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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 600 - (show annotations) (download)
Mon May 9 08:54:11 2011 UTC (3 years, 6 months ago) by ph10
File size: 200596 byte(s)
Fix backup bug for \R with greedy quantifier.

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

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