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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 574 - (show annotations) (download)
Sat Nov 20 17:47:27 2010 UTC (4 years ago) by ph10
File size: 195798 byte(s)
Give error if \c is followed by a byte > 127 (in ASCII/UTF-8 modes).

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

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