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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 612 - (show annotations) (download)
Sat Jul 2 15:20:59 2011 UTC (3 years, 2 months ago) by ph10
File size: 204337 byte(s)
Fix two study bugs concerned with minimum subject lengths; add features to 
pcretest so that all tests can be run with or without study; adjust tests so 
that this happens.

1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.13 30-Apr-2011
5 ------------------------
6
7 1. The Unicode data tables have been updated to Unicode 6.0.0.
8
9 2. Two minor typos in pcre_internal.h have been fixed.
10
11 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
12 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
13 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
14
15 4. There were a number of related bugs in the code for matching backrefences
16 caselessly in UTF-8 mode when codes for the characters concerned were
17 different numbers of bytes. For example, U+023A and U+2C65 are an upper
18 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
19 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
20 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
21 2-byte code at the end of the subject (it thought there wasn't enough data
22 left).
23
24 5. Comprehensive information about what went wrong is now returned by
25 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
26 as the output vector has at least 2 elements. The offset of the start of
27 the failing character and a reason code are placed in the vector.
28
29 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
30 now returned is for the first byte of the failing character, instead of the
31 last byte inspected. This is an incompatible change, but I hope it is small
32 enough not to be a problem. It makes the returned offset consistent with
33 pcre_exec() and pcre_dfa_exec().
34
35 7. pcretest now gives a text phrase as well as the error number when
36 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
37 failure, the offset and reason code are output.
38
39 8. When \R was used with a maximizing quantifier it failed to skip backwards
40 over a \r\n pair if the subsequent match failed. Instead, it just skipped
41 back over a single character (\n). This seems wrong (because it treated the
42 two characters as a single entity when going forwards), conflicts with the
43 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
44 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
45 has been changed.
46
47 9. Some internal refactoring has changed the processing so that the handling
48 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
49 time (the PCRE_DOTALL option was changed this way some time ago: version
50 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
51 which was always a bit of a fudge. It also means that there is one less
52 argument for the match() function, which reduces its stack requirements
53 slightly. This change also fixes an incompatibility with Perl: the pattern
54 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
55
56 10. More internal refactoring has drastically reduced the number of recursive
57 calls to match() for possessively repeated groups such as (abc)++ when
58 using pcre_exec().
59
60 11. While implementing 10, a number of bugs in the handling of groups were
61 discovered and fixed:
62
63 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
64 (a|)*(?1) gave a compile-time internal error.
65 ((a|)+)+ did not notice that the outer group could match an empty string.
66 (^a|^)+ was not marked as anchored.
67 (.*a|.*)+ was not marked as matching at start or after a newline.
68
69 12. Yet more internal refactoring has removed another argument from the match()
70 function. Special calls to this function are now indicated by setting a
71 value in a variable in the "match data" data block.
72
73 13. Be more explicit in pcre_study() instead of relying on "default" for
74 opcodes that mean there is no starting character; this means that when new
75 ones are added and accidentally left out of pcre_study(), testing should
76 pick them up.
77
78 14. The -s option of pcretest has been documented for ages as being an old
79 synonym of -m (show memory usage). I have changed it to mean "force study
80 for every regex", that is, assume /S for every regex. This is similar to -i
81 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
82 using it. It makes it easier to run collections of tests with and without
83 study enabled, and thereby test pcre_study() more easily. All the standard
84 tests are now run with and without -s (but some patterns can be marked as
85 "never study" - see 20 below).
86
87 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
88 restoration of the capturing data to the outer values was not happening
89 correctly.
90
91 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
92 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
93 pattern had matched an empty string, and so incorrectly returned a no
94 match.
95
96 17. There was optimizing code for the last branch of non-capturing parentheses,
97 and also for the obeyed branch of a conditional subexpression, which used
98 tail recursion to cut down on stack usage. Unfortunately, not that there is
99 the possibility of (*THEN) occurring in these branches, tail recursion is
100 no longer possible because the return has to be checked for (*THEN). These
101 two optimizations have therefore been removed.
102
103 18. If a pattern containing \R was studied, it was assumed that \R always
104 matched two bytes, thus causing the minimum subject length to be
105 incorrectly computed because \R can also match just one byte.
106
107 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
108 was incorrectly computed.
109
110 20. If /S is present twice on a test pattern in pcretest input, it *disables*
111 studying, thereby overriding the use of -s on the command line. This is
112 necessary for one or two tests to keep the output identical in both cases.
113
114
115 Version 8.12 15-Jan-2011
116 ------------------------
117
118 1. Fixed some typos in the markup of the man pages, and wrote a script that
119 checks for such things as part of the documentation building process.
120
121 2. On a big-endian 64-bit system, pcregrep did not correctly process the
122 --match-limit and --recursion-limit options (added for 8.11). In
123 particular, this made one of the standard tests fail. (The integer value
124 went into the wrong half of a long int.)
125
126 3. If the --colour option was given to pcregrep with -v (invert match), it
127 did strange things, either producing crazy output, or crashing. It should,
128 of course, ignore a request for colour when reporting lines that do not
129 match.
130
131 4. Another pcregrep bug caused similar problems if --colour was specified with
132 -M (multiline) and the pattern match finished with a line ending.
133
134 5. In pcregrep, when a pattern that ended with a literal newline sequence was
135 matched in multiline mode, the following line was shown as part of the
136 match. This seems wrong, so I have changed it.
137
138 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
139 the check for further matches in the same line (so they could be coloured)
140 to overrun the end of the current line. If another match was found, it was
141 incorrectly shown (and then shown again when found in the next line).
142
143 7. If pcregrep was compiled under Windows, there was a reference to the
144 function pcregrep_exit() before it was defined. I am assuming this was
145 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
146 reported by a user. I've moved the definition above the reference.
147
148
149 Version 8.11 10-Dec-2010
150 ------------------------
151
152 1. (*THEN) was not working properly if there were untried alternatives prior
153 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
154 backtracked to try for "b" instead of moving to the next alternative branch
155 at the same level (in this case, to look for "c"). The Perl documentation
156 is clear that when (*THEN) is backtracked onto, it goes to the "next
157 alternative in the innermost enclosing group".
158
159 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
160 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
161 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
162 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
163 (*THEN).
164
165 3. If \s appeared in a character class, it removed the VT character from
166 the class, even if it had been included by some previous item, for example
167 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
168 of \s, but is part of the POSIX "space" class.)
169
170 4. A partial match never returns an empty string (because you can always
171 match an empty string at the end of the subject); however the checking for
172 an empty string was starting at the "start of match" point. This has been
173 changed to the "earliest inspected character" point, because the returned
174 data for a partial match starts at this character. This means that, for
175 example, /(?<=abc)def/ gives a partial match for the subject "abc"
176 (previously it gave "no match").
177
178 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
179 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
180 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
181 has an implication that the given string is incomplete (because a partial
182 match is preferred over a full match). For this reason, these items now
183 give a partial match in this situation. [Aside: previously, the one case
184 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
185 match rather than a full match, which was wrong by the old rules, but is
186 now correct.]
187
188 6. There was a bug in the handling of #-introduced comments, recognized when
189 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
190 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
191 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
192 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
193 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
194 places in pcre_compile().
195
196 7. Related to (6) above, when pcre_compile() was skipping #-introduced
197 comments when looking ahead for named forward references to subpatterns,
198 the only newline sequence it recognized was NL. It now handles newlines
199 according to the set newline convention.
200
201 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
202 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
203 cater for a lack of strerror(). These oversights have been fixed.
204
205 9. Added --match-limit and --recursion-limit to pcregrep.
206
207 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
208
209 11. When the -o option was used, pcregrep was setting a return code of 1, even
210 when matches were found, and --line-buffered was not being honoured.
211
212 12. Added an optional parentheses number to the -o and --only-matching options
213 of pcregrep.
214
215 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
216 can match an empty string. The code to do it in pcretest and pcredemo
217 needed fixing:
218
219 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
220 only one byte after an empty string match just before CRLF (this case
221 just got forgotten; "any" and "anycrlf" were OK).
222
223 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
224 mode when an empty string match preceded an ASCII character followed by
225 a non-ASCII character. (The code for advancing by one character rather
226 than one byte was nonsense.)
227
228 (c) The pcredemo.c sample program did not have any code at all to handle
229 the cases when CRLF is a valid newline sequence.
230
231 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
232 as a starting offset was within the subject string. There is now a new
233 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
234 negative or greater than the length of the string. In order to test this,
235 pcretest is extended to allow the setting of negative starting offsets.
236
237 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
238 starting offset points to the beginning of a UTF-8 character was
239 unnecessarily clumsy. I tidied it up.
240
241 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
242 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
243
244 17. Nobody had reported that the --include_dir option, which was added in
245 release 7.7 should have been called --include-dir (hyphen, not underscore)
246 for compatibility with GNU grep. I have changed it to --include-dir, but
247 left --include_dir as an undocumented synonym, and the same for
248 --exclude-dir, though that is not available in GNU grep, at least as of
249 release 2.5.4.
250
251 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
252 characters from a string of bytes) have been redefined so as not to use
253 loops, in order to improve performance in some environments. At the same
254 time, I abstracted some of the common code into auxiliary macros to save
255 repetition (this should not affect the compiled code).
256
257 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
258 compile-time error is now given if \c is not followed by an ASCII
259 character, that is, a byte less than 128. (In EBCDIC mode, the code is
260 different, and any byte value is allowed.)
261
262 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
263 START_OPTIMIZE option, which is now allowed at compile time - but just
264 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
265 to pcregrep and other applications that have no direct access to PCRE
266 options. The new /Y option in pcretest sets this option when calling
267 pcre_compile().
268
269 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
270 back references. Groups containing recursive back references were forced to
271 be atomic by that change, but in the case of named groups, the amount of
272 memory required was incorrectly computed, leading to "Failed: internal
273 error: code overflow". This has been fixed.
274
275 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
276 pcretest.c, to avoid build problems in some Borland environments.
277
278
279 Version 8.10 25-Jun-2010
280 ------------------------
281
282 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
283 THEN.
284
285 2. (*ACCEPT) was not working when inside an atomic group.
286
287 3. Inside a character class, \B is treated as a literal by default, but
288 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
289 causes the error). The code is unchanged, but I tidied the documentation.
290
291 4. Inside a character class, PCRE always treated \R and \X as literals,
292 whereas Perl faults them if its -w option is set. I have changed PCRE so
293 that it faults them when PCRE_EXTRA is set.
294
295 5. Added support for \N, which always matches any character other than
296 newline. (It is the same as "." when PCRE_DOTALL is not set.)
297
298 6. When compiling pcregrep with newer versions of gcc which may have
299 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
300 declared with attribute warn_unused_result" were given. Just casting the
301 result to (void) does not stop the warnings; a more elaborate fudge is
302 needed. I've used a macro to implement this.
303
304 7. Minor change to pcretest.c to avoid a compiler warning.
305
306 8. Added four artifical Unicode properties to help with an option to make
307 \s etc use properties (see next item). The new properties are: Xan
308 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
309
310 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
311 use Unicode properties. (*UCP) at the start of a pattern can be used to set
312 this option. Modified pcretest to add /W to test this facility. Added
313 REG_UCP to make it available via the POSIX interface.
314
315 10. Added --line-buffered to pcregrep.
316
317 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
318 studied, and the match started with a letter with a code point greater than
319 127 whose first byte was different to the first byte of the other case of
320 the letter, the other case of this starting letter was not recognized
321 (#976).
322
323 12. If a pattern that was studied started with a repeated Unicode property
324 test, for example, \p{Nd}+, there was the theoretical possibility of
325 setting up an incorrect bitmap of starting bytes, but fortunately it could
326 not have actually happened in practice until change 8 above was made (it
327 added property types that matched character-matching opcodes).
328
329 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
330 possible starting bytes for non-anchored patterns.
331
332 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
333 \R, and also a number of cases that involve Unicode properties, both
334 explicit and implicit when PCRE_UCP is set.
335
336 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
337 input, it could crash or give wrong results if characters with values
338 greater than 0xc0 were present in the subject string. (Detail: it assumed
339 UTF-8 input when processing these items.)
340
341 16. Added a lot of (int) casts to avoid compiler warnings in systems where
342 size_t is 64-bit (#991).
343
344 17. Added a check for running out of memory when PCRE is compiled with
345 --disable-stack-for-recursion (#990).
346
347 18. If the last data line in a file for pcretest does not have a newline on
348 the end, a newline was missing in the output.
349
350 19. The default pcre_chartables.c file recognizes only ASCII characters (values
351 less than 128) in its various bitmaps. However, there is a facility for
352 generating tables according to the current locale when PCRE is compiled. It
353 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
354 characters, are recognized by isspace() and therefore were getting set in
355 these tables, and indeed these tables seem to approximate to ISO 8859. This
356 caused a problem in UTF-8 mode when pcre_study() was used to create a list
357 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
358 which of course cannot start UTF-8 characters. I have changed the code so
359 that only real ASCII characters (less than 128) and the correct starting
360 bytes for UTF-8 encodings are set for characters greater than 127 when in
361 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
362 altogether.)
363
364 20. Added the /T option to pcretest so as to be able to run tests with non-
365 standard character tables, thus making it possible to include the tests
366 used for 19 above in the standard set of tests.
367
368 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
369 reference to a subpattern the other side of a comment that contains an
370 opening parenthesis caused either an internal compiling error, or a
371 reference to the wrong subpattern.
372
373
374 Version 8.02 19-Mar-2010
375 ------------------------
376
377 1. The Unicode data tables have been updated to Unicode 5.2.0.
378
379 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
380 configured.
381
382 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
383 original author of that file, following a query about its status.
384
385 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
386 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
387
388 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
389 quantifier applied to a forward-referencing subroutine call, could compile
390 incorrect code or give the error "internal error: previously-checked
391 referenced subpattern not found".
392
393 6. Both MS Visual Studio and Symbian OS have problems with initializing
394 variables to point to external functions. For these systems, therefore,
395 pcre_malloc etc. are now initialized to local functions that call the
396 relevant global functions.
397
398 7. There were two entries missing in the vectors called coptable and poptable
399 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
400 I've fixed the data, and added a kludgy way of testing at compile time that
401 the lengths are correct (equal to the number of opcodes).
402
403 8. Following on from 7, I added a similar kludge to check the length of the
404 eint vector in pcreposix.c.
405
406 9. Error texts for pcre_compile() are held as one long string to avoid too
407 much relocation at load time. To find a text, the string is searched,
408 counting zeros. There was no check for running off the end of the string,
409 which could happen if a new error number was added without updating the
410 string.
411
412 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
413
414 11. \K was not working if it appeared in an atomic group or in a group that
415 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
416 \K is "not well defined" if used in an assertion. PCRE now accepts it if
417 the assertion is positive, but not if it is negative.
418
419 12. Change 11 fortuitously reduced the size of the stack frame used in the
420 "match()" function of pcre_exec.c by one pointer. Forthcoming
421 implementation of support for (*MARK) will need an extra pointer on the
422 stack; I have reserved it now, so that the stack frame size does not
423 decrease.
424
425 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
426 item in branch that calls a recursion is a subroutine call - as in the
427 second branch in the above example - was incorrectly given the compile-
428 time error "recursive call could loop indefinitely" because pcre_compile()
429 was not correctly checking the subroutine for matching a non-empty string.
430
431 14. The checks for overrunning compiling workspace could trigger after an
432 overrun had occurred. This is a "should never occur" error, but it can be
433 triggered by pathological patterns such as hundreds of nested parentheses.
434 The checks now trigger 100 bytes before the end of the workspace.
435
436 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
437
438
439 Version 8.01 19-Jan-2010
440 ------------------------
441
442 1. If a pattern contained a conditional subpattern with only one branch (in
443 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
444 computed the wrong minimum data length (which is of course zero for such
445 subpatterns). This could cause incorrect "no match" results.
446
447 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
448 the pattern is reset in the first branch, pcre_compile() failed with
449 "internal error: code overflow at offset...". This happened only when
450 the reset was to the original external option setting. (An optimization
451 abstracts leading options settings into an external setting, which was the
452 cause of this.)
453
454 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
455 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
456 assertion pattern did not match (meaning that the assertion was true), it
457 was incorrectly treated as false if the SKIP had been reached during the
458 matching. This also applied to assertions used as conditions.
459
460 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
461 assertion subpattern, including such a pattern used as a condition,
462 unpredictable results occurred, instead of the error return
463 PCRE_ERROR_DFA_UITEM.
464
465 5. The C++ GlobalReplace function was not working like Perl for the special
466 situation when an empty string is matched. It now does the fancy magic
467 stuff that is necessary.
468
469 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
470 removed. (These were left over from very, very early versions of PCRE.)
471
472 7. Some cosmetic changes to the code to make life easier when compiling it
473 as part of something else:
474
475 (a) Change DEBUG to PCRE_DEBUG.
476
477 (b) In pcre_compile(), rename the member of the "branch_chain" structure
478 called "current" as "current_branch", to prevent a collision with the
479 Linux macro when compiled as a kernel module.
480
481 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
482 prevent a collision with the Linux macro when compiled as a kernel
483 module.
484
485 8. In pcre_compile() there are some checks for integer overflows that used to
486 cast potentially large values to (double). This has been changed to that
487 when building, a check for int64_t is made, and if it is found, it is used
488 instead, thus avoiding the use of floating point arithmetic. (There is no
489 other use of FP in PCRE.) If int64_t is not found, the fallback is to
490 double.
491
492 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
493 2005 (difference between two addresses compared to an unsigned value).
494
495 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
496 custom one, because of the following reported problem in Windows:
497
498 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
499 under Win32.
500 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
501 therefore missing the function definition.
502 - The compiler thus generates a "C" signature for the test function.
503 - The linker fails to find the "C" function.
504 - PCRE fails to configure if asked to do so against libbz2.
505
506 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
507 messages were output:
508
509 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
510 rerunning libtoolize, to keep the correct libtool macros in-tree.
511 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
512
513 I have done both of these things.
514
515 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
516 most of the time, it *can* run out if it is given a pattern that contains a
517 runaway infinite recursion. I updated the discussion in the pcrestack man
518 page.
519
520 13. Now that we have gone to the x.xx style of version numbers, the minor
521 version may start with zero. Using 08 or 09 is a bad idea because users
522 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
523 interpreted as invalid octal numbers. I've updated the previous comment in
524 configure.ac, and also added a check that gives an error if 08 or 09 are
525 used.
526
527 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
528 causing partial matching to fail when the end of the subject matched \W
529 in a UTF-8 pattern where \W was quantified with a minimum of 3.
530
531 15. There were some discrepancies between the declarations in pcre_internal.h
532 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
533 their definitions. The declarations used "const uschar *" and the
534 definitions used USPTR. Even though USPTR is normally defined as "const
535 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
536 reported that: "This difference in casting confuses some C++ compilers, for
537 example, SunCC recognizes above declarations as different functions and
538 generates broken code for hbpcre." I have changed the declarations to use
539 USPTR.
540
541 16. GNU libtool is named differently on some systems. The autogen.sh script now
542 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
543 (FreeBSD).
544
545 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
546 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
547 comment: "Figure out how to create a longlong from a string: strtoll and
548 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
549 instance, but it only takes 2 args instead of 3!"
550
551 18. A subtle bug concerned with back references has been fixed by a change of
552 specification, with a corresponding code fix. A pattern such as
553 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
554 refers, was giving matches when it shouldn't. For example, xa=xaaa would
555 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
556 same bug. Such groups have to be quantified to be useful, or contained
557 inside another quantified group. (If there's no repetition, the reference
558 can never match.) The problem arises because, having left the group and
559 moved on to the rest of the pattern, a later failure that backtracks into
560 the group uses the captured value from the final iteration of the group
561 rather than the correct earlier one. I have fixed this in PCRE by forcing
562 any group that contains a reference to itself to be an atomic group; that
563 is, there cannot be any backtracking into it once it has completed. This is
564 similar to recursive and subroutine calls.
565
566
567 Version 8.00 19-Oct-09
568 ----------------------
569
570 1. The table for translating pcre_compile() error codes into POSIX error codes
571 was out-of-date, and there was no check on the pcre_compile() error code
572 being within the table. This could lead to an OK return being given in
573 error.
574
575 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
576 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
577 in a Windows environment.
578
579 3. The pcregrep --count option prints the count for each file even when it is
580 zero, as does GNU grep. However, pcregrep was also printing all files when
581 --files-with-matches was added. Now, when both options are given, it prints
582 counts only for those files that have at least one match. (GNU grep just
583 prints the file name in this circumstance, but including the count seems
584 more useful - otherwise, why use --count?) Also ensured that the
585 combination -clh just lists non-zero counts, with no names.
586
587 4. The long form of the pcregrep -F option was incorrectly implemented as
588 --fixed_strings instead of --fixed-strings. This is an incompatible change,
589 but it seems right to fix it, and I didn't think it was worth preserving
590 the old behaviour.
591
592 5. The command line items --regex=pattern and --regexp=pattern were not
593 recognized by pcregrep, which required --regex pattern or --regexp pattern
594 (with a space rather than an '='). The man page documented the '=' forms,
595 which are compatible with GNU grep; these now work.
596
597 6. No libpcreposix.pc file was created for pkg-config; there was just
598 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
599
600 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
601 when UCP support is not needed, by modifying the Python script that
602 generates it from Unicode data files. This should not matter if the module
603 is correctly used as a library, but I received one complaint about 50K of
604 unwanted data. My guess is that the person linked everything into his
605 program rather than using a library. Anyway, it does no harm.
606
607 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
608 was a minimum greater than 1 for a wide character in a possessive
609 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
610 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
611 character. Chaos in the form of incorrect output or a compiling loop could
612 result.
613
614 9. The restrictions on what a pattern can contain when partial matching is
615 requested for pcre_exec() have been removed. All patterns can now be
616 partially matched by this function. In addition, if there are at least two
617 slots in the offset vector, the offset of the earliest inspected character
618 for the match and the offset of the end of the subject are set in them when
619 PCRE_ERROR_PARTIAL is returned.
620
621 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
622 synonymous with PCRE_PARTIAL, for backwards compatibility, and
623 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
624 and may be more useful for multi-segment matching.
625
626 11. Partial matching with pcre_exec() is now more intuitive. A partial match
627 used to be given if ever the end of the subject was reached; now it is
628 given only if matching could not proceed because another character was
629 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
630 string "Z", which now yields "no match" instead of "partial match". In the
631 case of pcre_dfa_exec(), "no match" is given if every matching path for the
632 final character ended with (*FAIL).
633
634 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
635 if the pattern had a "must contain" character that was already found in the
636 earlier partial match, unless partial matching was again requested. For
637 example, with the pattern /dog.(body)?/, the "must contain" character is
638 "g". If the first part-match was for the string "dog", restarting with
639 "sbody" failed. This bug has been fixed.
640
641 13. The string returned by pcre_dfa_exec() after a partial match has been
642 changed so that it starts at the first inspected character rather than the
643 first character of the match. This makes a difference only if the pattern
644 starts with a lookbehind assertion or \b or \B (\K is not supported by
645 pcre_dfa_exec()). It's an incompatible change, but it makes the two
646 matching functions compatible, and I think it's the right thing to do.
647
648 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
649 so that the demonstration program is easily available in environments where
650 PCRE has not been installed from source.
651
652 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
653 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
654 library.
655
656 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
657 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
658 is not the first non-POSIX option to be added. Clearly some people find
659 these options useful.
660
661 17. If a caller to the POSIX matching function regexec() passes a non-zero
662 value for nmatch with a NULL value for pmatch, the value of
663 nmatch is forced to zero.
664
665 18. RunGrepTest did not have a test for the availability of the -u option of
666 the diff command, as RunTest does. It now checks in the same way as
667 RunTest, and also checks for the -b option.
668
669 19. If an odd number of negated classes containing just a single character
670 interposed, within parentheses, between a forward reference to a named
671 subpattern and the definition of the subpattern, compilation crashed with
672 an internal error, complaining that it could not find the referenced
673 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
674 [The bug was that it was starting one character too far in when skipping
675 over the character class, thus treating the ] as data rather than
676 terminating the class. This meant it could skip too much.]
677
678 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
679 /g option in pcretest when the pattern contains \K, which makes it possible
680 to have an empty string match not at the start, even when the pattern is
681 anchored. Updated pcretest and pcredemo to use this option.
682
683 21. If the maximum number of capturing subpatterns in a recursion was greater
684 than the maximum at the outer level, the higher number was returned, but
685 with unset values at the outer level. The correct (outer level) value is
686 now given.
687
688 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
689 PCRE did not set those parentheses (unlike Perl). I have now found a way to
690 make it do so. The string so far is captured, making this feature
691 compatible with Perl.
692
693 23. The tests have been re-organized, adding tests 11 and 12, to make it
694 possible to check the Perl 5.10 features against Perl 5.10.
695
696 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
697 pattern matches a fixed length string. PCRE did not allow this; now it
698 does. Neither allows recursion.
699
700 25. I finally figured out how to implement a request to provide the minimum
701 length of subject string that was needed in order to match a given pattern.
702 (It was back references and recursion that I had previously got hung up
703 on.) This code has now been added to pcre_study(); it finds a lower bound
704 to the length of subject needed. It is not necessarily the greatest lower
705 bound, but using it to avoid searching strings that are too short does give
706 some useful speed-ups. The value is available to calling programs via
707 pcre_fullinfo().
708
709 26. While implementing 25, I discovered to my embarrassment that pcretest had
710 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
711 study optimizations had never been tested with that matching function.
712 Oops. What is worse, even when it was passed study data, there was a bug in
713 pcre_dfa_exec() that meant it never actually used it. Double oops. There
714 were also very few tests of studied patterns with pcre_dfa_exec().
715
716 27. If (?| is used to create subpatterns with duplicate numbers, they are now
717 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
718 on the other side of the coin, they are no longer allowed to have different
719 names, because these cannot be distinguished in PCRE, and this has caused
720 confusion. (This is a difference from Perl.)
721
722 28. When duplicate subpattern names are present (necessarily with different
723 numbers, as required by 27 above), and a test is made by name in a
724 conditional pattern, either for a subpattern having been matched, or for
725 recursion in such a pattern, all the associated numbered subpatterns are
726 tested, and the overall condition is true if the condition is true for any
727 one of them. This is the way Perl works, and is also more like the way
728 testing by number works.
729
730
731 Version 7.9 11-Apr-09
732 ---------------------
733
734 1. When building with support for bzlib/zlib (pcregrep) and/or readline
735 (pcretest), all targets were linked against these libraries. This included
736 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
737 libraries. This caused unwanted dependencies to be created. This problem
738 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
739 pcretest is linked with readline.
740
741 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
742 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
743 moved outside it again, because FALSE and TRUE are already defined in AIX,
744 but BOOL is not.
745
746 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
747 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
748
749 4. The pcregrep documentation said spaces were inserted as well as colons (or
750 hyphens) following file names and line numbers when outputting matching
751 lines. This is not true; no spaces are inserted. I have also clarified the
752 wording for the --colour (or --color) option.
753
754 5. In pcregrep, when --colour was used with -o, the list of matching strings
755 was not coloured; this is different to GNU grep, so I have changed it to be
756 the same.
757
758 6. When --colo(u)r was used in pcregrep, only the first matching substring in
759 each matching line was coloured. Now it goes on to look for further matches
760 of any of the test patterns, which is the same behaviour as GNU grep.
761
762 7. A pattern that could match an empty string could cause pcregrep to loop; it
763 doesn't make sense to accept an empty string match in pcregrep, so I have
764 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
765 seems to be how GNU grep behaves.
766
767 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
768 start or after a newline", because the conditional assertion was not being
769 correctly handled. The rule now is that both the assertion and what follows
770 in the first alternative must satisfy the test.
771
772 9. If auto-callout was enabled in a pattern with a conditional group whose
773 condition was an assertion, PCRE could crash during matching, both with
774 pcre_exec() and pcre_dfa_exec().
775
776 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
777 used for matching.
778
779 11. Unicode property support in character classes was not working for
780 characters (bytes) greater than 127 when not in UTF-8 mode.
781
782 12. Added the -M command line option to pcretest.
783
784 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
785
786 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
787
788 16. Added comments and documentation about mis-use of no_arg in the C++
789 wrapper.
790
791 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
792 from Martin Jerabek that uses macro names for all relevant character and
793 string constants.
794
795 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
796 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
797 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
798 these, but not everybody uses configure.
799
800 19. A conditional group that had only one branch was not being correctly
801 recognized as an item that could match an empty string. This meant that an
802 enclosing group might also not be so recognized, causing infinite looping
803 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
804 with the subject "ab", where knowledge that the repeated group can match
805 nothing is needed in order to break the loop.
806
807 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
808 exec(), but without supplying a callout function, matching went wrong.
809
810 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
811 leak if the size of the offset vector was greater than 30. When the vector
812 is smaller, the saved offsets during recursion go onto a local stack
813 vector, but for larger vectors malloc() is used. It was failing to free
814 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
815 error, in fact).
816
817 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
818 heapframe that is used only when UTF-8 support is enabled. This caused no
819 problem, but was untidy.
820
821 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
822 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
823 included within another project.
824
825 24. Steven Van Ingelgem's patches to add more options to the CMake support,
826 slightly modified by me:
827
828 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
829 not building pcregrep.
830
831 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
832 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
833
834 25. Forward references, both numeric and by name, in patterns that made use of
835 duplicate group numbers, could behave incorrectly or give incorrect errors,
836 because when scanning forward to find the reference group, PCRE was not
837 taking into account the duplicate group numbers. A pattern such as
838 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
839
840 26. Changed a few more instances of "const unsigned char *" to USPTR, making
841 the feature of a custom pointer more persuasive (as requested by a user).
842
843 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
844 pcretest.c, inside #ifndefs, because it seems they are sometimes already
845 pre-defined.
846
847 28. Added support for (*UTF8) at the start of a pattern.
848
849 29. Arrange for flags added by the "release type" setting in CMake to be shown
850 in the configuration summary.
851
852
853 Version 7.8 05-Sep-08
854 ---------------------
855
856 1. Replaced UCP searching code with optimized version as implemented for Ad
857 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
858 stage table and inline lookup instead of a function, giving speed ups of 2
859 to 5 times on some simple patterns that I tested. Permission was given to
860 distribute the MultiStage2.py script that generates the tables (it's not in
861 the tarball, but is in the Subversion repository).
862
863 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
864 scripts.
865
866 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
867 a group with a zero qualifier. The result of the study could be incorrect,
868 or the function might crash, depending on the pattern.
869
870 4. Caseless matching was not working for non-ASCII characters in back
871 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
872 It now works when Unicode Property Support is available.
873
874 5. In pcretest, an escape such as \x{de} in the data was always generating
875 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
876 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
877 truncation.
878
879 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
880
881 7. Added two (int) casts to pcregrep when printing the difference of two
882 pointers, in case they are 64-bit values.
883
884 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
885 test 2 if it fails.
886
887 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
888 and a #define of that name to empty if it is not externally set. This is to
889 allow users of MSVC to set it if necessary.
890
891 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
892 the convenience functions in the pcre_get.c source file.
893
894 11. An option change at the start of a pattern that had top-level alternatives
895 could cause overwriting and/or a crash. This command provoked a crash in
896 some environments:
897
898 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
899
900 This potential security problem was recorded as CVE-2008-2371.
901
902 12. For a pattern where the match had to start at the beginning or immediately
903 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
904 pcre_dfa_exec() could read past the end of the passed subject if there was
905 no match. To help with detecting such bugs (e.g. with valgrind), I modified
906 pcretest so that it places the subject at the end of its malloc-ed buffer.
907
908 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
909 exec() might read past the end of the data buffer in UTF-8 mode.
910
911 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
912 the data contained the byte 0x85 as part of a UTF-8 character within its
913 first line. This applied both to normal and DFA matching.
914
915 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
916 /^[^d]*?$/8 failed to match "abc".
917
918 16. Added a missing copyright notice to pcrecpp_internal.h.
919
920 17. Make it more clear in the documentation that values returned from
921 pcre_exec() in ovector are byte offsets, not character counts.
922
923 18. Tidied a few places to stop certain compilers from issuing warnings.
924
925 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
926 supplied by Stefan Weber. I made a further small update for 7.8 because
927 there is a change of source arrangements: the pcre_searchfuncs.c module is
928 replaced by pcre_ucd.c.
929
930
931 Version 7.7 07-May-08
932 ---------------------
933
934 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
935 a string to a long long, pretend we don't even have a long long." This is
936 done by checking for the strtoq, strtoll, and _strtoi64 functions.
937
938 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
939 pre-7.6 versions, which defined a global no_arg variable instead of putting
940 it in the RE class. (See also #8 below.)
941
942 3. Remove a line of dead code, identified by coverity and reported by Nuno
943 Lopes.
944
945 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
946
947 (1) The include/exclude patterns were being applied to the whole pathnames
948 of files, instead of just to the final components.
949
950 (2) If there was more than one level of directory, the subdirectories were
951 skipped unless they satisfied the include/exclude conditions. This is
952 inconsistent with GNU grep (and could even be seen as contrary to the
953 pcregrep specification - which I improved to make it absolutely clear).
954 The action now is always to scan all levels of directory, and just
955 apply the include/exclude patterns to regular files.
956
957 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
958 --exclude_dir in the tests to avoid scanning .svn directories.
959
960 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
961 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
962 doesn't support NULs in patterns.
963
964 7. Added some missing "const"s to declarations of static tables in
965 pcre_compile.c and pcre_dfa_exec.c.
966
967 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
968 caused by fix #2 above. (Subsequently also a second patch to fix the
969 first patch. And a third patch - this was a messy problem.)
970
971 9. Applied Craig's patch to remove the use of push_back().
972
973 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
974 matching function regexec().
975
976 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
977 which, however, unlike Perl's \g{...}, are subroutine calls, not back
978 references. PCRE supports relative numbers with this syntax (I don't think
979 Oniguruma does).
980
981 12. Previously, a group with a zero repeat such as (...){0} was completely
982 omitted from the compiled regex. However, this means that if the group
983 was called as a subroutine from elsewhere in the pattern, things went wrong
984 (an internal error was given). Such groups are now left in the compiled
985 pattern, with a new opcode that causes them to be skipped at execution
986 time.
987
988 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
989 to the way PCRE behaves:
990
991 (a) A lone ] character is dis-allowed (Perl treats it as data).
992
993 (b) A back reference to an unmatched subpattern matches an empty string
994 (Perl fails the current match path).
995
996 (c) A data ] in a character class must be notated as \] because if the
997 first data character in a class is ], it defines an empty class. (In
998 Perl it is not possible to have an empty class.) The empty class []
999 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1000 The negative empty class [^] matches any one character, independently
1001 of the DOTALL setting.
1002
1003 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1004 non-existent subpattern following a character class starting with ']' and
1005 containing () gave an internal compiling error instead of "reference to
1006 non-existent subpattern". Fortunately, when the pattern did exist, the
1007 compiled code was correct. (When scanning forwards to check for the
1008 existencd of the subpattern, it was treating the data ']' as terminating
1009 the class, so got the count wrong. When actually compiling, the reference
1010 was subsequently set up correctly.)
1011
1012 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1013 it was being rejected as not supported by pcre_dfa_exec(), even though
1014 other assertions are supported. I have made pcre_dfa_exec() support
1015 (*FAIL).
1016
1017 16. The implementation of 13c above involved the invention of a new opcode,
1018 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1019 cannot be changed at match time, I realized I could make a small
1020 improvement to matching performance by compiling OP_ALLANY instead of
1021 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1022 on the OP_ANY path.
1023
1024 17. Compiling pcretest on Windows with readline support failed without the
1025 following two fixes: (1) Make the unistd.h include conditional on
1026 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1027
1028 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1029 ncurses library to be included for pcretest when ReadLine support is
1030 requested, but also to allow for it to be overridden. This patch came from
1031 Daniel Bergström.
1032
1033 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1034 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1035 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1036 spotting this.
1037
1038
1039 Version 7.6 28-Jan-08
1040 ---------------------
1041
1042 1. A character class containing a very large number of characters with
1043 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1044 overflow.
1045
1046 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1047 HAVE_LONG_LONG is not defined.
1048
1049 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1050 bring them up to date and include new features. This patch includes:
1051
1052 - Fixed PH's badly added libz and libbz2 support.
1053 - Fixed a problem with static linking.
1054 - Added pcredemo. [But later removed - see 7 below.]
1055 - Fixed dftables problem and added an option.
1056 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1057 HAVE_LONG_LONG.
1058 - Added readline support for pcretest.
1059 - Added an listing of the option settings after cmake has run.
1060
1061 4. A user submitted a patch to Makefile that makes it easy to create
1062 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1063 Makefile.am that cause it to include this special target, without
1064 affecting anything else. Note that the same mingw target plus all
1065 the other distribution libraries and programs are now supported
1066 when configuring with CMake (see 6 below) instead of with
1067 Configure/Make.
1068
1069 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1070 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1071 exported in the Windows port". It has not yet been confirmed that the patch
1072 solves the problem, but it does no harm.
1073
1074 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1075 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1076 with CMake, and also correct the comment about stack recursion.
1077
1078 7. Remove the automatic building of pcredemo from the ./configure system and
1079 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1080 of a program that users should build themselves after PCRE is installed, so
1081 building it automatically is not really right. What is more, it gave
1082 trouble in some build environments.
1083
1084 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1085
1086
1087 Version 7.5 10-Jan-08
1088 ---------------------
1089
1090 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1091 values in parens when parsing an RE using the C++ wrapper."
1092
1093 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1094 Characters greater than 255 were excluded from the class instead of being
1095 included.
1096
1097 3. The same bug as (2) above applied to negated POSIX classes such as
1098 [:^space:].
1099
1100 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1101 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1102 I have changed it.
1103
1104 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1105 first named subpattern) and a construct such as (?&a) would reference the
1106 first named subpattern whose name started with "a" (in other words, the
1107 length check was missing). Both these problems are fixed. "Subpattern name
1108 expected" is now given for (?&) (a zero-length name), and this patch also
1109 makes it give the same error for \k'' (previously it complained that that
1110 was a reference to a non-existent subpattern).
1111
1112 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1113 this is right because (?- can be followed by option settings as well as by
1114 digits. I have, however, made the messages clearer.
1115
1116 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1117 than the number used in the conditional) now cause a compile-time error.
1118 This is actually not compatible with Perl, which accepts such patterns, but
1119 treats the conditional as always being FALSE (as PCRE used to), but it
1120 seems to me that giving a diagnostic is better.
1121
1122 8. Change "alphameric" to the more common word "alphanumeric" in comments
1123 and messages.
1124
1125 9. Fix two occurrences of "backslash" in comments that should have been
1126 "backspace".
1127
1128 10. Remove two redundant lines of code that can never be obeyed (their function
1129 was moved elsewhere).
1130
1131 11. The program that makes PCRE's Unicode character property table had a bug
1132 which caused it to generate incorrect table entries for sequences of
1133 characters that have the same character type, but are in different scripts.
1134 It amalgamated them into a single range, with the script of the first of
1135 them. In other words, some characters were in the wrong script. There were
1136 thirteen such cases, affecting characters in the following ranges:
1137
1138 U+002b0 - U+002c1
1139 U+0060c - U+0060d
1140 U+0061e - U+00612
1141 U+0064b - U+0065e
1142 U+0074d - U+0076d
1143 U+01800 - U+01805
1144 U+01d00 - U+01d77
1145 U+01d9b - U+01dbf
1146 U+0200b - U+0200f
1147 U+030fc - U+030fe
1148 U+03260 - U+0327f
1149 U+0fb46 - U+0fbb1
1150 U+10450 - U+1049d
1151
1152 12. The -o option (show only the matching part of a line) for pcregrep was not
1153 compatible with GNU grep in that, if there was more than one match in a
1154 line, it showed only the first of them. It now behaves in the same way as
1155 GNU grep.
1156
1157 13. If the -o and -v options were combined for pcregrep, it printed a blank
1158 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1159 does the same. The return code can be used to tell if there were any
1160 non-matching lines.
1161
1162 14. Added --file-offsets and --line-offsets to pcregrep.
1163
1164 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1165 infinitely looping recursion. The bug was that positive lookaheads were not
1166 being skipped when checking for a possible empty match (negative lookaheads
1167 and both kinds of lookbehind were skipped).
1168
1169 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1170 inclusion of <windows.h> to before rather than after the definition of
1171 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1172
1173 17. Specifying a possessive quantifier with a specific limit for a Unicode
1174 character property caused pcre_compile() to compile bad code, which led at
1175 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1176 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1177 caused the error; without that there was no problem.
1178
1179 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1180
1181 19. Added --enable-pcretest-libreadline.
1182
1183 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1184 RE::GlobalReplace(). As a result, the number of replacements returned was
1185 double what it should be. I removed one of the increments, but Craig sent a
1186 later patch that removed the other one (the right fix) and added unit tests
1187 that check the return values (which was not done before).
1188
1189 21. Several CMake things:
1190
1191 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1192 the names libpcre and libpcreposix, not just pcre and pcreposix.
1193
1194 (2) The above change means that pcretest and pcregrep are now correctly
1195 linked with the newly-built libraries, not previously installed ones.
1196
1197 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1198
1199 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1200 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1201 UTF-8 newline character). The key issue is that the pattern starts .*;
1202 this means that the match must be either at the beginning, or after a
1203 newline. The bug was in the code for advancing after a failed match and
1204 checking that the new position followed a newline. It was not taking
1205 account of UTF-8 characters correctly.
1206
1207 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1208 character classes. PCRE was not treating the sequence [:...:] as a
1209 character class unless the ... were all letters. Perl, however, seems to
1210 allow any characters between [: and :], though of course it rejects as
1211 unknown any "names" that contain non-letters, because all the known class
1212 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1213 for example, whereas PCRE did not - it did not recognize a POSIX character
1214 class. This seemed a bit dangerous, so the code has been changed to be
1215 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1216 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1217 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1218 Perl does, and where it didn't before.
1219
1220 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1221 Windows environments %n is disabled by default.
1222
1223
1224 Version 7.4 21-Sep-07
1225 ---------------------
1226
1227 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1228 means that a class such as [\s] counted as "explicit reference to CR or
1229 LF". That isn't really right - the whole point of the change was to try to
1230 help when there was an actual mention of one of the two characters. So now
1231 the change happens only if \r or \n (or a literal CR or LF) character is
1232 encountered.
1233
1234 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1235 of both had grown to the point where there were only 3 bits left.
1236 Fortunately, there was spare space in the data structure, and so I have
1237 moved the internal flags into a new 16-bit field to free up more option
1238 bits.
1239
1240 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1241 but did not set the internal JCHANGED flag - either of these is enough to
1242 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1243 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1244 start sets both bits.
1245
1246 4. Added options (at build time, compile time, exec time) to change \R from
1247 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1248
1249 5. doc/pcresyntax.html was missing from the distribution.
1250
1251 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1252 compatibility, even though it is no longer used.
1253
1254 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1255 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1256 windows.h file is present (where different names are used). [This was
1257 reversed later after testing - see 16 below.]
1258
1259 8. Changed all #include <config.h> to #include "config.h". There were also
1260 some further <pcre.h> cases that I changed to "pcre.h".
1261
1262 9. When pcregrep was used with the --colour option, it missed the line ending
1263 sequence off the lines that it output.
1264
1265 10. It was pointed out to me that arrays of string pointers cause lots of
1266 relocations when a shared library is dynamically loaded. A technique of
1267 using a single long string with a table of offsets can drastically reduce
1268 these. I have refactored PCRE in four places to do this. The result is
1269 dramatic:
1270
1271 Originally: 290
1272 After changing UCP table: 187
1273 After changing error message table: 43
1274 After changing table of "verbs" 36
1275 After changing table of Posix names 22
1276
1277 Thanks to the folks working on Gregex for glib for this insight.
1278
1279 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1280 unicode-properties was also set.
1281
1282 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1283
1284 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1285 checked only for CRLF.
1286
1287 14. Added casts to pcretest.c to avoid compiler warnings.
1288
1289 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1290
1291 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1292 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1293 entirely. This removes changes made in 7 above.
1294
1295 17. The CMake files have been updated, and there is now more information about
1296 building with CMake in the NON-UNIX-USE document.
1297
1298
1299 Version 7.3 28-Aug-07
1300 ---------------------
1301
1302 1. In the rejigging of the build system that eventually resulted in 7.1, the
1303 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1304 brackets there is not right, since it causes compilers to look for an
1305 installed pcre.h, not the version that is in the source that is being
1306 compiled (which of course may be different). I have changed it back to:
1307
1308 #include "pcre.h"
1309
1310 I have a vague recollection that the change was concerned with compiling in
1311 different directories, but in the new build system, that is taken care of
1312 by the VPATH setting the Makefile.
1313
1314 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1315 when the subject happened to end in the byte 0x85 (e.g. if the last
1316 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1317 characters but of course it shouldn't be taken as a newline when it is part
1318 of another character. The bug was that, for an unlimited repeat of . in
1319 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1320 characters when looking for a newline.
1321
1322 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1323
1324 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1325 in debug output.
1326
1327 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1328 long printing in the pcrecpp unittest when running under MinGW.
1329
1330 6. ESC_K was left out of the EBCDIC table.
1331
1332 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1333 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1334 limit also applies to "virtual nesting" when a pattern is recursive, and in
1335 this case 1000 isn't so big. I have been able to remove this limit at the
1336 expense of backing off one optimization in certain circumstances. Normally,
1337 when pcre_exec() would call its internal match() function recursively and
1338 immediately return the result unconditionally, it uses a "tail recursion"
1339 feature to save stack. However, when a subpattern that can match an empty
1340 string has an unlimited repetition quantifier, it no longer makes this
1341 optimization. That gives it a stack frame in which to save the data for
1342 checking that an empty string has been matched. Previously this was taken
1343 from the 1000-entry workspace that had been reserved. So now there is no
1344 explicit limit, but more stack is used.
1345
1346 8. Applied Daniel's patches to solve problems with the import/export magic
1347 syntax that is required for Windows, and which was going wrong for the
1348 pcreposix and pcrecpp parts of the library. These were overlooked when this
1349 problem was solved for the main library.
1350
1351 9. There were some crude static tests to avoid integer overflow when computing
1352 the size of patterns that contain repeated groups with explicit upper
1353 limits. As the maximum quantifier is 65535, the maximum group length was
1354 set at 30,000 so that the product of these two numbers did not overflow a
1355 32-bit integer. However, it turns out that people want to use groups that
1356 are longer than 30,000 bytes (though not repeat them that many times).
1357 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1358 made it possible to implement the integer overflow checks in a much more
1359 dynamic way, which I have now done. The artificial limitation on group
1360 length has been removed - we now have only the limit on the total length of
1361 the compiled pattern, which depends on the LINK_SIZE setting.
1362
1363 10. Fixed a bug in the documentation for get/copy named substring when
1364 duplicate names are permitted. If none of the named substrings are set, the
1365 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1366 empty string.
1367
1368 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1369 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1370 because the ] is interpreted as the first data character and the
1371 terminating ] is not found. PCRE has been made compatible with Perl in this
1372 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1373 cause memory overwriting.
1374
1375 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1376 string has been matched (to stop an infinite loop). It was not recognizing
1377 a conditional subpattern that could match an empty string if that
1378 subpattern was within another subpattern. For example, it looped when
1379 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1380 condition was not nested. This bug has been fixed.
1381
1382 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1383 past the start of the subject in the presence of bytes with the top bit
1384 set, for example "\x8aBCD".
1385
1386 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1387 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1388
1389 14. Optimized (?!) to (*FAIL).
1390
1391 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1392 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1393 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1394 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1395 does: it's just the validity check that is more restrictive.
1396
1397 16. Inserted checks for integer overflows during escape sequence (backslash)
1398 processing, and also fixed erroneous offset values for syntax errors during
1399 backslash processing.
1400
1401 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1402 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1403
1404 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1405 caused an overrun.
1406
1407 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1408 something other than just ASCII characters) inside a group that had an
1409 unlimited repeat caused a loop at compile time (while checking to see
1410 whether the group could match an empty string).
1411
1412 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1413 [\P{Any}] did so. (Error in the code for printing property names.)
1414
1415 21. An orphan \E inside a character class could cause a crash.
1416
1417 22. A repeated capturing bracket such as (A)? could cause a wild memory
1418 reference during compilation.
1419
1420 23. There are several functions in pcre_compile() that scan along a compiled
1421 expression for various reasons (e.g. to see if it's fixed length for look
1422 behind). There were bugs in these functions when a repeated \p or \P was
1423 present in the pattern. These operators have additional parameters compared
1424 with \d, etc, and these were not being taken into account when moving along
1425 the compiled data. Specifically:
1426
1427 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1428 length.
1429
1430 (b) An item such as \pL+ within a repeated group could cause crashes or
1431 loops.
1432
1433 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1434 "reference to non-existent subpattern" error.
1435
1436 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1437
1438 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1439 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1440
1441 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1442
1443 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1444 character were causing crashes (broken optimization).
1445
1446 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1447 \p or \P) caused a compile-time loop.
1448
1449 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1450 break. For example, the unstudied pattern [\r\n]A does not match the string
1451 "\r\nA" because change 7.0/46 below moves the current point on by two
1452 characters after failing to match at the start. However, the pattern \nA
1453 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1454 the same is true. There doesn't seem any very clean way out of this, but
1455 what I have chosen to do makes the common cases work: PCRE now takes note
1456 of whether there can be an explicit match for \r or \n anywhere in the
1457 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1458 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1459 pattern has explicit CR or LF references.
1460
1461 29. Added (*CR) etc for changing newline setting at start of pattern.
1462
1463
1464 Version 7.2 19-Jun-07
1465 ---------------------
1466
1467 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1468 which is apparently normally available under Windows.
1469
1470 2. Re-jig the pcregrep tests with different newline settings in an attempt
1471 to make them independent of the local environment's newline setting.
1472
1473 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1474
1475 4. Some of the "internals" tests were previously cut out when the link size
1476 was not 2, because the output contained actual offsets. The recent new
1477 "Z" feature of pcretest means that these can be cut out, making the tests
1478 usable with all link sizes.
1479
1480 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1481 stack recursion. This gives a massive performance boost under BSD, but just
1482 a small improvement under Linux. However, it saves one field in the frame
1483 in all cases.
1484
1485 6. Added more features from the forthcoming Perl 5.10:
1486
1487 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1488 recursion call. It refers to the nth most recently opened parentheses.
1489
1490 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1491 to be opened parentheses.
1492
1493 (c) Conditions that refer to capturing parentheses can be specified
1494 relatively, for example, (?(-2)... or (?(+3)...
1495
1496 (d) \K resets the start of the current match so that everything before
1497 is not part of it.
1498
1499 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1500
1501 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1502 reference syntax.
1503
1504 (g) (?| introduces a group in which the numbering of parentheses in each
1505 alternative starts with the same number.
1506
1507 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1508
1509 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1510 PCRE_INFO_JCHANGED.
1511
1512 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1513 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1514 for detecting groups that can match an empty string.
1515
1516 9. A pattern with a very large number of alternatives (more than several
1517 hundred) was running out of internal workspace during the pre-compile
1518 phase, where pcre_compile() figures out how much memory will be needed. A
1519 bit of new cunning has reduced the workspace needed for groups with
1520 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1521 workspace instead of running out of the 4096 that are available.
1522
1523 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1524
1525 11. Applied patch from Google to remove an optimization that didn't quite work.
1526 The report of the bug said:
1527
1528 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1529 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1530 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1531
1532 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1533 it matched the wrong number of bytes.
1534
1535
1536 Version 7.1 24-Apr-07
1537 ---------------------
1538
1539 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1540 that is more "standard", making use of automake and other Autotools. There
1541 is some re-arrangement of the files and adjustment of comments consequent
1542 on this.
1543
1544 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1545 for recursive directory scanning broke on some systems because the files
1546 are not scanned in any specific order and on different systems the order
1547 was different. A call to "sort" has been inserted into RunGrepTest for the
1548 approprate test as a short-term fix. In the longer term there may be an
1549 alternative.
1550
1551 3. I had an email from Eric Raymond about problems translating some of PCRE's
1552 man pages to HTML (despite the fact that I distribute HTML pages, some
1553 people do their own conversions for various reasons). The problems
1554 concerned the use of low-level troff macros .br and .in. I have therefore
1555 removed all such uses from the man pages (some were redundant, some could
1556 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1557 HTML has been updated to handle .nf/.fi and to complain if it encounters
1558 .br or .in.
1559
1560 4. Updated comments in configure.ac that get placed in config.h.in and also
1561 arranged for config.h to be included in the distribution, with the name
1562 config.h.generic, for the benefit of those who have to compile without
1563 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1564
1565 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1566 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1567 makevp.bat for the current PCRE, using the additional files
1568 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1569
1570 6. A Windows user reported a minor discrepancy with test 2, which turned out
1571 to be caused by a trailing space on an input line that had got lost in his
1572 copy. The trailing space was an accident, so I've just removed it.
1573
1574 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1575 that is needed.
1576
1577 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1578 as "const" (a) because they are and (b) because it helps the PHP
1579 maintainers who have recently made a script to detect big data structures
1580 in the php code that should be moved to the .rodata section. I remembered
1581 to update Builducptable as well, so it won't revert if ucptable.h is ever
1582 re-created.
1583
1584 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1585 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1586 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1587 support is not required. This saves 1.5-2K of code, which is important in
1588 some applications.
1589
1590 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1591 so as not to refer to the tables, even though these functions will never be
1592 called when UTF-8 support is disabled. Otherwise there are problems with a
1593 shared library.
1594
1595 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1596
1597 (a) It was defining its arguments as char * instead of void *.
1598
1599 (b) It was assuming that all moves were upwards in memory; this was true
1600 a long time ago when I wrote it, but is no longer the case.
1601
1602 The emulated memove() is provided for those environments that have neither
1603 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1604 is clearly not the case, as these two bugs were recently reported.
1605
1606 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1607 and Detrail to create the HTML documentation, the .txt form of the man
1608 pages, and it removes trailing spaces from listed files. It also creates
1609 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1610 case, it wraps all the #defines with #ifndefs. This script should be run
1611 before "make dist".
1612
1613 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1614 with Unicode property support.
1615
1616 (a) For a maximizing quantifier, if the two different cases of the
1617 character were of different lengths in their UTF-8 codings (there are
1618 some cases like this - I found 11), and the matching function had to
1619 back up over a mixture of the two cases, it incorrectly assumed they
1620 were both the same length.
1621
1622 (b) When PCRE was configured to use the heap rather than the stack for
1623 recursion during matching, it was not correctly preserving the data for
1624 the other case of a UTF-8 character when checking ahead for a match
1625 while processing a minimizing repeat. If the check also involved
1626 matching a wide character, but failed, corruption could cause an
1627 erroneous result when trying to check for a repeat of the original
1628 character.
1629
1630 13. Some tidying changes to the testing mechanism:
1631
1632 (a) The RunTest script now detects the internal link size and whether there
1633 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1634 values substituted by "configure". (The RunGrepTest script already did
1635 this for UTF-8.) The configure.ac script no longer substitutes the
1636 relevant variables.
1637
1638 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1639 with length and offset values. This means that the output is different
1640 for different internal link sizes. Test 2 is skipped for link sizes
1641 other than 2 because of this, bypassing the problem. Unfortunately,
1642 there was also a test in test 3 (the locale tests) that used /B and
1643 failed for link sizes other than 2. Rather than cut the whole test out,
1644 I have added a new /Z option to pcretest that replaces the length and
1645 offset values with spaces. This is now used to make test 3 independent
1646 of link size. (Test 2 will be tidied up later.)
1647
1648 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1649 segmentation fault instead of returning the appropriate error message.
1650
1651 15. In multiline mode when the newline sequence was set to "any", the pattern
1652 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1653 This doesn't seem right; it now treats the CRLF combination as the line
1654 ending, and so does not match in that case. It's only a pattern such as ^$
1655 that would hit this one: something like ^ABC$ would have failed after \r
1656 and then tried again after \r\n.
1657
1658 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1659 in an attempt to make files that differ only in their line terminators
1660 compare equal. This works on Linux.
1661
1662 17. Under certain error circumstances pcregrep might try to free random memory
1663 as it exited. This is now fixed, thanks to valgrind.
1664
1665 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1666 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1667 was because its rules for how to advance for /g after matching an empty
1668 string at the end of a line did not allow for this case. They now check for
1669 it specially.
1670
1671 20. pcretest is supposed to handle patterns and data of any length, by
1672 extending its buffers when necessary. It was getting this wrong when the
1673 buffer for a data line had to be extended.
1674
1675 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1676 CRLF as a newline sequence.
1677
1678 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1679 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1680 I have nevertheless tidied it up.
1681
1682 23. Added some casts to kill warnings from HP-UX ia64 compiler.
1683
1684 24. Added a man page for pcre-config.
1685
1686
1687 Version 7.0 19-Dec-06
1688 ---------------------
1689
1690 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1691 moving to gcc 4.1.1.
1692
1693 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1694 sys/time.h, which is documented as needed for this function. It doesn't
1695 seem to matter on Linux, but it showed up on some releases of OS X.
1696
1697 3. It seems that there are systems where bytes whose values are greater than
1698 127 match isprint() in the "C" locale. The "C" locale should be the
1699 default when a C program starts up. In most systems, only ASCII printing
1700 characters match isprint(). This difference caused the output from pcretest
1701 to vary, making some of the tests fail. I have changed pcretest so that:
1702
1703 (a) When it is outputting text in the compiled version of a pattern, bytes
1704 other than 32-126 are always shown as hex escapes.
1705
1706 (b) When it is outputting text that is a matched part of a subject string,
1707 it does the same, unless a different locale has been set for the match
1708 (using the /L modifier). In this case, it uses isprint() to decide.
1709
1710 4. Fixed a major bug that caused incorrect computation of the amount of memory
1711 required for a compiled pattern when options that changed within the
1712 pattern affected the logic of the preliminary scan that determines the
1713 length. The relevant options are -x, and -i in UTF-8 mode. The result was
1714 that the computed length was too small. The symptoms of this bug were
1715 either the PCRE error "internal error: code overflow" from pcre_compile(),
1716 or a glibc crash with a message such as "pcretest: free(): invalid next
1717 size (fast)". Examples of patterns that provoked this bug (shown in
1718 pcretest format) are:
1719
1720 /(?-x: )/x
1721 /(?x)(?-x: \s*#\s*)/
1722 /((?i)[\x{c0}])/8
1723 /(?i:[\x{c0}])/8
1724
1725 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1726 is now done differently.
1727
1728 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1729 wrapper classes; (b) implement a new function in the C++ scanner that is
1730 more efficient than the old way of doing things because it avoids levels of
1731 recursion in the regex matching; (c) add a paragraph to the documentation
1732 for the FullMatch() function.
1733
1734 6. The escape sequence \n was being treated as whatever was defined as
1735 "newline". Not only was this contrary to the documentation, which states
1736 that \n is character 10 (hex 0A), but it also went horribly wrong when
1737 "newline" was defined as CRLF. This has been fixed.
1738
1739 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
1740 was being set to -1 for the "end of line" case (supposedly a value that no
1741 character can have). Though this value is never used (the check for end of
1742 line is "zero bytes in current character"), it caused compiler complaints.
1743 I've changed it to 0xffffffff.
1744
1745 8. In pcre_version.c, the version string was being built by a sequence of
1746 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
1747 string (as it is for production releases) called a macro with an empty
1748 argument. The C standard says the result of this is undefined. The gcc
1749 compiler treats it as an empty string (which was what was wanted) but it is
1750 reported that Visual C gives an error. The source has been hacked around to
1751 avoid this problem.
1752
1753 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
1754 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
1755 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
1756 of them did).
1757
1758 10. Originally, pcretest opened its input and output without "b"; then I was
1759 told that "b" was needed in some environments, so it was added for release
1760 5.0 to both the input and output. (It makes no difference on Unix-like
1761 systems.) Later I was told that it is wrong for the input on Windows. I've
1762 now abstracted the modes into two macros, to make it easier to fiddle with
1763 them, and removed "b" from the input mode under Windows.
1764
1765 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
1766
1767 12. Added -help and --help to pcretest as an official way of being reminded
1768 of the options.
1769
1770 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
1771 and pcrecpp.cc because they annoy compilers at high warning levels.
1772
1773 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
1774
1775 15. Fixed an occurrence of == in configure.ac that should have been = (shell
1776 scripts are not C programs :-) and which was not noticed because it works
1777 on Linux.
1778
1779 16. pcretest is supposed to handle any length of pattern and data line (as one
1780 line or as a continued sequence of lines) by extending its input buffer if
1781 necessary. This feature was broken for very long pattern lines, leading to
1782 a string of junk being passed to pcre_compile() if the pattern was longer
1783 than about 50K.
1784
1785 17. I have done a major re-factoring of the way pcre_compile() computes the
1786 amount of memory needed for a compiled pattern. Previously, there was code
1787 that made a preliminary scan of the pattern in order to do this. That was
1788 OK when PCRE was new, but as the facilities have expanded, it has become
1789 harder and harder to keep it in step with the real compile phase, and there
1790 have been a number of bugs (see for example, 4 above). I have now found a
1791 cunning way of running the real compile function in a "fake" mode that
1792 enables it to compute how much memory it would need, while actually only
1793 ever using a few hundred bytes of working memory and without too many
1794 tests of the mode. This should make future maintenance and development
1795 easier. A side effect of this work is that the limit of 200 on the nesting
1796 depth of parentheses has been removed (though this was never a serious
1797 limitation, I suspect). However, there is a downside: pcre_compile() now
1798 runs more slowly than before (30% or more, depending on the pattern). I
1799 hope this isn't a big issue. There is no effect on runtime performance.
1800
1801 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
1802 newline (only possible for the last line of a file) and it was a
1803 pattern that set a locale (followed by /Lsomething), pcretest crashed.
1804
1805 19. Added additional timing features to pcretest. (1) The -tm option now times
1806 matching only, not compiling. (2) Both -t and -tm can be followed, as a
1807 separate command line item, by a number that specifies the number of
1808 repeats to use when timing. The default is 50000; this gives better
1809 precision, but takes uncomfortably long for very large patterns.
1810
1811 20. Extended pcre_study() to be more clever in cases where a branch of a
1812 subpattern has no definite first character. For example, (a*|b*)[cd] would
1813 previously give no result from pcre_study(). Now it recognizes that the
1814 first character must be a, b, c, or d.
1815
1816 21. There was an incorrect error "recursive call could loop indefinitely" if
1817 a subpattern (or the entire pattern) that was being tested for matching an
1818 empty string contained only one non-empty item after a nested subpattern.
1819 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
1820 incorrectly, because the \d was being skipped in the check.
1821
1822 22. The pcretest program now has a new pattern option /B and a command line
1823 option -b, which is equivalent to adding /B to every pattern. This causes
1824 it to show the compiled bytecode, without the additional information that
1825 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
1826 is the same as /B/I).
1827
1828 23. A new optimization is now able automatically to treat some sequences such
1829 as a*b as a*+b. More specifically, if something simple (such as a character
1830 or a simple class like \d) has an unlimited quantifier, and is followed by
1831 something that cannot possibly match the quantified thing, the quantifier
1832 is automatically "possessified".
1833
1834 24. A recursive reference to a subpattern whose number was greater than 39
1835 went wrong under certain circumstances in UTF-8 mode. This bug could also
1836 have affected the operation of pcre_study().
1837
1838 25. Realized that a little bit of performance could be had by replacing
1839 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
1840
1841 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
1842
1843 27. Possessive quantifiers such as a++ were previously implemented by turning
1844 them into atomic groups such as ($>a+). Now they have their own opcodes,
1845 which improves performance. This includes the automatically created ones
1846 from 23 above.
1847
1848 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
1849 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
1850 the first matched character to be a colon. This applied both to named and
1851 numbered groups.
1852
1853 29. The ucpinternal.h header file was missing its idempotency #ifdef.
1854
1855 30. I was sent a "project" file called libpcre.a.dev which I understand makes
1856 building PCRE on Windows easier, so I have included it in the distribution.
1857
1858 31. There is now a check in pcretest against a ridiculously large number being
1859 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
1860 loop, the loop is abandoned.
1861
1862 32. Forward references to subpatterns in conditions such as (?(2)...) where
1863 subpattern 2 is defined later cause pcre_compile() to search forwards in
1864 the pattern for the relevant set of parentheses. This search went wrong
1865 when there were unescaped parentheses in a character class, parentheses
1866 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
1867
1868 33. "Subroutine" calls and backreferences were previously restricted to
1869 referencing subpatterns earlier in the regex. This restriction has now
1870 been removed.
1871
1872 34. Added a number of extra features that are going to be in Perl 5.10. On the
1873 whole, these are just syntactic alternatives for features that PCRE had
1874 previously implemented using the Python syntax or my own invention. The
1875 other formats are all retained for compatibility.
1876
1877 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
1878 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
1879 also .NET compatible.
1880
1881 (b) A recursion or subroutine call to a named group can now be defined as
1882 (?&name) as well as (?P>name).
1883
1884 (c) A backreference to a named group can now be defined as \k<name> or
1885 \k'name' as well as (?P=name). The new forms, as well as being in Perl
1886 5.10, are also .NET compatible.
1887
1888 (d) A conditional reference to a named group can now use the syntax
1889 (?(<name>) or (?('name') as well as (?(name).
1890
1891 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
1892 groups (named and numbered) that are never evaluated inline, but can be
1893 called as "subroutines" from elsewhere. In effect, the DEFINE condition
1894 is always false. There may be only one alternative in such a group.
1895
1896 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
1897 as the simple (?(R). The condition is true only if the most recent
1898 recursion is that of the given number or name. It does not search out
1899 through the entire recursion stack.
1900
1901 (g) The escape \gN or \g{N} has been added, where N is a positive or
1902 negative number, specifying an absolute or relative reference.
1903
1904 35. Tidied to get rid of some further signed/unsigned compiler warnings and
1905 some "unreachable code" warnings.
1906
1907 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
1908 things, this adds five new scripts.
1909
1910 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
1911 There were also incompatibilities regarding the handling of \Q..\E inside
1912 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
1913 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
1914
1915 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
1916 matches an empty string, and forcibly breaks the loop. There were bugs in
1917 this code in non-simple cases. For a pattern such as ^(a()*)* matched
1918 against aaaa the result was just "a" rather than "aaaa", for example. Two
1919 separate and independent bugs (that affected different cases) have been
1920 fixed.
1921
1922 39. Refactored the code to abolish the use of different opcodes for small
1923 capturing bracket numbers. This is a tidy that I avoided doing when I
1924 removed the limit on the number of capturing brackets for 3.5 back in 2001.
1925 The new approach is not only tidier, it makes it possible to reduce the
1926 memory needed to fix the previous bug (38).
1927
1928 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
1929 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
1930 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
1931 mode.
1932
1933 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
1934 report.
1935
1936 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
1937 copy construction and assignment in the C++ wrapper.
1938
1939 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
1940 couple of bugs that could have given wrong results in the "--newline=crlf"
1941 case.
1942
1943 44. Added a number of casts and did some reorganization of signed/unsigned int
1944 variables following suggestions from Dair Grant. Also renamed the variable
1945 "this" as "item" because it is a C++ keyword.
1946
1947 45. Arranged for dftables to add
1948
1949 #include "pcre_internal.h"
1950
1951 to pcre_chartables.c because without it, gcc 4.x may remove the array
1952 definition from the final binary if PCRE is built into a static library and
1953 dead code stripping is activated.
1954
1955 46. For an unanchored pattern, if a match attempt fails at the start of a
1956 newline sequence, and the newline setting is CRLF or ANY, and the next two
1957 characters are CRLF, advance by two characters instead of one.
1958
1959
1960 Version 6.7 04-Jul-06
1961 ---------------------
1962
1963 1. In order to handle tests when input lines are enormously long, pcretest has
1964 been re-factored so that it automatically extends its buffers when
1965 necessary. The code is crude, but this _is_ just a test program. The
1966 default size has been increased from 32K to 50K.
1967
1968 2. The code in pcre_study() was using the value of the re argument before
1969 testing it for NULL. (Of course, in any sensible call of the function, it
1970 won't be NULL.)
1971
1972 3. The memmove() emulation function in pcre_internal.h, which is used on
1973 systems that lack both memmove() and bcopy() - that is, hardly ever -
1974 was missing a "static" storage class specifier.
1975
1976 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
1977 containing an extended class (one that cannot be represented by a bitmap
1978 because it contains high-valued characters or Unicode property items, e.g.
1979 [\pZ]). Almost always one would set UTF-8 mode when processing such a
1980 pattern, but PCRE should not loop if you do not (it no longer does).
1981 [Detail: two cases were found: (a) a repeated subpattern containing an
1982 extended class; (b) a recursive reference to a subpattern that followed a
1983 previous extended class. It wasn't skipping over the extended class
1984 correctly when UTF-8 mode was not set.]
1985
1986 5. A negated single-character class was not being recognized as fixed-length
1987 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
1988 compile error "lookbehind assertion is not fixed length".
1989
1990 6. The RunPerlTest auxiliary script was showing an unexpected difference
1991 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
1992 write a Perl script that can interpret lines of an input file either as
1993 byte characters or as UTF-8, which is what "perltest" was being required to
1994 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
1995 can't do is switch easily at run time between having the "use utf8;" pragma
1996 or not. In the end, I fudged it by using the RunPerlTest script to insert
1997 "use utf8;" explicitly for the UTF-8 tests.
1998
1999 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2000 the end of the subject string, contrary to the documentation and to what
2001 Perl does. This was true of both matching functions. Now it matches only at
2002 the start of the subject and immediately after *internal* newlines.
2003
2004 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2005 a pointer to an int instead of a pointer to an unsigned long int. This
2006 caused problems on 64-bit systems.
2007
2008 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2009 instance of the 'standard' template library not being so standard".
2010
2011 10. There was no check on the number of named subpatterns nor the maximum
2012 length of a subpattern name. The product of these values is used to compute
2013 the size of the memory block for a compiled pattern. By supplying a very
2014 long subpattern name and a large number of named subpatterns, the size
2015 computation could be caused to overflow. This is now prevented by limiting
2016 the length of names to 32 characters, and the number of named subpatterns
2017 to 10,000.
2018
2019 11. Subpatterns that are repeated with specific counts have to be replicated in
2020 the compiled pattern. The size of memory for this was computed from the
2021 length of the subpattern and the repeat count. The latter is limited to
2022 65535, but there was no limit on the former, meaning that integer overflow
2023 could in principle occur. The compiled length of a repeated subpattern is
2024 now limited to 30,000 bytes in order to prevent this.
2025
2026 12. Added the optional facility to have named substrings with the same name.
2027
2028 13. Added the ability to use a named substring as a condition, using the
2029 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2030 are numbers (not recommended). Forward references are permitted.
2031
2032 14. Added forward references in named backreferences (if you see what I mean).
2033
2034 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2035 pattern could run off the end of the subject. For example, the pattern
2036 "(?s)(.{1,5})"8 did this with the subject "ab".
2037
2038 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2039 PCRE_CASELESS was set when matching characters that were quantified with ?
2040 or *.
2041
2042 17. A character class other than a single negated character that had a minimum
2043 but no maximum quantifier - for example [ab]{6,} - was not handled
2044 correctly by pce_dfa_exec(). It would match only one character.
2045
2046 18. A valid (though odd) pattern that looked like a POSIX character
2047 class but used an invalid character after [ (for example [[,abc,]]) caused
2048 pcre_compile() to give the error "Failed: internal error: code overflow" or
2049 in some cases to crash with a glibc free() error. This could even happen if
2050 the pattern terminated after [[ but there just happened to be a sequence of
2051 letters, a binary zero, and a closing ] in the memory that followed.
2052
2053 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2054 over the years. Originally (before any Unicode support), just the bottom 8
2055 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2056 output from "man perlunicode" includes this:
2057
2058 The regular expression compiler produces polymorphic opcodes. That
2059 is, the pattern adapts to the data and automatically switches to
2060 the Unicode character scheme when presented with Unicode data--or
2061 instead uses a traditional byte scheme when presented with byte
2062 data.
2063
2064 Sadly, a wide octal escape does not cause a switch, and in a string with
2065 no other multibyte characters, these octal escapes are treated as before.
2066 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2067 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2068 Unicode string.
2069
2070 I have not perpetrated such confusion in PCRE. Up till now, it took just
2071 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2072 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2073 translate to the appropriate multibyte character.
2074
2075 29. Applied some refactoring to reduce the number of warnings from Microsoft
2076 and Borland compilers. This has included removing the fudge introduced
2077 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2078 a warning about an unused variable.
2079
2080 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2081 characters since release 4.0, because Perl (from release 5.004) does not.
2082 [Or at least, is documented not to: some releases seem to be in conflict
2083 with the documentation.] However, when a pattern was studied with
2084 pcre_study() and all its branches started with \s, PCRE still included VT
2085 as a possible starting character. Of course, this did no harm; it just
2086 caused an unnecessary match attempt.
2087
2088 22. Removed a now-redundant internal flag bit that recorded the fact that case
2089 dependency changed within the pattern. This was once needed for "required
2090 byte" processing, but is no longer used. This recovers a now-scarce options
2091 bit. Also moved the least significant internal flag bit to the most-
2092 significant bit of the word, which was not previously used (hangover from
2093 the days when it was an int rather than a uint) to free up another bit for
2094 the future.
2095
2096 23. Added support for CRLF line endings as well as CR and LF. As well as the
2097 default being selectable at build time, it can now be changed at runtime
2098 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2099 specify that it is scanning data with non-default line endings.
2100
2101 24. Changed the definition of CXXLINK to make it agree with the definition of
2102 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2103
2104 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2105 recursions. This makes a big different to stack usage for some patterns.
2106
2107 26. If a subpattern containing a named recursion or subroutine reference such
2108 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2109 the space required for the compiled pattern went wrong and gave too small a
2110 value. Depending on the environment, this could lead to "Failed: internal
2111 error: code overflow at offset 49" or "glibc detected double free or
2112 corruption" errors.
2113
2114 27. Applied patches from Google (a) to support the new newline modes and (b) to
2115 advance over multibyte UTF-8 characters in GlobalReplace.
2116
2117 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2118 difference for some implementation of PCRE in some Windows version.
2119
2120 29. Added some extra testing facilities to pcretest:
2121
2122 \q<number> in a data line sets the "match limit" value
2123 \Q<number> in a data line sets the "match recursion limt" value
2124 -S <number> sets the stack size, where <number> is in megabytes
2125
2126 The -S option isn't available for Windows.
2127
2128
2129 Version 6.6 06-Feb-06
2130 ---------------------
2131
2132 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2133 in pcreposix.h. I have copied the definition from pcre.h.
2134
2135 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2136 because pcre.h is no longer a built file.
2137
2138 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2139 not normally included in the compiled code.
2140
2141
2142 Version 6.5 01-Feb-06
2143 ---------------------
2144
2145 1. When using the partial match feature with pcre_dfa_exec(), it was not
2146 anchoring the second and subsequent partial matches at the new starting
2147 point. This could lead to incorrect results. For example, with the pattern
2148 /1234/, partially matching against "123" and then "a4" gave a match.
2149
2150 2. Changes to pcregrep:
2151
2152 (a) All non-match returns from pcre_exec() were being treated as failures
2153 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2154 error message is output. Some extra information is given for the
2155 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2156 probably the only errors that are likely to be caused by users (by
2157 specifying a regex that has nested indefinite repeats, for instance).
2158 If there are more than 20 of these errors, pcregrep is abandoned.
2159
2160 (b) A binary zero was treated as data while matching, but terminated the
2161 output line if it was written out. This has been fixed: binary zeroes
2162 are now no different to any other data bytes.
2163
2164 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2165 used to set a locale for matching. The --locale=xxxx long option has
2166 been added (no short equivalent) to specify a locale explicitly on the
2167 pcregrep command, overriding the environment variables.
2168
2169 (d) When -B was used with -n, some line numbers in the output were one less
2170 than they should have been.
2171
2172 (e) Added the -o (--only-matching) option.
2173
2174 (f) If -A or -C was used with -c (count only), some lines of context were
2175 accidentally printed for the final match.
2176
2177 (g) Added the -H (--with-filename) option.
2178
2179 (h) The combination of options -rh failed to suppress file names for files
2180 that were found from directory arguments.
2181
2182 (i) Added the -D (--devices) and -d (--directories) options.
2183
2184 (j) Added the -F (--fixed-strings) option.
2185
2186 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2187
2188 (l) Added the --colo(u)r option.
2189
2190 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2191 is not present by default.
2192
2193 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2194 items such as (?R) or (?1), when the recursion could match a number of
2195 alternatives. If it matched one of the alternatives, but subsequently,
2196 outside the recursion, there was a failure, the code tried to back up into
2197 the recursion. However, because of the way PCRE is implemented, this is not
2198 possible, and the result was an incorrect result from the match.
2199
2200 In order to prevent this happening, the specification of recursion has
2201 been changed so that all such subpatterns are automatically treated as
2202 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2203
2204 4. I had overlooked the fact that, in some locales, there are characters for
2205 which isalpha() is true but neither isupper() nor islower() are true. In
2206 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2207 and ordfeminine) are like this. This affected the treatment of \w and \W
2208 when they appeared in character classes, but not when they appeared outside
2209 a character class. The bit map for "word" characters is now created
2210 separately from the results of isalnum() instead of just taking it from the
2211 upper, lower, and digit maps. (Plus the underscore character, of course.)
2212
2213 5. The above bug also affected the handling of POSIX character classes such as
2214 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2215 permanent tables. Instead, the bit maps for such a class were previously
2216 created as the appropriate unions of the upper, lower, and digit bitmaps.
2217 Now they are created by subtraction from the [[:word:]] class, which has
2218 its own bitmap.
2219
2220 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2221 It is created by subtracting the vertical space characters (\x09, \x0a,
2222 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2223 subtraction was done in the overall bitmap for a character class, meaning
2224 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2225 be recognized. This bug has been fixed.
2226
2227 7. Patches from the folks at Google:
2228
2229 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2230 real life, but is still worth protecting against".
2231
2232 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2233 regular expressions".
2234
2235 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2236 have it.
2237
2238 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2239 "configure" and the latter not, in order to fix a problem somebody had
2240 with compiling the Arg class on HP-UX.
2241
2242 (e) Improve the error-handling of the C++ wrapper a little bit.
2243
2244 (f) New tests for checking recursion limiting.
2245
2246 8. The pcre_memmove() function, which is used only if the environment does not
2247 have a standard memmove() function (and is therefore rarely compiled),
2248 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2249 returning a result (though PCRE never actually uses the result).
2250
2251 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2252 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2253 returned instead of calling malloc() with an overflowing number that would
2254 most likely cause subsequent chaos.
2255
2256 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2257
2258 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2259 with this option is matched, the nmatch and pmatch options of regexec() are
2260 ignored.
2261
2262 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2263 provided in case anyone wants to the the POSIX interface with UTF-8
2264 strings.
2265
2266 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2267 C++ linking (needed for some HP-UX environments).
2268
2269 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2270 (unused parameter) and in the pcre_printint() function (omitted "default"
2271 switch label when the default is to do nothing).
2272
2273 15. Added some code to make it possible, when PCRE is compiled as a C++
2274 library, to replace subject pointers for pcre_exec() with a smart pointer
2275 class, thus making it possible to process discontinuous strings.
2276
2277 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2278 much the same function. They were added by different people who were trying
2279 to make PCRE easy to compile on non-Unix systems. It has been suggested
2280 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2281 for compiling on Windows systems. I have therefore replaced it with
2282 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2283 defaults to "extern" for C or "extern C" for C++, which works fine on
2284 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2285 SCOPE with something explicit in config.h. In addition:
2286
2287 (a) pcreposix.h still had just "extern" instead of either of these macros;
2288 I have replaced it with PCRE_DATA_SCOPE.
2289
2290 (b) Functions such as _pcre_xclass(), which are internal to the library,
2291 but external in the C sense, all had PCRE_EXPORT in their definitions.
2292 This is apparently wrong for the Windows case, so I have removed it.
2293 (It makes no difference on Unix-like systems.)
2294
2295 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2296 of recursive calls to match(). This is different to MATCH_LIMIT because
2297 that limits the total number of calls to match(), not all of which increase
2298 the depth of recursion. Limiting the recursion depth limits the amount of
2299 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2300 when PCRE is compiled, and changed at run time. A patch from Google adds
2301 this functionality to the C++ interface.
2302
2303 18. Changes to the handling of Unicode character properties:
2304
2305 (a) Updated the table to Unicode 4.1.0.
2306
2307 (b) Recognize characters that are not in the table as "Cn" (undefined).
2308
2309 (c) I revised the way the table is implemented to a much improved format
2310 which includes recognition of ranges. It now supports the ranges that
2311 are defined in UnicodeData.txt, and it also amalgamates other
2312 characters into ranges. This has reduced the number of entries in the
2313 table from around 16,000 to around 3,000, thus reducing its size
2314 considerably. I realized I did not need to use a tree structure after
2315 all - a binary chop search is just as efficient. Having reduced the
2316 number of entries, I extended their size from 6 bytes to 8 bytes to
2317 allow for more data.
2318
2319 (d) Added support for Unicode script names via properties such as \p{Han}.
2320
2321 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2322 matching that character.
2323
2324 20. When matching a repeated Unicode property with a minimum greater than zero,
2325 (for example \pL{2,}), PCRE could look past the end of the subject if it
2326 reached it while seeking the minimum number of characters. This could
2327 happen only if some of the characters were more than one byte long, because
2328 there is a check for at least the minimum number of bytes.
2329
2330 21. Refactored the implementation of \p and \P so as to be more general, to
2331 allow for more different types of property in future. This has changed the
2332 compiled form incompatibly. Anybody with saved compiled patterns that use
2333 \p or \P will have to recompile them.
2334
2335 22. Added "Any" and "L&" to the supported property types.
2336
2337 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2338 but give a compile time error if the value is greater than 0xff.
2339
2340 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2341 accidentally not being installed or uninstalled.
2342
2343 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2344 made were to insert the current release number. This seemed silly, because
2345 it made things harder for people building PCRE on systems that don't run
2346 "configure". I have turned pcre.h into a distributed file, no longer built
2347 by "configure", with the version identification directly included. There is
2348 no longer a pcre.h.in file.
2349
2350 However, this change necessitated a change to the pcre-config script as
2351 well. It is built from pcre-config.in, and one of the substitutions was the
2352 release number. I have updated configure.ac so that ./configure now finds
2353 the release number by grepping pcre.h.
2354
2355 26. Added the ability to run the tests under valgrind.
2356
2357
2358 Version 6.4 05-Sep-05
2359 ---------------------
2360
2361 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2362 "--" to be printed when multiple files were scanned, even when none of the
2363 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2364 consider it to be a bug, and have restored the previous behaviour.
2365
2366 2. A couple of code tidies to get rid of compiler warnings.
2367
2368 3. The pcretest program used to cheat by referring to symbols in the library
2369 whose names begin with _pcre_. These are internal symbols that are not
2370 really supposed to be visible externally, and in some environments it is
2371 possible to suppress them. The cheating is now confined to including
2372 certain files from the library's source, which is a bit cleaner.
2373
2374 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2375 file's purpose clearer.
2376
2377 5. Reorganized pcre_ucp_findchar().
2378
2379
2380 Version 6.3 15-Aug-05
2381 ---------------------
2382
2383 1. The file libpcre.pc.in did not have general read permission in the tarball.
2384
2385 2. There were some problems when building without C++ support:
2386
2387 (a) If C++ support was not built, "make install" and "make test" still
2388 tried to test it.
2389
2390 (b) There were problems when the value of CXX was explicitly set. Some
2391 changes have been made to try to fix these, and ...
2392
2393 (c) --disable-cpp can now be used to explicitly disable C++ support.
2394
2395 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2396 backslash in a target when C++ was disabled. This confuses some
2397 versions of "make", apparently. Using an intermediate variable solves
2398 this. (Same for CPP_LOBJ.)
2399
2400 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2401 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2402 necessary on certain architectures.
2403
2404 4. Added a setting of -export-symbols-regex to the link command to remove
2405 those symbols that are exported in the C sense, but actually are local
2406 within the library, and not documented. Their names all begin with
2407 "_pcre_". This is not a perfect job, because (a) we have to except some
2408 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2409 available (and never for static libraries). I have made a note to try to
2410 find a way round (a) in the future.
2411
2412
2413 Version 6.2 01-Aug-05
2414 ---------------------
2415
2416 1. There was no test for integer overflow of quantifier values. A construction
2417 such as {1111111111111111} would give undefined results. What is worse, if
2418 a minimum quantifier for a parenthesized subpattern overflowed and became
2419 negative, the calculation of the memory size went wrong. This could have
2420 led to memory overwriting.
2421
2422 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2423
2424 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2425 operating environments where this matters.
2426
2427 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2428 PCRE options from within the C++ wrapper.
2429
2430 5. Named capturing subpatterns were not being correctly counted when a pattern
2431 was compiled. This caused two problems: (a) If there were more than 100
2432 such subpatterns, the calculation of the memory needed for the whole
2433 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2434 back references of the form \12, where the number was greater than 9, were
2435 not recognized as back references, even though there were sufficient
2436 previous subpatterns.
2437
2438 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2439 versions of gcc, e.g. 2.95.4.
2440
2441
2442 Version 6.1 21-Jun-05
2443 ---------------------
2444
2445 1. There was one reference to the variable "posix" in pcretest.c that was not
2446 surrounded by "#if !defined NOPOSIX".
2447
2448 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2449 the cross-check on the old pcre_info() function, for the benefit of the
2450 cut-down version of PCRE that is currently imported into Exim.
2451
2452 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2453 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2454 patterns that start (?i) but I don't think that matters. The use of (?i) is
2455 just an example; this all applies to the other options as well.
2456
2457 4. Since libtool seems to echo the compile commands it is issuing, the output
2458 from "make" can be reduced a bit by putting "@" in front of each libtool
2459 compile command.
2460
2461 5. Patch from the folks at Google for configure.in to be a bit more thorough
2462 in checking for a suitable C++ installation before trying to compile the
2463 C++ stuff. This should fix a reported problem when a compiler was present,
2464 but no suitable headers.
2465
2466 6. The man pages all had just "PCRE" as their title. I have changed them to
2467 be the relevant file name. I have also arranged that these names are
2468 retained in the file doc/pcre.txt, which is a concatenation in text format
2469 of all the man pages except the little individual ones for each function.
2470
2471 7. The NON-UNIX-USE file had not been updated for the different set of source
2472 files that come with release 6. I also added a few comments about the C++
2473 wrapper.
2474
2475
2476 Version 6.0 07-Jun-05
2477 ---------------------
2478
2479 1. Some minor internal re-organization to help with my DFA experiments.
2480
2481 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2482 didn't matter for the library itself when fully configured, but did matter
2483 when compiling without UCP support, or within Exim, where the ucp files are
2484 not imported.
2485
2486 3. Refactoring of the library code to split up the various functions into
2487 different source modules. The addition of the new DFA matching code (see
2488 below) to a single monolithic source would have made it really too
2489 unwieldy, quite apart from causing all the code to be include in a
2490 statically linked application, when only some functions are used. This is
2491 relevant even without the DFA addition now that patterns can be compiled in
2492 one application and matched in another.
2493
2494 The downside of splitting up is that there have to be some external
2495 functions and data tables that are used internally in different modules of
2496 the library but which are not part of the API. These have all had their
2497 names changed to start with "_pcre_" so that they are unlikely to clash
2498 with other external names.
2499
2500 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2501 a different (DFA) algorithm. Although it is slower than the original
2502 function, it does have some advantages for certain types of matching
2503 problem.
2504
2505 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2506 including restarting after a partial match.
2507
2508 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2509 defined when compiling for Windows was sent to me. I have put it into the
2510 code, though I have no means of testing or verifying it.
2511
2512 7. Added the pcre_refcount() auxiliary function.
2513
2514 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2515 match before or at the first newline in the subject string. In pcretest,
2516 the /f option on a pattern can be used to set this.
2517
2518 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2519 would behave wrongly. This has been present in PCRE since release 4.0.
2520
2521 10. A number of changes to the pcregrep command:
2522
2523 (a) Refactored how -x works; insert ^(...)$ instead of setting
2524 PCRE_ANCHORED and checking the length, in preparation for adding
2525 something similar for -w.
2526
2527 (b) Added the -w (match as a word) option.
2528
2529 (c) Refactored the way lines are read and buffered so as to have more
2530 than one at a time available.
2531
2532 (d) Implemented a pcregrep test script.
2533
2534 (e) Added the -M (multiline match) option. This allows patterns to match
2535 over several lines of the subject. The buffering ensures that at least
2536 8K, or the rest of the document (whichever is the shorter) is available
2537 for matching (and similarly the previous 8K for lookbehind assertions).
2538
2539 (f) Changed the --help output so that it now says
2540
2541 -w, --word-regex(p)
2542
2543 instead of two lines, one with "regex" and the other with "regexp"
2544 because that confused at least one person since the short forms are the
2545 same. (This required a bit of code, as the output is generated
2546 automatically from a table. It wasn't just a text change.)
2547
2548 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2549 option but starts with a hyphen. Could be a pattern or a path name
2550 starting with a hyphen, for instance.
2551
2552 (h) "-" can be given as a file name to represent stdin.
2553
2554 (i) When file names are being printed, "(standard input)" is used for
2555 the standard input, for compatibility with GNU grep. Previously
2556 "<stdin>" was used.
2557
2558 (j) The option --label=xxx can be used to supply a name to be used for
2559 stdin when file names are being printed. There is no short form.
2560
2561 (k) Re-factored the options decoding logic because we are going to add
2562 two more options that take data. Such options can now be given in four
2563 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2564
2565 (l) Added the -A, -B, and -C options for requesting that lines of context
2566 around matches be printed.
2567
2568 (m) Added the -L option to print the names of files that do not contain
2569 any matching lines, that is, the complement of -l.
2570
2571 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2572 continue to scan other files.
2573
2574 (o) The -s option was incorrectly implemented. For compatibility with other
2575 greps, it now suppresses the error message for a non-existent or non-
2576 accessible file (but not the return code). There is a new option called
2577 -q that suppresses the output of matching lines, which was what -s was
2578 previously doing.
2579
2580 (p) Added --include and --exclude options to specify files for inclusion
2581 and exclusion when recursing.
2582
2583 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2584 Hopefully, it now does.
2585
2586 12. Missing cast in pcre_study().
2587
2588 13. Added an "uninstall" target to the makefile.
2589
2590 14. Replaced "extern" in the function prototypes in Makefile.in with
2591 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2592 world, but is set differently for Windows.
2593
2594 15. Added a second compiling function called pcre_compile2(). The only
2595 difference is that it has an extra argument, which is a pointer to an
2596 integer error code. When there is a compile-time failure, this is set
2597 non-zero, in addition to the error test pointer being set to point to an
2598 error message. The new argument may be NULL if no error number is required
2599 (but then you may as well call pcre_compile(), which is now just a
2600 wrapper). This facility is provided because some applications need a
2601 numeric error indication, but it has also enabled me to tidy up the way
2602 compile-time errors are handled in the POSIX wrapper.
2603
2604 16. Added VPATH=.libs to the makefile; this should help when building with one
2605 prefix path and installing with another. (Or so I'm told by someone who
2606 knows more about this stuff than I do.)
2607
2608 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2609 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2610 match everything, including newlines. This is not POSIX-compatible, but
2611 somebody wanted the feature. From pcretest it can be activated by using
2612 both the P and the s flags.
2613
2614 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2615
2616 19. libpcre.pc was being incorrectly installed as executable.
2617
2618 20. A couple of places in pcretest check for end-of-line by looking for '\n';
2619 it now also looks for '\r' so that it will work unmodified on Windows.
2620
2621 21. Added Google's contributed C++ wrapper to the distribution.
2622
2623 22. Added some untidy missing memory free() calls in pcretest, to keep
2624 Electric Fence happy when testing.
2625
2626
2627
2628 Version 5.0 13-Sep-04
2629 ---------------------
2630
2631 1. Internal change: literal characters are no longer packed up into items
2632 containing multiple characters in a single byte-string. Each character
2633 is now matched using a separate opcode. However, there may be more than one
2634 byte in the character in UTF-8 mode.
2635
2636 2. The pcre_callout_block structure has two new fields: pattern_position and
2637 next_item_length. These contain the offset in the pattern to the next match
2638 item, and its length, respectively.
2639
2640 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2641 insertion of callouts before each pattern item. Added the /C option to
2642 pcretest to make use of this.
2643
2644 4. On the advice of a Windows user, the lines
2645
2646 #if defined(_WIN32) || defined(WIN32)
2647 _setmode( _fileno( stdout ), 0x8000 );
2648 #endif /* defined(_WIN32) || defined(WIN32) */
2649
2650 have been added to the source of pcretest. This apparently does useful
2651 magic in relation to line terminators.
2652
2653 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
2654 for the benefit of those environments where the "b" makes a difference.
2655
2656 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
2657 to know about it. I have put a hack into configure.in that adds in code
2658 to set GCC=yes if CC=icc. This seems to end up at a point in the
2659 generated configure script that is early enough to affect the setting of
2660 compiler options, which is what is needed, but I have no means of testing
2661 whether it really works. (The user who reported this had patched the
2662 generated configure script, which of course I cannot do.)
2663
2664 LATER: After change 22 below (new libtool files), the configure script
2665 seems to know about icc (and also ecc). Therefore, I have commented out
2666 this hack in configure.in.
2667
2668 7. Added support for pkg-config (2 patches were sent in).
2669
2670 8. Negated POSIX character classes that used a combination of internal tables
2671 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
2672 [[:^ascii]]. Typically, they would match almost any characters. The other
2673 POSIX classes were not broken in this way.
2674
2675 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
2676 to find the match, as PCRE was deluded into thinking that the match had to
2677 start at the start point or following a newline. The same bug applied to
2678 patterns with negative forward assertions or any backward assertions
2679 preceding ".*" at the start, unless the pattern required a fixed first
2680 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
2681
2682 10. In UTF-8 mode, when moving forwards in the subject after a failed match
2683 starting at the last subject character, bytes beyond the end of the subject
2684 string were read.
2685
2686 11. Renamed the variable "class" as "classbits" to make life easier for C++
2687 users. (Previously there was a macro definition, but it apparently wasn't
2688 enough.)
2689
2690 12. Added the new field "tables" to the extra data so that tables can be passed
2691 in at exec time, or the internal tables can be re-selected. This allows
2692 a compiled regex to be saved and re-used at a later time by a different
2693 program that might have everything at different addresses.
2694
2695 13. Modified the pcre-config script so that, when run on Solaris, it shows a
2696 -R library as well as a -L library.
2697
2698 14. The debugging options of pcretest (-d on the command line or D on a
2699 pattern) showed incorrect output for anything following an extended class
2700 that contained multibyte characters and which was followed by a quantifier.
2701
2702 15. Added optional support for general category Unicode character properties
2703 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
2704 support. It adds about 90K to the size of the library. The meanings of the
2705 inbuilt class escapes such as \d and \s have NOT been changed.
2706
2707 16. Updated pcredemo.c to include calls to free() to release the memory for the
2708 compiled pattern.
2709
2710 17. The generated file chartables.c was being created in the source directory
2711 instead of in the building directory. This caused the build to fail if the
2712 source directory was different from the building directory, and was
2713 read-only.
2714
2715 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
2716 file. No doubt somebody will tell me if they don't make sense... Also added
2717 Dan Mooney's comments about building on OpenVMS.
2718
2719 19. Added support for partial matching via the PCRE_PARTIAL option for
2720 pcre_exec() and the \P data escape in pcretest.
2721
2722 20. Extended pcretest with 3 new pattern features:
2723
2724 (i) A pattern option of the form ">rest-of-line" causes pcretest to
2725 write the compiled pattern to the file whose name is "rest-of-line".
2726 This is a straight binary dump of the data, with the saved pointer to
2727 the character tables forced to be NULL. The study data, if any, is
2728 written too. After writing, pcretest reads a new pattern.
2729
2730 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
2731 compiled pattern from the given file. There must not be any
2732 occurrences of "<" in the file name (pretty unlikely); if there are,
2733 pcretest will instead treat the initial "<" as a pattern delimiter.
2734 After reading in the pattern, pcretest goes on to read data lines as
2735 usual.
2736
2737 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
2738 and 16-bit fields in a compiled pattern, to simulate a pattern that
2739 was compiled on a host of opposite endianness.
2740
2741 21. The pcre-exec() function can now cope with patterns that were compiled on
2742 hosts of opposite endianness, with this restriction:
2743
2744 As for any compiled expression that is saved and used later, the tables
2745 pointer field cannot be preserved; the extra_data field in the arguments
2746 to pcre_exec() should be used to pass in a tables address if a value
2747 other than the default internal tables were used at compile time.
2748
2749 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
2750 now diagnosed as an error. Previously, most of the time, a negative number
2751 would have been treated as zero, but if in addition "ovector" was passed as
2752 NULL, a crash could occur.
2753
2754 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
2755 new versions from the libtool 1.5 distribution (the last one is a copy of
2756 a file called libtool.m4). This seems to have fixed the need to patch
2757 "configure" to support Darwin 1.3 (which I used to do). However, I still
2758 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
2759 workstation).
2760
2761 24. Changed the PCRE licence to be the more standard "BSD" licence.
2762
2763
2764 Version 4.5 01-Dec-03
2765 ---------------------
2766
2767 1. There has been some re-arrangement of the code for the match() function so
2768 that it can be compiled in a version that does not call itself recursively.
2769 Instead, it keeps those local variables that need separate instances for
2770 each "recursion" in a frame on the heap, and gets/frees frames whenever it
2771 needs to "recurse". Keeping track of where control must go is done by means
2772 of setjmp/longjmp. The whole thing is implemented by a set of macros that
2773 hide most of the details from the main code, and operates only if
2774 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
2775 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
2776 operating.
2777
2778 To make it easier for callers to provide specially tailored get/free
2779 functions for this usage, two new functions, pcre_stack_malloc, and
2780 pcre_stack_free, are used. They are always called in strict stacking order,
2781 and the size of block requested is always the same.
2782
2783 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
2784 PCRE has been compiled to use the stack or the heap for recursion. The
2785 -C option of pcretest uses this to show which version is compiled.
2786
2787 A new data escape \S, is added to pcretest; it causes the amounts of store
2788 obtained and freed by both kinds of malloc/free at match time to be added
2789 to the output.
2790
2791 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
2792 what's available on my current Linux desktop machine.
2793
2794 3. When matching a UTF-8 string, the test for a valid string at the start has
2795 been extended. If start_offset is not zero, PCRE now checks that it points
2796 to a byte that is the start of a UTF-8 character. If not, it returns
2797 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
2798 this is necessary because there may be backward assertions in the pattern.
2799 When matching the same subject several times, it may save resources to use
2800 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
2801
2802 4. The code for checking the validity of UTF-8 strings has been tightened so
2803 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
2804 containing "overlong sequences".
2805
2806 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
2807 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
2808 should have been "&", but it just so happened that all the cases this let
2809 through by mistake were picked up later in the function.
2810
2811 6. I had used a variable called "isblank" - this is a C99 function, causing
2812 some compilers to warn. To avoid this, I renamed it (as "blankclass").
2813
2814 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
2815 prompting; (b) run "./pcretest /dev/null" at the start of the test script
2816 so the version is shown; (c) stop "make test" echoing "./RunTest".
2817
2818 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
2819
2820 9. The prototype for memmove() for systems that don't have it was using
2821 size_t, but the inclusion of the header that defines size_t was later. I've
2822 moved the #includes for the C headers earlier to avoid this.
2823
2824 10. Added some adjustments to the code to make it easier to compiler on certain
2825 special systems:
2826
2827 (a) Some "const" qualifiers were missing.
2828 (b) Added the macro EXPORT before all exported functions; by default this
2829 is defined to be empty.
2830 (c) Changed the dftables auxiliary program (that builds chartables.c) so
2831 that it reads its output file name as an argument instead of writing
2832 to the standard output and assuming this can be redirected.
2833
2834 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
2835 class containing characters with values greater than 255, PCRE compilation
2836 went into a loop.
2837
2838 12. A recursive reference to a subpattern that was within another subpattern
2839 that had a minimum quantifier of zero caused PCRE to crash. For example,
2840 (x(y(?2))z)? provoked this bug with a subject that got as far as the
2841 recursion. If the recursively-called subpattern itself had a zero repeat,
2842 that was OK.
2843
2844 13. In pcretest, the buffer for reading a data line was set at 30K, but the
2845 buffer into which it was copied (for escape processing) was still set at
2846 1024, so long lines caused crashes.
2847
2848 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
2849 "internal error: code overflow...". This applied to any character class
2850 that was followed by a possessive quantifier.
2851
2852 15. Modified the Makefile to add libpcre.la as a prerequisite for
2853 libpcreposix.la because I was told this is needed for a parallel build to
2854 work.
2855
2856 16. If a pattern that contained .* following optional items at the start was
2857 studied, the wrong optimizing data was generated, leading to matching
2858 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
2859 matching string must start with a or b or c. The correct conclusion for
2860 this pattern is that a match can start with any character.
2861
2862
2863 Version 4.4 13-Aug-03
2864 ---------------------
2865
2866 1. In UTF-8 mode, a character class containing characters with values between
2867 127 and 255 was not handled correctly if the compiled pattern was studied.
2868 In fixing this, I have also improved the studying algorithm for such
2869 classes (slightly).
2870
2871 2. Three internal functions had redundant arguments passed to them. Removal
2872 might give a very teeny performance improvement.
2873
2874 3. Documentation bug: the value of the capture_top field in a callout is *one
2875 more than* the number of the hightest numbered captured substring.
2876
2877 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
2878 in incorrectly linking with a previously installed version. They now link
2879 explicitly with libpcre.la.
2880
2881 5. configure.in no longer needs to recognize Cygwin specially.
2882
2883 6. A problem in pcre.in for Windows platforms is fixed.
2884
2885 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
2886 pcretest, it used to include the size of the study block as part of its
2887 output. Unfortunately, the structure contains a field that has a different
2888 size on different hardware architectures. This meant that the tests that
2889 showed this size failed. As the block is currently always of a fixed size,
2890 this information isn't actually particularly useful in pcretest output, so
2891 I have just removed it.
2892
2893 8. Three pre-processor statements accidentally did not start in column 1.
2894 Sadly, there are *still* compilers around that complain, even though
2895 standard C has not required this for well over a decade. Sigh.
2896
2897 9. In pcretest, the code for checking callouts passed small integers in the
2898 callout_data field, which is a void * field. However, some picky compilers
2899 complained about the casts involved for this on 64-bit systems. Now
2900 pcretest passes the address of the small integer instead, which should get
2901 rid of the warnings.
2902
2903 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
2904 both compile and run time, and gives an error if an invalid UTF-8 sequence
2905 is found. There is a option for disabling this check in cases where the
2906 string is known to be correct and/or the maximum performance is wanted.
2907
2908 11. In response to a bug report, I changed one line in Makefile.in from
2909
2910 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
2911 to
2912 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
2913
2914 to look similar to other lines, but I have no way of telling whether this
2915 is the right thing to do, as I do not use Windows. No doubt I'll get told
2916 if it's wrong...
2917
2918
2919 Version 4.3 21-May-03
2920 ---------------------
2921
2922 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
2923 Makefile.
2924
2925 2. Some refactoring to improve the quality of the code:
2926
2927 (i) The utf8_table... variables are now declared "const".
2928
2929 (ii) The code for \cx, which used the "case flipping" table to upper case
2930 lower case letters, now just substracts 32. This is ASCII-specific,
2931 but the whole concept of \cx is ASCII-specific, so it seems
2932 reasonable.
2933
2934 (iii) PCRE was using its character types table to recognize decimal and
2935 hexadecimal digits in the pattern. This is silly, because it handles
2936 only 0-9, a-f, and A-F, but the character types table is locale-
2937 specific, which means strange things might happen. A private
2938 table is now used for this - though it costs 256 bytes, a table is
2939 much faster than multiple explicit tests. Of course, the standard
2940 character types table is still used for matching digits in subject
2941 strings against \d.
2942
2943 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
2944 ending in _t are). So I've renamed it as ESC_tee.
2945
2946 3. The first argument for regexec() in the POSIX wrapper should have been
2947 defined as "const".
2948
2949 4. Changed pcretest to use malloc() for its buffers so that they can be
2950 Electric Fenced for debugging.
2951
2952 5. There were several places in the code where, in UTF-8 mode, PCRE would try
2953 to read one or more bytes before the start of the subject string. Often this
2954 had no effect on PCRE's behaviour, but in some circumstances it could
2955 provoke a segmentation fault.
2956
2957 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
2958 to try to read one or more bytes before the start of the subject string.
2959
2960 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
2961 UTF-8 support could misbehave in various ways if the subject string
2962 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
2963 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
2964 back over UTF-8 characters.)
2965
2966
2967 Version 4.2 14-Apr-03
2968 ---------------------
2969
2970 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
2971
2972 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
2973 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
2974 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
2975 [WIN_PREFIX]: new variable, "cyg" for Cygwin
2976 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
2977 and BUILD_EXEEXT
2978 Note: automatic setting of the BUILD variables is not yet working
2979 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
2980 compile-time but not at link-time
2981 [LINK]: use for linking executables only
2982 make different versions for Windows and non-Windows
2983 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
2984 libraries
2985 [LINK_FOR_BUILD]: new variable
2986 [OBJEXT]: use throughout
2987 [EXEEXT]: use throughout
2988 <winshared>: new target
2989 <wininstall>: new target
2990 <dftables.o>: use native compiler
2991 <dftables>: use native linker
2992 <install>: handle Windows platform correctly
2993 <clean>: ditto
2994 <check>: ditto
2995 copy DLL to top builddir before testing
2996
2997 As part of these changes, -no-undefined was removed again. This was reported
2998 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
2999 in any case.
3000
3001 3. Some tidies to get rid of compiler warnings:
3002
3003 . In the match_data structure, match_limit was an unsigned long int, whereas
3004 match_call_count was an int. I've made them both unsigned long ints.
3005
3006 . In pcretest the fact that a const uschar * doesn't automatically cast to
3007 a void * provoked a warning.
3008
3009 . Turning on some more compiler warnings threw up some "shadow" variables
3010 and a few more missing casts.
3011
3012 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3013 option, a class that contained a single character with a value between 128
3014 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3015
3016 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3017 option, a class that contained several characters, but with at least one
3018 whose value was between 128 and 255 caused PCRE to crash.
3019
3020
3021 Version 4.1 12-Mar-03
3022 ---------------------
3023
3024 1. Compiling with gcc -pedantic found a couple of places where casts were
3025 needed, and a string in dftables.c that was longer than standard compilers are
3026 required to support.
3027
3028 2. Compiling with Sun's compiler found a few more places where the code could
3029 be tidied up in order to avoid warnings.
3030
3031 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3032 first of these names is deprecated in the latest Autoconf in favour of the name
3033 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3034 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3035 analogy I have changed it to CFLAGS_FOR_BUILD.
3036
3037 4. Added -no-undefined to the linking command in the Makefile, because this is
3038 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3039 linking step for the pcreposix library.
3040
3041 5. PCRE was failing to diagnose the case of two named groups with the same
3042 name.
3043
3044 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3045 literal character that is needed in the subject for a match, and scans along to
3046 ensure that it is present before embarking on the full matching process. This
3047 saves time in cases of nested unlimited repeats that are never going to match.
3048 Problem: the scan can take a lot of time if the subject is very long (e.g.
3049 megabytes), thus penalizing straightforward matches. It is now done only if the
3050 amount of subject to be scanned is less than 1000 bytes.
3051
3052 7. A lesser problem with the same optimization is that it was recording the
3053 first character of an anchored pattern as "needed", thus provoking a search
3054 right along the subject, even when the first match of the pattern was going to
3055 fail. The "needed" character is now not set for anchored patterns, unless it
3056 follows something in the pattern that is of non-fixed length. Thus, it still
3057 fulfils its original purpose of finding quick non-matches in cases of nested
3058 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3059
3060
3061 Version 4.0 17-Feb-03
3062 ---------------------
3063
3064 1. If a comment in an extended regex that started immediately after a meta-item
3065 extended to the end of string, PCRE compiled incorrect data. This could lead to
3066 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3067
3068 2. Moved to autoconf 2.53 and libtool 1.4.2.
3069
3070 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3071 the special perltest8 script is no longer needed - all the tests can be run
3072 from a single perltest script.
3073
3074 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3075 by \s. It has now been removed in PCRE. This means it isn't recognized as
3076 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3077 class [:space:] *does* include VT, thereby creating a mess.
3078
3079 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3080 space and tab.
3081
3082 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3083 its new features into the main test script, reducing the number of scripts.
3084
3085 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3086 were backward compatible, and made the (?i) apply to the whole pattern, as if
3087 /i were given. Now it behaves more logically, and applies the option setting
3088 only to what follows. PCRE has been changed to follow suit. However, if it
3089 finds options settings right at the start of the pattern, it extracts them into
3090 the global options, as before. Thus, they show up in the info data.
3091
3092 8. Added support for the \Q...\E escape sequence. Characters in between are
3093 treated as literals. This is slightly different from Perl in that $ and @ are
3094 also handled as literals inside the quotes. In Perl, they will cause variable
3095 interpolation. Note the following examples:
3096
3097 Pattern PCRE matches Perl matches
3098
3099 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3100 \Qabc\$xyz\E abc\$xyz abc\$xyz
3101 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3102
3103 For compatibility with Perl, \Q...\E sequences are recognized inside character
3104 classes as well as outside them.
3105
3106 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3107 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3108 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3109 signed/unsigned warnings.
3110
3111 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3112 option for pcretest, so I've replaced it by a simple function that does just
3113 that job.
3114
3115 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3116 "pcregrep -".
3117
3118 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3119 Java package. This provides some syntactic sugar for simple cases of what my
3120 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3121 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3122 item, you can use this simplified notation. Note that only makes sense with
3123 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3124 greediness, whatever the setting of the PCRE_UNGREEDY option.
3125
3126 13. A change of greediness default within a pattern was not taking effect at
3127 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3128 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3129 was abstracted outside.
3130
3131 14. PCRE now supports the \G assertion. It is true when the current matching
3132 position is at the start point of the match. This differs from \A when the
3133 starting offset is non-zero. Used with the /g option of pcretest (or similar
3134 code), it works in the same way as it does for Perl's /g option. If all
3135 alternatives of a regex begin with \G, the expression is anchored to the start
3136 match position, and the "anchored" flag is set in the compiled expression.
3137
3138 15. Some bugs concerning the handling of certain option changes within patterns
3139 have been fixed. These applied to options other than (?ims). For example,
3140 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3141 been the other way round. Some of this was related to change 7 above.
3142
3143 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3144 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3145 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3146 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3147
3148 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3149 mode. Unlike ".", it always matches newline, whatever the setting of
3150 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3151 assertions. Perl allows it, but it doesn't (in general) work because it can't
3152 calculate the length of the lookbehind. At least, that's the case for Perl
3153 5.8.0 - I've been told they are going to document that it doesn't work in
3154 future.
3155
3156 18. Added an error diagnosis for escapes that PCRE does not support: these are
3157 \L, \l, \N, \P, \p, \U, \u, and \X.
3158
3159 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3160 reading past the end of the pattern in cases such as /[abcd/.
3161
3162 20. PCRE was getting more memory than necessary for patterns with classes that
3163 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3164
3165 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3166 compiling PCRE for use with Virtual Pascal.
3167
3168 22. Small fix to the Makefile to make it work properly if the build is done
3169 outside the source tree.
3170
3171 23. Added a new extension: a condition to go with recursion. If a conditional
3172 subpattern starts with (?(R) the "true" branch is used if recursion has
3173 happened, whereas the "false" branch is used only at the top level.
3174
3175 24. When there was a very long string of literal characters (over 255 bytes
3176 without UTF support, over 250 bytes with UTF support), the computation of how
3177 much memory was required could be incorrect, leading to segfaults or other
3178 strange effects.
3179
3180 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3181 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3182 there was a subsequent back reference to those brackets. This meant that, for
3183 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3184 possible to check for precisely this case. All we can do is abandon the
3185 optimization if .* occurs inside capturing brackets when there are any back
3186 references whatsoever. (See below for a better fix that came later.)
3187
3188 26. The handling of the optimization for finding the first character of a
3189 non-anchored pattern, and for finding a character that is required later in the
3190 match were failing in some cases. This didn't break the matching; it just
3191 failed to optimize when it could. The way this is done has been re-implemented.
3192
3193 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3194
3195 28. Added a new feature that provides some of the functionality that Perl
3196 provides with (?{...}). The facility is termed a "callout". The way it is done
3197 in PCRE is for the caller to provide an optional function, by setting
3198 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3199 global variable. By default it is unset, which disables all calling out. To get
3200 the function called, the regex must include (?C) at appropriate points. This
3201 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3202 This provides a means of identifying different callout points. When PCRE
3203 reaches such a point in the regex, if pcre_callout has been set, the external
3204 function is called. It is provided with data in a structure called
3205 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3206 matching continues; if it returns a non-zero value, the match at the current
3207 point fails. However, backtracking will occur if possible. [This was changed
3208 later and other features added - see item 49 below.]
3209
3210 29. pcretest is upgraded to test the callout functionality. It provides a
3211 callout function that displays information. By default, it shows the start of
3212 the match and the current position in the text. There are some new data escapes
3213 to vary what happens:
3214
3215 \C+ in addition, show current contents of captured substrings
3216 \C- do not supply a callout function
3217 \C!n return 1 when callout number n is reached
3218 \C!n!m return 1 when callout number n is reached for the mth time
3219
3220 30. If pcregrep was called with the -l option and just a single file name, it
3221 output "<stdin>" if a match was found, instead of the file name.
3222
3223 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3224 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3225 pcre_exec(). This saves a malloc/free per call. The default value of
3226 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3227 when configuring.
3228
3229 32. The default maximum size of a compiled pattern is 64K. There have been a
3230 few cases of people hitting this limit. The code now uses macros to handle the
3231 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3232 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3233 configuring. Tests 2 and 5 work only with 2-byte links because they output
3234 debugging information about compiled patterns.
3235
3236 33. Internal code re-arrangements:
3237
3238 (a) Moved the debugging function for printing out a compiled regex into
3239 its own source file (printint.c) and used #include to pull it into
3240 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3241 separate copies.
3242
3243 (b) Defined the list of op-code names for debugging as a macro in
3244 internal.h so that it is next to the definition of the opcodes.
3245
3246 (c) Defined a table of op-code lengths for simpler skipping along compiled
3247 code. This is again a macro in internal.h so that it is next to the
3248 definition of the opcodes.
3249
3250 34. Added support for recursive calls to individual subpatterns, along the
3251 lines of Robin Houston's patch (but implemented somewhat differently).
3252
3253 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3254 allow it to read and process whole directories in Win32. This code was
3255 contributed by Lionel Fourquaux; it has not been tested by me.
3256
3257 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3258 used to name a group. Names consist of alphanumerics and underscores, and must
3259 be unique. Back references use the syntax (?P=name) and recursive calls use
3260 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3261 numbers. The function pcre_fullinfo() can be used after compilation to extract
3262 a name/number map. There are three relevant calls:
3263
3264 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3265 PCRE_INFO_NAMECOUNT yields the number of entries
3266 PCRE_INFO_NAMETABLE yields a pointer to the map.
3267
3268 The map is a vector of fixed-size entries. The size of each entry depends on
3269 the length of the longest name used. The first two bytes of each entry are the
3270 group number, most significant byte first. There follows the corresponding
3271 name, zero terminated. The names are in alphabetical order.
3272
3273 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3274 case instead of 255. Making it the same both with and without UTF-8 support
3275 means that the same test output works with both.
3276
3277 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3278 calling malloc() with a zero argument.
3279
3280 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3281 optimization. I've improved things by keeping a bitmap of backreferences with
3282 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3283 fact referenced, the optimization can be applied. It is unlikely that a
3284 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3285 the match to follow \n) will appear inside brackets with a number greater than
3286 31, but if it does, any back reference > 31 suppresses the optimization.
3287
3288 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3289 of disabling numbered capturing parentheses. Any opening parenthesis that is
3290 not followed by ? behaves as if it were followed by ?: but named parentheses
3291 can still be used for capturing (and they will acquire numbers in the usual
3292 way).
3293
3294 41. Redesigned the return codes from the match() function into yes/no/error so
3295 that errors can be passed back from deep inside the nested calls. A malloc
3296 failure while inside a recursive subpattern call now causes the
3297 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3298
3299 42. It is now possible to set a limit on the number of times the match()
3300 function is called in a call to pcre_exec(). This facility makes it possible to
3301 limit the amount of recursion and backtracking, though not in a directly
3302 obvious way, because the match() function is used in a number of different
3303 circumstances. The count starts from zero for each position in the subject
3304 string (for non-anchored patterns). The default limit is, for compatibility, a
3305 large number, namely 10 000 000. You can change this in two ways:
3306
3307 (a) When configuring PCRE before making, you can use --with-match-limit=n
3308 to set a default value for the compiled library.
3309
3310 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3311 a different value is set. See 45 below.
3312
3313 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3314
3315 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3316 of things that can be changed at compile time. The first argument specifies
3317 what is wanted and the second points to where the information is to be placed.
3318 The current list of available information is:
3319
3320 PCRE_CONFIG_UTF8
3321
3322 The output is an integer that is set to one if UTF-8 support is available;
3323 otherwise it is set to zero.
3324
3325 PCRE_CONFIG_NEWLINE
3326
3327 The output is an integer that it set to the value of the code that is used for
3328 newline. It is either LF (10) or CR (13).
3329
3330 PCRE_CONFIG_LINK_SIZE
3331
3332 The output is an integer that contains the number of bytes used for internal
3333 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3334
3335 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3336
3337 The output is an integer that contains the threshold above which the POSIX
3338 interface uses malloc() for output vectors. See item 31 above.
3339
3340 PCRE_CONFIG_MATCH_LIMIT
3341
3342 The output is an unsigned integer that contains the default limit of the number
3343 of match() calls in a pcre_exec() execution. See 42 above.
3344
3345 44. pcretest has been upgraded by the addition of the -C option. This causes it
3346 to extract all the available output from the new pcre_config() function, and to
3347 output it. The program then exits immediately.
3348
3349 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3350 order to support additional features. One way would have been to define
3351 pcre_exec2() (for example) with extra arguments, but this would not have been
3352 extensible, and would also have required all calls to the original function to
3353 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3354 is used for passing in "extra" data from pcre_study().
3355
3356 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3357 contains the following fields:
3358
3359 flags a bitmap indicating which of the following fields are set
3360 study_data opaque data from pcre_study()
3361 match_limit a way of specifying a limit on match() calls for a specific
3362 call to pcre_exec()
3363 callout_data data for callouts (see 49 below)
3364
3365 The flag bits are also defined in pcre.h, and are
3366
3367 PCRE_EXTRA_STUDY_DATA
3368 PCRE_EXTRA_MATCH_LIMIT
3369 PCRE_EXTRA_CALLOUT_DATA
3370
3371 The pcre_study() function now returns one of these new pcre_extra blocks, with
3372 the actual study data pointed to by the study_data field, and the
3373 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3374 before. That is, this change is entirely upwards-compatible and requires no
3375 change to existing code.
3376
3377 If you want to pass in additional data to pcre_exec(), you can either place it
3378 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3379 block.
3380
3381 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3382 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3383 times with different match limits, until it finds the minimum value needed for
3384 pcre_exec() to complete. The value is then output. This can be instructive; for
3385 most simple matches the number is quite small, but for pathological cases it
3386 gets very large very quickly.
3387
3388 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3389 returns the size of the data block pointed to by the study_data field in a
3390 pcre_extra block, that is, the value that was passed as the argument to
3391 pcre_malloc() when PCRE was getting memory in which to place the information
3392 created by pcre_study(). The fourth argument should point to a size_t variable.
3393 pcretest has been extended so that this information is shown after a successful
3394 pcre_study() call when information about the compiled regex is being displayed.
3395
3396 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3397 because what follows is always an absolute path. (Later: it turns out that this
3398 is more than cosmetic for MinGW, because it doesn't like empty path
3399 components.)
3400
3401 49. Some changes have been made to the callout feature (see 28 above):
3402
3403 (i) A callout function now has three choices for what it returns:
3404
3405 0 => success, carry on matching
3406 > 0 => failure at this point, but backtrack if possible
3407 < 0 => serious error, return this value from pcre_exec()
3408
3409 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3410 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3411 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3412 use by callout functions. It will never be used by PCRE itself.
3413
3414 (ii) The pcre_extra structure (see 45 above) has a void * field called
3415 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3416 pcre_callout_block structure has a field of the same name. The contents of
3417 the field passed in the pcre_extra structure are passed to the callout
3418 function in the corresponding field in the callout block. This makes it
3419 easier to use the same callout-containing regex from multiple threads. For
3420 testing, the pcretest program has a new data escape
3421
3422 \C*n pass the number n (may be negative) as callout_data
3423
3424 If the callout function in pcretest receives a non-zero value as
3425 callout_data, it returns that value.
3426
3427 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3428 there were some redundant $(CFLAGS) in commands that are now specified as
3429 $(LINK), which already includes $(CFLAGS).
3430
3431 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3432 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3433 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3434 one-byte characters throughout. Note that case-insensitive matching applies
3435 only to characters whose values are less than 256. PCRE doesn't support the
3436 notion of cases for higher-valued characters.
3437
3438 (i) A character class whose characters are all within 0-255 is handled as
3439 a bit map, and the map is inverted for negative classes. Previously, a
3440 character > 255 always failed to match such a class; however it should
3441 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3442
3443 (ii) A negated character class with a single character < 255 is coded as
3444 "not this character" (OP_NOT). This wasn't working properly when the test
3445 character was multibyte, either singly or repeated.
3446
3447 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3448 mode, for example: \x{100}{2,3}.
3449
3450 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3451 singly or repeated) now correctly test multibyte characters. However,
3452 PCRE doesn't recognize any characters with values greater than 255 as
3453 digits, spaces, or word characters. Such characters always match \D, \S,
3454 and \W, and never match \d, \s, or \w.
3455
3456 (v) Classes may now contain characters and character ranges with values
3457 greater than 255. For example: [ab\x{100}-\x{400}].
3458
3459 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3460 PCRE in UTF-8 mode.
3461
3462 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3463 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3464 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3465 value.)
3466
3467 53. The single man page has become too large. I have therefore split it up into
3468 a number of separate man pages. These also give rise to individual HTML pages;
3469 these are now put in a separate directory, and there is an index.html page that
3470 lists them all. Some hyperlinking between the pages has been installed.
3471
3472 54. Added convenience functions for handling named capturing parentheses.
3473
3474 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3475 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3476 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3477 are faulted.
3478
3479 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3480 calling configure. These values are used when compiling the dftables.c program
3481 which is run to generate the source of the default character tables. They
3482 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3483 you will need to set these values.
3484
3485 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3486
3487
3488 Version 3.9 02-Jan-02
3489 ---------------------
3490
3491 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3492
3493 2. If --disable-static was given, the building process failed when trying to
3494 build pcretest and pcregrep. (For some reason it was using libtool to compile
3495 them, which is not right, as they aren't part of the library.)
3496
3497
3498 Version 3.8 18-Dec-01
3499 ---------------------
3500
3501 1. The experimental UTF-8 code was completely screwed up. It was packing the
3502 bytes in the wrong order. How dumb can you get?
3503
3504
3505 Version 3.7 29-Oct-01
3506 ---------------------
3507
3508 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3509 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3510 this didn't happen under Solaris 8, where I normally test things.
3511
3512 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3513 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3514 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3515 this fix an hour or so after the initial 3.7 release.)
3516
3517
3518 Version 3.6 23-Oct-01
3519 ---------------------
3520
3521 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3522 offsets passed as NULL with zero offset count.
3523
3524 2. The config.guess and config.sub files had not been updated when I moved to
3525 the latest autoconf.
3526
3527
3528 Version 3.5 15-Aug-01
3529 ---------------------
3530
3531 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3532 had been forgotten.
3533
3534 2. By using declared but undefined structures, we can avoid using "void"
3535 definitions in pcre.h while keeping the internal definitions of the structures
3536 private.
3537
3538 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3539 user point of view, this means that both static and shared libraries are built
3540 by default, but this can be individually controlled. More of the work of
3541 handling this static/shared cases is now inside libtool instead of PCRE's make
3542 file.
3543
3544 4. The pcretest utility is now installed along with pcregrep because it is
3545 useful for users (to test regexs) and by doing this, it automatically gets
3546 relinked by libtool. The documentation has been turned into a man page, so
3547 there are now .1, .txt, and .html versions in /doc.
3548
3549 5. Upgrades to pcregrep:
3550 (i) Added long-form option names like gnu grep.
3551 (ii) Added --help to list all options with an explanatory phrase.
3552 (iii) Added -r, --recursive to recurse into sub-directories.
3553 (iv) Added -f, --file to read patterns from a file.
3554
3555 6. pcre_exec() was referring to its "code" argument before testing that
3556 argument for NULL (and giving an error if it was NULL).
3557
3558 7. Upgraded Makefile.in to allow for compiling in a different directory from
3559 the source directory.
3560
3561 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3562 options bits, the pointer it was passed was to an int instead of to an unsigned
3563 long int. This mattered only on 64-bit systems.
3564
3565 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3566 generated) instead of pcre.in, which it its source. Also made the same change
3567 in several of the .c files.
3568
3569 10. A new release of gcc defines printf() as a macro, which broke pcretest
3570 because it had an ifdef in the middle of a string argument for printf(). Fixed
3571 by using separate calls to printf().
3572
3573 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3574 script, to force use of CR or LF instead of \n in the source. On non-Unix
3575 systems, the value can be set in config.h.
3576
3577 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3578 absolute limit. Changed the text of the error message to make this clear, and
3579 likewise updated the man page.
3580
3581 13. The limit of 99 on the number of capturing subpatterns has been removed.
3582 The new limit is 65535, which I hope will not be a "real" limit.
3583
3584
3585 Version 3.4 22-Aug-00
3586 ---------------------
3587
3588 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3589
3590 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3591
3592
3593 Version 3.3 01-Aug-00
3594 ---------------------
3595
3596 1. If an octal character was given, but the value was greater than \377, it
3597 was not getting masked to the least significant bits, as documented. This could
3598 lead to crashes in some systems.
3599
3600 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3601 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3602
3603 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3604 These just pass their arguments on to (pcre_free)(), but they are provided
3605 because some uses of PCRE bind it to non-C systems that can call its functions,
3606 but cannot call free() or pcre_free() directly.
3607
3608 4. Add "make test" as a synonym for "make check". Corrected some comments in
3609 the Makefile.
3610
3611 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3612 Makefile.
3613
3614 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3615 command called pgrep for grepping around the active processes.
3616
3617 7. Added the beginnings of support for UTF-8 character strings.
3618
3619 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
3620 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
3621 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
3622 out for the ar command.)
3623
3624
3625 Version 3.2 12-May-00
3626 ---------------------
3627
3628 This is purely a bug fixing release.
3629
3630 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3631 of ZA. This was just one example of several cases that could provoke this bug,
3632 which was introduced by change 9 of version 2.00. The code for breaking
3633 infinite loops after an iteration that matches an empty string was't working
3634 correctly.
3635
3636 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3637 when matched against abbab (for example). After matching an empty string, it
3638 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3639 caused it to match further down the string than it should.
3640
3641 3. The code contained an inclusion of sys/types.h. It isn't clear why this
3642 was there because it doesn't seem to be needed, and it causes trouble on some
3643 systems, as it is not a Standard C header. It has been removed.
3644
3645 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
3646 were reported on the Macintosh. The changes were from
3647
3648 while ((c = *(++ptr)) != 0 && c != '\n');
3649 to
3650 while ((c = *(++ptr)) != 0 && c != '\n') ;
3651
3652 Totally extraordinary, but if that's what it takes...
3653
3654 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
3655 available. Added HAVE_BCOPY and an autoconf test for it; if neither
3656 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
3657 assumes the way PCRE uses memmove() (always moving upwards).
3658
3659 6. PCRE is being used in one environment where strchr() is not available. There
3660 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
3661 faster code anyway.
3662
3663
3664 Version 3.1 09-Feb-00
3665 ---------------------
3666
3667 The only change in this release is the fixing of some bugs in Makefile.in for
3668 the "install" target:
3669
3670 (1) It was failing to install pcreposix.h.
3671
3672 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
3673
3674
3675 Version 3.0 01-Feb-00
3676 ---------------------
3677
3678 1. Add support for the /+ modifier to perltest (to output $` like it does in
3679 pcretest).
3680
3681 2. Add support for the /g modifier to perltest.
3682
3683 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
3684 matches null strings.
3685
3686 4. Fix perltest so that it doesn't do unwanted things when fed an empty
3687 pattern. Perl treats empty patterns specially - it reuses the most recent
3688 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
3689 effect.
3690
3691 5. The POSIX interface was broken in that it was just handing over the POSIX
3692 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
3693 required a bigger vector, with some working space on the end. This means that
3694 the POSIX wrapper now has to get and free some memory, and copy the results.
3695
3696 6. Added some simple autoconf support, placing the test data and the
3697 documentation in separate directories, re-organizing some of the
3698 information files, and making it build pcre-config (a GNU standard). Also added
3699 libtool support for building PCRE as a shared library, which is now the
3700 default.
3701
3702 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
3703 09 are not valid octal constants. Single digits will be used for minor values
3704 less than 10.
3705
3706 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
3707 existing programs that set these in the POSIX interface can use PCRE without
3708 modification.
3709
3710 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
3711 return all that pcre_info() returns, plus additional data. The pcre_info()
3712 function is retained for compatibility, but is considered to be obsolete.
3713
3714 10. Added experimental recursion feature (?R) to handle one common case that
3715 Perl 5.6 will be able to do with (?p{...}).
3716
3717 11. Added support for POSIX character classes like [:alpha:], which Perl is
3718 adopting.
3719
3720
3721 Version 2.08 31-Aug-99
3722 ----------------------
3723
3724 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
3725 trying to match at the startoffset position, but instead was moving forward to
3726 the next newline as if a previous match had failed.
3727
3728 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
3729 and could get into a loop if a null string was matched other than at the start
3730 of the subject.
3731
3732 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
3733 be distinguished at compile time, and for completeness also added PCRE_DATE.
3734
3735 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
3736 in GnuWin32 environments.
3737
3738
3739 Version 2.07 29-Jul-99
3740 ----------------------
3741
3742 1. The documentation is now supplied in plain text form and HTML as well as in
3743 the form of man page sources.
3744
3745 2. C++ compilers don't like assigning (void *) values to other pointer types.
3746 In particular this affects malloc(). Although there is no problem in Standard
3747 C, I've put in casts to keep C++ compilers happy.
3748
3749 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
3750 should be (const char *).
3751
3752 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
3753 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
3754 However, I haven't made this a standard facility. The documentation doesn't
3755 mention it, and the Makefile doesn't support it.
3756
3757 5. The Makefile now contains an "install" target, with editable destinations at
3758 the top of the file. The pcretest program is not installed.
3759
3760 6. pgrep -V now gives the PCRE version number and date.
3761
3762 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
3763 causing the entire string to be ignored, instead of just the last character.
3764
3765 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
3766 non-matching string, it can take a very, very long time, even for strings of
3767 quite modest length, because of the nested recursion. PCRE now does better in
3768 some of these cases. It does this by remembering the last required literal
3769 character in the pattern, and pre-searching the subject to ensure it is present
3770 before running the real match. In other words, it applies a heuristic to detect
3771 some types of certain failure quickly, and in the above example, if presented
3772 with a string that has no trailing " it gives "no match" very quickly.
3773
3774 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
3775 other alternatives are tried instead.
3776
3777
3778 Version 2.06 09-Jun-99
3779 ----------------------
3780
3781 1. Change pcretest's output for amount of store used to show just the code
3782 space, because the remainder (the data block) varies in size between 32-bit and
3783 64-bit systems.
3784
3785 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
3786 start matching at. This allows lookbehinds to work when searching for multiple
3787 occurrences in a string.
3788
3789 3. Added additional options to pcretest for testing multiple occurrences:
3790
3791 /+ outputs the rest of the string that follows a match
3792 /g loops for multiple occurrences, using the new startoffset argument
3793 /G loops for multiple occurrences by passing an incremented pointer
3794
3795 4. PCRE wasn't doing the "first character" optimization for patterns starting
3796 with \b or \B, though it was doing it for other lookbehind assertions. That is,
3797 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
3798 the letter 'x'. On long subject strings, this gives a significant speed-up.
3799
3800
3801 Version 2.05 21-Apr-99
3802 ----------------------
3803
3804 1. Changed the type of magic_number from int to long int so that it works
3805 properly on 16-bit systems.
3806
3807 2. Fixed a bug which caused patterns starting with .* not to work correctly
3808 when the subject string contained newline characters. PCRE was assuming
3809 anchoring for such patterns in all cases, which is not correct because .* will
3810 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
3811 DOTALL is set at top level; otherwise it knows that patterns starting with .*
3812 must be retried after every newline in the subject.
3813
3814
3815 Version 2.04 18-Feb-99
3816 ----------------------
3817
3818 1. For parenthesized subpatterns with repeats whose minimum was zero, the
3819 computation of the store needed to hold the pattern was incorrect (too large).
3820 If such patterns were nested a few deep, this could multiply and become a real
3821 problem.
3822
3823 2. Added /M option to pcretest to show the memory requirement of a specific
3824 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
3825
3826 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
3827 compiled in such a way that the backtracking after subsequent failure was
3828 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
3829 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
3830
3831
3832 Version 2.03 02-Feb-99
3833 ----------------------
3834
3835 1. Fixed typo and small mistake in man page.
3836
3837 2. Added 4th condition (GPL supersedes if conflict) and created separate
3838 LICENCE file containing the conditions.
3839
3840 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
3841 Perl, that is the internal \ allows the delimiter to be included in the
3842 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
3843 the final delimiter, add \ to the end of the pattern (to test the error).
3844
3845 4. Added the convenience functions for extracting substrings after a successful
3846 match. Updated pcretest to make it able to test these functions.
3847
3848
3849 Version 2.02 14-Jan-99
3850 ----------------------
3851
3852 1. Initialized the working variables associated with each extraction so that
3853 their saving and restoring doesn't refer to uninitialized store.
3854
3855 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
3856 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
3857 fix the problem.
3858
3859 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
3860 calls, and wasn't printing the correct value for compiling calls. Increased the
3861 default value of LOOPREPEAT, and the number of significant figures in the
3862 times.
3863
3864 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
3865
3866 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
3867 a building problem on Windows NT with a FAT file system.
3868
3869
3870 Version 2.01 21-Oct-98
3871 ----------------------
3872
3873 1. Changed the API for pcre_compile() to allow for the provision of a pointer
3874 to character tables built by pcre_maketables() in the current locale. If NULL
3875 is passed, the default tables are used.
3876
3877
3878 Version 2.00 24-Sep-98
3879 ----------------------
3880
3881 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
3882 it any more.
3883
3884 2. Allow quantification of (?>) groups, and make it work correctly.
3885
3886 3. The first character computation wasn't working for (?>) groups.
3887
3888 4. Correct the implementation of \Z (it is permitted to match on the \n at the
3889 end of the subject) and add 5.005's \z, which really does match only at the
3890 very end of the subject.
3891
3892 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
3893
3894 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
3895 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
3896 localized options. All options to pcre_study() were also removed.
3897
3898 7. Add other new features from 5.005:
3899
3900 $(?<= positive lookbehind
3901 $(?<! negative lookbehind
3902 (?imsx-imsx) added the unsetting capability
3903 such a setting is global if at outer level; local otherwise
3904 (?imsx-imsx:) non-capturing groups with option setting
3905 (?(cond)re|re) conditional pattern matching
3906
3907 A backreference to itself in a repeated group matches the previous
3908 captured string.
3909
3910 8. General tidying up of studying (both automatic and via "study")
3911 consequential on the addition of new assertions.
3912
3913 9. As in 5.005, unlimited repeated groups that could match an empty substring
3914 are no longer faulted at compile time. Instead, the loop is forcibly broken at
3915 runtime if any iteration does actually match an empty substring.
3916
3917 10. Include the RunTest script in the distribution.
3918
3919 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
3920 discrepancies, some of which were old and were also with respect to 5.004. They
3921 have now been fixed.
3922
3923
3924 Version 1.09 28-Apr-98
3925 ----------------------
3926
3927 1. A negated single character class followed by a quantifier with a minimum
3928 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
3929 program crashes, or just wrong answers. This did not apply to negated classes
3930 containing more than one character, or to minima other than one.
3931
3932
3933 Version 1.08 27-Mar-98
3934 ----------------------
3935
3936 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
3937
3938 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
3939 latter must appear before anything that relies on it in the pattern.
3940
3941
3942 Version 1.07 16-Feb-98
3943 ----------------------
3944
3945 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
3946 repeat of a potentially empty string).
3947
3948
3949 Version 1.06 23-Jan-98
3950 ----------------------
3951
3952 1. Added Markus Oberhumer's little patches for C++.
3953
3954 2. Literal strings longer than 255 characters were broken.
3955
3956
3957 Version 1.05 23-Dec-97
3958 ----------------------
3959
3960 1. Negated character classes containing more than one character were failing if
3961 PCRE_CASELESS was set at run time.
3962
3963
3964 Version 1.04 19-Dec-97
3965 ----------------------
3966
3967 1. Corrected the man page, where some "const" qualifiers had been omitted.
3968
3969 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
3970 input syntax.
3971
3972 3. Fixed memory leak which occurred when a regex with back references was
3973 matched with an offsets vector that wasn't big enough. The temporary memory
3974 that is used in this case wasn't being freed if the match failed.
3975
3976 4. Tidied pcretest to ensure it frees memory that it gets.
3977
3978 5. Temporary memory was being obtained in the case where the passed offsets
3979 vector was exactly big enough.
3980
3981 6. Corrected definition of offsetof() from change 5 below.
3982
3983 7. I had screwed up change 6 below and broken the rules for the use of
3984 setjmp(). Now fixed.
3985
3986
3987 Version 1.03 18-Dec-97
3988 ----------------------
3989
3990 1. A erroneous regex with a missing opening parenthesis was correctly
3991 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
3992 on some systems.
3993
3994 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
3995 it was reported that one broken compiler failed on the former because "code" is
3996 also an independent variable.
3997
3998 3. The erroneous regex a[]b caused an array overrun reference.
3999
4000 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4001 fail on data ending with that character. (It was going on too far, and checking
4002 the next character, typically a binary zero.) This was specific to the
4003 optimized code for single-character negative classes.
4004
4005 5. Added a contributed patch from the TIN world which does the following:
4006
4007 + Add an undef for memmove, in case the the system defines a macro for it.
4008
4009 + Add a definition of offsetof(), in case there isn't one. (I don't know
4010 the reason behind this - offsetof() is part of the ANSI standard - but
4011 it does no harm).
4012
4013 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4014 most of the places where whitespace preceded '#'. I have given up and
4015 allowed the remaining 2 cases to be at the margin.
4016
4017 + Rename some variables in pcre to eliminate shadowing. This seems very
4018 pedantic, but does no harm, of course.
4019
4020 6. Moved the call to setjmp() into its own function, to get rid of warnings
4021 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4022
4023 7. Constructs such as \d{8,} were compiling into the equivalent of
4024 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4025 outcome, but in this particular case used more store than had been allocated,
4026 which caused the bug to be discovered because it threw up an internal error.
4027
4028 8. The debugging code in both pcre and pcretest for outputting the compiled
4029 form of a regex was going wrong in the case of back references followed by
4030 curly-bracketed repeats.
4031
4032
4033 Version 1.02 12-Dec-97
4034 ----------------------
4035
4036 1. Typos in pcre.3 and comments in the source fixed.
4037
4038 2. Applied a contributed patch to get rid of places where it used to remove
4039 'const' from variables, and fixed some signed/unsigned and uninitialized
4040 variable warnings.
4041
4042 3. Added the "runtest" target to Makefile.
4043
4044 4. Set default compiler flag to -O2 rather than just -O.
4045
4046
4047 Version 1.01 19-Nov-97
4048 ----------------------
4049
4050 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4051 like /([ab]*)*/, that is, for classes with more than one character in them.
4052
4053 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4054 as /((?>a*))*/ (a PCRE_EXTRA facility).
4055
4056
4057 Version 1.00 18-Nov-97
4058 ----------------------
4059
4060 1. Added compile-time macros to support systems such as SunOS4 which don't have
4061 memmove() or strerror() but have other things that can be used instead.
4062
4063 2. Arranged that "make clean" removes the executables.
4064
4065
4066 Version 0.99 27-Oct-97
4067 ----------------------
4068
4069 1. Fixed bug in code for optimizing classes with only one character. It was
4070 initializing a 32-byte map regardless, which could cause it to run off the end
4071 of the memory it had got.
4072
4073 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4074
4075
4076 Version 0.98 22-Oct-97
4077 ----------------------
4078
4079 1. Fixed bug in code for handling temporary memory usage when there are more
4080 back references than supplied space in the ovector. This could cause segfaults.
4081
4082
4083 Version 0.97 21-Oct-97
4084 ----------------------
4085
4086 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4087
4088 2. Optimized negated single characters not to use a bit map.
4089
4090 3. Brought error texts together as macro definitions; clarified some of them;
4091 fixed one that was wrong - it said "range out of order" when it meant "invalid
4092 escape sequence".
4093
4094 4. Changed some char * arguments to const char *.
4095
4096 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4097
4098 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4099 pcretest.
4100
4101
4102 Version 0.96 16-Oct-97
4103 ----------------------
4104
4105 1. Added a simple "pgrep" utility to the distribution.
4106
4107 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4108 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4109 where "ddd" means "one or more decimal digits".
4110
4111 3. Fixed serious bug. If a pattern had a back reference, but the call to
4112 pcre_exec() didn't supply a large enough ovector to record the related
4113 identifying subpattern, the match always failed. PCRE now remembers the number
4114 of the largest back reference, and gets some temporary memory in which to save
4115 the offsets during matching if necessary, in order to ensure that
4116 backreferences always work.
4117
4118 4. Increased the compatibility with Perl in a number of ways:
4119
4120 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4121 to request this handling. The option can be set at compile or exec time.
4122
4123 (b) $ matches before a terminating newline by default; an option
4124 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4125 mode). The option can be set at compile or exec time.
4126
4127 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4128 the same as Perl's. If the decimal number it represents is less than 10
4129 or there aren't that many previous left capturing parentheses, an octal
4130 escape is read. Inside a character class, it's always an octal escape,
4131 even if it is a single digit.
4132
4133 (d) An escaped but undefined alphabetic character is taken as a literal,
4134 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4135 escapes.
4136
4137 (e) {0} is now permitted. (The previous item is removed from the compiled
4138 pattern).
4139
4140 5. Changed all the names of code files so that the basic parts are no longer
4141 than 10 characters, and abolished the teeny "globals.c" file.
4142
4143 6. Changed the handling of character classes; they are now done with a 32-byte
4144 bit map always.
4145
4146 7. Added the -d and /D options to pcretest to make it possible to look at the
4147 internals of compilation without having to recompile pcre.
4148
4149
4150 Version 0.95 23-Sep-97
4151 ----------------------
4152
4153 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4154 \x20 at the start of a run of normal characters. These were being treated as
4155 real characters, instead of the source characters being re-checked.
4156
4157
4158 Version 0.94 18-Sep-97
4159 ----------------------
4160
4161 1. The functions are now thread-safe, with the caveat that the global variables
4162 containing pointers to malloc() and free() or alternative functions are the
4163 same for all threads.
4164
4165 2. Get pcre_study() to generate a bitmap of initial characters for non-
4166 anchored patterns when this is possible, and use it if passed to pcre_exec().
4167
4168
4169 Version 0.93 15-Sep-97
4170 ----------------------
4171
4172 1. /(b)|(:+)/ was computing an incorrect first character.
4173
4174 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4175 but not actually doing anything yet.
4176
4177 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4178 as Perl does (e.g. [-az] or [az-]).
4179
4180 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4181 all possible positions.
4182
4183 5. Split up into different modules to avoid including unneeded functions in a
4184 compiled binary. However, compile and exec are still in one module. The "study"
4185 function is split off.
4186
4187 6. The character tables are now in a separate module whose source is generated
4188 by an auxiliary program - but can then be edited by hand if required. There are
4189 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4190 toupper() in the code.
4191
4192 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4193 make them global. Abolish the function for setting them, as the caller can now
4194 set them directly.
4195
4196
4197 Version 0.92 11-Sep-97
4198 ----------------------
4199
4200 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4201 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4202
4203 2. Caseless matching was not working in character classes if the characters in
4204 the pattern were in upper case.
4205
4206 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4207
4208 4. Make PCRE_ANCHORED public and accept as a compile option.
4209
4210 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4211 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4212 pass them.
4213
4214 6. Give an error if bad option bits passed at compile or run time.
4215
4216 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4217 pcretest to cause it to pass that flag.
4218
4219 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4220 options, and the first character, if set.
4221
4222 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4223
4224
4225 Version 0.91 10-Sep-97
4226 ----------------------
4227
4228 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4229 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4230
4231 2. PCRE was looping on encountering an indefinitely repeated back reference to
4232 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4233 Perl does - treats the match as successful.
4234
4235 ****

Properties

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

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12