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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 541 - (show annotations) (download)
Mon Jun 14 15:19:33 2010 UTC (4 years, 1 month ago) by ph10
File size: 189290 byte(s)
Add /T option to pcretest and additional tests with non-standard tables.

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

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