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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1538 - (show annotations)
Sun Mar 29 11:22:24 2015 UTC (4 years, 6 months ago) by ph10
File size: 275431 byte(s)
Error occurred while calculating annotation data.
Fix comment between subroutine call and quantifier bug.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.37 xx-xxx-2015
5 ------------------------
6
7 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
8 for those parentheses to be closed with whatever has been captured so far.
9 However, it was failing to mark any other groups between the hightest
10 capture so far and the currrent group as "unset". Thus, the ovector for
11 those groups contained whatever was previously there. An example is the
12 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
13
14 2. If an assertion condition was quantified with a minimum of zero (an odd
15 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
16
17 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
18 unrecognized modifier, a crash could occur.
19
20 4. An attempt to do global matching in pcretest with a zero-length ovector
21 caused a crash.
22
23 5. Fixed a memory leak during matching that could occur for a subpattern
24 subroutine call (recursive or otherwise) if the number of captured groups
25 that had to be saved was greater than ten.
26
27 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
28 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
29 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
30
31 7. A UTF pattern containing a "not" match of a non-ASCII character and a
32 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
33
34 8. When a pattern is compiled, it remembers the highest back reference so that
35 when matching, if the ovector is too small, extra memory can be obtained to
36 use instead. A conditional subpattern whose condition is a check on a
37 capture having happened, such as, for example in the pattern
38 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
39 setting the highest backreference number. This mattered only if pcre_exec()
40 was called with an ovector that was too small to hold the capture, and there
41 was no other kind of back reference (a situation which is probably quite
42 rare). The effect of the bug was that the condition was always treated as
43 FALSE when the capture could not be consulted, leading to a incorrect
44 behaviour by pcre2_match(). This bug has been fixed.
45
46 9. A reference to a duplicated named group (either a back reference or a test
47 for being set in a conditional) that occurred in a part of the pattern where
48 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
49 to be incorrectly calculated, leading to overwriting.
50
51 10. A mutually recursive set of back references such as (\2)(\1) caused a
52 segfault at study time (while trying to find the minimum matching length).
53 The infinite loop is now broken (with the minimum length unset, that is,
54 zero).
55
56 11. If an assertion that was used as a condition was quantified with a minimum
57 of zero, matching went wrong. In particular, if the whole group had
58 unlimited repetition and could match an empty string, a segfault was
59 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
60 assertions to be quantified, but not if they are being used as conditions,
61 so the above pattern is faulted by Perl. PCRE has now been changed so that
62 it also rejects such patterns.
63
64 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
65 failed to allow the zero-repeat case if pcre2_exec() was called with an
66 ovector too small to capture the group.
67
68 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
69 Red Hat Product Security:
70
71 (a) A crash if /K and /F were both set with the option to save the compiled
72 pattern.
73
74 (b) Another crash if the option to print captured substrings in a callout
75 was combined with setting a null ovector, for example \O\C+ as a subject
76 string.
77
78 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
79 forward reference repeated a large (but limited) number of times within a
80 repeated outer group that has a zero minimum quantifier, caused incorrect
81 code to be compiled, leading to the error "internal error:
82 previously-checked referenced subpattern not found" when an incorrect
83 memory address was read. This bug was reported as "heap overflow",
84 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
85 CVE-2015-2325.
86
87 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
88 call within a group that also contained a recursive back reference caused
89 incorrect code to be compiled. This bug was reported as "heap overflow",
90 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
91 number CVE-2015-2326.
92
93 24. Computing the size of the JIT read-only data in advance has been a source
94 of various issues, and new ones are still appear unfortunately. To fix
95 existing and future issues, size computation is eliminated from the code,
96 and replaced by on-demand memory allocation.
97
98 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
99 adjacent to the end of the range, and the range contained characters with
100 more than one other case, caused incorrect behaviour when compiled in UTF
101 mode. In that example, the range a-j was left out of the class.
102
103 26. Fix JIT compilation of conditional blocks, which assertion
104 is converted to (*FAIL). E.g: /(?(?!))/.
105
106 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
107 discovered by the LLVM fuzzer.
108
109 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
110 when this assertion was used as a condition, for example (?(?!)a|b). In
111 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
112 error about an unsupported item.
113
114 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
115 possessification code could take exponential time to complete. A recursion
116 depth limit of 1000 has been imposed to limit the resources used by this
117 optimization.
118
119 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
120 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
121 because \S ensures they are all in the class. The code for doing this was
122 interacting badly with the code for computing the amount of space needed to
123 compile the pattern, leading to a buffer overflow. This bug was discovered
124 by the LLVM fuzzer.
125
126 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
127 other kinds of group caused stack overflow at compile time. This bug was
128 discovered by the LLVM fuzzer.
129
130 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
131 between a subroutine call and its quantifier was incorrectly compiled,
132 leading to buffer overflow or other errors. This bug was discovered by the
133 LLVM fuzzer.
134
135
136 Version 8.36 26-September-2014
137 ------------------------------
138
139 1. Got rid of some compiler warnings in the C++ modules that were shown up by
140 -Wmissing-field-initializers and -Wunused-parameter.
141
142 2. The tests for quantifiers being too big (greater than 65535) were being
143 applied after reading the number, and stupidly assuming that integer
144 overflow would give a negative number. The tests are now applied as the
145 numbers are read.
146
147 3. Tidy code in pcre_exec.c where two branches that used to be different are
148 now the same.
149
150 4. The JIT compiler did not generate match limit checks for certain
151 bracketed expressions with quantifiers. This may lead to exponential
152 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
153 issue should be resolved now.
154
155 5. Fixed an issue, which occures when nested alternatives are optimized
156 with table jumps.
157
158 6. Inserted two casts and changed some ints to size_t in the light of some
159 reported 64-bit compiler warnings (Bugzilla 1477).
160
161 7. Fixed a bug concerned with zero-minimum possessive groups that could match
162 an empty string, which sometimes were behaving incorrectly in the
163 interpreter (though correctly in the JIT matcher). This pcretest input is
164 an example:
165
166 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
167 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
168
169 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
170 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
171 for an empty string was breaking the inner loop and carrying on at a lower
172 level, when possessive repeated groups should always return to a higher
173 level as they have no backtrack points in them. The empty string test now
174 occurs at the outer level.
175
176 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
177 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
178
179 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
180 doesn't).
181
182 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
183 to hold a pattern to be incorrectly computed (too small) when there were
184 named back references to duplicated names. This could cause "internal
185 error: code overflow" or "double free or corruption" or other memory
186 handling errors.
187
188 11. When named subpatterns had the same prefixes, back references could be
189 confused. For example, in this pattern:
190
191 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
192
193 the reference to 'Name' was incorrectly treated as a reference to a
194 duplicate name.
195
196 12. A pattern such as /^s?c/mi8 where the optional character has more than
197 one "other case" was incorrectly compiled such that it would only try to
198 match starting at "c".
199
200 13. When a pattern starting with \s was studied, VT was not included in the
201 list of possible starting characters; this should have been part of the
202 8.34/18 patch.
203
204 14. If a character class started [\Qx]... where x is any character, the class
205 was incorrectly terminated at the ].
206
207 15. If a pattern that started with a caseless match for a character with more
208 than one "other case" was studied, PCRE did not set up the starting code
209 unit bit map for the list of possible characters. Now it does. This is an
210 optimization improvement, not a bug fix.
211
212 16. The Unicode data tables have been updated to Unicode 7.0.0.
213
214 17. Fixed a number of memory leaks in pcregrep.
215
216 18. Avoid a compiler warning (from some compilers) for a function call with
217 a cast that removes "const" from an lvalue by using an intermediate
218 variable (to which the compiler does not object).
219
220 19. Incorrect code was compiled if a group that contained an internal recursive
221 back reference was optional (had quantifier with a minimum of zero). This
222 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
223 caused segmentation faults because of stack overflows at compile time.
224
225 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
226 group that is quantified with an indefinite repeat, caused a compile-time
227 loop which used up all the system stack and provoked a segmentation fault.
228 This was not the same bug as 19 above.
229
230 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
231 Patch by Mike Frysinger.
232
233
234 Version 8.35 04-April-2014
235 --------------------------
236
237 1. A new flag is set, when property checks are present in an XCLASS.
238 When this flag is not set, PCRE can perform certain optimizations
239 such as studying these XCLASS-es.
240
241 2. The auto-possessification of character sets were improved: a normal
242 and an extended character set can be compared now. Furthermore
243 the JIT compiler optimizes more character set checks.
244
245 3. Got rid of some compiler warnings for potentially uninitialized variables
246 that show up only when compiled with -O2.
247
248 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
249 of a match later then the end of the match. The pcretest program was not
250 handling the case sensibly - it was outputting from the start to the next
251 binary zero. It now reports this situation in a message, and outputs the
252 text from the end to the start.
253
254 5. Fast forward search is improved in JIT. Instead of the first three
255 characters, any three characters with fixed position can be searched.
256 Search order: first, last, middle.
257
258 6. Improve character range checks in JIT. Characters are read by an inprecise
259 function now, which returns with an unknown value if the character code is
260 above a certain threshold (e.g: 256). The only limitation is that the value
261 must be bigger than the threshold as well. This function is useful when
262 the characters above the threshold are handled in the same way.
263
264 7. The macros whose names start with RAWUCHAR are placeholders for a future
265 mode in which only the bottom 21 bits of 32-bit data items are used. To
266 make this more memorable for those maintaining the code, the names have
267 been changed to start with UCHAR21, and an extensive comment has been added
268 to their definition.
269
270 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
271 to the export list in Makefile.am (they were accidentally omitted from the
272 8.34 tarball).
273
274 9. The informational output from pcretest used the phrase "starting byte set"
275 which is inappropriate for the 16-bit and 32-bit libraries. As the output
276 for "first char" and "need char" really means "non-UTF-char", I've changed
277 "byte" to "char", and slightly reworded the output. The documentation about
278 these values has also been (I hope) clarified.
279
280 10. Another JIT related optimization: use table jumps for selecting the correct
281 backtracking path, when more than four alternatives are present inside a
282 bracket.
283
284 11. Empty match is not possible, when the minimum length is greater than zero,
285 and there is no \K in the pattern. JIT should avoid empty match checks in
286 such cases.
287
288 12. In a caseless character class with UCP support, when a character with more
289 than one alternative case was not the first character of a range, not all
290 the alternative cases were added to the class. For example, s and \x{17f}
291 are both alternative cases for S: the class [RST] was handled correctly,
292 but [R-T] was not.
293
294 13. The configure.ac file always checked for pthread support when JIT was
295 enabled. This is not used in Windows, so I have put this test inside a
296 check for the presence of windows.h (which was already tested for).
297
298 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
299 The algorithm provides a way to skip certain starting offsets, and usually
300 faster than linear prefix searches.
301
302 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
303 as for 'fr_FR' and 'french'. For some reason, however, it then used the
304 Windows-specific input and output files, which have 'french' screwed in.
305 So this could never have worked. One of the problems with locales is that
306 they aren't always the same. I have now updated RunTest so that it checks
307 the output of the locale test (test 3) against three different output
308 files, and it allows the test to pass if any one of them matches. With luck
309 this should make the test pass on some versions of Solaris where it was
310 failing. Because of the uncertainty, the script did not used to stop if
311 test 3 failed; it now does. If further versions of a French locale ever
312 come to light, they can now easily be added.
313
314 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
315 there was a message during ./configure, but it did not stop. This now
316 provokes an error. The invalid example in README has been corrected.
317 If a value less than the minimum is given, the minimum value has always
318 been used, but now a warning is given.
319
320 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
321 was a bug in the test system, which is now fixed. Also, the list of various
322 configurations that are tested for each release did not have one with both
323 16/32 bits and --enable-bar-anycrlf. It now does.
324
325 18. pcretest was missing "-C bsr" for displaying the \R default setting.
326
327 19. Little endian PowerPC systems are supported now by the JIT compiler.
328
329 20. The fast forward newline mechanism could enter to an infinite loop on
330 certain invalid UTF-8 input. Although we don't support these cases
331 this issue can be fixed by a performance optimization.
332
333 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
334 not take account if existing stack usage. There is now a new global
335 variable called pcre_stack_guard that can be set to point to an external
336 function to check stack availability. It is called at the start of
337 processing every parenthesized group.
338
339 22. A typo in the code meant that in ungreedy mode the max/min qualifier
340 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
341 not match "ab".
342
343 23. When UTF was disabled, the JIT program reported some incorrect compile
344 errors. These messages are silenced now.
345
346 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
347 compiler.
348
349 25. Change all the temporary files used in RunGrepTest to be different to those
350 used by RunTest so that the tests can be run simultaneously, for example by
351 "make -j check".
352
353
354 Version 8.34 15-December-2013
355 -----------------------------
356
357 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
358 executable memory. Patch inspired by Carsten Klein.
359
360 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
361 this macro is never tested and has no effect, because the work to support
362 coverage involves only compiling and linking options and special targets in
363 the Makefile. The comment in config.h implied that defining the macro would
364 enable coverage support, which is totally false. There was also support for
365 setting this macro in the CMake files (my fault, I just copied it from
366 configure). SUPPORT_GCOV has now been removed.
367
368 3. Make a small performance improvement in strlen16() and strlen32() in
369 pcretest.
370
371 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
372 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
373 these cases). There was also one in pcretest.c.
374
375 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
376
377 6. In UTF mode, the code for checking whether a group could match an empty
378 string (which is used for indefinitely repeated groups to allow for
379 breaking an infinite loop) was broken when the group contained a repeated
380 negated single-character class with a character that occupied more than one
381 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
382 UTF-8 mode). The effect was undefined: the group might or might not be
383 deemed as matching an empty string, or the program might have crashed.
384
385 7. The code for checking whether a group could match an empty string was not
386 recognizing that \h, \H, \v, \V, and \R must match a character.
387
388 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
389 an empty string. If it can, pcretest shows this in its information output.
390
391 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
392 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
393 matched by pcre_exec() without using JIT:
394
395 (a) If the rest of the pattern did not match after a maximal run of
396 grapheme clusters, the code for backing up to try with fewer of them
397 did not always back up over a full grapheme when characters that do not
398 have the modifier quality were involved, e.g. Hangul syllables.
399
400 (b) If the match point in a subject started with modifier character, and
401 there was no match, the code could incorrectly back up beyond the match
402 point, and potentially beyond the first character in the subject,
403 leading to a segfault or an incorrect match result.
404
405 10. A conditional group with an assertion condition could lead to PCRE
406 recording an incorrect first data item for a match if no other first data
407 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
408 first data item, and therefore matched "ca" after "c" instead of at the
409 start.
410
411 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
412 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
413
414 12. The source of pcregrep now includes z/OS-specific code so that it can be
415 compiled for z/OS as part of the special z/OS distribution.
416
417 13. Added the -T and -TM options to pcretest.
418
419 14. The code in pcre_compile.c for creating the table of named capturing groups
420 has been refactored. Instead of creating the table dynamically during the
421 actual compiling pass, the information is remembered during the pre-compile
422 pass (on the stack unless there are more than 20 named groups, in which
423 case malloc() is used) and the whole table is created before the actual
424 compile happens. This has simplified the code (it is now nearly 150 lines
425 shorter) and prepared the way for better handling of references to groups
426 with duplicate names.
427
428 15. A back reference to a named subpattern when there is more than one of the
429 same name now checks them in the order in which they appear in the pattern.
430 The first one that is set is used for the reference. Previously only the
431 first one was inspected. This change makes PCRE more compatible with Perl.
432
433 16. Unicode character properties were updated from Unicode 6.3.0.
434
435 17. The compile-time code for auto-possessification has been refactored, based
436 on a patch by Zoltan Herczeg. It now happens after instead of during
437 compilation. The code is cleaner, and more cases are handled. The option
438 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
439 options in pcretest are provided to set it. It can also be set by
440 (*NO_AUTO_POSSESS) at the start of a pattern.
441
442 18. The character VT has been added to the default ("C" locale) set of
443 characters that match \s and are generally treated as white space,
444 following this same change in Perl 5.18. There is now no difference between
445 "Perl space" and "POSIX space". Whether VT is treated as white space in
446 other locales depends on the locale.
447
448 19. The code for checking named groups as conditions, either for being set or
449 for being recursed, has been refactored (this is related to 14 and 15
450 above). Processing unduplicated named groups should now be as fast at
451 numerical groups, and processing duplicated groups should be faster than
452 before.
453
454 20. Two patches to the CMake build system, by Alexander Barkov:
455
456 (1) Replace the "source" command by "." in CMakeLists.txt because
457 "source" is a bash-ism.
458
459 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
460 without these the CMake build does not work on Solaris.
461
462 21. Perl has changed its handling of \8 and \9. If there is no previously
463 encountered capturing group of those numbers, they are treated as the
464 literal characters 8 and 9 instead of a binary zero followed by the
465 literals. PCRE now does the same.
466
467 22. Following Perl, added \o{} to specify codepoints in octal, making it
468 possible to specify values greater than 0777 and also making them
469 unambiguous.
470
471 23. Perl now gives an error for missing closing braces after \x{... instead of
472 treating the string as literal. PCRE now does the same.
473
474 24. RunTest used to grumble if an inappropriate test was selected explicitly,
475 but just skip it when running all tests. This make it awkward to run ranges
476 of tests when one of them was inappropriate. Now it just skips any
477 inappropriate tests, as it always did when running all tests.
478
479 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
480 character types such as \d or \w, too many callouts were inserted, and the
481 data that they returned was rubbish.
482
483 26. In UCP mode, \s was not matching two of the characters that Perl matches,
484 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
485 were matched by \h. The code has now been refactored so that the lists of
486 the horizontal and vertical whitespace characters used for \h and \v (which
487 are defined only in one place) are now also used for \s.
488
489 27. Add JIT support for the 64 bit TileGX architecture.
490 Patch by Jiong Wang (Tilera Corporation).
491
492 28. Possessive quantifiers for classes (both explicit and automatically
493 generated) now use special opcodes instead of wrapping in ONCE brackets.
494
495 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
496 (because it's meaningless), this was not happening when PCRE_CASELESS was
497 set. Not wrong, but inefficient.
498
499 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
500 properties for \w, \d, etc) is present in a test regex. Otherwise if the
501 test contains no characters greater than 255, Perl doesn't realise it
502 should be using Unicode semantics.
503
504 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
505 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
506 does in Unicode mode.
507
508 32. Added the "forbid" facility to pcretest so that putting tests into the
509 wrong test files can sometimes be quickly detected.
510
511 33. There is now a limit (default 250) on the depth of nesting of parentheses.
512 This limit is imposed to control the amount of system stack used at compile
513 time. It can be changed at build time by --with-parens-nest-limit=xxx or
514 the equivalent in CMake.
515
516 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
517 errors. Perl warns for these when in warning mode, but PCRE has no facility
518 for giving warnings.
519
520 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
521 However, this was not working for (?!) because it is optimized to (*FAIL),
522 for which PCRE does not allow quantifiers. The optimization is now disabled
523 when a quantifier follows (?!). I can't see any use for this, but it makes
524 things uniform.
525
526 36. Perl no longer allows group names to start with digits, so I have made this
527 change also in PCRE. It simplifies the code a bit.
528
529 37. In extended mode, Perl ignores spaces before a + that indicates a
530 possessive quantifier. PCRE allowed a space before the quantifier, but not
531 before the possessive +. It now does.
532
533 38. The use of \K (reset reported match start) within a repeated possessive
534 group such as (a\Kb)*+ was not working.
535
536 40. Document that the same character tables must be used at compile time and
537 run time, and that the facility to pass tables to pcre_exec() and
538 pcre_dfa_exec() is for use only with saved/restored patterns.
539
540 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
541 features for Builds with MSVC:
542
543 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
544 builds). The .rc files can be used to set FileDescription and many other
545 attributes.
546
547 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
548 This allows higher-level build scripts which want .pdb files to avoid
549 hard-coding the exact files needed."
550
551 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
552 mean "start of word" and "end of word", respectively, as a transition aid.
553
554 43. A minimizing repeat of a class containing codepoints greater than 255 in
555 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
556 compiled to use the heap for recursion.
557
558 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
559 is configured.
560
561
562 Version 8.33 28-May-2013
563 ------------------------
564
565 1. Added 'U' to some constants that are compared to unsigned integers, to
566 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
567 variables that are added to signed variables, to ensure the result is
568 signed and can be negated.
569
570 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
571 CMake config files.
572
573 3. Revise the creation of config.h.generic so that all boolean macros are
574 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
575 overriding via -D on the command line possible.
576
577 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
578 to unsigned int is reported to make a quite noticeable speed difference in
579 a specific Windows environment. Testing on Linux did also appear to show
580 some benefit (and it is clearly not harmful). Also fixed the definition of
581 Xop which should be unsigned.
582
583 5. Related to (4), changing the definition of the intermediate variable cc
584 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
585 improvements.
586
587 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
588 unnecessary spaces.
589
590 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
591 and later.
592
593 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
594
595 9. Optimizing fast_forward_start_bits in JIT.
596
597 10. Adding support for callouts in JIT, and fixing some issues revealed
598 during this work. Namely:
599
600 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
601
602 (b) Minimum length was not checked before the matching is started.
603
604 11. The value of capture_last that is passed to callouts was incorrect in some
605 cases when there was a capture on one path that was subsequently abandoned
606 after a backtrack. Also, the capture_last value is now reset after a
607 recursion, since all captures are also reset in this case.
608
609 12. The interpreter no longer returns the "too many substrings" error in the
610 case when an overflowing capture is in a branch that is subsequently
611 abandoned after a backtrack.
612
613 13. In the pathological case when an offset vector of size 2 is used, pcretest
614 now prints out the matched string after a yield of 0 or 1.
615
616 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
617 Only supported by the JIT compiler at the moment.
618
619 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
620
621 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
622 offset of the starting point of the matching process, provided the offsets
623 vector is large enough.
624
625 17. The \A escape now records a lookbehind value of 1, though its execution
626 does not actually inspect the previous character. This is to ensure that,
627 in partial multi-segment matching, at least one character from the old
628 segment is retained when a new segment is processed. Otherwise, if there
629 are no lookbehinds in the pattern, \A might match incorrectly at the start
630 of a new segment.
631
632 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
633
634 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
635 gives some modest performance improvement in 8-bit mode.
636
637 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
638 be expressed in certain programming languages using Universal Character
639 Names.
640
641 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
642 which points out that "non characters" are not "characters that may not
643 appear in Unicode strings" but rather "characters that are reserved for
644 internal use and have only local meaning".
645
646 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
647 there was a conditional group that depended on an assertion, if the
648 assertion was false, the callout that immediately followed the alternation
649 in the condition was skipped when pcre_exec() was used for matching.
650
651 23. Allow an explicit callout to be inserted before an assertion that is the
652 condition for a conditional group, for compatibility with automatic
653 callouts, which always insert a callout at this point.
654
655 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
656 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
657
658 25. (*PRUNE) is now supported by the JIT compiler.
659
660 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
661
662 27. Fix the case where there are two or more SKIPs with arguments that may be
663 ignored.
664
665 28. (*SKIP) is now supported by the JIT compiler.
666
667 29. (*THEN) is now supported by the JIT compiler.
668
669 30. Update RunTest with additional test selector options.
670
671 31. The way PCRE handles backtracking verbs has been changed in two ways.
672
673 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
674 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
675 backtracking. In some cases this makes it more Perl-compatible, but Perl's
676 rather obscure rules do not always do the same thing.
677
678 (2) Previously, backtracking verbs were confined within assertions. This is
679 no longer the case for positive assertions, except for (*ACCEPT). Again,
680 this sometimes improves Perl compatibility, and sometimes does not.
681
682 32. A number of tests that were in test 2 because Perl did things differently
683 have been moved to test 1, because either Perl or PCRE has changed, and
684 these tests are now compatible.
685
686 32. Backtracking control verbs are now handled in the same way in JIT and
687 interpreter.
688
689 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
690 contained a forward subroutine reference caused a compile error.
691
692 34. Auto-detect and optimize limited repetitions in JIT.
693
694 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
695 blocking (*UTF) etc.
696
697 36. In the interpreter, maximizing pattern repetitions for characters and
698 character types now use tail recursion, which reduces stack usage.
699
700 37. The value of the max lookbehind was not correctly preserved if a compiled
701 and saved regex was reloaded on a host of different endianness.
702
703 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
704 of the compiled pattern block, expand the flags field from 16 to 32 bits
705 because it was almost full.
706
707 39. Try madvise first before posix_madvise.
708
709 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
710 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
711 original change locked out all matches of empty strings. This has been
712 changed so that one match of an empty string per line is recognized.
713 Subsequent searches on the same line (for colouring or for --only-matching,
714 for example) do not recognize empty strings.
715
716 41. Applied a user patch to fix a number of spelling mistakes in comments.
717
718 42. Data lines longer than 65536 caused pcretest to crash.
719
720 43. Clarified the data type for length and startoffset arguments for pcre_exec
721 and pcre_dfa_exec in the function-specific man pages, where they were
722 explicitly stated to be in bytes, never having been updated. I also added
723 some clarification to the pcreapi man page.
724
725 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
726 a segmentation fault.
727
728
729 Version 8.32 30-November-2012
730 -----------------------------
731
732 1. Improved JIT compiler optimizations for first character search and single
733 character iterators.
734
735 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
736 Patch by Daniel Richard G.
737
738 3. Single character iterator optimizations in the JIT compiler.
739
740 4. Improved JIT compiler optimizations for character ranges.
741
742 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
743 Reported by Giuseppe D'Angelo.
744
745 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
746 of a line, was being set incorrectly in cases where .* appeared inside
747 atomic brackets at the start of a pattern, or where there was a subsequent
748 *PRUNE or *SKIP.
749
750 7. Improved instruction cache flush for POWER/PowerPC.
751 Patch by Daniel Richard G.
752
753 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
754 grep:
755
756 (a) There is now no limit to the number of patterns to be matched.
757
758 (b) An error is given if a pattern is too long.
759
760 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
761 are now supported.
762
763 (d) --exclude-from and --include-from (multiple use) have been added.
764
765 (e) Exclusions and inclusions now apply to all files and directories, not
766 just to those obtained from scanning a directory recursively.
767
768 (f) Multiple uses of -f and --file-list are now supported.
769
770 (g) In a Windows environment, the default for -d has been changed from
771 "read" (the GNU grep default) to "skip", because otherwise the presence
772 of a directory in the file list provokes an error.
773
774 (h) The documentation has been revised and clarified in places.
775
776 9. Improve the matching speed of capturing brackets.
777
778 10. Changed the meaning of \X so that it now matches a Unicode extended
779 grapheme cluster.
780
781 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
782 out POSIX threads when JIT support is configured.
783
784 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
785
786 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
787 structure could go wrong in environments where size_t is not the same size
788 as int.
789
790 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
791 set.
792
793 15. The EBCDIC support had decayed; later updates to the code had included
794 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
795 general tidy up of EBCDIC-related issues, and the documentation was also
796 not quite right. There is now a test that can be run on ASCII systems to
797 check some of the EBCDIC-related things (but is it not a full test).
798
799 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
800 in a small tidy to the code.
801
802 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
803
804 18. If the --only-matching (-o) option in pcregrep is specified multiple
805 times, each one causes appropriate output. For example, -o1 -o2 outputs the
806 substrings matched by the 1st and 2nd capturing parentheses. A separating
807 string can be specified by --om-separator (default empty).
808
809 19. Improving the first n character searches.
810
811 20. Turn case lists for horizontal and vertical white space into macros so that
812 they are defined only once.
813
814 21. This set of changes together give more compatible Unicode case-folding
815 behaviour for characters that have more than one other case when UCP
816 support is available.
817
818 (a) The Unicode property table now has offsets into a new table of sets of
819 three or more characters that are case-equivalent. The MultiStage2.py
820 script that generates these tables (the pcre_ucd.c file) now scans
821 CaseFolding.txt instead of UnicodeData.txt for character case
822 information.
823
824 (b) The code for adding characters or ranges of characters to a character
825 class has been abstracted into a generalized function that also handles
826 case-independence. In UTF-mode with UCP support, this uses the new data
827 to handle characters with more than one other case.
828
829 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
830 whose other case is greater than 256 are now correctly matched
831 caselessly. Previously, the high codepoint matched the low one, but not
832 vice versa.
833
834 (d) The processing of \h, \H, \v, and \ in character classes now makes use
835 of the new class addition function, using character lists defined as
836 macros alongside the case definitions of 20 above.
837
838 (e) Caseless back references now work with characters that have more than
839 one other case.
840
841 (f) General caseless matching of characters with more than one other case
842 is supported.
843
844 22. Unicode character properties were updated from Unicode 6.2.0
845
846 23. Improved CMake support under Windows. Patch by Daniel Richard G.
847
848 24. Add support for 32-bit character strings, and UTF-32
849
850 25. Major JIT compiler update (code refactoring and bugfixing).
851 Experimental Sparc 32 support is added.
852
853 26. Applied a modified version of Daniel Richard G's patch to create
854 pcre.h.generic and config.h.generic by "make" instead of in the
855 PrepareRelease script.
856
857 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
858 pcre_compile.c when checking for a zero character.
859
860 28. Introducing a native interface for JIT. Through this interface, the compiled
861 machine code can be directly executed. The purpose of this interface is to
862 provide fast pattern matching, so several sanity checks are not performed.
863 However, feature tests are still performed. The new interface provides
864 1.4x speedup compared to the old one.
865
866 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
867 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
868 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
869
870 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
871 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
872 the "old" RFC 2279). Instead, it ended up passing a negative length to
873 pcre_exec().
874
875 31. Add support for GCC's visibility feature to hide internal functions.
876
877 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
878 "unknown -C option" after outputting 0 or 1.
879
880 33. There is now support for generating a code coverage report for the test
881 suite in environments where gcc is the compiler and lcov is installed. This
882 is mainly for the benefit of the developers.
883
884 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
885 unaddressable using valgrind annotations, allowing valgrind to detect
886 invalid memory accesses. This is mainly for the benefit of the developers.
887
888 25. (*UTF) can now be used to start a pattern in any of the three libraries.
889
890 26. Give configure error if --enable-cpp but no C++ compiler found.
891
892
893 Version 8.31 06-July-2012
894 -------------------------
895
896 1. Fixing a wrong JIT test case and some compiler warnings.
897
898 2. Removed a bashism from the RunTest script.
899
900 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
901 to unsigned type, result still unsigned" that was given by an MS compiler
902 on encountering the code "-sizeof(xxx)".
903
904 4. Partial matching support is added to the JIT compiler.
905
906 5. Fixed several bugs concerned with partial matching of items that consist
907 of more than one character:
908
909 (a) /^(..)\1/ did not partially match "aba" because checking references was
910 done on an "all or nothing" basis. This also applied to repeated
911 references.
912
913 (b) \R did not give a hard partial match if \r was found at the end of the
914 subject.
915
916 (c) \X did not give a hard partial match after matching one or more
917 characters at the end of the subject.
918
919 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
920 a partial match for the string "\r".
921
922 (e) When newline was set to CRLF, the metacharacter "." did not recognize
923 a partial match for a CR character at the end of the subject string.
924
925 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
926 running pcretest, the text "(JIT)" added to the output whenever JIT is
927 actually used to run the match.
928
929 7. Individual JIT compile options can be set in pcretest by following -s+[+]
930 or /S+[+] with a digit between 1 and 7.
931
932 8. OP_NOT now supports any UTF character not just single-byte ones.
933
934 9. (*MARK) control verb is now supported by the JIT compiler.
935
936 10. The command "./RunTest list" lists the available tests without actually
937 running any of them. (Because I keep forgetting what they all are.)
938
939 11. Add PCRE_INFO_MAXLOOKBEHIND.
940
941 12. Applied a (slightly modified) user-supplied patch that improves performance
942 when the heap is used for recursion (compiled with --disable-stack-for-
943 recursion). Instead of malloc and free for each heap frame each time a
944 logical recursion happens, frames are retained on a chain and re-used where
945 possible. This sometimes gives as much as 30% improvement.
946
947 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
948 call.
949
950 14. As documented, (*COMMIT) is now confined to within a positive assertion.
951
952 15. It is now possible to link pcretest with libedit as an alternative to
953 libreadline.
954
955 16. (*COMMIT) control verb is now supported by the JIT compiler.
956
957 17. The Unicode data tables have been updated to Unicode 6.1.0.
958
959 18. Added --file-list option to pcregrep.
960
961 19. Added binary file support to pcregrep, including the -a, --binary-files,
962 -I, and --text options.
963
964 20. The madvise function is renamed for posix_madvise for QNX compatibility
965 reasons. Fixed by Giuseppe D'Angelo.
966
967 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
968
969 22. Changed the option for creating symbolic links for 16-bit man pages from
970 -s to -sf so that re-installing does not cause issues.
971
972 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
973
974 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
975 matches in certain environments (the workspace was not being correctly
976 retained). Also added to pcre_dfa_exec() a simple plausibility check on
977 some of the workspace data at the beginning of a restart.
978
979 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
980 was not doing so when it should - probably a typo introduced by SVN 528
981 (change 8.10/14).
982
983 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
984 \w+ when the character tables indicated that \x{c4} was a word character.
985 There were several related cases, all because the tests for doing a table
986 lookup were testing for characters less than 127 instead of 255.
987
988 27. If a pattern contains capturing parentheses that are not used in a match,
989 their slots in the ovector are set to -1. For those that are higher than
990 any matched groups, this happens at the end of processing. In the case when
991 there were back references that the ovector was too small to contain
992 (causing temporary malloc'd memory to be used during matching), and the
993 highest capturing number was not used, memory off the end of the ovector
994 was incorrectly being set to -1. (It was using the size of the temporary
995 memory instead of the true size.)
996
997 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
998 ovector size, it uses memory at the end of the block that it has got.
999
1000 29. Check for an overlong MARK name and give an error at compile time. The
1001 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1002
1003 30. JIT compiler update.
1004
1005 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1006 Rill for the patch.
1007
1008 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1009
1010 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1011
1012 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1013 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1014 was enabled.
1015
1016 35. Improve JIT code generation for greedy plus quantifier.
1017
1018 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1019 1 to "aa" instead of to an empty string. The bug affected repeated groups
1020 that could potentially match an empty string.
1021
1022 37. Optimizing single character iterators in JIT.
1023
1024 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1025 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1026 codepoints that are too big for the mode are faulted, and in a UTF mode,
1027 disallowed codepoints are also faulted.
1028
1029 39. If PCRE was compiled with UTF support, in three places in the DFA
1030 matcher there was code that should only have been obeyed in UTF mode, but
1031 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1032 processing when bytes with values greater than 127 were present. In 16-bit
1033 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1034 both cases the values are those that cannot be the first data item in a UTF
1035 character. The three items that might have provoked this were recursions,
1036 possessively repeated groups, and atomic groups.
1037
1038 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1039 and pcregrep, because some OS require shared objects to be explicitly
1040 passed to ld, causing the link step to fail if they are not.
1041
1042 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1043 mode, patterns that started with \h* or \R* might be incorrectly matched.
1044
1045
1046 Version 8.30 04-February-2012
1047 -----------------------------
1048
1049 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1050 name is defined in ctype.h.
1051
1052 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1053 only in quite long subpatterns.
1054
1055 3. Removed the function pcre_info(), which has been obsolete and deprecated
1056 since it was replaced by pcre_fullinfo() in February 2000.
1057
1058 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1059 match a (*MARK), and the match failed at the start of the subject, a
1060 reference to memory before the start of the subject could occur. This bug
1061 was introduced by fix 17 of release 8.21.
1062
1063 5. A reference to an unset group with zero minimum repetition was giving
1064 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1065 /(another)?(\1?)test/ matched against "hello world test". This bug was
1066 introduced in release 8.13.
1067
1068 6. Add support for 16-bit character strings (a large amount of work involving
1069 many changes and refactorings).
1070
1071 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1072 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1073 from a file.
1074
1075 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1076 rounding is not applied in this particular case).
1077
1078 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1079 if they appear, or are escaped, in patterns.
1080
1081 10. Get rid of a number of -Wunused-but-set-variable warnings.
1082
1083 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1084 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1085 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1086 also returns the mark "x". This bug applied to capturing parentheses,
1087 non-capturing parentheses, and atomic parentheses. It also applied to some
1088 assertions.
1089
1090 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1091 information out of configure.ac instead of relying on pcre.h.generic, which
1092 is not stored in the repository.
1093
1094 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1095 -lreadline.
1096
1097 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1098
1099 15. Applied Graycode's patch to put the top-level frame on the stack rather
1100 than the heap when not using the stack for recursion. This gives a
1101 performance improvement in many cases when recursion is not deep.
1102
1103 16. Experimental code added to "pcretest -C" to output the stack frame size.
1104
1105
1106 Version 8.21 12-Dec-2011
1107 ------------------------
1108
1109 1. Updating the JIT compiler.
1110
1111 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1112 are added as well.
1113
1114 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1115 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1116 calling _pcre_jit_exec. Some extra comments are added.
1117
1118 4. (*MARK) settings inside atomic groups that do not contain any capturing
1119 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1120 was introduced by change 18 for 8.20.
1121
1122 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1123 ECMA-262 standard.
1124
1125 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1126 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1127 This bug was probably introduced by change 9 of 8.13.
1128
1129 7. While fixing 6 above, I noticed that a number of other items were being
1130 incorrectly rejected as "not fixed length". This arose partly because newer
1131 opcodes had not been added to the fixed-length checking code. I have (a)
1132 corrected the bug and added tests for these items, and (b) arranged for an
1133 error to occur if an unknown opcode is encountered while checking for fixed
1134 length instead of just assuming "not fixed length". The items that were
1135 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1136 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1137 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1138
1139 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1140 being incorrectly compiled and would have given unpredicatble results.
1141
1142 9. A possessively repeated subpattern with minimum repeat count greater than
1143 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1144 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1145 the first (A) could occur when it should not.
1146
1147 10. Add a cast and remove a redundant test from the code.
1148
1149 11. JIT should use pcre_malloc/pcre_free for allocation.
1150
1151 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1152 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1153 is anything other than /usr, -L is still shown).
1154
1155 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1156
1157 14. Perl does not support \N without a following name in a [] class; PCRE now
1158 also gives an error.
1159
1160 15. If a forward reference was repeated with an upper limit of around 2000,
1161 it caused the error "internal error: overran compiling workspace". The
1162 maximum number of forward references (including repeats) was limited by the
1163 internal workspace, and dependent on the LINK_SIZE. The code has been
1164 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1165 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1166 of around 200,000 forward references. While doing this, I also speeded up
1167 the filling in of repeated forward references.
1168
1169 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1170 incorrectly expecting the subject to contain another "a" after the start.
1171
1172 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1173 in the match, the SKIP should be ignored. This was not happening; instead
1174 the SKIP was being treated as NOMATCH. For patterns such as
1175 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1176 tested.
1177
1178 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1179 now much more compatible with Perl, in particular in cases where the result
1180 is a non-match for a non-anchored pattern. For example, if
1181 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1182 "m", where previously it did not return a name. A side effect of this
1183 change is that for partial matches, the last encountered mark name is
1184 returned, as for non matches. A number of tests that were previously not
1185 Perl-compatible have been moved into the Perl-compatible test files. The
1186 refactoring has had the pleasing side effect of removing one argument from
1187 the match() function, thus reducing its stack requirements.
1188
1189 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1190 subsequent uses of /S (without +) incorrectly behaved like /S+.
1191
1192 21. Retrieve executable code size support for the JIT compiler and fixing
1193 some warnings.
1194
1195 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1196 not work when the shorter character appeared right at the end of the
1197 subject string.
1198
1199 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1200 systems.
1201
1202 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1203 output it when the /M option is used in pcretest.
1204
1205 25. The CheckMan script was not being included in the distribution. Also, added
1206 an explicit "perl" to run Perl scripts from the PrepareRelease script
1207 because this is reportedly needed in Windows.
1208
1209 26. If study data was being save in a file and studying had not found a set of
1210 "starts with" bytes for the pattern, the data written to the file (though
1211 never used) was taken from uninitialized memory and so caused valgrind to
1212 complain.
1213
1214 27. Updated RunTest.bat as provided by Sheri Pierce.
1215
1216 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1217
1218 29. Computation of memory usage for the table of capturing group names was
1219 giving an unnecessarily large value.
1220
1221
1222 Version 8.20 21-Oct-2011
1223 ------------------------
1224
1225 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1226 a POSIX class. After further experiments with Perl, which convinced me that
1227 Perl has bugs and confusions, a closing square bracket is no longer allowed
1228 in a POSIX name. This bug also affected patterns with classes that started
1229 with full stops.
1230
1231 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1232 captured substring, but while checking the failing first alternative,
1233 substring 1 is temporarily captured. If the output vector supplied to
1234 pcre_exec() was not big enough for this capture, the yield of the function
1235 was still zero ("insufficient space for captured substrings"). This cannot
1236 be totally fixed without adding another stack variable, which seems a lot
1237 of expense for a edge case. However, I have improved the situation in cases
1238 such as /(a)(b)x|abc/ matched against "abc", where the return code
1239 indicates that fewer than the maximum number of slots in the ovector have
1240 been set.
1241
1242 3. Related to (2) above: when there are more back references in a pattern than
1243 slots in the output vector, pcre_exec() uses temporary memory during
1244 matching, and copies in the captures as far as possible afterwards. It was
1245 using the entire output vector, but this conflicts with the specification
1246 that only 2/3 is used for passing back captured substrings. Now it uses
1247 only the first 2/3, for compatibility. This is, of course, another edge
1248 case.
1249
1250 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1251 main code base, and can be used by building with --enable-jit. When this is
1252 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1253 runtime --no-jit option is given.
1254
1255 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1256 ovector, the return from the function was zero, implying that there were
1257 other matches that did not fit. The correct "exactly full" value is now
1258 returned.
1259
1260 6. If a subpattern that was called recursively or as a subroutine contained
1261 (*PRUNE) or any other control that caused it to give a non-standard return,
1262 invalid errors such as "Error -26 (nested recursion at the same subject
1263 position)" or even infinite loops could occur.
1264
1265 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1266 computing the minimum length on reaching *ACCEPT, and so ended up with the
1267 wrong value of 1 rather than 0. Further investigation indicates that
1268 computing a minimum subject length in the presence of *ACCEPT is difficult
1269 (think back references, subroutine calls), and so I have changed the code
1270 so that no minimum is registered for a pattern that contains *ACCEPT.
1271
1272 8. If (*THEN) was present in the first (true) branch of a conditional group,
1273 it was not handled as intended. [But see 16 below.]
1274
1275 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1276 Sheri Pierce.
1277
1278 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1279 the first byte in a match must be "a".
1280
1281 11. Change 17 for 8.13 increased the recursion depth for patterns like
1282 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1283 pattern contains any instances of (*THEN). If it does not, the old
1284 optimizations are restored. It would be nice to do this on a per-group
1285 basis, but at the moment that is not feasible.
1286
1287 12. In some environments, the output of pcretest -C is CRLF terminated. This
1288 broke RunTest's code that checks for the link size. A single white space
1289 character after the value is now allowed for.
1290
1291 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1292 For "fr", it uses the Windows-specific input and output files.
1293
1294 14. If (*THEN) appeared in a group that was called recursively or as a
1295 subroutine, it did not work as intended. [But see next item.]
1296
1297 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1298 pattern fragments (but not containing any | characters). If A and B are
1299 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1300 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1301 D. In other words, Perl considers parentheses that do not contain any |
1302 characters to be part of a surrounding alternative, whereas PCRE was
1303 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1304 differently. PCRE now behaves in the same way as Perl, except in the case
1305 of subroutine/recursion calls such as (?1) which have in any case always
1306 been different (but PCRE had them first :-).
1307
1308 16. Related to 15 above: Perl does not treat the | in a conditional group as
1309 creating alternatives. Such a group is treated in the same way as an
1310 ordinary group without any | characters when processing (*THEN). PCRE has
1311 been changed to match Perl's behaviour.
1312
1313 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1314 RunGrepTest script failed.
1315
1316 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1317 inevitable for groups that contain captures, but it can lead to a lot of
1318 stack use in large patterns. The old behaviour has been restored for atomic
1319 groups that do not contain any capturing parentheses.
1320
1321 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1322 suppress the check for a minimum subject length at run time. (If it was
1323 given to pcre_exec() or pcre_dfa_exec() it did work.)
1324
1325 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1326 fail to work when decoding hex characters in data strings in EBCDIC
1327 environments.
1328
1329 21. It appears that in at least one Mac OS environment, the isxdigit() function
1330 is implemented as a macro that evaluates to its argument more than once,
1331 contravening the C 90 Standard (I haven't checked a later standard). There
1332 was an instance in pcretest which caused it to go wrong when processing
1333 \x{...} escapes in subject strings. The has been rewritten to avoid using
1334 things like p++ in the argument of isxdigit().
1335
1336
1337 Version 8.13 16-Aug-2011
1338 ------------------------
1339
1340 1. The Unicode data tables have been updated to Unicode 6.0.0.
1341
1342 2. Two minor typos in pcre_internal.h have been fixed.
1343
1344 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1345 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1346 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1347
1348 4. There were a number of related bugs in the code for matching backrefences
1349 caselessly in UTF-8 mode when codes for the characters concerned were
1350 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1351 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1352 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1353 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1354 2-byte code at the end of the subject (it thought there wasn't enough data
1355 left).
1356
1357 5. Comprehensive information about what went wrong is now returned by
1358 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1359 as the output vector has at least 2 elements. The offset of the start of
1360 the failing character and a reason code are placed in the vector.
1361
1362 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1363 now returned is for the first byte of the failing character, instead of the
1364 last byte inspected. This is an incompatible change, but I hope it is small
1365 enough not to be a problem. It makes the returned offset consistent with
1366 pcre_exec() and pcre_dfa_exec().
1367
1368 7. pcretest now gives a text phrase as well as the error number when
1369 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1370 failure, the offset and reason code are output.
1371
1372 8. When \R was used with a maximizing quantifier it failed to skip backwards
1373 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1374 back over a single character (\n). This seems wrong (because it treated the
1375 two characters as a single entity when going forwards), conflicts with the
1376 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1377 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1378 has been changed.
1379
1380 9. Some internal refactoring has changed the processing so that the handling
1381 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1382 time (the PCRE_DOTALL option was changed this way some time ago: version
1383 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1384 which was always a bit of a fudge. It also means that there is one less
1385 argument for the match() function, which reduces its stack requirements
1386 slightly. This change also fixes an incompatibility with Perl: the pattern
1387 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1388
1389 10. More internal refactoring has drastically reduced the number of recursive
1390 calls to match() for possessively repeated groups such as (abc)++ when
1391 using pcre_exec().
1392
1393 11. While implementing 10, a number of bugs in the handling of groups were
1394 discovered and fixed:
1395
1396 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1397 (a|)*(?1) gave a compile-time internal error.
1398 ((a|)+)+ did not notice that the outer group could match an empty string.
1399 (^a|^)+ was not marked as anchored.
1400 (.*a|.*)+ was not marked as matching at start or after a newline.
1401
1402 12. Yet more internal refactoring has removed another argument from the match()
1403 function. Special calls to this function are now indicated by setting a
1404 value in a variable in the "match data" data block.
1405
1406 13. Be more explicit in pcre_study() instead of relying on "default" for
1407 opcodes that mean there is no starting character; this means that when new
1408 ones are added and accidentally left out of pcre_study(), testing should
1409 pick them up.
1410
1411 14. The -s option of pcretest has been documented for ages as being an old
1412 synonym of -m (show memory usage). I have changed it to mean "force study
1413 for every regex", that is, assume /S for every regex. This is similar to -i
1414 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1415 using it. It makes it easier to run collections of tests with and without
1416 study enabled, and thereby test pcre_study() more easily. All the standard
1417 tests are now run with and without -s (but some patterns can be marked as
1418 "never study" - see 20 below).
1419
1420 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1421 restoration of the capturing data to the outer values was not happening
1422 correctly.
1423
1424 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1425 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1426 pattern had matched an empty string, and so incorrectly returned a no
1427 match.
1428
1429 17. There was optimizing code for the last branch of non-capturing parentheses,
1430 and also for the obeyed branch of a conditional subexpression, which used
1431 tail recursion to cut down on stack usage. Unfortunately, now that there is
1432 the possibility of (*THEN) occurring in these branches, tail recursion is
1433 no longer possible because the return has to be checked for (*THEN). These
1434 two optimizations have therefore been removed. [But see 8.20/11 above.]
1435
1436 18. If a pattern containing \R was studied, it was assumed that \R always
1437 matched two bytes, thus causing the minimum subject length to be
1438 incorrectly computed because \R can also match just one byte.
1439
1440 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1441 was incorrectly computed.
1442
1443 20. If /S is present twice on a test pattern in pcretest input, it now
1444 *disables* studying, thereby overriding the use of -s on the command line
1445 (see 14 above). This is necessary for one or two tests to keep the output
1446 identical in both cases.
1447
1448 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1449 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1450
1451 22. When an atomic group that contained a capturing parenthesis was
1452 successfully matched, but the branch in which it appeared failed, the
1453 capturing was not being forgotten if a higher numbered group was later
1454 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1455 group 1 to "a", when in fact it should be unset. This applied to multi-
1456 branched capturing and non-capturing groups, repeated or not, and also to
1457 positive assertions (capturing in negative assertions does not happen
1458 in PCRE) and also to nested atomic groups.
1459
1460 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1461 subject after a captured substring, to make it easier to tell which of a
1462 number of identical substrings has been captured.
1463
1464 24. The way atomic groups are processed by pcre_exec() has been changed so that
1465 if they are repeated, backtracking one repetition now resets captured
1466 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1467 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1468 "aaa". Previously, it would have been "a". As part of this code
1469 refactoring, the way recursive calls are handled has also been changed.
1470
1471 25. If an assertion condition captured any substrings, they were not passed
1472 back unless some other capturing happened later. For example, if
1473 (?(?=(a))a) was matched against "a", no capturing was returned.
1474
1475 26. When studying a pattern that contained subroutine calls or assertions,
1476 the code for finding the minimum length of a possible match was handling
1477 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1478 group 1 called group 2 while simultaneously a separate group 2 called group
1479 1). A stack overflow occurred in this case. I have fixed this by limiting
1480 the recursion depth to 10.
1481
1482 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1483 Fortmann. This supports explicit test numbers on the command line, and has
1484 argument validation and error reporting.
1485
1486 28. An instance of \X with an unlimited repeat could fail if at any point the
1487 first character it looked at was a mark character.
1488
1489 29. Some minor code refactoring concerning Unicode properties and scripts
1490 should reduce the stack requirement of match() slightly.
1491
1492 30. Added the '=' option to pcretest to check the setting of unused capturing
1493 slots at the end of the pattern, which are documented as being -1, but are
1494 not included in the return count.
1495
1496 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1497 compiled something random. Now it gives a compile-time error (as does
1498 Perl).
1499
1500 32. A *MARK encountered during the processing of a positive assertion is now
1501 recorded and passed back (compatible with Perl).
1502
1503 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1504 had alternative anchored branches, the search for a second match in a line
1505 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1506 matched the line "0102" twice. The same bug affected patterns that started
1507 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1508 twice.
1509
1510 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1511 does, and because of capturing effects, quantifying parenthesized
1512 assertions may at times be useful. Quantifiers are now allowed for
1513 parenthesized assertions.
1514
1515 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1516
1517 36. \g was being checked for fancy things in a character class, when it should
1518 just be a literal "g".
1519
1520 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1521 appearance of a nested POSIX class supersedes an apparent external class.
1522 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1523 unescaped square brackets may also appear as part of class names. For
1524 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1525 more like Perl. (But see 8.20/1 above.)
1526
1527 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1528 was because it thought it was \N{name}, which is not supported).
1529
1530 39. Add minix to OS list not supporting the -S option in pcretest.
1531
1532 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1533 cannot analyze patterns in sufficient detail to catch mutual recursions
1534 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1535 subgroup is called recursively as a subpattern for a second time at the
1536 same position in the subject string. In previous releases this might have
1537 been caught by the recursion limit, or it might have run out of stack.
1538
1539 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1540 happen only once. PCRE was, however incorrectly giving a compile time error
1541 "recursive call could loop indefinitely" because it cannot analyze the
1542 pattern in sufficient detail. The compile time test no longer happens when
1543 PCRE is compiling a conditional subpattern, but actual runaway loops are
1544 now caught at runtime (see 40 above).
1545
1546 42. It seems that Perl allows any characters other than a closing parenthesis
1547 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1548 has been changed to be the same.
1549
1550 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1551 as not to get warnings when autogen.sh is called. Also changed
1552 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1553
1554 44. To help people who use pcregrep to scan files containing exceedingly long
1555 lines, the following changes have been made:
1556
1557 (a) The default value of the buffer size parameter has been increased from
1558 8K to 20K. (The actual buffer used is three times this size.)
1559
1560 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1561 PCRE is built.
1562
1563 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1564 to be set at run time.
1565
1566 (d) Numerical values in pcregrep options can be followed by K or M, for
1567 example --buffer-size=50K.
1568
1569 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1570 given and the return code is set to 2.
1571
1572 45. Add a pointer to the latest mark to the callout data block.
1573
1574 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1575 partial match of an empty string instead of no match. This was specific to
1576 the use of ".".
1577
1578 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1579 complete match instead of a partial match. This bug was dependent on both
1580 the PCRE_UTF8 and PCRE_DOTALL options being set.
1581
1582 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1583 starting byte set, because \b was not being ignored.
1584
1585
1586 Version 8.12 15-Jan-2011
1587 ------------------------
1588
1589 1. Fixed some typos in the markup of the man pages, and wrote a script that
1590 checks for such things as part of the documentation building process.
1591
1592 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1593 --match-limit and --recursion-limit options (added for 8.11). In
1594 particular, this made one of the standard tests fail. (The integer value
1595 went into the wrong half of a long int.)
1596
1597 3. If the --colour option was given to pcregrep with -v (invert match), it
1598 did strange things, either producing crazy output, or crashing. It should,
1599 of course, ignore a request for colour when reporting lines that do not
1600 match.
1601
1602 4. Another pcregrep bug caused similar problems if --colour was specified with
1603 -M (multiline) and the pattern match finished with a line ending.
1604
1605 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1606 matched in multiline mode, the following line was shown as part of the
1607 match. This seems wrong, so I have changed it.
1608
1609 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1610 the check for further matches in the same line (so they could be coloured)
1611 to overrun the end of the current line. If another match was found, it was
1612 incorrectly shown (and then shown again when found in the next line).
1613
1614 7. If pcregrep was compiled under Windows, there was a reference to the
1615 function pcregrep_exit() before it was defined. I am assuming this was
1616 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1617 reported by a user. I've moved the definition above the reference.
1618
1619
1620 Version 8.11 10-Dec-2010
1621 ------------------------
1622
1623 1. (*THEN) was not working properly if there were untried alternatives prior
1624 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1625 backtracked to try for "b" instead of moving to the next alternative branch
1626 at the same level (in this case, to look for "c"). The Perl documentation
1627 is clear that when (*THEN) is backtracked onto, it goes to the "next
1628 alternative in the innermost enclosing group".
1629
1630 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1631 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1632 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1633 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1634 (*THEN).
1635
1636 3. If \s appeared in a character class, it removed the VT character from
1637 the class, even if it had been included by some previous item, for example
1638 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1639 of \s, but is part of the POSIX "space" class.)
1640
1641 4. A partial match never returns an empty string (because you can always
1642 match an empty string at the end of the subject); however the checking for
1643 an empty string was starting at the "start of match" point. This has been
1644 changed to the "earliest inspected character" point, because the returned
1645 data for a partial match starts at this character. This means that, for
1646 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1647 (previously it gave "no match").
1648
1649 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1650 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1651 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1652 has an implication that the given string is incomplete (because a partial
1653 match is preferred over a full match). For this reason, these items now
1654 give a partial match in this situation. [Aside: previously, the one case
1655 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1656 match rather than a full match, which was wrong by the old rules, but is
1657 now correct.]
1658
1659 6. There was a bug in the handling of #-introduced comments, recognized when
1660 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1661 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1662 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1663 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1664 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1665 places in pcre_compile().
1666
1667 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1668 comments when looking ahead for named forward references to subpatterns,
1669 the only newline sequence it recognized was NL. It now handles newlines
1670 according to the set newline convention.
1671
1672 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1673 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1674 cater for a lack of strerror(). These oversights have been fixed.
1675
1676 9. Added --match-limit and --recursion-limit to pcregrep.
1677
1678 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1679
1680 11. When the -o option was used, pcregrep was setting a return code of 1, even
1681 when matches were found, and --line-buffered was not being honoured.
1682
1683 12. Added an optional parentheses number to the -o and --only-matching options
1684 of pcregrep.
1685
1686 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1687 can match an empty string. The code to do it in pcretest and pcredemo
1688 needed fixing:
1689
1690 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1691 only one byte after an empty string match just before CRLF (this case
1692 just got forgotten; "any" and "anycrlf" were OK).
1693
1694 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1695 mode when an empty string match preceded an ASCII character followed by
1696 a non-ASCII character. (The code for advancing by one character rather
1697 than one byte was nonsense.)
1698
1699 (c) The pcredemo.c sample program did not have any code at all to handle
1700 the cases when CRLF is a valid newline sequence.
1701
1702 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1703 as a starting offset was within the subject string. There is now a new
1704 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1705 negative or greater than the length of the string. In order to test this,
1706 pcretest is extended to allow the setting of negative starting offsets.
1707
1708 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1709 starting offset points to the beginning of a UTF-8 character was
1710 unnecessarily clumsy. I tidied it up.
1711
1712 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1713 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1714
1715 17. Nobody had reported that the --include_dir option, which was added in
1716 release 7.7 should have been called --include-dir (hyphen, not underscore)
1717 for compatibility with GNU grep. I have changed it to --include-dir, but
1718 left --include_dir as an undocumented synonym, and the same for
1719 --exclude-dir, though that is not available in GNU grep, at least as of
1720 release 2.5.4.
1721
1722 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1723 characters from a string of bytes) have been redefined so as not to use
1724 loops, in order to improve performance in some environments. At the same
1725 time, I abstracted some of the common code into auxiliary macros to save
1726 repetition (this should not affect the compiled code).
1727
1728 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1729 compile-time error is now given if \c is not followed by an ASCII
1730 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1731 different, and any byte value is allowed.)
1732
1733 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1734 START_OPTIMIZE option, which is now allowed at compile time - but just
1735 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1736 to pcregrep and other applications that have no direct access to PCRE
1737 options. The new /Y option in pcretest sets this option when calling
1738 pcre_compile().
1739
1740 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1741 back references. Groups containing recursive back references were forced to
1742 be atomic by that change, but in the case of named groups, the amount of
1743 memory required was incorrectly computed, leading to "Failed: internal
1744 error: code overflow". This has been fixed.
1745
1746 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1747 pcretest.c, to avoid build problems in some Borland environments.
1748
1749
1750 Version 8.10 25-Jun-2010
1751 ------------------------
1752
1753 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1754 THEN.
1755
1756 2. (*ACCEPT) was not working when inside an atomic group.
1757
1758 3. Inside a character class, \B is treated as a literal by default, but
1759 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1760 causes the error). The code is unchanged, but I tidied the documentation.
1761
1762 4. Inside a character class, PCRE always treated \R and \X as literals,
1763 whereas Perl faults them if its -w option is set. I have changed PCRE so
1764 that it faults them when PCRE_EXTRA is set.
1765
1766 5. Added support for \N, which always matches any character other than
1767 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1768
1769 6. When compiling pcregrep with newer versions of gcc which may have
1770 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1771 declared with attribute warn_unused_result" were given. Just casting the
1772 result to (void) does not stop the warnings; a more elaborate fudge is
1773 needed. I've used a macro to implement this.
1774
1775 7. Minor change to pcretest.c to avoid a compiler warning.
1776
1777 8. Added four artifical Unicode properties to help with an option to make
1778 \s etc use properties (see next item). The new properties are: Xan
1779 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1780
1781 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1782 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1783 this option. Modified pcretest to add /W to test this facility. Added
1784 REG_UCP to make it available via the POSIX interface.
1785
1786 10. Added --line-buffered to pcregrep.
1787
1788 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1789 studied, and the match started with a letter with a code point greater than
1790 127 whose first byte was different to the first byte of the other case of
1791 the letter, the other case of this starting letter was not recognized
1792 (#976).
1793
1794 12. If a pattern that was studied started with a repeated Unicode property
1795 test, for example, \p{Nd}+, there was the theoretical possibility of
1796 setting up an incorrect bitmap of starting bytes, but fortunately it could
1797 not have actually happened in practice until change 8 above was made (it
1798 added property types that matched character-matching opcodes).
1799
1800 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1801 possible starting bytes for non-anchored patterns.
1802
1803 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1804 \R, and also a number of cases that involve Unicode properties, both
1805 explicit and implicit when PCRE_UCP is set.
1806
1807 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1808 input, it could crash or give wrong results if characters with values
1809 greater than 0xc0 were present in the subject string. (Detail: it assumed
1810 UTF-8 input when processing these items.)
1811
1812 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1813 size_t is 64-bit (#991).
1814
1815 17. Added a check for running out of memory when PCRE is compiled with
1816 --disable-stack-for-recursion (#990).
1817
1818 18. If the last data line in a file for pcretest does not have a newline on
1819 the end, a newline was missing in the output.
1820
1821 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1822 less than 128) in its various bitmaps. However, there is a facility for
1823 generating tables according to the current locale when PCRE is compiled. It
1824 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1825 characters, are recognized by isspace() and therefore were getting set in
1826 these tables, and indeed these tables seem to approximate to ISO 8859. This
1827 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1828 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1829 which of course cannot start UTF-8 characters. I have changed the code so
1830 that only real ASCII characters (less than 128) and the correct starting
1831 bytes for UTF-8 encodings are set for characters greater than 127 when in
1832 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1833 altogether.)
1834
1835 20. Added the /T option to pcretest so as to be able to run tests with non-
1836 standard character tables, thus making it possible to include the tests
1837 used for 19 above in the standard set of tests.
1838
1839 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1840 reference to a subpattern the other side of a comment that contains an
1841 opening parenthesis caused either an internal compiling error, or a
1842 reference to the wrong subpattern.
1843
1844
1845 Version 8.02 19-Mar-2010
1846 ------------------------
1847
1848 1. The Unicode data tables have been updated to Unicode 5.2.0.
1849
1850 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1851 configured.
1852
1853 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1854 original author of that file, following a query about its status.
1855
1856 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1857 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1858
1859 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1860 quantifier applied to a forward-referencing subroutine call, could compile
1861 incorrect code or give the error "internal error: previously-checked
1862 referenced subpattern not found".
1863
1864 6. Both MS Visual Studio and Symbian OS have problems with initializing
1865 variables to point to external functions. For these systems, therefore,
1866 pcre_malloc etc. are now initialized to local functions that call the
1867 relevant global functions.
1868
1869 7. There were two entries missing in the vectors called coptable and poptable
1870 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1871 I've fixed the data, and added a kludgy way of testing at compile time that
1872 the lengths are correct (equal to the number of opcodes).
1873
1874 8. Following on from 7, I added a similar kludge to check the length of the
1875 eint vector in pcreposix.c.
1876
1877 9. Error texts for pcre_compile() are held as one long string to avoid too
1878 much relocation at load time. To find a text, the string is searched,
1879 counting zeros. There was no check for running off the end of the string,
1880 which could happen if a new error number was added without updating the
1881 string.
1882
1883 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1884
1885 11. \K was not working if it appeared in an atomic group or in a group that
1886 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1887 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1888 the assertion is positive, but not if it is negative.
1889
1890 12. Change 11 fortuitously reduced the size of the stack frame used in the
1891 "match()" function of pcre_exec.c by one pointer. Forthcoming
1892 implementation of support for (*MARK) will need an extra pointer on the
1893 stack; I have reserved it now, so that the stack frame size does not
1894 decrease.
1895
1896 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1897 item in branch that calls a recursion is a subroutine call - as in the
1898 second branch in the above example - was incorrectly given the compile-
1899 time error "recursive call could loop indefinitely" because pcre_compile()
1900 was not correctly checking the subroutine for matching a non-empty string.
1901
1902 14. The checks for overrunning compiling workspace could trigger after an
1903 overrun had occurred. This is a "should never occur" error, but it can be
1904 triggered by pathological patterns such as hundreds of nested parentheses.
1905 The checks now trigger 100 bytes before the end of the workspace.
1906
1907 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1908
1909
1910 Version 8.01 19-Jan-2010
1911 ------------------------
1912
1913 1. If a pattern contained a conditional subpattern with only one branch (in
1914 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1915 computed the wrong minimum data length (which is of course zero for such
1916 subpatterns). This could cause incorrect "no match" results.
1917
1918 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1919 the pattern is reset in the first branch, pcre_compile() failed with
1920 "internal error: code overflow at offset...". This happened only when
1921 the reset was to the original external option setting. (An optimization
1922 abstracts leading options settings into an external setting, which was the
1923 cause of this.)
1924
1925 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1926 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1927 assertion pattern did not match (meaning that the assertion was true), it
1928 was incorrectly treated as false if the SKIP had been reached during the
1929 matching. This also applied to assertions used as conditions.
1930
1931 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1932 assertion subpattern, including such a pattern used as a condition,
1933 unpredictable results occurred, instead of the error return
1934 PCRE_ERROR_DFA_UITEM.
1935
1936 5. The C++ GlobalReplace function was not working like Perl for the special
1937 situation when an empty string is matched. It now does the fancy magic
1938 stuff that is necessary.
1939
1940 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1941 removed. (These were left over from very, very early versions of PCRE.)
1942
1943 7. Some cosmetic changes to the code to make life easier when compiling it
1944 as part of something else:
1945
1946 (a) Change DEBUG to PCRE_DEBUG.
1947
1948 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1949 called "current" as "current_branch", to prevent a collision with the
1950 Linux macro when compiled as a kernel module.
1951
1952 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1953 prevent a collision with the Linux macro when compiled as a kernel
1954 module.
1955
1956 8. In pcre_compile() there are some checks for integer overflows that used to
1957 cast potentially large values to (double). This has been changed to that
1958 when building, a check for int64_t is made, and if it is found, it is used
1959 instead, thus avoiding the use of floating point arithmetic. (There is no
1960 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1961 double.
1962
1963 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1964 2005 (difference between two addresses compared to an unsigned value).
1965
1966 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1967 custom one, because of the following reported problem in Windows:
1968
1969 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1970 under Win32.
1971 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1972 therefore missing the function definition.
1973 - The compiler thus generates a "C" signature for the test function.
1974 - The linker fails to find the "C" function.
1975 - PCRE fails to configure if asked to do so against libbz2.
1976
1977 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1978 messages were output:
1979
1980 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1981 rerunning libtoolize, to keep the correct libtool macros in-tree.
1982 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1983
1984 I have done both of these things.
1985
1986 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1987 most of the time, it *can* run out if it is given a pattern that contains a
1988 runaway infinite recursion. I updated the discussion in the pcrestack man
1989 page.
1990
1991 13. Now that we have gone to the x.xx style of version numbers, the minor
1992 version may start with zero. Using 08 or 09 is a bad idea because users
1993 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1994 interpreted as invalid octal numbers. I've updated the previous comment in
1995 configure.ac, and also added a check that gives an error if 08 or 09 are
1996 used.
1997
1998 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1999 causing partial matching to fail when the end of the subject matched \W
2000 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2001
2002 15. There were some discrepancies between the declarations in pcre_internal.h
2003 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2004 their definitions. The declarations used "const uschar *" and the
2005 definitions used USPTR. Even though USPTR is normally defined as "const
2006 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2007 reported that: "This difference in casting confuses some C++ compilers, for
2008 example, SunCC recognizes above declarations as different functions and
2009 generates broken code for hbpcre." I have changed the declarations to use
2010 USPTR.
2011
2012 16. GNU libtool is named differently on some systems. The autogen.sh script now
2013 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2014 (FreeBSD).
2015
2016 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2017 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2018 comment: "Figure out how to create a longlong from a string: strtoll and
2019 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2020 instance, but it only takes 2 args instead of 3!"
2021
2022 18. A subtle bug concerned with back references has been fixed by a change of
2023 specification, with a corresponding code fix. A pattern such as
2024 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2025 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2026 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2027 same bug. Such groups have to be quantified to be useful, or contained
2028 inside another quantified group. (If there's no repetition, the reference
2029 can never match.) The problem arises because, having left the group and
2030 moved on to the rest of the pattern, a later failure that backtracks into
2031 the group uses the captured value from the final iteration of the group
2032 rather than the correct earlier one. I have fixed this in PCRE by forcing
2033 any group that contains a reference to itself to be an atomic group; that
2034 is, there cannot be any backtracking into it once it has completed. This is
2035 similar to recursive and subroutine calls.
2036
2037
2038 Version 8.00 19-Oct-09
2039 ----------------------
2040
2041 1. The table for translating pcre_compile() error codes into POSIX error codes
2042 was out-of-date, and there was no check on the pcre_compile() error code
2043 being within the table. This could lead to an OK return being given in
2044 error.
2045
2046 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2047 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2048 in a Windows environment.
2049
2050 3. The pcregrep --count option prints the count for each file even when it is
2051 zero, as does GNU grep. However, pcregrep was also printing all files when
2052 --files-with-matches was added. Now, when both options are given, it prints
2053 counts only for those files that have at least one match. (GNU grep just
2054 prints the file name in this circumstance, but including the count seems
2055 more useful - otherwise, why use --count?) Also ensured that the
2056 combination -clh just lists non-zero counts, with no names.
2057
2058 4. The long form of the pcregrep -F option was incorrectly implemented as
2059 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2060 but it seems right to fix it, and I didn't think it was worth preserving
2061 the old behaviour.
2062
2063 5. The command line items --regex=pattern and --regexp=pattern were not
2064 recognized by pcregrep, which required --regex pattern or --regexp pattern
2065 (with a space rather than an '='). The man page documented the '=' forms,
2066 which are compatible with GNU grep; these now work.
2067
2068 6. No libpcreposix.pc file was created for pkg-config; there was just
2069 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2070
2071 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2072 when UCP support is not needed, by modifying the Python script that
2073 generates it from Unicode data files. This should not matter if the module
2074 is correctly used as a library, but I received one complaint about 50K of
2075 unwanted data. My guess is that the person linked everything into his
2076 program rather than using a library. Anyway, it does no harm.
2077
2078 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2079 was a minimum greater than 1 for a wide character in a possessive
2080 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2081 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2082 character. Chaos in the form of incorrect output or a compiling loop could
2083 result.
2084
2085 9. The restrictions on what a pattern can contain when partial matching is
2086 requested for pcre_exec() have been removed. All patterns can now be
2087 partially matched by this function. In addition, if there are at least two
2088 slots in the offset vector, the offset of the earliest inspected character
2089 for the match and the offset of the end of the subject are set in them when
2090 PCRE_ERROR_PARTIAL is returned.
2091
2092 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2093 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2094 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2095 and may be more useful for multi-segment matching.
2096
2097 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2098 used to be given if ever the end of the subject was reached; now it is
2099 given only if matching could not proceed because another character was
2100 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2101 string "Z", which now yields "no match" instead of "partial match". In the
2102 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2103 final character ended with (*FAIL).
2104
2105 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2106 if the pattern had a "must contain" character that was already found in the
2107 earlier partial match, unless partial matching was again requested. For
2108 example, with the pattern /dog.(body)?/, the "must contain" character is
2109 "g". If the first part-match was for the string "dog", restarting with
2110 "sbody" failed. This bug has been fixed.
2111
2112 13. The string returned by pcre_dfa_exec() after a partial match has been
2113 changed so that it starts at the first inspected character rather than the
2114 first character of the match. This makes a difference only if the pattern
2115 starts with a lookbehind assertion or \b or \B (\K is not supported by
2116 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2117 matching functions compatible, and I think it's the right thing to do.
2118
2119 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2120 so that the demonstration program is easily available in environments where
2121 PCRE has not been installed from source.
2122
2123 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2124 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2125 library.
2126
2127 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2128 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2129 is not the first non-POSIX option to be added. Clearly some people find
2130 these options useful.
2131
2132 17. If a caller to the POSIX matching function regexec() passes a non-zero
2133 value for nmatch with a NULL value for pmatch, the value of
2134 nmatch is forced to zero.
2135
2136 18. RunGrepTest did not have a test for the availability of the -u option of
2137 the diff command, as RunTest does. It now checks in the same way as
2138 RunTest, and also checks for the -b option.
2139
2140 19. If an odd number of negated classes containing just a single character
2141 interposed, within parentheses, between a forward reference to a named
2142 subpattern and the definition of the subpattern, compilation crashed with
2143 an internal error, complaining that it could not find the referenced
2144 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2145 [The bug was that it was starting one character too far in when skipping
2146 over the character class, thus treating the ] as data rather than
2147 terminating the class. This meant it could skip too much.]
2148
2149 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2150 /g option in pcretest when the pattern contains \K, which makes it possible
2151 to have an empty string match not at the start, even when the pattern is
2152 anchored. Updated pcretest and pcredemo to use this option.
2153
2154 21. If the maximum number of capturing subpatterns in a recursion was greater
2155 than the maximum at the outer level, the higher number was returned, but
2156 with unset values at the outer level. The correct (outer level) value is
2157 now given.
2158
2159 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2160 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2161 make it do so. The string so far is captured, making this feature
2162 compatible with Perl.
2163
2164 23. The tests have been re-organized, adding tests 11 and 12, to make it
2165 possible to check the Perl 5.10 features against Perl 5.10.
2166
2167 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2168 pattern matches a fixed length string. PCRE did not allow this; now it
2169 does. Neither allows recursion.
2170
2171 25. I finally figured out how to implement a request to provide the minimum
2172 length of subject string that was needed in order to match a given pattern.
2173 (It was back references and recursion that I had previously got hung up
2174 on.) This code has now been added to pcre_study(); it finds a lower bound
2175 to the length of subject needed. It is not necessarily the greatest lower
2176 bound, but using it to avoid searching strings that are too short does give
2177 some useful speed-ups. The value is available to calling programs via
2178 pcre_fullinfo().
2179
2180 26. While implementing 25, I discovered to my embarrassment that pcretest had
2181 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2182 study optimizations had never been tested with that matching function.
2183 Oops. What is worse, even when it was passed study data, there was a bug in
2184 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2185 were also very few tests of studied patterns with pcre_dfa_exec().
2186
2187 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2188 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2189 on the other side of the coin, they are no longer allowed to have different
2190 names, because these cannot be distinguished in PCRE, and this has caused
2191 confusion. (This is a difference from Perl.)
2192
2193 28. When duplicate subpattern names are present (necessarily with different
2194 numbers, as required by 27 above), and a test is made by name in a
2195 conditional pattern, either for a subpattern having been matched, or for
2196 recursion in such a pattern, all the associated numbered subpatterns are
2197 tested, and the overall condition is true if the condition is true for any
2198 one of them. This is the way Perl works, and is also more like the way
2199 testing by number works.
2200
2201
2202 Version 7.9 11-Apr-09
2203 ---------------------
2204
2205 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2206 (pcretest), all targets were linked against these libraries. This included
2207 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2208 libraries. This caused unwanted dependencies to be created. This problem
2209 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2210 pcretest is linked with readline.
2211
2212 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2213 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2214 moved outside it again, because FALSE and TRUE are already defined in AIX,
2215 but BOOL is not.
2216
2217 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2218 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2219
2220 4. The pcregrep documentation said spaces were inserted as well as colons (or
2221 hyphens) following file names and line numbers when outputting matching
2222 lines. This is not true; no spaces are inserted. I have also clarified the
2223 wording for the --colour (or --color) option.
2224
2225 5. In pcregrep, when --colour was used with -o, the list of matching strings
2226 was not coloured; this is different to GNU grep, so I have changed it to be
2227 the same.
2228
2229 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2230 each matching line was coloured. Now it goes on to look for further matches
2231 of any of the test patterns, which is the same behaviour as GNU grep.
2232
2233 7. A pattern that could match an empty string could cause pcregrep to loop; it
2234 doesn't make sense to accept an empty string match in pcregrep, so I have
2235 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2236 seems to be how GNU grep behaves. [But see later change 40 for release
2237 8.33.]
2238
2239 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2240 start or after a newline", because the conditional assertion was not being
2241 correctly handled. The rule now is that both the assertion and what follows
2242 in the first alternative must satisfy the test.
2243
2244 9. If auto-callout was enabled in a pattern with a conditional group whose
2245 condition was an assertion, PCRE could crash during matching, both with
2246 pcre_exec() and pcre_dfa_exec().
2247
2248 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2249 used for matching.
2250
2251 11. Unicode property support in character classes was not working for
2252 characters (bytes) greater than 127 when not in UTF-8 mode.
2253
2254 12. Added the -M command line option to pcretest.
2255
2256 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2257
2258 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2259
2260 16. Added comments and documentation about mis-use of no_arg in the C++
2261 wrapper.
2262
2263 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2264 from Martin Jerabek that uses macro names for all relevant character and
2265 string constants.
2266
2267 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2268 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2269 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2270 these, but not everybody uses configure.
2271
2272 19. A conditional group that had only one branch was not being correctly
2273 recognized as an item that could match an empty string. This meant that an
2274 enclosing group might also not be so recognized, causing infinite looping
2275 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2276 with the subject "ab", where knowledge that the repeated group can match
2277 nothing is needed in order to break the loop.
2278
2279 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2280 exec(), but without supplying a callout function, matching went wrong.
2281
2282 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2283 leak if the size of the offset vector was greater than 30. When the vector
2284 is smaller, the saved offsets during recursion go onto a local stack
2285 vector, but for larger vectors malloc() is used. It was failing to free
2286 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2287 error, in fact).
2288
2289 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2290 heapframe that is used only when UTF-8 support is enabled. This caused no
2291 problem, but was untidy.
2292
2293 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2294 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2295 included within another project.
2296
2297 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2298 slightly modified by me:
2299
2300 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2301 not building pcregrep.
2302
2303 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2304 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2305
2306 25. Forward references, both numeric and by name, in patterns that made use of
2307 duplicate group numbers, could behave incorrectly or give incorrect errors,
2308 because when scanning forward to find the reference group, PCRE was not
2309 taking into account the duplicate group numbers. A pattern such as
2310 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2311
2312 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2313 the feature of a custom pointer more persuasive (as requested by a user).
2314
2315 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2316 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2317 pre-defined.
2318
2319 28. Added support for (*UTF8) at the start of a pattern.
2320
2321 29. Arrange for flags added by the "release type" setting in CMake to be shown
2322 in the configuration summary.
2323
2324
2325 Version 7.8 05-Sep-08
2326 ---------------------
2327
2328 1. Replaced UCP searching code with optimized version as implemented for Ad
2329 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2330 stage table and inline lookup instead of a function, giving speed ups of 2
2331 to 5 times on some simple patterns that I tested. Permission was given to
2332 distribute the MultiStage2.py script that generates the tables (it's not in
2333 the tarball, but is in the Subversion repository).
2334
2335 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2336 scripts.
2337
2338 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2339 a group with a zero qualifier. The result of the study could be incorrect,
2340 or the function might crash, depending on the pattern.
2341
2342 4. Caseless matching was not working for non-ASCII characters in back
2343 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2344 It now works when Unicode Property Support is available.
2345
2346 5. In pcretest, an escape such as \x{de} in the data was always generating
2347 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2348 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2349 truncation.
2350
2351 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2352
2353 7. Added two (int) casts to pcregrep when printing the difference of two
2354 pointers, in case they are 64-bit values.
2355
2356 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2357 test 2 if it fails.
2358
2359 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2360 and a #define of that name to empty if it is not externally set. This is to
2361 allow users of MSVC to set it if necessary.
2362
2363 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2364 the convenience functions in the pcre_get.c source file.
2365
2366 11. An option change at the start of a pattern that had top-level alternatives
2367 could cause overwriting and/or a crash. This command provoked a crash in
2368 some environments:
2369
2370 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2371
2372 This potential security problem was recorded as CVE-2008-2371.
2373
2374 12. For a pattern where the match had to start at the beginning or immediately
2375 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2376 pcre_dfa_exec() could read past the end of the passed subject if there was
2377 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2378 pcretest so that it places the subject at the end of its malloc-ed buffer.
2379
2380 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2381 exec() might read past the end of the data buffer in UTF-8 mode.
2382
2383 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2384 the data contained the byte 0x85 as part of a UTF-8 character within its
2385 first line. This applied both to normal and DFA matching.
2386
2387 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2388 /^[^d]*?$/8 failed to match "abc".
2389
2390 16. Added a missing copyright notice to pcrecpp_internal.h.
2391
2392 17. Make it more clear in the documentation that values returned from
2393 pcre_exec() in ovector are byte offsets, not character counts.
2394
2395 18. Tidied a few places to stop certain compilers from issuing warnings.
2396
2397 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2398 supplied by Stefan Weber. I made a further small update for 7.8 because
2399 there is a change of source arrangements: the pcre_searchfuncs.c module is
2400 replaced by pcre_ucd.c.
2401
2402
2403 Version 7.7 07-May-08
2404 ---------------------
2405
2406 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2407 a string to a long long, pretend we don't even have a long long." This is
2408 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2409
2410 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2411 pre-7.6 versions, which defined a global no_arg variable instead of putting
2412 it in the RE class. (See also #8 below.)
2413
2414 3. Remove a line of dead code, identified by coverity and reported by Nuno
2415 Lopes.
2416
2417 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2418
2419 (1) The include/exclude patterns were being applied to the whole pathnames
2420 of files, instead of just to the final components.
2421
2422 (2) If there was more than one level of directory, the subdirectories were
2423 skipped unless they satisfied the include/exclude conditions. This is
2424 inconsistent with GNU grep (and could even be seen as contrary to the
2425 pcregrep specification - which I improved to make it absolutely clear).
2426 The action now is always to scan all levels of directory, and just
2427 apply the include/exclude patterns to regular files.
2428
2429 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2430 --exclude_dir in the tests to avoid scanning .svn directories.
2431
2432 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2433 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2434 doesn't support NULs in patterns.
2435
2436 7. Added some missing "const"s to declarations of static tables in
2437 pcre_compile.c and pcre_dfa_exec.c.
2438
2439 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2440 caused by fix #2 above. (Subsequently also a second patch to fix the
2441 first patch. And a third patch - this was a messy problem.)
2442
2443 9. Applied Craig's patch to remove the use of push_back().
2444
2445 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2446 matching function regexec().
2447
2448 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2449 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2450 references. PCRE supports relative numbers with this syntax (I don't think
2451 Oniguruma does).
2452
2453 12. Previously, a group with a zero repeat such as (...){0} was completely
2454 omitted from the compiled regex. However, this means that if the group
2455 was called as a subroutine from elsewhere in the pattern, things went wrong
2456 (an internal error was given). Such groups are now left in the compiled
2457 pattern, with a new opcode that causes them to be skipped at execution
2458 time.
2459
2460 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2461 to the way PCRE behaves:
2462
2463 (a) A lone ] character is dis-allowed (Perl treats it as data).
2464
2465 (b) A back reference to an unmatched subpattern matches an empty string
2466 (Perl fails the current match path).
2467
2468 (c) A data ] in a character class must be notated as \] because if the
2469 first data character in a class is ], it defines an empty class. (In
2470 Perl it is not possible to have an empty class.) The empty class []
2471 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2472 The negative empty class [^] matches any one character, independently
2473 of the DOTALL setting.
2474
2475 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2476 non-existent subpattern following a character class starting with ']' and
2477 containing () gave an internal compiling error instead of "reference to
2478 non-existent subpattern". Fortunately, when the pattern did exist, the
2479 compiled code was correct. (When scanning forwards to check for the
2480 existence of the subpattern, it was treating the data ']' as terminating
2481 the class, so got the count wrong. When actually compiling, the reference
2482 was subsequently set up correctly.)
2483
2484 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2485 it was being rejected as not supported by pcre_dfa_exec(), even though
2486 other assertions are supported. I have made pcre_dfa_exec() support
2487 (*FAIL).
2488
2489 16. The implementation of 13c above involved the invention of a new opcode,
2490 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2491 cannot be changed at match time, I realized I could make a small
2492 improvement to matching performance by compiling OP_ALLANY instead of
2493 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2494 on the OP_ANY path.
2495
2496 17. Compiling pcretest on Windows with readline support failed without the
2497 following two fixes: (1) Make the unistd.h include conditional on
2498 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2499
2500 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2501 ncurses library to be included for pcretest when ReadLine support is
2502 requested, but also to allow for it to be overridden. This patch came from
2503 Daniel Bergström.
2504
2505 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2506 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2507 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2508 spotting this.
2509
2510
2511 Version 7.6 28-Jan-08
2512 ---------------------
2513
2514 1. A character class containing a very large number of characters with
2515 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2516 overflow.
2517
2518 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2519 HAVE_LONG_LONG is not defined.
2520
2521 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2522 bring them up to date and include new features. This patch includes:
2523
2524 - Fixed PH's badly added libz and libbz2 support.
2525 - Fixed a problem with static linking.
2526 - Added pcredemo. [But later removed - see 7 below.]
2527 - Fixed dftables problem and added an option.
2528 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2529 HAVE_LONG_LONG.
2530 - Added readline support for pcretest.
2531 - Added an listing of the option settings after cmake has run.
2532
2533 4. A user submitted a patch to Makefile that makes it easy to create
2534 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2535 Makefile.am that cause it to include this special target, without
2536 affecting anything else. Note that the same mingw target plus all
2537 the other distribution libraries and programs are now supported
2538 when configuring with CMake (see 6 below) instead of with
2539 Configure/Make.
2540
2541 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2542 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2543 exported in the Windows port". It has not yet been confirmed that the patch
2544 solves the problem, but it does no harm.
2545
2546 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2547 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2548 with CMake, and also correct the comment about stack recursion.
2549
2550 7. Remove the automatic building of pcredemo from the ./configure system and
2551 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2552 of a program that users should build themselves after PCRE is installed, so
2553 building it automatically is not really right. What is more, it gave
2554 trouble in some build environments.
2555
2556 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2557
2558
2559 Version 7.5 10-Jan-08
2560 ---------------------
2561
2562 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2563 values in parens when parsing an RE using the C++ wrapper."
2564
2565 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2566 Characters greater than 255 were excluded from the class instead of being
2567 included.
2568
2569 3. The same bug as (2) above applied to negated POSIX classes such as
2570 [:^space:].
2571
2572 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2573 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2574 I have changed it.
2575
2576 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2577 first named subpattern) and a construct such as (?&a) would reference the
2578 first named subpattern whose name started with "a" (in other words, the
2579 length check was missing). Both these problems are fixed. "Subpattern name
2580 expected" is now given for (?&) (a zero-length name), and this patch also
2581 makes it give the same error for \k'' (previously it complained that that
2582 was a reference to a non-existent subpattern).
2583
2584 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2585 this is right because (?- can be followed by option settings as well as by
2586 digits. I have, however, made the messages clearer.
2587
2588 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2589 than the number used in the conditional) now cause a compile-time error.
2590 This is actually not compatible with Perl, which accepts such patterns, but
2591 treats the conditional as always being FALSE (as PCRE used to), but it
2592 seems to me that giving a diagnostic is better.
2593
2594 8. Change "alphameric" to the more common word "alphanumeric" in comments
2595 and messages.
2596
2597 9. Fix two occurrences of "backslash" in comments that should have been
2598 "backspace".
2599
2600 10. Remove two redundant lines of code that can never be obeyed (their function
2601 was moved elsewhere).
2602
2603 11. The program that makes PCRE's Unicode character property table had a bug
2604 which caused it to generate incorrect table entries for sequences of
2605 characters that have the same character type, but are in different scripts.
2606 It amalgamated them into a single range, with the script of the first of
2607 them. In other words, some characters were in the wrong script. There were
2608 thirteen such cases, affecting characters in the following ranges:
2609
2610 U+002b0 - U+002c1
2611 U+0060c - U+0060d
2612 U+0061e - U+00612
2613 U+0064b - U+0065e
2614 U+0074d - U+0076d
2615 U+01800 - U+01805
2616 U+01d00 - U+01d77
2617 U+01d9b - U+01dbf
2618 U+0200b - U+0200f
2619 U+030fc - U+030fe
2620 U+03260 - U+0327f
2621 U+0fb46 - U+0fbb1
2622 U+10450 - U+1049d
2623
2624 12. The -o option (show only the matching part of a line) for pcregrep was not
2625 compatible with GNU grep in that, if there was more than one match in a
2626 line, it showed only the first of them. It now behaves in the same way as
2627 GNU grep.
2628
2629 13. If the -o and -v options were combined for pcregrep, it printed a blank
2630 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2631 does the same. The return code can be used to tell if there were any
2632 non-matching lines.
2633
2634 14. Added --file-offsets and --line-offsets to pcregrep.
2635
2636 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2637 infinitely looping recursion. The bug was that positive lookaheads were not
2638 being skipped when checking for a possible empty match (negative lookaheads
2639 and both kinds of lookbehind were skipped).
2640
2641 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2642 inclusion of <windows.h> to before rather than after the definition of
2643 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2644
2645 17. Specifying a possessive quantifier with a specific limit for a Unicode
2646 character property caused pcre_compile() to compile bad code, which led at
2647 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2648 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2649 caused the error; without that there was no problem.
2650
2651 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2652
2653 19. Added --enable-pcretest-libreadline.
2654
2655 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2656 RE::GlobalReplace(). As a result, the number of replacements returned was
2657 double what it should be. I removed one of the increments, but Craig sent a
2658 later patch that removed the other one (the right fix) and added unit tests
2659 that check the return values (which was not done before).
2660
2661 21. Several CMake things:
2662
2663 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2664 the names libpcre and libpcreposix, not just pcre and pcreposix.
2665
2666 (2) The above change means that pcretest and pcregrep are now correctly
2667 linked with the newly-built libraries, not previously installed ones.
2668
2669 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2670
2671 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2672 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2673 UTF-8 newline character). The key issue is that the pattern starts .*;
2674 this means that the match must be either at the beginning, or after a
2675 newline. The bug was in the code for advancing after a failed match and
2676 checking that the new position followed a newline. It was not taking
2677 account of UTF-8 characters correctly.
2678
2679 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2680 character classes. PCRE was not treating the sequence [:...:] as a
2681 character class unless the ... were all letters. Perl, however, seems to
2682 allow any characters between [: and :], though of course it rejects as
2683 unknown any "names" that contain non-letters, because all the known class
2684 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2685 for example, whereas PCRE did not - it did not recognize a POSIX character
2686 class. This seemed a bit dangerous, so the code has been changed to be
2687 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2688 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2689 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2690 Perl does, and where it didn't before.
2691
2692 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2693 Windows environments %n is disabled by default.
2694
2695
2696 Version 7.4 21-Sep-07
2697 ---------------------
2698
2699 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2700 means that a class such as [\s] counted as "explicit reference to CR or
2701 LF". That isn't really right - the whole point of the change was to try to
2702 help when there was an actual mention of one of the two characters. So now
2703 the change happens only if \r or \n (or a literal CR or LF) character is
2704 encountered.
2705
2706 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2707 of both had grown to the point where there were only 3 bits left.
2708 Fortunately, there was spare space in the data structure, and so I have
2709 moved the internal flags into a new 16-bit field to free up more option
2710 bits.
2711
2712 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2713 but did not set the internal JCHANGED flag - either of these is enough to
2714 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2715 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2716 start sets both bits.
2717
2718 4. Added options (at build time, compile time, exec time) to change \R from
2719 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2720
2721 5. doc/pcresyntax.html was missing from the distribution.
2722
2723 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2724 compatibility, even though it is no longer used.
2725
2726 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2727 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2728 windows.h file is present (where different names are used). [This was
2729 reversed later after testing - see 16 below.]
2730
2731 8. Changed all #include <config.h> to #include "config.h". There were also
2732 some further <pcre.h> cases that I changed to "pcre.h".
2733
2734 9. When pcregrep was used with the --colour option, it missed the line ending
2735 sequence off the lines that it output.
2736
2737 10. It was pointed out to me that arrays of string pointers cause lots of
2738 relocations when a shared library is dynamically loaded. A technique of
2739 using a single long string with a table of offsets can drastically reduce
2740 these. I have refactored PCRE in four places to do this. The result is
2741 dramatic:
2742
2743 Originally: 290
2744 After changing UCP table: 187
2745 After changing error message table: 43
2746 After changing table of "verbs" 36
2747 After changing table of Posix names 22
2748
2749 Thanks to the folks working on Gregex for glib for this insight.
2750
2751 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2752 unicode-properties was also set.
2753
2754 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2755
2756 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2757 checked only for CRLF.
2758
2759 14. Added casts to pcretest.c to avoid compiler warnings.
2760
2761 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2762
2763 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2764 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2765 entirely. This removes changes made in 7 above.
2766
2767 17. The CMake files have been updated, and there is now more information about
2768 building with CMake in the NON-UNIX-USE document.
2769
2770
2771 Version 7.3 28-Aug-07
2772 ---------------------
2773
2774 1. In the rejigging of the build system that eventually resulted in 7.1, the
2775 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2776 brackets there is not right, since it causes compilers to look for an
2777 installed pcre.h, not the version that is in the source that is being
2778 compiled (which of course may be different). I have changed it back to:
2779
2780 #include "pcre.h"
2781
2782 I have a vague recollection that the change was concerned with compiling in
2783 different directories, but in the new build system, that is taken care of
2784 by the VPATH setting the Makefile.
2785
2786 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2787 when the subject happened to end in the byte 0x85 (e.g. if the last
2788 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2789 characters but of course it shouldn't be taken as a newline when it is part
2790 of another character. The bug was that, for an unlimited repeat of . in
2791 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2792 characters when looking for a newline.
2793
2794 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2795
2796 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2797 in debug output.
2798
2799 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2800 long printing in the pcrecpp unittest when running under MinGW.
2801
2802 6. ESC_K was left out of the EBCDIC table.
2803
2804 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2805 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2806 limit also applies to "virtual nesting" when a pattern is recursive, and in
2807 this case 1000 isn't so big. I have been able to remove this limit at the
2808 expense of backing off one optimization in certain circumstances. Normally,
2809 when pcre_exec() would call its internal match() function recursively and
2810 immediately return the result unconditionally, it uses a "tail recursion"
2811 feature to save stack. However, when a subpattern that can match an empty
2812 string has an unlimited repetition quantifier, it no longer makes this
2813 optimization. That gives it a stack frame in which to save the data for
2814 checking that an empty string has been matched. Previously this was taken
2815 from the 1000-entry workspace that had been reserved. So now there is no
2816 explicit limit, but more stack is used.
2817
2818 8. Applied Daniel's patches to solve problems with the import/export magic
2819 syntax that is required for Windows, and which was going wrong for the
2820 pcreposix and pcrecpp parts of the library. These were overlooked when this
2821 problem was solved for the main library.
2822
2823 9. There were some crude static tests to avoid integer overflow when computing
2824 the size of patterns that contain repeated groups with explicit upper
2825 limits. As the maximum quantifier is 65535, the maximum group length was
2826 set at 30,000 so that the product of these two numbers did not overflow a
2827 32-bit integer. However, it turns out that people want to use groups that
2828 are longer than 30,000 bytes (though not repeat them that many times).
2829 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2830 made it possible to implement the integer overflow checks in a much more
2831 dynamic way, which I have now done. The artificial limitation on group
2832 length has been removed - we now have only the limit on the total length of
2833 the compiled pattern, which depends on the LINK_SIZE setting.
2834
2835 10. Fixed a bug in the documentation for get/copy named substring when
2836 duplicate names are permitted. If none of the named substrings are set, the
2837 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2838 empty string.
2839
2840 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2841 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2842 because the ] is interpreted as the first data character and the
2843 terminating ] is not found. PCRE has been made compatible with Perl in this
2844 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2845 cause memory overwriting.
2846
2847 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2848 string has been matched (to stop an infinite loop). It was not recognizing
2849 a conditional subpattern that could match an empty string if that
2850 subpattern was within another subpattern. For example, it looped when
2851 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2852 condition was not nested. This bug has been fixed.
2853
2854 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2855 past the start of the subject in the presence of bytes with the top bit
2856 set, for example "\x8aBCD".
2857
2858 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2859 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2860
2861 14. Optimized (?!) to (*FAIL).
2862
2863 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2864 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2865 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2866 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2867 does: it's just the validity check that is more restrictive.
2868
2869 16. Inserted checks for integer overflows during escape sequence (backslash)
2870 processing, and also fixed erroneous offset values for syntax errors during
2871 backslash processing.
2872
2873 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2874 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2875
2876 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2877 caused an overrun.
2878
2879 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2880 something other than just ASCII characters) inside a group that had an
2881 unlimited repeat caused a loop at compile time (while checking to see
2882 whether the group could match an empty string).
2883
2884 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2885 [\P{Any}] did so. (Error in the code for printing property names.)
2886
2887 21. An orphan \E inside a character class could cause a crash.
2888
2889 22. A repeated capturing bracket such as (A)? could cause a wild memory
2890 reference during compilation.
2891
2892 23. There are several functions in pcre_compile() that scan along a compiled
2893 expression for various reasons (e.g. to see if it's fixed length for look
2894 behind). There were bugs in these functions when a repeated \p or \P was
2895 present in the pattern. These operators have additional parameters compared
2896 with \d, etc, and these were not being taken into account when moving along
2897 the compiled data. Specifically:
2898
2899 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2900 length.
2901
2902 (b) An item such as \pL+ within a repeated group could cause crashes or
2903 loops.
2904
2905 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2906 "reference to non-existent subpattern" error.
2907
2908 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2909
2910 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2911 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2912
2913 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2914
2915 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2916 character were causing crashes (broken optimization).
2917
2918 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2919 \p or \P) caused a compile-time loop.
2920
2921 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2922 break. For example, the unstudied pattern [\r\n]A does not match the string
2923 "\r\nA" because change 7.0/46 below moves the current point on by two
2924 characters after failing to match at the start. However, the pattern \nA
2925 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2926 the same is true. There doesn't seem any very clean way out of this, but
2927 what I have chosen to do makes the common cases work: PCRE now takes note
2928 of whether there can be an explicit match for \r or \n anywhere in the
2929 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2930 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2931 pattern has explicit CR or LF references.
2932
2933 29. Added (*CR) etc for changing newline setting at start of pattern.
2934
2935
2936 Version 7.2 19-Jun-07
2937 ---------------------
2938
2939 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2940 which is apparently normally available under Windows.
2941
2942 2. Re-jig the pcregrep tests with different newline settings in an attempt
2943 to make them independent of the local environment's newline setting.
2944
2945 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2946
2947 4. Some of the "internals" tests were previously cut out when the link size
2948 was not 2, because the output contained actual offsets. The recent new
2949 "Z" feature of pcretest means that these can be cut out, making the tests
2950 usable with all link sizes.
2951
2952 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2953 stack recursion. This gives a massive performance boost under BSD, but just
2954 a small improvement under Linux. However, it saves one field in the frame
2955 in all cases.
2956
2957 6. Added more features from the forthcoming Perl 5.10:
2958
2959 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2960 recursion call. It refers to the nth most recently opened parentheses.
2961
2962 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2963 to be opened parentheses.
2964
2965 (c) Conditions that refer to capturing parentheses can be specified
2966 relatively, for example, (?(-2)... or (?(+3)...
2967
2968 (d) \K resets the start of the current match so that everything before
2969 is not part of it.
2970
2971 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2972
2973 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2974 reference syntax.
2975
2976 (g) (?| introduces a group in which the numbering of parentheses in each
2977 alternative starts with the same number.
2978
2979 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2980
2981 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2982 PCRE_INFO_JCHANGED.
2983
2984 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2985 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2986 for detecting groups that can match an empty string.
2987
2988 9. A pattern with a very large number of alternatives (more than several
2989 hundred) was running out of internal workspace during the pre-compile
2990 phase, where pcre_compile() figures out how much memory will be needed. A
2991 bit of new cunning has reduced the workspace needed for groups with
2992 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2993 workspace instead of running out of the 4096 that are available.
2994
2995 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2996
2997 11. Applied patch from Google to remove an optimization that didn't quite work.
2998 The report of the bug said:
2999
3000 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3001 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3002 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3003
3004 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3005 it matched the wrong number of bytes.
3006
3007
3008 Version 7.1 24-Apr-07
3009 ---------------------
3010
3011 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3012 that is more "standard", making use of automake and other Autotools. There
3013 is some re-arrangement of the files and adjustment of comments consequent
3014 on this.
3015
3016 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3017 for recursive directory scanning broke on some systems because the files
3018 are not scanned in any specific order and on different systems the order
3019 was different. A call to "sort" has been inserted into RunGrepTest for the
3020 approprate test as a short-term fix. In the longer term there may be an
3021 alternative.
3022
3023 3. I had an email from Eric Raymond about problems translating some of PCRE's
3024 man pages to HTML (despite the fact that I distribute HTML pages, some
3025 people do their own conversions for various reasons). The problems
3026 concerned the use of low-level troff macros .br and .in. I have therefore
3027 removed all such uses from the man pages (some were redundant, some could
3028 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3029 HTML has been updated to handle .nf/.fi and to complain if it encounters
3030 .br or .in.
3031
3032 4. Updated comments in configure.ac that get placed in config.h.in and also
3033 arranged for config.h to be included in the distribution, with the name
3034 config.h.generic, for the benefit of those who have to compile without
3035 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3036
3037 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3038 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3039 makevp.bat for the current PCRE, using the additional files
3040 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3041
3042 6. A Windows user reported a minor discrepancy with test 2, which turned out
3043 to be caused by a trailing space on an input line that had got lost in his
3044 copy. The trailing space was an accident, so I've just removed it.
3045
3046 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3047 that is needed.
3048
3049 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3050 as "const" (a) because they are and (b) because it helps the PHP
3051 maintainers who have recently made a script to detect big data structures
3052 in the php code that should be moved to the .rodata section. I remembered
3053 to update Builducptable as well, so it won't revert if ucptable.h is ever
3054 re-created.
3055
3056 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3057 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3058 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3059 support is not required. This saves 1.5-2K of code, which is important in
3060 some applications.
3061
3062 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3063 so as not to refer to the tables, even though these functions will never be
3064 called when UTF-8 support is disabled. Otherwise there are problems with a
3065 shared library.
3066
3067 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3068
3069 (a) It was defining its arguments as char * instead of void *.
3070
3071 (b) It was assuming that all moves were upwards in memory; this was true
3072 a long time ago when I wrote it, but is no longer the case.
3073
3074 The emulated memove() is provided for those environments that have neither
3075 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3076 is clearly not the case, as these two bugs were recently reported.
3077
3078 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3079 and Detrail to create the HTML documentation, the .txt form of the man
3080 pages, and it removes trailing spaces from listed files. It also creates
3081 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3082 case, it wraps all the #defines with #ifndefs. This script should be run
3083 before "make dist".
3084
3085 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3086 with Unicode property support.
3087
3088 (a) For a maximizing quantifier, if the two different cases of the
3089 character were of different lengths in their UTF-8 codings (there are
3090 some cases like this - I found 11), and the matching function had to
3091 back up over a mixture of the two cases, it incorrectly assumed they
3092 were both the same length.
3093
3094 (b) When PCRE was configured to use the heap rather than the stack for
3095 recursion during matching, it was not correctly preserving the data for
3096 the other case of a UTF-8 character when checking ahead for a match
3097 while processing a minimizing repeat. If the check also involved
3098 matching a wide character, but failed, corruption could cause an
3099 erroneous result when trying to check for a repeat of the original
3100 character.
3101
3102 13. Some tidying changes to the testing mechanism:
3103
3104 (a) The RunTest script now detects the internal link size and whether there
3105 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3106 values substituted by "configure". (The RunGrepTest script already did
3107 this for UTF-8.) The configure.ac script no longer substitutes the
3108 relevant variables.
3109
3110 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3111 with length and offset values. This means that the output is different
3112 for different internal link sizes. Test 2 is skipped for link sizes
3113 other than 2 because of this, bypassing the problem. Unfortunately,
3114 there was also a test in test 3 (the locale tests) that used /B and
3115 failed for link sizes other than 2. Rather than cut the whole test out,
3116 I have added a new /Z option to pcretest that replaces the length and
3117 offset values with spaces. This is now used to make test 3 independent
3118 of link size. (Test 2 will be tidied up later.)
3119
3120 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3121 segmentation fault instead of returning the appropriate error message.
3122
3123 15. In multiline mode when the newline sequence was set to "any", the pattern
3124 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3125 This doesn't seem right; it now treats the CRLF combination as the line
3126 ending, and so does not match in that case. It's only a pattern such as ^$
3127 that would hit this one: something like ^ABC$ would have failed after \r
3128 and then tried again after \r\n.
3129
3130 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3131 in an attempt to make files that differ only in their line terminators
3132 compare equal. This works on Linux.
3133
3134 17. Under certain error circumstances pcregrep might try to free random memory
3135 as it exited. This is now fixed, thanks to valgrind.
3136
3137 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3138 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3139 was because its rules for how to advance for /g after matching an empty
3140 string at the end of a line did not allow for this case. They now check for
3141 it specially.
3142
3143 20. pcretest is supposed to handle patterns and data of any length, by
3144 extending its buffers when necessary. It was getting this wrong when the
3145 buffer for a data line had to be extended.
3146
3147 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3148 CRLF as a newline sequence.
3149
3150 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3151 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3152 I have nevertheless tidied it up.
3153
3154 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3155
3156 24. Added a man page for pcre-config.
3157
3158
3159 Version 7.0 19-Dec-06
3160 ---------------------
3161
3162 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3163 moving to gcc 4.1.1.
3164
3165 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3166 sys/time.h, which is documented as needed for this function. It doesn't
3167 seem to matter on Linux, but it showed up on some releases of OS X.
3168
3169 3. It seems that there are systems where bytes whose values are greater than
3170 127 match isprint() in the "C" locale. The "C" locale should be the
3171 default when a C program starts up. In most systems, only ASCII printing
3172 characters match isprint(). This difference caused the output from pcretest
3173 to vary, making some of the tests fail. I have changed pcretest so that:
3174
3175 (a) When it is outputting text in the compiled version of a pattern, bytes
3176 other than 32-126 are always shown as hex escapes.
3177
3178 (b) When it is outputting text that is a matched part of a subject string,
3179 it does the same, unless a different locale has been set for the match
3180 (using the /L modifier). In this case, it uses isprint() to decide.
3181
3182 4. Fixed a major bug that caused incorrect computation of the amount of memory
3183 required for a compiled pattern when options that changed within the
3184 pattern affected the logic of the preliminary scan that determines the
3185 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3186 that the computed length was too small. The symptoms of this bug were
3187 either the PCRE error "internal error: code overflow" from pcre_compile(),
3188 or a glibc crash with a message such as "pcretest: free(): invalid next
3189 size (fast)". Examples of patterns that provoked this bug (shown in
3190 pcretest format) are:
3191
3192 /(?-x: )/x
3193 /(?x)(?-x: \s*#\s*)/
3194 /((?i)[\x{c0}])/8
3195 /(?i:[\x{c0}])/8
3196
3197 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3198 is now done differently.
3199
3200 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3201 wrapper classes; (b) implement a new function in the C++ scanner that is
3202 more efficient than the old way of doing things because it avoids levels of
3203 recursion in the regex matching; (c) add a paragraph to the documentation
3204 for the FullMatch() function.
3205
3206 6. The escape sequence \n was being treated as whatever was defined as
3207 "newline". Not only was this contrary to the documentation, which states
3208 that \n is character 10 (hex 0A), but it also went horribly wrong when
3209 "newline" was defined as CRLF. This has been fixed.
3210
3211 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3212 was being set to -1 for the "end of line" case (supposedly a value that no
3213 character can have). Though this value is never used (the check for end of
3214 line is "zero bytes in current character"), it caused compiler complaints.
3215 I've changed it to 0xffffffff.
3216
3217 8. In pcre_version.c, the version string was being built by a sequence of
3218 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3219 string (as it is for production releases) called a macro with an empty
3220 argument. The C standard says the result of this is undefined. The gcc
3221 compiler treats it as an empty string (which was what was wanted) but it is
3222 reported that Visual C gives an error. The source has been hacked around to
3223 avoid this problem.
3224
3225 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3226 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3227 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3228 of them did).
3229
3230 10. Originally, pcretest opened its input and output without "b"; then I was
3231 told that "b" was needed in some environments, so it was added for release
3232 5.0 to both the input and output. (It makes no difference on Unix-like
3233 systems.) Later I was told that it is wrong for the input on Windows. I've
3234 now abstracted the modes into two macros, to make it easier to fiddle with
3235 them, and removed "b" from the input mode under Windows.
3236
3237 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3238
3239 12. Added -help and --help to pcretest as an official way of being reminded
3240 of the options.
3241
3242 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3243 and pcrecpp.cc because they annoy compilers at high warning levels.
3244
3245 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3246
3247 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3248 scripts are not C programs :-) and which was not noticed because it works
3249 on Linux.
3250
3251 16. pcretest is supposed to handle any length of pattern and data line (as one
3252 line or as a continued sequence of lines) by extending its input buffer if
3253 necessary. This feature was broken for very long pattern lines, leading to
3254 a string of junk being passed to pcre_compile() if the pattern was longer
3255 than about 50K.
3256
3257 17. I have done a major re-factoring of the way pcre_compile() computes the
3258 amount of memory needed for a compiled pattern. Previously, there was code
3259 that made a preliminary scan of the pattern in order to do this. That was
3260 OK when PCRE was new, but as the facilities have expanded, it has become
3261 harder and harder to keep it in step with the real compile phase, and there
3262 have been a number of bugs (see for example, 4 above). I have now found a
3263 cunning way of running the real compile function in a "fake" mode that
3264 enables it to compute how much memory it would need, while actually only
3265 ever using a few hundred bytes of working memory and without too many
3266 tests of the mode. This should make future maintenance and development
3267 easier. A side effect of this work is that the limit of 200 on the nesting
3268 depth of parentheses has been removed (though this was never a serious
3269 limitation, I suspect). However, there is a downside: pcre_compile() now
3270 runs more slowly than before (30% or more, depending on the pattern). I
3271 hope this isn't a big issue. There is no effect on runtime performance.
3272
3273 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3274 newline (only possible for the last line of a file) and it was a
3275 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3276
3277 19. Added additional timing features to pcretest. (1) The -tm option now times
3278 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3279 separate command line item, by a number that specifies the number of
3280 repeats to use when timing. The default is 50000; this gives better
3281 precision, but takes uncomfortably long for very large patterns.
3282
3283 20. Extended pcre_study() to be more clever in cases where a branch of a
3284 subpattern has no definite first character. For example, (a*|b*)[cd] would
3285 previously give no result from pcre_study(). Now it recognizes that the
3286 first character must be a, b, c, or d.
3287
3288 21. There was an incorrect error "recursive call could loop indefinitely" if
3289 a subpattern (or the entire pattern) that was being tested for matching an
3290 empty string contained only one non-empty item after a nested subpattern.
3291 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3292 incorrectly, because the \d was being skipped in the check.
3293
3294 22. The pcretest program now has a new pattern option /B and a command line
3295 option -b, which is equivalent to adding /B to every pattern. This causes
3296 it to show the compiled bytecode, without the additional information that
3297 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3298 is the same as /B/I).
3299
3300 23. A new optimization is now able automatically to treat some sequences such
3301 as a*b as a*+b. More specifically, if something simple (such as a character
3302 or a simple class like \d) has an unlimited quantifier, and is followed by
3303 something that cannot possibly match the quantified thing, the quantifier
3304 is automatically "possessified".
3305
3306 24. A recursive reference to a subpattern whose number was greater than 39
3307 went wrong under certain circumstances in UTF-8 mode. This bug could also
3308 have affected the operation of pcre_study().
3309
3310 25. Realized that a little bit of performance could be had by replacing
3311 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3312
3313 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3314
3315 27. Possessive quantifiers such as a++ were previously implemented by turning
3316 them into atomic groups such as ($>a+). Now they have their own opcodes,
3317 which improves performance. This includes the automatically created ones
3318 from 23 above.
3319
3320 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3321 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3322 the first matched character to be a colon. This applied both to named and
3323 numbered groups.
3324
3325 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3326
3327 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3328 building PCRE on Windows easier, so I have included it in the distribution.
3329
3330 31. There is now a check in pcretest against a ridiculously large number being
3331 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3332 loop, the loop is abandoned.
3333
3334 32. Forward references to subpatterns in conditions such as (?(2)...) where
3335 subpattern 2 is defined later cause pcre_compile() to search forwards in
3336 the pattern for the relevant set of parentheses. This search went wrong
3337 when there were unescaped parentheses in a character class, parentheses
3338 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3339
3340 33. "Subroutine" calls and backreferences were previously restricted to
3341 referencing subpatterns earlier in the regex. This restriction has now
3342 been removed.
3343
3344 34. Added a number of extra features that are going to be in Perl 5.10. On the
3345 whole, these are just syntactic alternatives for features that PCRE had
3346 previously implemented using the Python syntax or my own invention. The
3347 other formats are all retained for compatibility.
3348
3349 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3350 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3351 also .NET compatible.
3352
3353 (b) A recursion or subroutine call to a named group can now be defined as
3354 (?&name) as well as (?P>name).
3355
3356 (c) A backreference to a named group can now be defined as \k<name> or
3357 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3358 5.10, are also .NET compatible.
3359
3360 (d) A conditional reference to a named group can now use the syntax
3361 (?(<name>) or (?('name') as well as (?(name).
3362
3363 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3364 groups (named and numbered) that are never evaluated inline, but can be
3365 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3366 is always false. There may be only one alternative in such a group.
3367
3368 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3369 as the simple (?(R). The condition is true only if the most recent
3370 recursion is that of the given number or name. It does not search out
3371 through the entire recursion stack.
3372
3373 (g) The escape \gN or \g{N} has been added, where N is a positive or
3374 negative number, specifying an absolute or relative reference.
3375
3376 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3377 some "unreachable code" warnings.
3378
3379 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3380 things, this adds five new scripts.
3381
3382 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3383 There were also incompatibilities regarding the handling of \Q..\E inside
3384 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3385 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3386
3387 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3388 matches an empty string, and forcibly breaks the loop. There were bugs in
3389 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3390 against aaaa the result was just "a" rather than "aaaa", for example. Two
3391 separate and independent bugs (that affected different cases) have been
3392 fixed.
3393
3394 39. Refactored the code to abolish the use of different opcodes for small
3395 capturing bracket numbers. This is a tidy that I avoided doing when I
3396 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3397 The new approach is not only tidier, it makes it possible to reduce the
3398 memory needed to fix the previous bug (38).
3399
3400 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3401 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3402 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3403 mode.
3404
3405 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3406 report.
3407
3408 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3409 copy construction and assignment in the C++ wrapper.
3410
3411 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3412 couple of bugs that could have given wrong results in the "--newline=crlf"
3413 case.
3414
3415 44. Added a number of casts and did some reorganization of signed/unsigned int
3416 variables following suggestions from Dair Grant. Also renamed the variable
3417 "this" as "item" because it is a C++ keyword.
3418
3419 45. Arranged for dftables to add
3420
3421 #include "pcre_internal.h"
3422
3423 to pcre_chartables.c because without it, gcc 4.x may remove the array
3424 definition from the final binary if PCRE is built into a static library and
3425 dead code stripping is activated.
3426
3427 46. For an unanchored pattern, if a match attempt fails at the start of a
3428 newline sequence, and the newline setting is CRLF or ANY, and the next two
3429 characters are CRLF, advance by two characters instead of one.
3430
3431
3432 Version 6.7 04-Jul-06
3433 ---------------------
3434
3435 1. In order to handle tests when input lines are enormously long, pcretest has
3436 been re-factored so that it automatically extends its buffers when
3437 necessary. The code is crude, but this _is_ just a test program. The
3438 default size has been increased from 32K to 50K.
3439
3440 2. The code in pcre_study() was using the value of the re argument before
3441 testing it for NULL. (Of course, in any sensible call of the function, it
3442 won't be NULL.)
3443
3444 3. The memmove() emulation function in pcre_internal.h, which is used on
3445 systems that lack both memmove() and bcopy() - that is, hardly ever -
3446 was missing a "static" storage class specifier.
3447
3448 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3449 containing an extended class (one that cannot be represented by a bitmap
3450 because it contains high-valued characters or Unicode property items, e.g.
3451 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3452 pattern, but PCRE should not loop if you do not (it no longer does).
3453 [Detail: two cases were found: (a) a repeated subpattern containing an
3454 extended class; (b) a recursive reference to a subpattern that followed a
3455 previous extended class. It wasn't skipping over the extended class
3456 correctly when UTF-8 mode was not set.]
3457
3458 5. A negated single-character class was not being recognized as fixed-length
3459 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3460 compile error "lookbehind assertion is not fixed length".
3461
3462 6. The RunPerlTest auxiliary script was showing an unexpected difference
3463 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3464 write a Perl script that can interpret lines of an input file either as
3465 byte characters or as UTF-8, which is what "perltest" was being required to
3466 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3467 can't do is switch easily at run time between having the "use utf8;" pragma
3468 or not. In the end, I fudged it by using the RunPerlTest script to insert
3469 "use utf8;" explicitly for the UTF-8 tests.
3470
3471 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3472 the end of the subject string, contrary to the documentation and to what
3473 Perl does. This was true of both matching functions. Now it matches only at
3474 the start of the subject and immediately after *internal* newlines.
3475
3476 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3477 a pointer to an int instead of a pointer to an unsigned long int. This
3478 caused problems on 64-bit systems.
3479
3480 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3481 instance of the 'standard' template library not being so standard".
3482
3483 10. There was no check on the number of named subpatterns nor the maximum
3484 length of a subpattern name. The product of these values is used to compute
3485 the size of the memory block for a compiled pattern. By supplying a very
3486 long subpattern name and a large number of named subpatterns, the size
3487 computation could be caused to overflow. This is now prevented by limiting
3488 the length of names to 32 characters, and the number of named subpatterns
3489 to 10,000.
3490
3491 11. Subpatterns that are repeated with specific counts have to be replicated in
3492 the compiled pattern. The size of memory for this was computed from the
3493 length of the subpattern and the repeat count. The latter is limited to
3494 65535, but there was no limit on the former, meaning that integer overflow
3495 could in principle occur. The compiled length of a repeated subpattern is
3496 now limited to 30,000 bytes in order to prevent this.
3497
3498 12. Added the optional facility to have named substrings with the same name.
3499
3500 13. Added the ability to use a named substring as a condition, using the
3501 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3502 are numbers (not recommended). Forward references are permitted.
3503
3504 14. Added forward references in named backreferences (if you see what I mean).
3505
3506 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3507 pattern could run off the end of the subject. For example, the pattern
3508 "(?s)(.{1,5})"8 did this with the subject "ab".
3509
3510 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3511 PCRE_CASELESS was set when matching characters that were quantified with ?
3512 or *.
3513
3514 17. A character class other than a single negated character that had a minimum
3515 but no maximum quantifier - for example [ab]{6,} - was not handled
3516 correctly by pce_dfa_exec(). It would match only one character.
3517
3518 18. A valid (though odd) pattern that looked like a POSIX character
3519 class but used an invalid character after [ (for example [[,abc,]]) caused
3520 pcre_compile() to give the error "Failed: internal error: code overflow" or
3521 in some cases to crash with a glibc free() error. This could even happen if
3522 the pattern terminated after [[ but there just happened to be a sequence of
3523 letters, a binary zero, and a closing ] in the memory that followed.
3524
3525 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3526 over the years. Originally (before any Unicode support), just the bottom 8
3527 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3528 output from "man perlunicode" includes this:
3529
3530 The regular expression compiler produces polymorphic opcodes. That
3531 is, the pattern adapts to the data and automatically switches to
3532 the Unicode character scheme when presented with Unicode data--or
3533 instead uses a traditional byte scheme when presented with byte
3534 data.
3535
3536 Sadly, a wide octal escape does not cause a switch, and in a string with
3537 no other multibyte characters, these octal escapes are treated as before.
3538 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3539 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3540 Unicode string.
3541
3542 I have not perpetrated such confusion in PCRE. Up till now, it took just
3543 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3544 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3545 translate to the appropriate multibyte character.
3546
3547 29. Applied some refactoring to reduce the number of warnings from Microsoft
3548 and Borland compilers. This has included removing the fudge introduced
3549 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3550 a warning about an unused variable.
3551
3552 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3553 characters since release 4.0, because Perl (from release 5.004) does not.
3554 [Or at least, is documented not to: some releases seem to be in conflict
3555 with the documentation.] However, when a pattern was studied with
3556 pcre_study() and all its branches started with \s, PCRE still included VT
3557 as a possible starting character. Of course, this did no harm; it just
3558 caused an unnecessary match attempt.
3559
3560 22. Removed a now-redundant internal flag bit that recorded the fact that case
3561 dependency changed within the pattern. This was once needed for "required
3562 byte" processing, but is no longer used. This recovers a now-scarce options
3563 bit. Also moved the least significant internal flag bit to the most-
3564 significant bit of the word, which was not previously used (hangover from
3565 the days when it was an int rather than a uint) to free up another bit for
3566 the future.
3567
3568 23. Added support for CRLF line endings as well as CR and LF. As well as the
3569 default being selectable at build time, it can now be changed at runtime
3570 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3571 specify that it is scanning data with non-default line endings.
3572
3573 24. Changed the definition of CXXLINK to make it agree with the definition of
3574 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3575
3576 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3577 recursions. This makes a big different to stack usage for some patterns.
3578
3579 26. If a subpattern containing a named recursion or subroutine reference such
3580 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3581 the space required for the compiled pattern went wrong and gave too small a
3582 value. Depending on the environment, this could lead to "Failed: internal
3583 error: code overflow at offset 49" or "glibc detected double free or
3584 corruption" errors.
3585
3586 27. Applied patches from Google (a) to support the new newline modes and (b) to
3587 advance over multibyte UTF-8 characters in GlobalReplace.
3588
3589 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3590 difference for some implementation of PCRE in some Windows version.
3591
3592 29. Added some extra testing facilities to pcretest:
3593
3594 \q<number> in a data line sets the "match limit" value
3595 \Q<number> in a data line sets the "match recursion limt" value
3596 -S <number> sets the stack size, where <number> is in megabytes
3597
3598 The -S option isn't available for Windows.
3599
3600
3601 Version 6.6 06-Feb-06
3602 ---------------------
3603
3604 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3605 in pcreposix.h. I have copied the definition from pcre.h.
3606
3607 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3608 because pcre.h is no longer a built file.
3609
3610 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3611 not normally included in the compiled code.
3612
3613
3614 Version 6.5 01-Feb-06
3615 ---------------------
3616
3617 1. When using the partial match feature with pcre_dfa_exec(), it was not
3618 anchoring the second and subsequent partial matches at the new starting
3619 point. This could lead to incorrect results. For example, with the pattern
3620 /1234/, partially matching against "123" and then "a4" gave a match.
3621
3622 2. Changes to pcregrep:
3623
3624 (a) All non-match returns from pcre_exec() were being treated as failures
3625 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3626 error message is output. Some extra information is given for the
3627 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3628 probably the only errors that are likely to be caused by users (by
3629 specifying a regex that has nested indefinite repeats, for instance).
3630 If there are more than 20 of these errors, pcregrep is abandoned.
3631
3632 (b) A binary zero was treated as data while matching, but terminated the
3633 output line if it was written out. This has been fixed: binary zeroes
3634 are now no different to any other data bytes.
3635
3636 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3637 used to set a locale for matching. The --locale=xxxx long option has
3638 been added (no short equivalent) to specify a locale explicitly on the
3639 pcregrep command, overriding the environment variables.
3640
3641 (d) When -B was used with -n, some line numbers in the output were one less
3642 than they should have been.
3643
3644 (e) Added the -o (--only-matching) option.
3645
3646 (f) If -A or -C was used with -c (count only), some lines of context were
3647 accidentally printed for the final match.
3648
3649 (g) Added the -H (--with-filename) option.
3650
3651 (h) The combination of options -rh failed to suppress file names for files
3652 that were found from directory arguments.
3653
3654 (i) Added the -D (--devices) and -d (--directories) options.
3655
3656 (j) Added the -F (--fixed-strings) option.
3657
3658 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3659
3660 (l) Added the --colo(u)r option.
3661
3662 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3663 is not present by default.
3664
3665 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3666 items such as (?R) or (?1), when the recursion could match a number of
3667 alternatives. If it matched one of the alternatives, but subsequently,
3668 outside the recursion, there was a failure, the code tried to back up into
3669 the recursion. However, because of the way PCRE is implemented, this is not
3670 possible, and the result was an incorrect result from the match.
3671
3672 In order to prevent this happening, the specification of recursion has
3673 been changed so that all such subpatterns are automatically treated as
3674 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3675
3676 4. I had overlooked the fact that, in some locales, there are characters for
3677 which isalpha() is true but neither isupper() nor islower() are true. In
3678 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3679 and ordfeminine) are like this. This affected the treatment of \w and \W
3680 when they appeared in character classes, but not when they appeared outside
3681 a character class. The bit map for "word" characters is now created
3682 separately from the results of isalnum() instead of just taking it from the
3683 upper, lower, and digit maps. (Plus the underscore character, of course.)
3684
3685 5. The above bug also affected the handling of POSIX character classes such as
3686 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3687 permanent tables. Instead, the bit maps for such a class were previously
3688 created as the appropriate unions of the upper, lower, and digit bitmaps.
3689 Now they are created by subtraction from the [[:word:]] class, which has
3690 its own bitmap.
3691
3692 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3693 It is created by subtracting the vertical space characters (\x09, \x0a,
3694 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3695 subtraction was done in the overall bitmap for a character class, meaning
3696 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3697 be recognized. This bug has been fixed.
3698
3699 7. Patches from the folks at Google:
3700
3701 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3702 real life, but is still worth protecting against".
3703
3704 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3705 regular expressions".
3706
3707 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3708 have it.
3709
3710 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3711 "configure" and the latter not, in order to fix a problem somebody had
3712 with compiling the Arg class on HP-UX.
3713
3714 (e) Improve the error-handling of the C++ wrapper a little bit.
3715
3716 (f) New tests for checking recursion limiting.
3717
3718 8. The pcre_memmove() function, which is used only if the environment does not
3719 have a standard memmove() function (and is therefore rarely compiled),
3720 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3721 returning a result (though PCRE never actually uses the result).
3722
3723 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3724 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3725 returned instead of calling malloc() with an overflowing number that would
3726 most likely cause subsequent chaos.
3727
3728 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3729
3730 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3731 with this option is matched, the nmatch and pmatch options of regexec() are
3732 ignored.
3733
3734 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3735 provided in case anyone wants to the the POSIX interface with UTF-8
3736 strings.
3737
3738 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3739 C++ linking (needed for some HP-UX environments).
3740
3741 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3742 (unused parameter) and in the pcre_printint() function (omitted "default"
3743 switch label when the default is to do nothing).
3744
3745 15. Added some code to make it possible, when PCRE is compiled as a C++
3746 library, to replace subject pointers for pcre_exec() with a smart pointer
3747 class, thus making it possible to process discontinuous strings.
3748
3749 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3750 much the same function. They were added by different people who were trying
3751 to make PCRE easy to compile on non-Unix systems. It has been suggested
3752 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3753 for compiling on Windows systems. I have therefore replaced it with
3754 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3755 defaults to "extern" for C or "extern C" for C++, which works fine on
3756 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3757 SCOPE with something explicit in config.h. In addition:
3758
3759 (a) pcreposix.h still had just "extern" instead of either of these macros;
3760 I have replaced it with PCRE_DATA_SCOPE.
3761
3762 (b) Functions such as _pcre_xclass(), which are internal to the library,
3763 but external in the C sense, all had PCRE_EXPORT in their definitions.
3764 This is apparently wrong for the Windows case, so I have removed it.
3765 (It makes no difference on Unix-like systems.)
3766
3767 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3768 of recursive calls to match(). This is different to MATCH_LIMIT because
3769 that limits the total number of calls to match(), not all of which increase
3770 the depth of recursion. Limiting the recursion depth limits the amount of
3771 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3772 when PCRE is compiled, and changed at run time. A patch from Google adds
3773 this functionality to the C++ interface.
3774
3775 18. Changes to the handling of Unicode character properties:
3776
3777 (a) Updated the table to Unicode 4.1.0.
3778
3779 (b) Recognize characters that are not in the table as "Cn" (undefined).
3780
3781 (c) I revised the way the table is implemented to a much improved format
3782 which includes recognition of ranges. It now supports the ranges that
3783 are defined in UnicodeData.txt, and it also amalgamates other
3784 characters into ranges. This has reduced the number of entries in the
3785 table from around 16,000 to around 3,000, thus reducing its size
3786 considerably. I realized I did not need to use a tree structure after
3787 all - a binary chop search is just as efficient. Having reduced the
3788 number of entries, I extended their size from 6 bytes to 8 bytes to
3789 allow for more data.
3790
3791 (d) Added support for Unicode script names via properties such as \p{Han}.
3792
3793 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3794 matching that character.
3795
3796 20. When matching a repeated Unicode property with a minimum greater than zero,
3797 (for example \pL{2,}), PCRE could look past the end of the subject if it
3798 reached it while seeking the minimum number of characters. This could
3799 happen only if some of the characters were more than one byte long, because
3800 there is a check for at least the minimum number of bytes.
3801
3802 21. Refactored the implementation of \p and \P so as to be more general, to
3803 allow for more different types of property in future. This has changed the
3804 compiled form incompatibly. Anybody with saved compiled patterns that use
3805 \p or \P will have to recompile them.
3806
3807 22. Added "Any" and "L&" to the supported property types.
3808
3809 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3810 but give a compile time error if the value is greater than 0xff.
3811
3812 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3813 accidentally not being installed or uninstalled.
3814
3815 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3816 made were to insert the current release number. This seemed silly, because
3817 it made things harder for people building PCRE on systems that don't run
3818 "configure". I have turned pcre.h into a distributed file, no longer built
3819 by "configure", with the version identification directly included. There is
3820 no longer a pcre.h.in file.
3821
3822 However, this change necessitated a change to the pcre-config script as
3823 well. It is built from pcre-config.in, and one of the substitutions was the
3824 release number. I have updated configure.ac so that ./configure now finds
3825 the release number by grepping pcre.h.
3826
3827 26. Added the ability to run the tests under valgrind.
3828
3829
3830 Version 6.4 05-Sep-05
3831 ---------------------
3832
3833 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3834 "--" to be printed when multiple files were scanned, even when none of the
3835 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3836 consider it to be a bug, and have restored the previous behaviour.
3837
3838 2. A couple of code tidies to get rid of compiler warnings.
3839
3840 3. The pcretest program used to cheat by referring to symbols in the library
3841 whose names begin with _pcre_. These are internal symbols that are not
3842 really supposed to be visible externally, and in some environments it is
3843 possible to suppress them. The cheating is now confined to including
3844 certain files from the library's source, which is a bit cleaner.
3845
3846 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3847 file's purpose clearer.
3848
3849 5. Reorganized pcre_ucp_findchar().
3850
3851
3852 Version 6.3 15-Aug-05
3853 ---------------------
3854
3855 1. The file libpcre.pc.in did not have general read permission in the tarball.
3856
3857 2. There were some problems when building without C++ support:
3858
3859 (a) If C++ support was not built, "make install" and "make test" still
3860 tried to test it.
3861
3862 (b) There were problems when the value of CXX was explicitly set. Some
3863 changes have been made to try to fix these, and ...
3864
3865 (c) --disable-cpp can now be used to explicitly disable C++ support.
3866
3867 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3868 backslash in a target when C++ was disabled. This confuses some
3869 versions of "make", apparently. Using an intermediate variable solves
3870 this. (Same for CPP_LOBJ.)
3871
3872 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3873 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3874 necessary on certain architectures.
3875
3876 4. Added a setting of -export-symbols-regex to the link command to remove
3877 those symbols that are exported in the C sense, but actually are local
3878 within the library, and not documented. Their names all begin with
3879 "_pcre_". This is not a perfect job, because (a) we have to except some
3880 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3881 available (and never for static libraries). I have made a note to try to
3882 find a way round (a) in the future.
3883
3884
3885 Version 6.2 01-Aug-05
3886 ---------------------
3887
3888 1. There was no test for integer overflow of quantifier values. A construction
3889 such as {1111111111111111} would give undefined results. What is worse, if
3890 a minimum quantifier for a parenthesized subpattern overflowed and became
3891 negative, the calculation of the memory size went wrong. This could have
3892 led to memory overwriting.
3893
3894 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3895
3896 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3897 operating environments where this matters.
3898
3899 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3900 PCRE options from within the C++ wrapper.
3901
3902 5. Named capturing subpatterns were not being correctly counted when a pattern
3903 was compiled. This caused two problems: (a) If there were more than 100
3904 such subpatterns, the calculation of the memory needed for the whole
3905 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3906 back references of the form \12, where the number was greater than 9, were
3907 not recognized as back references, even though there were sufficient
3908 previous subpatterns.
3909
3910 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3911 versions of gcc, e.g. 2.95.4.
3912
3913
3914 Version 6.1 21-Jun-05
3915 ---------------------
3916
3917 1. There was one reference to the variable "posix" in pcretest.c that was not
3918 surrounded by "#if !defined NOPOSIX".
3919
3920 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3921 the cross-check on the old pcre_info() function, for the benefit of the
3922 cut-down version of PCRE that is currently imported into Exim.
3923
3924 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3925 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3926 patterns that start (?i) but I don't think that matters. The use of (?i) is
3927 just an example; this all applies to the other options as well.
3928
3929 4. Since libtool seems to echo the compile commands it is issuing, the output
3930 from "make" can be reduced a bit by putting "@" in front of each libtool
3931 compile command.
3932
3933 5. Patch from the folks at Google for configure.in to be a bit more thorough
3934 in checking for a suitable C++ installation before trying to compile the
3935 C++ stuff. This should fix a reported problem when a compiler was present,
3936 but no suitable headers.
3937
3938 6. The man pages all had just "PCRE" as their title. I have changed them to
3939 be the relevant file name. I have also arranged that these names are
3940 retained in the file doc/pcre.txt, which is a concatenation in text format
3941 of all the man pages except the little individual ones for each function.
3942
3943 7. The NON-UNIX-USE file had not been updated for the different set of source
3944 files that come with release 6. I also added a few comments about the C++
3945 wrapper.
3946
3947
3948 Version 6.0 07-Jun-05
3949 ---------------------
3950
3951 1. Some minor internal re-organization to help with my DFA experiments.
3952
3953 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3954 didn't matter for the library itself when fully configured, but did matter
3955 when compiling without UCP support, or within Exim, where the ucp files are
3956 not imported.
3957
3958 3. Refactoring of the library code to split up the various functions into
3959 different source modules. The addition of the new DFA matching code (see
3960 below) to a single monolithic source would have made it really too
3961 unwieldy, quite apart from causing all the code to be include in a
3962 statically linked application, when only some functions are used. This is
3963 relevant even without the DFA addition now that patterns can be compiled in
3964 one application and matched in another.
3965
3966 The downside of splitting up is that there have to be some external
3967 functions and data tables that are used internally in different modules of
3968 the library but which are not part of the API. These have all had their
3969 names changed to start with "_pcre_" so that they are unlikely to clash
3970 with other external names.
3971
3972 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3973 a different (DFA) algorithm. Although it is slower than the original
3974 function, it does have some advantages for certain types of matching
3975 problem.
3976
3977 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3978 including restarting after a partial match.
3979
3980 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3981 defined when compiling for Windows was sent to me. I have put it into the
3982 code, though I have no means of testing or verifying it.
3983
3984 7. Added the pcre_refcount() auxiliary function.
3985
3986 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3987 match before or at the first newline in the subject string. In pcretest,
3988 the /f option on a pattern can be used to set this.
3989
3990 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3991 would behave wrongly. This has been present in PCRE since release 4.0.
3992
3993 10. A number of changes to the pcregrep command:
3994
3995 (a) Refactored how -x works; insert ^(...)$ instead of setting
3996 PCRE_ANCHORED and checking the length, in preparation for adding
3997 something similar for -w.
3998
3999 (b) Added the -w (match as a word) option.
4000
4001 (c) Refactored the way lines are read and buffered so as to have more
4002 than one at a time available.
4003
4004 (d) Implemented a pcregrep test script.
4005
4006 (e) Added the -M (multiline match) option. This allows patterns to match
4007 over several lines of the subject. The buffering ensures that at least
4008 8K, or the rest of the document (whichever is the shorter) is available
4009 for matching (and similarly the previous 8K for lookbehind assertions).
4010
4011 (f) Changed the --help output so that it now says
4012
4013 -w, --word-regex(p)
4014
4015 instead of two lines, one with "regex" and the other with "regexp"
4016 because that confused at least one person since the short forms are the
4017 same. (This required a bit of code, as the output is generated
4018 automatically from a table. It wasn't just a text change.)
4019
4020 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4021 option but starts with a hyphen. Could be a pattern or a path name
4022 starting with a hyphen, for instance.
4023
4024 (h) "-" can be given as a file name to represent stdin.
4025
4026 (i) When file names are being printed, "(standard input)" is used for
4027 the standard input, for compatibility with GNU grep. Previously
4028 "<stdin>" was used.
4029
4030 (j) The option --label=xxx can be used to supply a name to be used for
4031 stdin when file names are being printed. There is no short form.
4032
4033 (k) Re-factored the options decoding logic because we are going to add
4034 two more options that take data. Such options can now be given in four
4035 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4036
4037 (l) Added the -A, -B, and -C options for requesting that lines of context
4038 around matches be printed.
4039
4040 (m) Added the -L option to print the names of files that do not contain
4041 any matching lines, that is, the complement of -l.
4042
4043 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4044 continue to scan other files.
4045
4046 (o) The -s option was incorrectly implemented. For compatibility with other
4047 greps, it now suppresses the error message for a non-existent or non-
4048 accessible file (but not the return code). There is a new option called
4049 -q that suppresses the output of matching lines, which was what -s was
4050 previously doing.
4051
4052 (p) Added --include and --exclude options to specify files for inclusion
4053 and exclusion when recursing.
4054
4055 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4056 Hopefully, it now does.
4057
4058 12. Missing cast in pcre_study().
4059
4060 13. Added an "uninstall" target to the makefile.
4061
4062 14. Replaced "extern" in the function prototypes in Makefile.in with
4063 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4064 world, but is set differently for Windows.
4065
4066 15. Added a second compiling function called pcre_compile2(). The only
4067 difference is that it has an extra argument, which is a pointer to an
4068 integer error code. When there is a compile-time failure, this is set
4069 non-zero, in addition to the error test pointer being set to point to an
4070 error message. The new argument may be NULL if no error number is required
4071 (but then you may as well call pcre_compile(), which is now just a
4072 wrapper). This facility is provided because some applications need a
4073 numeric error indication, but it has also enabled me to tidy up the way
4074 compile-time errors are handled in the POSIX wrapper.
4075
4076 16. Added VPATH=.libs to the makefile; this should help when building with one
4077 prefix path and installing with another. (Or so I'm told by someone who
4078 knows more about this stuff than I do.)
4079
4080 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4081 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4082 match everything, including newlines. This is not POSIX-compatible, but
4083 somebody wanted the feature. From pcretest it can be activated by using
4084 both the P and the s flags.
4085
4086 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4087
4088 19. libpcre.pc was being incorrectly installed as executable.
4089
4090 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4091 it now also looks for '\r' so that it will work unmodified on Windows.
4092
4093 21. Added Google's contributed C++ wrapper to the distribution.
4094
4095 22. Added some untidy missing memory free() calls in pcretest, to keep
4096 Electric Fence happy when testing.
4097
4098
4099
4100 Version 5.0 13-Sep-04
4101 ---------------------
4102
4103 1. Internal change: literal characters are no longer packed up into items
4104 containing multiple characters in a single byte-string. Each character
4105 is now matched using a separate opcode. However, there may be more than one
4106 byte in the character in UTF-8 mode.
4107
4108 2. The pcre_callout_block structure has two new fields: pattern_position and
4109 next_item_length. These contain the offset in the pattern to the next match
4110 item, and its length, respectively.
4111
4112 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4113 insertion of callouts before each pattern item. Added the /C option to
4114 pcretest to make use of this.
4115
4116 4. On the advice of a Windows user, the lines
4117
4118 #if defined(_WIN32) || defined(WIN32)
4119 _setmode( _fileno( stdout ), 0x8000 );
4120 #endif /* defined(_WIN32) || defined(WIN32) */
4121
4122 have been added to the source of pcretest. This apparently does useful
4123 magic in relation to line terminators.
4124
4125 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4126 for the benefit of those environments where the "b" makes a difference.
4127
4128 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4129 to know about it. I have put a hack into configure.in that adds in code
4130 to set GCC=yes if CC=icc. This seems to end up at a point in the
4131 generated configure script that is early enough to affect the setting of
4132 compiler options, which is what is needed, but I have no means of testing
4133 whether it really works. (The user who reported this had patched the
4134 generated configure script, which of course I cannot do.)
4135
4136 LATER: After change 22 below (new libtool files), the configure script
4137 seems to know about icc (and also ecc). Therefore, I have commented out
4138 this hack in configure.in.
4139
4140 7. Added support for pkg-config (2 patches were sent in).
4141
4142 8. Negated POSIX character classes that used a combination of internal tables
4143 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4144 [[:^ascii]]. Typically, they would match almost any characters. The other
4145 POSIX classes were not broken in this way.
4146
4147 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4148 to find the match, as PCRE was deluded into thinking that the match had to
4149 start at the start point or following a newline. The same bug applied to
4150 patterns with negative forward assertions or any backward assertions
4151 preceding ".*" at the start, unless the pattern required a fixed first
4152 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4153
4154 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4155 starting at the last subject character, bytes beyond the end of the subject
4156 string were read.
4157
4158 11. Renamed the variable "class" as "classbits" to make life easier for C++
4159 users. (Previously there was a macro definition, but it apparently wasn't
4160 enough.)
4161
4162 12. Added the new field "tables" to the extra data so that tables can be passed
4163 in at exec time, or the internal tables can be re-selected. This allows
4164 a compiled regex to be saved and re-used at a later time by a different
4165 program that might have everything at different addresses.
4166
4167 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4168 -R library as well as a -L library.
4169
4170 14. The debugging options of pcretest (-d on the command line or D on a
4171 pattern) showed incorrect output for anything following an extended class
4172 that contained multibyte characters and which was followed by a quantifier.
4173
4174 15. Added optional support for general category Unicode character properties
4175 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4176 support. It adds about 90K to the size of the library. The meanings of the
4177 inbuilt class escapes such as \d and \s have NOT been changed.
4178
4179 16. Updated pcredemo.c to include calls to free() to release the memory for the
4180 compiled pattern.
4181
4182 17. The generated file chartables.c was being created in the source directory
4183 instead of in the building directory. This caused the build to fail if the
4184 source directory was different from the building directory, and was
4185 read-only.
4186
4187 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4188 file. No doubt somebody will tell me if they don't make sense... Also added
4189 Dan Mooney's comments about building on OpenVMS.
4190
4191 19. Added support for partial matching via the PCRE_PARTIAL option for
4192 pcre_exec() and the \P data escape in pcretest.
4193
4194 20. Extended pcretest with 3 new pattern features:
4195
4196 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4197 write the compiled pattern to the file whose name is "rest-of-line".
4198 This is a straight binary dump of the data, with the saved pointer to
4199 the character tables forced to be NULL. The study data, if any, is
4200 written too. After writing, pcretest reads a new pattern.
4201
4202 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4203 compiled pattern from the given file. There must not be any
4204 occurrences of "<" in the file name (pretty unlikely); if there are,
4205 pcretest will instead treat the initial "<" as a pattern delimiter.
4206 After reading in the pattern, pcretest goes on to read data lines as
4207 usual.
4208
4209 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4210 and 16-bit fields in a compiled pattern, to simulate a pattern that
4211 was compiled on a host of opposite endianness.
4212
4213 21. The pcre-exec() function can now cope with patterns that were compiled on
4214 hosts of opposite endianness, with this restriction:
4215
4216 As for any compiled expression that is saved and used later, the tables
4217 pointer field cannot be preserved; the extra_data field in the arguments
4218 to pcre_exec() should be used to pass in a tables address if a value
4219 other than the default internal tables were used at compile time.
4220
4221 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4222 now diagnosed as an error. Previously, most of the time, a negative number
4223 would have been treated as zero, but if in addition "ovector" was passed as
4224 NULL, a crash could occur.
4225
4226 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4227 new versions from the libtool 1.5 distribution (the last one is a copy of
4228 a file called libtool.m4). This seems to have fixed the need to patch
4229 "configure" to support Darwin 1.3 (which I used to do). However, I still
4230 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4231 workstation).
4232
4233 24. Changed the PCRE licence to be the more standard "BSD" licence.
4234
4235
4236 Version 4.5 01-Dec-03
4237 ---------------------
4238
4239 1. There has been some re-arrangement of the code for the match() function so
4240 that it can be compiled in a version that does not call itself recursively.
4241 Instead, it keeps those local variables that need separate instances for
4242 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4243 needs to "recurse". Keeping track of where control must go is done by means
4244 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4245 hide most of the details from the main code, and operates only if
4246 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4247 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4248 operating.
4249
4250 To make it easier for callers to provide specially tailored get/free
4251 functions for this usage, two new functions, pcre_stack_malloc, and
4252 pcre_stack_free, are used. They are always called in strict stacking order,
4253 and the size of block requested is always the same.
4254
4255 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4256 PCRE has been compiled to use the stack or the heap for recursion. The
4257 -C option of pcretest uses this to show which version is compiled.
4258
4259 A new data escape \S, is added to pcretest; it causes the amounts of store
4260 obtained and freed by both kinds of malloc/free at match time to be added
4261 to the output.
4262
4263 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4264 what's available on my current Linux desktop machine.
4265
4266 3. When matching a UTF-8 string, the test for a valid string at the start has
4267 been extended. If start_offset is not zero, PCRE now checks that it points
4268 to a byte that is the start of a UTF-8 character. If not, it returns
4269 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4270 this is necessary because there may be backward assertions in the pattern.
4271 When matching the same subject several times, it may save resources to use
4272 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4273
4274 4. The code for checking the validity of UTF-8 strings has been tightened so
4275 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4276 containing "overlong sequences".
4277
4278 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4279 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4280 should have been "&", but it just so happened that all the cases this let
4281 through by mistake were picked up later in the function.
4282
4283 6. I had used a variable called "isblank" - this is a C99 function, causing
4284 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4285
4286 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4287 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4288 so the version is shown; (c) stop "make test" echoing "./RunTest".
4289
4290 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4291
4292 9. The prototype for memmove() for systems that don't have it was using
4293 size_t, but the inclusion of the header that defines size_t was later. I've
4294 moved the #includes for the C headers earlier to avoid this.
4295
4296 10. Added some adjustments to the code to make it easier to compiler on certain
4297 special systems:
4298
4299 (a) Some "const" qualifiers were missing.
4300 (b) Added the macro EXPORT before all exported functions; by default this
4301 is defined to be empty.
4302 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4303 that it reads its output file name as an argument instead of writing
4304 to the standard output and assuming this can be redirected.
4305
4306 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4307 class containing characters with values greater than 255, PCRE compilation
4308 went into a loop.
4309
4310 12. A recursive reference to a subpattern that was within another subpattern
4311 that had a minimum quantifier of zero caused PCRE to crash. For example,
4312 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4313 recursion. If the recursively-called subpattern itself had a zero repeat,
4314 that was OK.
4315
4316 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4317 buffer into which it was copied (for escape processing) was still set at
4318 1024, so long lines caused crashes.
4319
4320 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4321 "internal error: code overflow...". This applied to any character class
4322 that was followed by a possessive quantifier.
4323
4324 15. Modified the Makefile to add libpcre.la as a prerequisite for
4325 libpcreposix.la because I was told this is needed for a parallel build to
4326 work.
4327
4328 16. If a pattern that contained .* following optional items at the start was
4329 studied, the wrong optimizing data was generated, leading to matching
4330 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4331 matching string must start with a or b or c. The correct conclusion for
4332 this pattern is that a match can start with any character.
4333
4334
4335 Version 4.4 13-Aug-03
4336 ---------------------
4337
4338 1. In UTF-8 mode, a character class containing characters with values between
4339 127 and 255 was not handled correctly if the compiled pattern was studied.
4340 In fixing this, I have also improved the studying algorithm for such
4341 classes (slightly).
4342
4343 2. Three internal functions had redundant arguments passed to them. Removal
4344 might give a very teeny performance improvement.
4345
4346 3. Documentation bug: the value of the capture_top field in a callout is *one
4347 more than* the number of the hightest numbered captured substring.
4348
4349 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4350 in incorrectly linking with a previously installed version. They now link
4351 explicitly with libpcre.la.
4352
4353 5. configure.in no longer needs to recognize Cygwin specially.
4354
4355 6. A problem in pcre.in for Windows platforms is fixed.
4356
4357 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4358 pcretest, it used to include the size of the study block as part of its
4359 output. Unfortunately, the structure contains a field that has a different
4360 size on different hardware architectures. This meant that the tests that
4361 showed this size failed. As the block is currently always of a fixed size,
4362 this information isn't actually particularly useful in pcretest output, so
4363 I have just removed it.
4364
4365 8. Three pre-processor statements accidentally did not start in column 1.
4366 Sadly, there are *still* compilers around that complain, even though
4367 standard C has not required this for well over a decade. Sigh.
4368
4369 9. In pcretest, the code for checking callouts passed small integers in the
4370 callout_data field, which is a void * field. However, some picky compilers
4371 complained about the casts involved for this on 64-bit systems. Now
4372 pcretest passes the address of the small integer instead, which should get
4373 rid of the warnings.
4374
4375 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4376 both compile and run time, and gives an error if an invalid UTF-8 sequence
4377 is found. There is a option for disabling this check in cases where the
4378 string is known to be correct and/or the maximum performance is wanted.
4379
4380 11. In response to a bug report, I changed one line in Makefile.in from
4381
4382 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4383 to
4384 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4385
4386 to look similar to other lines, but I have no way of telling whether this
4387 is the right thing to do, as I do not use Windows. No doubt I'll get told
4388 if it's wrong...
4389
4390
4391 Version 4.3 21-May-03
4392 ---------------------
4393
4394 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4395 Makefile.
4396
4397 2. Some refactoring to improve the quality of the code:
4398
4399 (i) The utf8_table... variables are now declared "const".
4400
4401 (ii) The code for \cx, which used the "case flipping" table to upper case
4402 lower case letters, now just substracts 32. This is ASCII-specific,
4403 but the whole concept of \cx is ASCII-specific, so it seems
4404 reasonable.
4405
4406 (iii) PCRE was using its character types table to recognize decimal and
4407 hexadecimal digits in the pattern. This is silly, because it handles
4408 only 0-9, a-f, and A-F, but the character types table is locale-
4409 specific, which means strange things might happen. A private
4410 table is now used for this - though it costs 256 bytes, a table is
4411 much faster than multiple explicit tests. Of course, the standard
4412 character types table is still used for matching digits in subject
4413 strings against \d.
4414
4415 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4416 ending in _t are). So I've renamed it as ESC_tee.
4417
4418 3. The first argument for regexec() in the POSIX wrapper should have been
4419 defined as "const".
4420
4421 4. Changed pcretest to use malloc() for its buffers so that they can be
4422 Electric Fenced for debugging.
4423
4424 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4425 to read one or more bytes before the start of the subject string. Often this
4426 had no effect on PCRE's behaviour, but in some circumstances it could
4427 provoke a segmentation fault.
4428
4429 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4430 to try to read one or more bytes before the start of the subject string.
4431
4432 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4433 UTF-8 support could misbehave in various ways if the subject string
4434 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4435 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4436 back over UTF-8 characters.)
4437
4438
4439 Version 4.2 14-Apr-03
4440 ---------------------
4441
4442 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4443
4444 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4445 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4446 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4447 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4448 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4449 and BUILD_EXEEXT
4450 Note: automatic setting of the BUILD variables is not yet working
4451 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4452 compile-time but not at link-time
4453 [LINK]: use for linking executables only
4454 make different versions for Windows and non-Windows
4455 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4456 libraries
4457 [LINK_FOR_BUILD]: new variable
4458 [OBJEXT]: use throughout
4459 [EXEEXT]: use throughout
4460 <winshared>: new target
4461 <wininstall>: new target
4462 <dftables.o>: use native compiler
4463 <dftables>: use native linker
4464 <install>: handle Windows platform correctly
4465 <clean>: ditto
4466 <check>: ditto
4467 copy DLL to top builddir before testing
4468
4469 As part of these changes, -no-undefined was removed again. This was reported
4470 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4471 in any case.
4472
4473 3. Some tidies to get rid of compiler warnings:
4474
4475 . In the match_data structure, match_limit was an unsigned long int, whereas
4476 match_call_count was an int. I've made them both unsigned long ints.
4477
4478 . In pcretest the fact that a const uschar * doesn't automatically cast to
4479 a void * provoked a warning.
4480
4481 . Turning on some more compiler warnings threw up some "shadow" variables
4482 and a few more missing casts.
4483
4484 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4485 option, a class that contained a single character with a value between 128
4486 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4487
4488 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4489 option, a class that contained several characters, but with at least one
4490 whose value was between 128 and 255 caused PCRE to crash.
4491
4492
4493 Version 4.1 12-Mar-03
4494 ---------------------
4495
4496 1. Compiling with gcc -pedantic found a couple of places where casts were
4497 needed, and a string in dftables.c that was longer than standard compilers are
4498 required to support.
4499
4500 2. Compiling with Sun's compiler found a few more places where the code could
4501 be tidied up in order to avoid warnings.
4502
4503 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4504 first of these names is deprecated in the latest Autoconf in favour of the name
4505 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4506 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4507 analogy I have changed it to CFLAGS_FOR_BUILD.
4508
4509 4. Added -no-undefined to the linking command in the Makefile, because this is
4510 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4511 linking step for the pcreposix library.
4512
4513 5. PCRE was failing to diagnose the case of two named groups with the same
4514 name.
4515
4516 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4517 literal character that is needed in the subject for a match, and scans along to
4518 ensure that it is present before embarking on the full matching process. This
4519 saves time in cases of nested unlimited repeats that are never going to match.
4520 Problem: the scan can take a lot of time if the subject is very long (e.g.
4521 megabytes), thus penalizing straightforward matches. It is now done only if the
4522 amount of subject to be scanned is less than 1000 bytes.
4523
4524 7. A lesser problem with the same optimization is that it was recording the
4525 first character of an anchored pattern as "needed", thus provoking a search
4526 right along the subject, even when the first match of the pattern was going to
4527 fail. The "needed" character is now not set for anchored patterns, unless it
4528 follows something in the pattern that is of non-fixed length. Thus, it still
4529 fulfils its original purpose of finding quick non-matches in cases of nested
4530 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4531
4532
4533 Version 4.0 17-Feb-03
4534 ---------------------
4535
4536 1. If a comment in an extended regex that started immediately after a meta-item
4537 extended to the end of string, PCRE compiled incorrect data. This could lead to
4538 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4539
4540 2. Moved to autoconf 2.53 and libtool 1.4.2.
4541
4542 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4543 the special perltest8 script is no longer needed - all the tests can be run
4544 from a single perltest script.
4545
4546 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4547 by \s. It has now been removed in PCRE. This means it isn't recognized as
4548 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4549 class [:space:] *does* include VT, thereby creating a mess.
4550
4551 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4552 space and tab.
4553
4554 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4555 its new features into the main test script, reducing the number of scripts.
4556
4557 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4558 were backward compatible, and made the (?i) apply to the whole pattern, as if
4559 /i were given. Now it behaves more logically, and applies the option setting
4560 only to what follows. PCRE has been changed to follow suit. However, if it
4561 finds options settings right at the start of the pattern, it extracts them into
4562 the global options, as before. Thus, they show up in the info data.
4563
4564 8. Added support for the \Q...\E escape sequence. Characters in between are
4565 treated as literals. This is slightly different from Perl in that $ and @ are
4566 also handled as literals inside the quotes. In Perl, they will cause variable
4567 interpolation. Note the following examples:
4568
4569 Pattern PCRE matches Perl matches
4570
4571 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4572 \Qabc\$xyz\E abc\$xyz abc\$xyz
4573 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4574
4575 For compatibility with Perl, \Q...\E sequences are recognized inside character
4576 classes as well as outside them.
4577
4578 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4579 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4580 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4581 signed/unsigned warnings.
4582
4583 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4584 option for pcretest, so I've replaced it by a simple function that does just
4585 that job.
4586
4587 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4588 "pcregrep -".
4589
4590 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4591 Java package. This provides some syntactic sugar for simple cases of what my
4592 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4593 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4594 item, you can use this simplified notation. Note that only makes sense with
4595 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4596 greediness, whatever the setting of the PCRE_UNGREEDY option.
4597
4598 13. A change of greediness default within a pattern was not taking effect at
4599 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4600 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4601 was abstracted outside.
4602
4603 14. PCRE now supports the \G assertion. It is true when the current matching
4604 position is at the start point of the match. This differs from \A when the
4605 starting offset is non-zero. Used with the /g option of pcretest (or similar
4606 code), it works in the same way as it does for Perl's /g option. If all
4607 alternatives of a regex begin with \G, the expression is anchored to the start
4608 match position, and the "anchored" flag is set in the compiled expression.
4609
4610 15. Some bugs concerning the handling of certain option changes within patterns
4611 have been fixed. These applied to options other than (?ims). For example,
4612 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4613 been the other way round. Some of this was related to change 7 above.
4614
4615 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4616 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4617 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4618 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4619
4620 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4621 mode. Unlike ".", it always matches newline, whatever the setting of
4622 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4623 assertions. Perl allows it, but it doesn't (in general) work because it can't
4624 calculate the length of the lookbehind. At least, that's the case for Perl
4625 5.8.0 - I've been told they are going to document that it doesn't work in
4626 future.
4627
4628 18. Added an error diagnosis for escapes that PCRE does not support: these are
4629 \L, \l, \N, \P, \p, \U, \u, and \X.
4630
4631 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4632 reading past the end of the pattern in cases such as /[abcd/.
4633
4634 20. PCRE was getting more memory than necessary for patterns with classes that
4635 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4636
4637 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4638 compiling PCRE for use with Virtual Pascal.
4639
4640 22. Small fix to the Makefile to make it work properly if the build is done
4641 outside the source tree.
4642
4643 23. Added a new extension: a condition to go with recursion. If a conditional
4644 subpattern starts with (?(R) the "true" branch is used if recursion has
4645 happened, whereas the "false" branch is used only at the top level.
4646
4647 24. When there was a very long string of literal characters (over 255 bytes
4648 without UTF support, over 250 bytes with UTF support), the computation of how
4649 much memory was required could be incorrect, leading to segfaults or other
4650 strange effects.
4651
4652 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4653 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4654 there was a subsequent back reference to those brackets. This meant that, for
4655 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4656 possible to check for precisely this case. All we can do is abandon the
4657 optimization if .* occurs inside capturing brackets when there are any back
4658 references whatsoever. (See below for a better fix that came later.)
4659
4660 26. The handling of the optimization for finding the first character of a
4661 non-anchored pattern, and for finding a character that is required later in the
4662 match were failing in some cases. This didn't break the matching; it just
4663 failed to optimize when it could. The way this is done has been re-implemented.
4664
4665 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4666
4667 28. Added a new feature that provides some of the functionality that Perl
4668 provides with (?{...}). The facility is termed a "callout". The way it is done
4669 in PCRE is for the caller to provide an optional function, by setting
4670 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4671 global variable. By default it is unset, which disables all calling out. To get
4672 the function called, the regex must include (?C) at appropriate points. This
4673 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4674 This provides a means of identifying different callout points. When PCRE
4675 reaches such a point in the regex, if pcre_callout has been set, the external
4676 function is called. It is provided with data in a structure called
4677 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4678 matching continues; if it returns a non-zero value, the match at the current
4679 point fails. However, backtracking will occur if possible. [This was changed
4680 later and other features added - see item 49 below.]
4681
4682 29. pcretest is upgraded to test the callout functionality. It provides a
4683 callout function that displays information. By default, it shows the start of
4684 the match and the current position in the text. There are some new data escapes
4685 to vary what happens:
4686
4687 \C+ in addition, show current contents of captured substrings
4688 \C- do not supply a callout function
4689 \C!n return 1 when callout number n is reached
4690 \C!n!m return 1 when callout number n is reached for the mth time
4691
4692 30. If pcregrep was called with the -l option and just a single file name, it
4693 output "<stdin>" if a match was found, instead of the file name.
4694
4695 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4696 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4697 pcre_exec(). This saves a malloc/free per call. The default value of
4698 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4699 when configuring.
4700
4701 32. The default maximum size of a compiled pattern is 64K. There have been a
4702 few cases of people hitting this limit. The code now uses macros to handle the
4703 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4704 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4705 configuring. Tests 2 and 5 work only with 2-byte links because they output
4706 debugging information about compiled patterns.
4707
4708 33. Internal code re-arrangements:
4709
4710 (a) Moved the debugging function for printing out a compiled regex into
4711 its own source file (printint.c) and used #include to pull it into
4712 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4713 separate copies.
4714
4715 (b) Defined the list of op-code names for debugging as a macro in
4716 internal.h so that it is next to the definition of the opcodes.
4717
4718 (c) Defined a table of op-code lengths for simpler skipping along compiled
4719 code. This is again a macro in internal.h so that it is next to the
4720 definition of the opcodes.
4721
4722 34. Added support for recursive calls to individual subpatterns, along the
4723 lines of Robin Houston's patch (but implemented somewhat differently).
4724
4725 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4726 allow it to read and process whole directories in Win32. This code was
4727 contributed by Lionel Fourquaux; it has not been tested by me.
4728
4729 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4730 used to name a group. Names consist of alphanumerics and underscores, and must
4731 be unique. Back references use the syntax (?P=name) and recursive calls use
4732 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4733 numbers. The function pcre_fullinfo() can be used after compilation to extract
4734 a name/number map. There are three relevant calls:
4735
4736 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4737 PCRE_INFO_NAMECOUNT yields the number of entries
4738 PCRE_INFO_NAMETABLE yields a pointer to the map.
4739
4740 The map is a vector of fixed-size entries. The size of each entry depends on
4741 the length of the longest name used. The first two bytes of each entry are the
4742 group number, most significant byte first. There follows the corresponding
4743 name, zero terminated. The names are in alphabetical order.
4744
4745 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4746 case instead of 255. Making it the same both with and without UTF-8 support
4747 means that the same test output works with both.
4748
4749 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4750 calling malloc() with a zero argument.
4751
4752 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4753 optimization. I've improved things by keeping a bitmap of backreferences with
4754 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4755 fact referenced, the optimization can be applied. It is unlikely that a
4756 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4757 the match to follow \n) will appear inside brackets with a number greater than
4758 31, but if it does, any back reference > 31 suppresses the optimization.
4759
4760 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4761 of disabling numbered capturing parentheses. Any opening parenthesis that is
4762 not followed by ? behaves as if it were followed by ?: but named parentheses
4763 can still be used for capturing (and they will acquire numbers in the usual
4764 way).
4765
4766 41. Redesigned the return codes from the match() function into yes/no/error so
4767 that errors can be passed back from deep inside the nested calls. A malloc
4768 failure while inside a recursive subpattern call now causes the
4769 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4770
4771 42. It is now possible to set a limit on the number of times the match()
4772 function is called in a call to pcre_exec(). This facility makes it possible to
4773 limit the amount of recursion and backtracking, though not in a directly
4774 obvious way, because the match() function is used in a number of different
4775 circumstances. The count starts from zero for each position in the subject
4776 string (for non-anchored patterns). The default limit is, for compatibility, a
4777 large number, namely 10 000 000. You can change this in two ways:
4778
4779 (a) When configuring PCRE before making, you can use --with-match-limit=n
4780 to set a default value for the compiled library.
4781
4782 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4783 a different value is set. See 45 below.
4784
4785 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4786
4787 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4788 of things that can be changed at compile time. The first argument specifies
4789 what is wanted and the second points to where the information is to be placed.
4790 The current list of available information is:
4791
4792 PCRE_CONFIG_UTF8
4793
4794 The output is an integer that is set to one if UTF-8 support is available;
4795 otherwise it is set to zero.
4796
4797 PCRE_CONFIG_NEWLINE
4798
4799 The output is an integer that it set to the value of the code that is used for
4800 newline. It is either LF (10) or CR (13).
4801
4802 PCRE_CONFIG_LINK_SIZE
4803
4804 The output is an integer that contains the number of bytes used for internal
4805 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4806
4807 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4808
4809 The output is an integer that contains the threshold above which the POSIX
4810 interface uses malloc() for output vectors. See item 31 above.
4811
4812 PCRE_CONFIG_MATCH_LIMIT
4813
4814 The output is an unsigned integer that contains the default limit of the number
4815 of match() calls in a pcre_exec() execution. See 42 above.
4816
4817 44. pcretest has been upgraded by the addition of the -C option. This causes it
4818 to extract all the available output from the new pcre_config() function, and to
4819 output it. The program then exits immediately.
4820
4821 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4822 order to support additional features. One way would have been to define
4823 pcre_exec2() (for example) with extra arguments, but this would not have been
4824 extensible, and would also have required all calls to the original function to
4825 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4826 is used for passing in "extra" data from pcre_study().
4827
4828 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4829 contains the following fields:
4830
4831 flags a bitmap indicating which of the following fields are set
4832 study_data opaque data from pcre_study()
4833 match_limit a way of specifying a limit on match() calls for a specific
4834 call to pcre_exec()
4835 callout_data data for callouts (see 49 below)
4836
4837 The flag bits are also defined in pcre.h, and are
4838
4839 PCRE_EXTRA_STUDY_DATA
4840 PCRE_EXTRA_MATCH_LIMIT
4841 PCRE_EXTRA_CALLOUT_DATA
4842
4843 The pcre_study() function now returns one of these new pcre_extra blocks, with
4844 the actual study data pointed to by the study_data field, and the
4845 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4846 before. That is, this change is entirely upwards-compatible and requires no
4847 change to existing code.
4848
4849 If you want to pass in additional data to pcre_exec(), you can either place it
4850 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4851 block.
4852
4853 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4854 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4855 times with different match limits, until it finds the minimum value needed for
4856 pcre_exec() to complete. The value is then output. This can be instructive; for
4857 most simple matches the number is quite small, but for pathological cases it
4858 gets very large very quickly.
4859
4860 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4861 returns the size of the data block pointed to by the study_data field in a
4862 pcre_extra block, that is, the value that was passed as the argument to
4863 pcre_malloc() when PCRE was getting memory in which to place the information
4864 created by pcre_study(). The fourth argument should point to a size_t variable.
4865 pcretest has been extended so that this information is shown after a successful
4866 pcre_study() call when information about the compiled regex is being displayed.
4867
4868 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4869 because what follows is always an absolute path. (Later: it turns out that this
4870 is more than cosmetic for MinGW, because it doesn't like empty path
4871 components.)
4872
4873 49. Some changes have been made to the callout feature (see 28 above):
4874
4875 (i) A callout function now has three choices for what it returns:
4876
4877 0 => success, carry on matching
4878 > 0 => failure at this point, but backtrack if possible
4879 < 0 => serious error, return this value from pcre_exec()
4880
4881 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4882 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4883 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4884 use by callout functions. It will never be used by PCRE itself.
4885
4886 (ii) The pcre_extra structure (see 45 above) has a void * field called
4887 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4888 pcre_callout_block structure has a field of the same name. The contents of
4889 the field passed in the pcre_extra structure are passed to the callout
4890 function in the corresponding field in the callout block. This makes it
4891 easier to use the same callout-containing regex from multiple threads. For
4892 testing, the pcretest program has a new data escape
4893
4894 \C*n pass the number n (may be negative) as callout_data
4895
4896 If the callout function in pcretest receives a non-zero value as
4897 callout_data, it returns that value.
4898
4899 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4900 there were some redundant $(CFLAGS) in commands that are now specified as
4901 $(LINK), which already includes $(CFLAGS).
4902
4903 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4904 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4905 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4906 one-byte characters throughout. Note that case-insensitive matching applies
4907 only to characters whose values are less than 256. PCRE doesn't support the
4908 notion of cases for higher-valued characters.
4909
4910 (i) A character class whose characters are all within 0-255 is handled as
4911 a bit map, and the map is inverted for negative classes. Previously, a
4912 character > 255 always failed to match such a class; however it should
4913 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4914
4915 (ii) A negated character class with a single character < 255 is coded as
4916 "not this character" (OP_NOT). This wasn't working properly when the test
4917 character was multibyte, either singly or repeated.
4918
4919 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4920 mode, for example: \x{100}{2,3}.
4921
4922 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4923 singly or repeated) now correctly test multibyte characters. However,
4924 PCRE doesn't recognize any characters with values greater than 255 as
4925 digits, spaces, or word characters. Such characters always match \D, \S,
4926 and \W, and never match \d, \s, or \w.
4927
4928 (v) Classes may now contain characters and character ranges with values
4929 greater than 255. For example: [ab\x{100}-\x{400}].
4930
4931 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4932 PCRE in UTF-8 mode.
4933
4934 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4935 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4936 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4937 value.)
4938
4939 53. The single man page has become too large. I have therefore split it up into
4940 a number of separate man pages. These also give rise to individual HTML pages;
4941 these are now put in a separate directory, and there is an index.html page that
4942 lists them all. Some hyperlinking between the pages has been installed.
4943
4944 54. Added convenience functions for handling named capturing parentheses.
4945
4946 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4947 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4948 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4949 are faulted.
4950
4951 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4952 calling configure. These values are used when compiling the dftables.c program
4953 which is run to generate the source of the default character tables. They
4954 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4955 you will need to set these values.
4956
4957 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4958
4959
4960 Version 3.9 02-Jan-02
4961 ---------------------
4962
4963 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4964
4965 2. If --disable-static was given, the building process failed when trying to
4966 build pcretest and pcregrep. (For some reason it was using libtool to compile
4967 them, which is not right, as they aren't part of the library.)
4968
4969
4970 Version 3.8 18-Dec-01
4971 ---------------------
4972
4973 1. The experimental UTF-8 code was completely screwed up. It was packing the
4974 bytes in the wrong order. How dumb can you get?
4975
4976
4977 Version 3.7 29-Oct-01
4978 ---------------------
4979
4980 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4981 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4982 this didn't happen under Solaris 8, where I normally test things.
4983
4984 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4985 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4986 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4987 this fix an hour or so after the initial 3.7 release.)
4988
4989
4990 Version 3.6 23-Oct-01
4991 ---------------------
4992
4993 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4994 offsets passed as NULL with zero offset count.
4995
4996 2. The config.guess and config.sub files had not been updated when I moved to
4997 the latest autoconf.
4998
4999
5000 Version 3.5 15-Aug-01
5001 ---------------------
5002
5003 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5004 had been forgotten.
5005
5006 2. By using declared but undefined structures, we can avoid using "void"
5007 definitions in pcre.h while keeping the internal definitions of the structures
5008 private.
5009
5010 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5011 user point of view, this means that both static and shared libraries are built
5012 by default, but this can be individually controlled. More of the work of
5013 handling this static/shared cases is now inside libtool instead of PCRE's make
5014 file.
5015
5016 4. The pcretest utility is now installed along with pcregrep because it is
5017 useful for users (to test regexs) and by doing this, it automatically gets
5018 relinked by libtool. The documentation has been turned into a man page, so
5019 there are now .1, .txt, and .html versions in /doc.
5020
5021 5. Upgrades to pcregrep:
5022 (i) Added long-form option names like gnu grep.
5023 (ii) Added --help to list all options with an explanatory phrase.
5024 (iii) Added -r, --recursive to recurse into sub-directories.
5025 (iv) Added -f, --file to read patterns from a file.
5026
5027 6. pcre_exec() was referring to its "code" argument before testing that
5028 argument for NULL (and giving an error if it was NULL).
5029
5030 7. Upgraded Makefile.in to allow for compiling in a different directory from
5031 the source directory.
5032
5033 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5034 options bits, the pointer it was passed was to an int instead of to an unsigned
5035 long int. This mattered only on 64-bit systems.
5036
5037 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5038 generated) instead of pcre.in, which it its source. Also made the same change
5039 in several of the .c files.
5040
5041 10. A new release of gcc defines printf() as a macro, which broke pcretest
5042 because it had an ifdef in the middle of a string argument for printf(). Fixed
5043 by using separate calls to printf().
5044
5045 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5046 script, to force use of CR or LF instead of \n in the source. On non-Unix
5047 systems, the value can be set in config.h.
5048
5049 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5050 absolute limit. Changed the text of the error message to make this clear, and
5051 likewise updated the man page.
5052
5053 13. The limit of 99 on the number of capturing subpatterns has been removed.
5054 The new limit is 65535, which I hope will not be a "real" limit.
5055
5056
5057 Version 3.4 22-Aug-00
5058 ---------------------
5059
5060 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5061
5062 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5063
5064
5065 Version 3.3 01-Aug-00
5066 ---------------------
5067
5068 1. If an octal character was given, but the value was greater than \377, it
5069 was not getting masked to the least significant bits, as documented. This could
5070 lead to crashes in some systems.
5071
5072 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5073 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5074
5075 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5076 These just pass their arguments on to (pcre_free)(), but they are provided
5077 because some uses of PCRE bind it to non-C systems that can call its functions,
5078 but cannot call free() or pcre_free() directly.
5079
5080 4. Add "make test" as a synonym for "make check". Corrected some comments in
5081 the Makefile.
5082
5083 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5084 Makefile.
5085
5086 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5087 command called pgrep for grepping around the active processes.
5088
5089 7. Added the beginnings of support for UTF-8 character strings.
5090
5091 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5092 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5093 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5094 out for the ar command.)
5095
5096
5097 Version 3.2 12-May-00
5098 ---------------------
5099
5100 This is purely a bug fixing release.
5101
5102 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5103 of ZA. This was just one example of several cases that could provoke this bug,
5104 which was introduced by change 9 of version 2.00. The code for breaking
5105 infinite loops after an iteration that matches an empty string was't working
5106 correctly.
5107
5108 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5109 when matched against abbab (for example). After matching an empty string, it
5110 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5111 caused it to match further down the string than it should.
5112
5113 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5114 was there because it doesn't seem to be needed, and it causes trouble on some
5115 systems, as it is not a Standard C header. It has been removed.
5116
5117 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5118 were reported on the Macintosh. The changes were from
5119
5120 while ((c = *(++ptr)) != 0 && c != '\n');
5121 to
5122 while ((c = *(++ptr)) != 0 && c != '\n') ;
5123
5124 Totally extraordinary, but if that's what it takes...
5125
5126 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5127 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5128 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5129 assumes the way PCRE uses memmove() (always moving upwards).
5130
5131 6. PCRE is being used in one environment where strchr() is not available. There
5132 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5133 faster code anyway.
5134
5135
5136 Version 3.1 09-Feb-00
5137 ---------------------
5138
5139 The only change in this release is the fixing of some bugs in Makefile.in for
5140 the "install" target:
5141
5142 (1) It was failing to install pcreposix.h.
5143
5144 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5145
5146
5147 Version 3.0 01-Feb-00
5148 ---------------------
5149
5150 1. Add support for the /+ modifier to perltest (to output $` like it does in
5151 pcretest).
5152
5153 2. Add support for the /g modifier to perltest.
5154
5155 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5156 matches null strings.
5157
5158 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5159 pattern. Perl treats empty patterns specially - it reuses the most recent
5160 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5161 effect.
5162
5163 5. The POSIX interface was broken in that it was just handing over the POSIX
5164 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5165 required a bigger vector, with some working space on the end. This means that
5166 the POSIX wrapper now has to get and free some memory, and copy the results.
5167
5168 6. Added some simple autoconf support, placing the test data and the
5169 documentation in separate directories, re-organizing some of the
5170 information files, and making it build pcre-config (a GNU standard). Also added
5171 libtool support for building PCRE as a shared library, which is now the
5172 default.
5173
5174 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5175 09 are not valid octal constants. Single digits will be used for minor values
5176 less than 10.
5177
5178 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5179 existing programs that set these in the POSIX interface can use PCRE without
5180 modification.
5181
5182 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5183 return all that pcre_info() returns, plus additional data. The pcre_info()
5184 function is retained for compatibility, but is considered to be obsolete.
5185
5186 10. Added experimental recursion feature (?R) to handle one common case that
5187 Perl 5.6 will be able to do with (?p{...}).
5188
5189 11. Added support for POSIX character classes like [:alpha:], which Perl is
5190 adopting.
5191
5192
5193 Version 2.08 31-Aug-99
5194 ----------------------
5195
5196 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5197 trying to match at the startoffset position, but instead was moving forward to
5198 the next newline as if a previous match had failed.
5199
5200 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5201 and could get into a loop if a null string was matched other than at the start
5202 of the subject.
5203
5204 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5205 be distinguished at compile time, and for completeness also added PCRE_DATE.
5206
5207 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5208 in GnuWin32 environments.
5209
5210
5211 Version 2.07 29-Jul-99
5212 ----------------------
5213
5214 1. The documentation is now supplied in plain text form and HTML as well as in
5215 the form of man page sources.
5216
5217 2. C++ compilers don't like assigning (void *) values to other pointer types.
5218 In particular this affects malloc(). Although there is no problem in Standard
5219 C, I've put in casts to keep C++ compilers happy.
5220
5221 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5222 should be (const char *).
5223
5224 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5225 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5226 However, I haven't made this a standard facility. The documentation doesn't
5227 mention it, and the Makefile doesn't support it.
5228
5229 5. The Makefile now contains an "install" target, with editable destinations at
5230 the top of the file. The pcretest program is not installed.
5231
5232 6. pgrep -V now gives the PCRE version number and date.
5233
5234 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5235 causing the entire string to be ignored, instead of just the last character.
5236
5237 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5238 non-matching string, it can take a very, very long time, even for strings of
5239 quite modest length, because of the nested recursion. PCRE now does better in
5240 some of these cases. It does this by remembering the last required literal
5241 character in the pattern, and pre-searching the subject to ensure it is present
5242 before running the real match. In other words, it applies a heuristic to detect
5243 some types of certain failure quickly, and in the above example, if presented
5244 with a string that has no trailing " it gives "no match" very quickly.
5245
5246 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5247 other alternatives are tried instead.
5248
5249
5250 Version 2.06 09-Jun-99
5251 ----------------------
5252
5253 1. Change pcretest's output for amount of store used to show just the code
5254 space, because the remainder (the data block) varies in size between 32-bit and
5255 64-bit systems.
5256
5257 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5258 start matching at. This allows lookbehinds to work when searching for multiple
5259 occurrences in a string.
5260
5261 3. Added additional options to pcretest for testing multiple occurrences:
5262
5263 /+ outputs the rest of the string that follows a match
5264 /g loops for multiple occurrences, using the new startoffset argument
5265 /G loops for multiple occurrences by passing an incremented pointer
5266
5267 4. PCRE wasn't doing the "first character" optimization for patterns starting
5268 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5269 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5270 the letter 'x'. On long subject strings, this gives a significant speed-up.
5271
5272
5273 Version 2.05 21-Apr-99
5274 ----------------------
5275
5276 1. Changed the type of magic_number from int to long int so that it works
5277 properly on 16-bit systems.
5278
5279 2. Fixed a bug which caused patterns starting with .* not to work correctly
5280 when the subject string contained newline characters. PCRE was assuming
5281 anchoring for such patterns in all cases, which is not correct because .* will
5282 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5283 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5284 must be retried after every newline in the subject.
5285
5286
5287 Version 2.04 18-Feb-99
5288 ----------------------
5289
5290 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5291 computation of the store needed to hold the pattern was incorrect (too large).
5292 If such patterns were nested a few deep, this could multiply and become a real
5293 problem.
5294
5295 2. Added /M option to pcretest to show the memory requirement of a specific
5296 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5297
5298 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5299 compiled in such a way that the backtracking after subsequent failure was
5300 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5301 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5302
5303
5304 Version 2.03 02-Feb-99
5305 ----------------------
5306
5307 1. Fixed typo and small mistake in man page.
5308
5309 2. Added 4th condition (GPL supersedes if conflict) and created separate
5310 LICENCE file containing the conditions.
5311
5312 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5313 Perl, that is the internal \ allows the delimiter to be included in the
5314 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5315 the final delimiter, add \ to the end of the pattern (to test the error).
5316
5317 4. Added the convenience functions for extracting substrings after a successful
5318 match. Updated pcretest to make it able to test these functions.
5319
5320
5321 Version 2.02 14-Jan-99
5322 ----------------------
5323
5324 1. Initialized the working variables associated with each extraction so that
5325 their saving and restoring doesn't refer to uninitialized store.
5326
5327 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5328 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5329 fix the problem.
5330
5331 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5332 calls, and wasn't printing the correct value for compiling calls. Increased the
5333 default value of LOOPREPEAT, and the number of significant figures in the
5334 times.
5335
5336 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5337
5338 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5339 a building problem on Windows NT with a FAT file system.
5340
5341
5342 Version 2.01 21-Oct-98
5343 ----------------------
5344
5345 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5346 to character tables built by pcre_maketables() in the current locale. If NULL
5347 is passed, the default tables are used.
5348
5349
5350 Version 2.00 24-Sep-98
5351 ----------------------
5352
5353 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5354 it any more.
5355
5356 2. Allow quantification of (?>) groups, and make it work correctly.
5357
5358 3. The first character computation wasn't working for (?>) groups.
5359
5360 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5361 end of the subject) and add 5.005's \z, which really does match only at the
5362 very end of the subject.
5363
5364 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5365
5366 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5367 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5368 localized options. All options to pcre_study() were also removed.
5369
5370 7. Add other new features from 5.005:
5371
5372 $(?<= positive lookbehind
5373 $(?<! negative lookbehind
5374 (?imsx-imsx) added the unsetting capability
5375 such a setting is global if at outer level; local otherwise
5376 (?imsx-imsx:) non-capturing groups with option setting
5377 (?(cond)re|re) conditional pattern matching
5378
5379 A backreference to itself in a repeated group matches the previous
5380 captured string.
5381
5382 8. General tidying up of studying (both automatic and via "study")
5383 consequential on the addition of new assertions.
5384
5385 9. As in 5.005, unlimited repeated groups that could match an empty substring
5386 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5387 runtime if any iteration does actually match an empty substring.
5388
5389 10. Include the RunTest script in the distribution.
5390
5391 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5392 discrepancies, some of which were old and were also with respect to 5.004. They
5393 have now been fixed.
5394
5395
5396 Version 1.09 28-Apr-98
5397 ----------------------
5398
5399 1. A negated single character class followed by a quantifier with a minimum
5400 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5401 program crashes, or just wrong answers. This did not apply to negated classes
5402 containing more than one character, or to minima other than one.
5403
5404
5405 Version 1.08 27-Mar-98
5406 ----------------------
5407
5408 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5409
5410 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5411 latter must appear before anything that relies on it in the pattern.
5412
5413
5414 Version 1.07 16-Feb-98
5415 ----------------------
5416
5417 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5418 repeat of a potentially empty string).
5419
5420
5421 Version 1.06 23-Jan-98
5422 ----------------------
5423
5424 1. Added Markus Oberhumer's little patches for C++.
5425
5426 2. Literal strings longer than 255 characters were broken.
5427
5428
5429 Version 1.05 23-Dec-97
5430 ----------------------
5431
5432 1. Negated character classes containing more than one character were failing if
5433 PCRE_CASELESS was set at run time.
5434
5435
5436 Version 1.04 19-Dec-97
5437 ----------------------
5438
5439 1. Corrected the man page, where some "const" qualifiers had been omitted.
5440
5441 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5442 input syntax.
5443
5444 3. Fixed memory leak which occurred when a regex with back references was
5445 matched with an offsets vector that wasn't big enough. The temporary memory
5446 that is used in this case wasn't being freed if the match failed.
5447
5448 4. Tidied pcretest to ensure it frees memory that it gets.
5449
5450 5. Temporary memory was being obtained in the case where the passed offsets
5451 vector was exactly big enough.
5452
5453 6. Corrected definition of offsetof() from change 5 below.
5454
5455 7. I had screwed up change 6 below and broken the rules for the use of
5456 setjmp(). Now fixed.
5457
5458
5459 Version 1.03 18-Dec-97
5460 ----------------------
5461
5462 1. A erroneous regex with a missing opening parenthesis was correctly
5463 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5464 on some systems.
5465
5466 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5467 it was reported that one broken compiler failed on the former because "code" is
5468 also an independent variable.
5469
5470 3. The erroneous regex a[]b caused an array overrun reference.
5471
5472 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5473 fail on data ending with that character. (It was going on too far, and checking
5474 the next character, typically a binary zero.) This was specific to the
5475 optimized code for single-character negative classes.
5476
5477 5. Added a contributed patch from the TIN world which does the following:
5478
5479 + Add an undef for memmove, in case the the system defines a macro for it.
5480
5481 + Add a definition of offsetof(), in case there isn't one. (I don't know
5482 the reason behind this - offsetof() is part of the ANSI standard - but
5483 it does no harm).
5484
5485 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5486 most of the places where whitespace preceded '#'. I have given up and
5487 allowed the remaining 2 cases to be at the margin.
5488
5489 + Rename some variables in pcre to eliminate shadowing. This seems very
5490 pedantic, but does no harm, of course.
5491
5492 6. Moved the call to setjmp() into its own function, to get rid of warnings
5493 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5494
5495 7. Constructs such as \d{8,} were compiling into the equivalent of
5496 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5497 outcome, but in this particular case used more store than had been allocated,
5498 which caused the bug to be discovered because it threw up an internal error.
5499
5500 8. The debugging code in both pcre and pcretest for outputting the compiled
5501 form of a regex was going wrong in the case of back references followed by
5502 curly-bracketed repeats.
5503
5504
5505 Version 1.02 12-Dec-97
5506 ----------------------
5507
5508 1. Typos in pcre.3 and comments in the source fixed.
5509
5510 2. Applied a contributed patch to get rid of places where it used to remove
5511 'const' from variables, and fixed some signed/unsigned and uninitialized
5512 variable warnings.
5513
5514 3. Added the "runtest" target to Makefile.
5515
5516 4. Set default compiler flag to -O2 rather than just -O.
5517
5518
5519 Version 1.01 19-Nov-97
5520 ----------------------
5521
5522 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5523 like /([ab]*)*/, that is, for classes with more than one character in them.
5524
5525 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5526 as /((?>a*))*/ (a PCRE_EXTRA facility).
5527
5528
5529 Version 1.00 18-Nov-97
5530 ----------------------
5531
5532 1. Added compile-time macros to support systems such as SunOS4 which don't have
5533 memmove() or strerror() but have other things that can be used instead.
5534
5535 2. Arranged that "make clean" removes the executables.
5536
5537
5538 Version 0.99 27-Oct-97
5539 ----------------------
5540
5541 1. Fixed bug in code for optimizing classes with only one character. It was
5542 initializing a 32-byte map regardless, which could cause it to run off the end
5543 of the memory it had got.
5544
5545 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5546
5547
5548 Version 0.98 22-Oct-97
5549 ----------------------
5550
5551 1. Fixed bug in code for handling temporary memory usage when there are more
5552 back references than supplied space in the ovector. This could cause segfaults.
5553
5554
5555 Version 0.97 21-Oct-97
5556 ----------------------
5557
5558 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5559
5560 2. Optimized negated single characters not to use a bit map.
5561
5562 3. Brought error texts together as macro definitions; clarified some of them;
5563 fixed one that was wrong - it said "range out of order" when it meant "invalid
5564 escape sequence".
5565
5566 4. Changed some char * arguments to const char *.
5567
5568 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5569
5570 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5571 pcretest.
5572
5573
5574 Version 0.96 16-Oct-97
5575 ----------------------
5576
5577 1. Added a simple "pgrep" utility to the distribution.
5578
5579 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5580 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5581 where "ddd" means "one or more decimal digits".
5582
5583 3. Fixed serious bug. If a pattern had a back reference, but the call to
5584 pcre_exec() didn't supply a large enough ovector to record the related
5585 identifying subpattern, the match always failed. PCRE now remembers the number
5586 of the largest back reference, and gets some temporary memory in which to save
5587 the offsets during matching if necessary, in order to ensure that
5588 backreferences always work.
5589
5590 4. Increased the compatibility with Perl in a number of ways:
5591
5592 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5593 to request this handling. The option can be set at compile or exec time.
5594
5595 (b) $ matches before a terminating newline by default; an option
5596 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5597 mode). The option can be set at compile or exec time.
5598
5599 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5600 the same as Perl's. If the decimal number it represents is less than 10
5601 or there aren't that many previous left capturing parentheses, an octal
5602 escape is read. Inside a character class, it's always an octal escape,
5603 even if it is a single digit.
5604
5605 (d) An escaped but undefined alphabetic character is taken as a literal,
5606 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5607 escapes.
5608
5609 (e) {0} is now permitted. (The previous item is removed from the compiled
5610 pattern).
5611
5612 5. Changed all the names of code files so that the basic parts are no longer
5613 than 10 characters, and abolished the teeny "globals.c" file.
5614
5615 6. Changed the handling of character classes; they are now done with a 32-byte
5616 bit map always.
5617
5618 7. Added the -d and /D options to pcretest to make it possible to look at the
5619 internals of compilation without having to recompile pcre.
5620
5621
5622 Version 0.95 23-Sep-97
5623 ----------------------
5624
5625 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5626 \x20 at the start of a run of normal characters. These were being treated as
5627 real characters, instead of the source characters being re-checked.
5628
5629
5630 Version 0.94 18-Sep-97
5631 ----------------------
5632
5633 1. The functions are now thread-safe, with the caveat that the global variables
5634 containing pointers to malloc() and free() or alternative functions are the
5635 same for all threads.
5636
5637 2. Get pcre_study() to generate a bitmap of initial characters for non-
5638 anchored patterns when this is possible, and use it if passed to pcre_exec().
5639
5640
5641 Version 0.93 15-Sep-97
5642 ----------------------
5643
5644 1. /(b)|(:+)/ was computing an incorrect first character.
5645
5646 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5647 but not actually doing anything yet.
5648
5649 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5650 as Perl does (e.g. [-az] or [az-]).
5651
5652 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5653 all possible positions.
5654
5655 5. Split up into different modules to avoid including unneeded functions in a
5656 compiled binary. However, compile and exec are still in one module. The "study"
5657 function is split off.
5658
5659 6. The character tables are now in a separate module whose source is generated
5660 by an auxiliary program - but can then be edited by hand if required. There are
5661 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5662 toupper() in the code.
5663
5664 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5665 make them global. Abolish the function for setting them, as the caller can now
5666 set them directly.
5667
5668
5669 Version 0.92 11-Sep-97
5670 ----------------------
5671
5672 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5673 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5674
5675 2. Caseless matching was not working in character classes if the characters in
5676 the pattern were in upper case.
5677
5678 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5679
5680 4. Make PCRE_ANCHORED public and accept as a compile option.
5681
5682 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5683 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5684 pass them.
5685
5686 6. Give an error if bad option bits passed at compile or run time.
5687
5688 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5689 pcretest to cause it to pass that flag.
5690
5691 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5692 options, and the first character, if set.
5693
5694 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5695
5696
5697 Version 0.91 10-Sep-97
5698 ----------------------
5699
5700 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5701 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5702
5703 2. PCRE was looping on encountering an indefinitely repeated back reference to
5704 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5705 Perl does - treats the match as successful.
5706
5707 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5