loop if a null string was matched other than at the start
5252 of the subject.
5253
5254 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5255 be distinguished at compile time, and for completeness also added PCRE_DATE.
5256
5257 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5258 in GnuWin32 environments.
5259
5260
5261 Version 2.07 29-Jul-99
5262 ----------------------
5263
5264 1. The documentation is now supplied in plain text form and HTML as well as in
5265 the form of man page sources.
5266
5267 2. C++ compilers don't like assigning (void *) values to other pointer types.
5268 In particular this affects malloc(). Although there is no problem in Standard
5269 C, I've put in casts to keep C++ compilers happy.
5270
5271 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5272 should be (const char *).
5273
5274 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5275 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5276 However, I haven't made this a standard facility. The documentation doesn't
5277 mention it, and the Makefile doesn't support it.
5278
5279 5. The Makefile now contains an "install" target, with editable destinations at
5280 the top of the file. The pcretest program is not installed.
5281
5282 6. pgrep -V now gives the PCRE version number and date.
5283
5284 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5285 causing the entire string to be ignored, instead of just the last character.
5286
5287 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5288 non-matching string, it can take a very, very long time, even for strings of
5289 quite modest length, because of the nested recursion. PCRE now does better in
5290 some of these cases. It does this by remembering the last required literal
5291 character in the pattern, and pre-searching the subject to ensure it is present
5292 before running the real match. In other words, it applies a heuristic to detect
5293 some types of certain failure quickly, and in the above example, if presented
5294 with a string that has no trailing " it gives "no match" very quickly.
5295
5296 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5297 other alternatives are tried instead.
5298
5299
5300 Version 2.06 09-Jun-99
5301 ----------------------
5302
5303 1. Change pcretest's output for amount of store used to show just the code
5304 space, because the remainder (the data block) varies in size between 32-bit and
5305 64-bit systems.
5306
5307 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5308 start matching at. This allows lookbehinds to work when searching for multiple
5309 occurrences in a string.
5310
5311 3. Added additional options to pcretest for testing multiple occurrences:
5312
5313 /+ outputs the rest of the string that follows a match
5314 /g loops for multiple occurrences, using the new startoffset argument
5315 /G loops for multiple occurrences by passing an incremented pointer
5316
5317 4. PCRE wasn't doing the "first character" optimization for patterns starting
5318 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5319 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5320 the letter 'x'. On long subject strings, this gives a significant speed-up.
5321
5322
5323 Version 2.05 21-Apr-99
5324 ----------------------
5325
5326 1. Changed the type of magic_number from int to long int so that it works
5327 properly on 16-bit systems.
5328
5329 2. Fixed a bug which caused patterns starting with .* not to work correctly
5330 when the subject string contained newline characters. PCRE was assuming
5331 anchoring for such patterns in all cases, which is not correct because .* will
5332 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5333 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5334 must be retried after every newline in the subject.
5335
5336
5337 Version 2.04 18-Feb-99
5338 ----------------------
5339
5340 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5341 computation of the store needed to hold the pattern was incorrect (too large).
5342 If such patterns were nested a few deep, this could multiply and become a real
5343 problem.
5344
5345 2. Added /M option to pcretest to show the memory requirement of a specific
5346 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5347
5348 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5349 compiled in such a way that the backtracking after subsequent failure was
5350 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5351 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5352
5353
5354 Version 2.03 02-Feb-99
5355 ----------------------
5356
5357 1. Fixed typo and small mistake in man page.
5358
5359 2. Added 4th condition (GPL supersedes if conflict) and created separate
5360 LICENCE file containing the conditions.
5361
5362 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5363 Perl, that is the internal \ allows the delimiter to be included in the
5364 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5365 the final delimiter, add \ to the end of the pattern (to test the error).
5366
5367 4. Added the convenience functions for extracting substrings after a successful
5368 match. Updated pcretest to make it able to test these functions.
5369
5370
5371 Version 2.02 14-Jan-99
5372 ----------------------
5373
5374 1. Initialized the working variables associated with each extraction so that
5375 their saving and restoring doesn't refer to uninitialized store.
5376
5377 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5378 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5379 fix the problem.
5380
5381 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5382 calls, and wasn't printing the correct value for compiling calls. Increased the
5383 default value of LOOPREPEAT, and the number of significant figures in the
5384 times.
5385
5386 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5387
5388 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5389 a building problem on Windows NT with a FAT file system.
5390
5391
5392 Version 2.01 21-Oct-98
5393 ----------------------
5394
5395 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5396 to character tables built by pcre_maketables() in the current locale. If NULL
5397 is passed, the default tables are used.
5398
5399
5400 Version 2.00 24-Sep-98
5401 ----------------------
5402
5403 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5404 it any more.
5405
5406 2. Allow quantification of (?>) groups, and make it work correctly.
5407
5408 3. The first character computation wasn't working for (?>) groups.
5409
5410 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5411 end of the subject) and add 5.005's \z, which really does match only at the
5412 very end of the subject.
5413
5414 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5415
5416 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5417 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5418 localized options. All options to pcre_study() were also removed.
5419
5420 7. Add other new features from 5.005:
5421
5422 $(?<= positive lookbehind
5423 $(?<! negative lookbehind
5424 (?imsx-imsx) added the unsetting capability
5425 such a setting is global if at outer level; local otherwise
5426 (?imsx-imsx:) non-capturing groups with option setting
5427 (?(cond)re|re) conditional pattern matching
5428
5429 A backreference to itself in a repeated group matches the previous
5430 captured string.
5431
5432 8. General tidying up of studying (both automatic and via "study")
5433 consequential on the addition of new assertions.
5434
5435 9. As in 5.005, unlimited repeated groups that could match an empty substring
5436 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5437 runtime if any iteration does actually match an empty substring.
5438
5439 10. Include the RunTest script in the distribution.
5440
5441 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5442 discrepancies, some of which were old and were also with respect to 5.004. They
5443 have now been fixed.
5444
5445
5446 Version 1.09 28-Apr-98
5447 ----------------------
5448
5449 1. A negated single character class followed by a quantifier with a minimum
5450 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5451 program crashes, or just wrong answers. This did not apply to negated classes
5452 containing more than one character, or to minima other than one.
5453
5454
5455 Version 1.08 27-Mar-98
5456 ----------------------
5457
5458 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5459
5460 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5461 latter must appear before anything that relies on it in the pattern.
5462
5463
5464 Version 1.07 16-Feb-98
5465 ----------------------
5466
5467 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5468 repeat of a potentially empty string).
5469
5470
5471 Version 1.06 23-Jan-98
5472 ----------------------
5473
5474 1. Added Markus Oberhumer's little patches for C++.
5475
5476 2. Literal strings longer than 255 characters were broken.
5477
5478
5479 Version 1.05 23-Dec-97
5480 ----------------------
5481
5482 1. Negated character classes containing more than one character were failing if
5483 PCRE_CASELESS was set at run time.
5484
5485
5486 Version 1.04 19-Dec-97
5487 ----------------------
5488
5489 1. Corrected the man page, where some "const" qualifiers had been omitted.
5490
5491 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5492 input syntax.
5493
5494 3. Fixed memory leak which occurred when a regex with back references was
5495 matched with an offsets vector that wasn't big enough. The temporary memory
5496 that is used in this case wasn't being freed if the match failed.
5497
5498 4. Tidied pcretest to ensure it frees memory that it gets.
5499
5500 5. Temporary memory was being obtained in the case where the passed offsets
5501 vector was exactly big enough.
5502
5503 6. Corrected definition of offsetof() from change 5 below.
5504
5505 7. I had screwed up change 6 below and broken the rules for the use of
5506 setjmp(). Now fixed.
5507
5508
5509 Version 1.03 18-Dec-97
5510 ----------------------
5511
5512 1. A erroneous regex with a missing opening parenthesis was correctly
5513 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5514 on some systems.
5515
5516 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5517 it was reported that one broken compiler failed on the former because "code" is
5518 also an independent variable.
5519
5520 3. The erroneous regex a[]b caused an array overrun reference.
5521
5522 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5523 fail on data ending with that character. (It was going on too far, and checking
5524 the next character, typically a binary zero.) This was specific to the
5525 optimized code for single-character negative classes.
5526
5527 5. Added a contributed patch from the TIN world which does the following:
5528
5529 + Add an undef for memmove, in case the the system defines a macro for it.
5530
5531 + Add a definition of offsetof(), in case there isn't one. (I don't know
5532 the reason behind this - offsetof() is part of the ANSI standard - but
5533 it does no harm).
5534
5535 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5536 most of the places where whitespace preceded '#'. I have given up and
5537 allowed the remaining 2 cases to be at the margin.
5538
5539 + Rename some variables in pcre to eliminate shadowing. This seems very
5540 pedantic, but does no harm, of course.
5541
5542 6. Moved the call to setjmp() into its own function, to get rid of warnings
5543 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5544
5545 7. Constructs such as \d{8,} were compiling into the equivalent of
5546 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5547 outcome, but in this particular case used more store than had been allocated,
5548 which caused the bug to be discovered because it threw up an internal error.
5549
5550 8. The debugging code in both pcre and pcretest for outputting the compiled
5551 form of a regex was going wrong in the case of back references followed by
5552 curly-bracketed repeats.
5553
5554
5555 Version 1.02 12-Dec-97
5556 ----------------------
5557
5558 1. Typos in pcre.3 and comments in the source fixed.
5559
5560 2. Applied a contributed patch to get rid of places where it used to remove
5561 'const' from variables, and fixed some signed/unsigned and uninitialized
5562 variable warnings.
5563
5564 3. Added the "runtest" target to Makefile.
5565
5566 4. Set default compiler flag to -O2 rather than just -O.
5567
5568
5569 Version 1.01 19-Nov-97
5570 ----------------------
5571
5572 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5573 like /([ab]*)*/, that is, for classes with more than one character in them.
5574
5575 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5576 as /((?>a*))*/ (a PCRE_EXTRA facility).
5577
5578
5579 Version 1.00 18-Nov-97
5580 ----------------------
5581
5582 1. Added compile-time macros to support systems such as SunOS4 which don't have
5583 memmove() or strerror() but have other things that can be used instead.
5584
5585 2. Arranged that "make clean" removes the executables.
5586
5587
5588 Version 0.99 27-Oct-97
5589 ----------------------
5590
5591 1. Fixed bug in code for optimizing classes with only one character. It was
5592 initializing a 32-byte map regardless, which could cause it to run off the end
5593 of the memory it had got.
5594
5595 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5596
5597
5598 Version 0.98 22-Oct-97
5599 ----------------------
5600
5601 1. Fixed bug in code for handling temporary memory usage when there are more
5602 back references than supplied space in the ovector. This could cause segfaults.
5603
5604
5605 Version 0.97 21-Oct-97
5606 ----------------------
5607
5608 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5609
5610 2. Optimized negated single characters not to use a bit map.
5611
5612 3. Brought error texts together as macro definitions; clarified some of them;
5613 fixed one that was wrong - it said "range out of order" when it meant "invalid
5614 escape sequence".
5615
5616 4. Changed some char * arguments to const char *.
5617
5618 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5619
5620 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5621 pcretest.
5622
5623
5624 Version 0.96 16-Oct-97
5625 ----------------------
5626
5627 1. Added a simple "pgrep" utility to the distribution.
5628
5629 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5630 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5631 where "ddd" means "one or more decimal digits".
5632
5633 3. Fixed serious bug. If a pattern had a back reference, but the call to
5634 pcre_exec() didn't supply a large enough ovector to record the related
5635 identifying subpattern, the match always failed. PCRE now remembers the number
5636 of the largest back reference, and gets some temporary memory in which to save
5637 the offsets during matching if necessary, in order to ensure that
5638 backreferences always work.
5639
5640 4. Increased the compatibility with Perl in a number of ways:
5641
5642 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5643 to request this handling. The option can be set at compile or exec time.
5644
5645 (b) $ matches before a terminating newline by default; an option
5646 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5647 mode). The option can be set at compile or exec time.
5648
5649 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5650 the same as Perl's. If the decimal number it represents is less than 10
5651 or there aren't that many previous left capturing parentheses, an octal
5652 escape is read. Inside a character class, it's always an octal escape,
5653 even if it is a single digit.
5654
5655 (d) An escaped but undefined alphabetic character is taken as a literal,
5656 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5657 escapes.
5658
5659 (e) {0} is now permitted. (The previous item is removed from the compiled
5660 pattern).
5661
5662 5. Changed all the names of code files so that the basic parts are no longer
5663 than 10 characters, and abolished the teeny "globals.c" file.
5664
5665 6. Changed the handling of character classes; they are now done with a 32-byte
5666 bit map always.
5667
5668 7. Added the -d and /D options to pcretest to make it possible to look at the
5669 internals of compilation without having to recompile pcre.
5670
5671
5672 Version 0.95 23-Sep-97
5673 ----------------------
5674
5675 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5676 \x20 at the start of a run of normal characters. These were being treated as
5677 real characters, instead of the source characters being re-checked.
5678
5679
5680 Version 0.94 18-Sep-97
5681 ----------------------
5682
5683 1. The functions are now thread-safe, with the caveat that the global variables
5684 containing pointers to malloc() and free() or alternative functions are the
5685 same for all threads.
5686
5687 2. Get pcre_study() to generate a bitmap of initial characters for non-
5688 anchored patterns when this is possible, and use it if passed to pcre_exec().
5689
5690
5691 Version 0.93 15-Sep-97
5692 ----------------------
5693
5694 1. /(b)|(:+)/ was computing an incorrect first character.
5695
5696 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5697 but not actually doing anything yet.
5698
5699 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5700 as Perl does (e.g. [-az] or [az-]).
5701
5702 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5703 all possible positions.
5704
5705 5. Split up into different modules to avoid including unneeded functions in a
5706 compiled binary. However, compile and exec are still in one module. The "study"
5707 function is split off.
5708
5709 6. The character tables are now in a separate module whose source is generated
5710 by an auxiliary program - but can then be edited by hand if required. There are
5711 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5712 toupper() in the code.
5713
5714 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5715 make them global. Abolish the function for setting them, as the caller can now
5716 set them directly.
5717
5718
5719 Version 0.92 11-Sep-97
5720 ----------------------
5721
5722 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5723 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5724
5725 2. Caseless matching was not working in character classes if the characters in
5726 the pattern were in upper case.
5727
5728 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5729
5730 4. Make PCRE_ANCHORED public and accept as a compile option.
5731
5732 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5733 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5734 pass them.
5735
5736 6. Give an error if bad option bits passed at compile or run time.
5737
5738 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5739 pcretest to cause it to pass that flag.
5740
5741 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5742 options, and the first character, if set.
5743
5744 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5745
5746
5747 Version 0.91 10-Sep-97
5748 ----------------------
5749
5750 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5751 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5752
5753 2. PCRE was looping on encountering an indefinitely repeated back reference to
5754 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5755 Perl does - treats the match as successful.
5756
5757 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5