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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 510 - (show annotations) (download)
Sat Mar 27 17:45:29 2010 UTC (4 years, 3 months ago) by ph10
File size: 184866 byte(s)
Add support for *MARK and names for *PRUNE, *SKIP, *THEN.

1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.03 26-Mar-2010
5 ------------------------
6
7 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
8 THEN.
9
10
11 Version 8.02 19-Mar-2010
12 ------------------------
13
14 1. The Unicode data tables have been updated to Unicode 5.2.0.
15
16 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
17 configured.
18
19 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
20 original author of that file, following a query about its status.
21
22 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
23 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
24
25 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
26 quantifier applied to a forward-referencing subroutine call, could compile
27 incorrect code or give the error "internal error: previously-checked
28 referenced subpattern not found".
29
30 6. Both MS Visual Studio and Symbian OS have problems with initializing
31 variables to point to external functions. For these systems, therefore,
32 pcre_malloc etc. are now initialized to local functions that call the
33 relevant global functions.
34
35 7. There were two entries missing in the vectors called coptable and poptable
36 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
37 I've fixed the data, and added a kludgy way of testing at compile time that
38 the lengths are correct (equal to the number of opcodes).
39
40 8. Following on from 7, I added a similar kludge to check the length of the
41 eint vector in pcreposix.c.
42
43 9. Error texts for pcre_compile() are held as one long string to avoid too
44 much relocation at load time. To find a text, the string is searched,
45 counting zeros. There was no check for running off the end of the string,
46 which could happen if a new error number was added without updating the
47 string.
48
49 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
50
51 11. \K was not working if it appeared in an atomic group or in a group that
52 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
53 \K is "not well defined" if used in an assertion. PCRE now accepts it if
54 the assertion is positive, but not if it is negative.
55
56 12. Change 11 fortuitously reduced the size of the stack frame used in the
57 "match()" function of pcre_exec.c by one pointer. Forthcoming
58 implementation of support for (*MARK) will need an extra pointer on the
59 stack; I have reserved it now, so that the stack frame size does not
60 decrease.
61
62 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
63 item in branch that calls a recursion is a subroutine call - as in the
64 second branch in the above example - was incorrectly given the compile-
65 time error "recursive call could loop indefinitely" because pcre_compile()
66 was not correctly checking the subroutine for matching a non-empty string.
67
68 14. The checks for overrunning compiling workspace could trigger after an
69 overrun had occurred. This is a "should never occur" error, but it can be
70 triggered by pathological patterns such as hundreds of nested parentheses.
71 The checks now trigger 100 bytes before the end of the workspace.
72
73 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
74
75
76 Version 8.01 19-Jan-2010
77 ------------------------
78
79 1. If a pattern contained a conditional subpattern with only one branch (in
80 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
81 computed the wrong minimum data length (which is of course zero for such
82 subpatterns). This could cause incorrect "no match" results.
83
84 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
85 the pattern is reset in the first branch, pcre_compile() failed with
86 "internal error: code overflow at offset...". This happened only when
87 the reset was to the original external option setting. (An optimization
88 abstracts leading options settings into an external setting, which was the
89 cause of this.)
90
91 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
92 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
93 assertion pattern did not match (meaning that the assertion was true), it
94 was incorrectly treated as false if the SKIP had been reached during the
95 matching. This also applied to assertions used as conditions.
96
97 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
98 assertion subpattern, including such a pattern used as a condition,
99 unpredictable results occurred, instead of the error return
100 PCRE_ERROR_DFA_UITEM.
101
102 5. The C++ GlobalReplace function was not working like Perl for the special
103 situation when an empty string is matched. It now does the fancy magic
104 stuff that is necessary.
105
106 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
107 removed. (These were left over from very, very early versions of PCRE.)
108
109 7. Some cosmetic changes to the code to make life easier when compiling it
110 as part of something else:
111
112 (a) Change DEBUG to PCRE_DEBUG.
113
114 (b) In pcre_compile(), rename the member of the "branch_chain" structure
115 called "current" as "current_branch", to prevent a collision with the
116 Linux macro when compiled as a kernel module.
117
118 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
119 prevent a collision with the Linux macro when compiled as a kernel
120 module.
121
122 8. In pcre_compile() there are some checks for integer overflows that used to
123 cast potentially large values to (double). This has been changed to that
124 when building, a check for int64_t is made, and if it is found, it is used
125 instead, thus avoiding the use of floating point arithmetic. (There is no
126 other use of FP in PCRE.) If int64_t is not found, the fallback is to
127 double.
128
129 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
130 2005 (difference between two addresses compared to an unsigned value).
131
132 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
133 custom one, because of the following reported problem in Windows:
134
135 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
136 under Win32.
137 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
138 therefore missing the function definition.
139 - The compiler thus generates a "C" signature for the test function.
140 - The linker fails to find the "C" function.
141 - PCRE fails to configure if asked to do so against libbz2.
142
143 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
144 messages were output:
145
146 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
147 rerunning libtoolize, to keep the correct libtool macros in-tree.
148 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
149
150 I have done both of these things.
151
152 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
153 most of the time, it *can* run out if it is given a pattern that contains a
154 runaway infinite recursion. I updated the discussion in the pcrestack man
155 page.
156
157 13. Now that we have gone to the x.xx style of version numbers, the minor
158 version may start with zero. Using 08 or 09 is a bad idea because users
159 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
160 interpreted as invalid octal numbers. I've updated the previous comment in
161 configure.ac, and also added a check that gives an error if 08 or 09 are
162 used.
163
164 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
165 causing partial matching to fail when the end of the subject matched \W
166 in a UTF-8 pattern where \W was quantified with a minimum of 3.
167
168 15. There were some discrepancies between the declarations in pcre_internal.h
169 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
170 their definitions. The declarations used "const uschar *" and the
171 definitions used USPTR. Even though USPTR is normally defined as "const
172 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
173 reported that: "This difference in casting confuses some C++ compilers, for
174 example, SunCC recognizes above declarations as different functions and
175 generates broken code for hbpcre." I have changed the declarations to use
176 USPTR.
177
178 16. GNU libtool is named differently on some systems. The autogen.sh script now
179 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
180 (FreeBSD).
181
182 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
183 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
184 comment: "Figure out how to create a longlong from a string: strtoll and
185 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
186 instance, but it only takes 2 args instead of 3!"
187
188 18. A subtle bug concerned with back references has been fixed by a change of
189 specification, with a corresponding code fix. A pattern such as
190 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
191 refers, was giving matches when it shouldn't. For example, xa=xaaa would
192 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
193 same bug. Such groups have to be quantified to be useful, or contained
194 inside another quantified group. (If there's no repetition, the reference
195 can never match.) The problem arises because, having left the group and
196 moved on to the rest of the pattern, a later failure that backtracks into
197 the group uses the captured value from the final iteration of the group
198 rather than the correct earlier one. I have fixed this in PCRE by forcing
199 any group that contains a reference to itself to be an atomic group; that
200 is, there cannot be any backtracking into it once it has completed. This is
201 similar to recursive and subroutine calls.
202
203
204 Version 8.00 19-Oct-09
205 ----------------------
206
207 1. The table for translating pcre_compile() error codes into POSIX error codes
208 was out-of-date, and there was no check on the pcre_compile() error code
209 being within the table. This could lead to an OK return being given in
210 error.
211
212 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
213 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
214 in a Windows environment.
215
216 3. The pcregrep --count option prints the count for each file even when it is
217 zero, as does GNU grep. However, pcregrep was also printing all files when
218 --files-with-matches was added. Now, when both options are given, it prints
219 counts only for those files that have at least one match. (GNU grep just
220 prints the file name in this circumstance, but including the count seems
221 more useful - otherwise, why use --count?) Also ensured that the
222 combination -clh just lists non-zero counts, with no names.
223
224 4. The long form of the pcregrep -F option was incorrectly implemented as
225 --fixed_strings instead of --fixed-strings. This is an incompatible change,
226 but it seems right to fix it, and I didn't think it was worth preserving
227 the old behaviour.
228
229 5. The command line items --regex=pattern and --regexp=pattern were not
230 recognized by pcregrep, which required --regex pattern or --regexp pattern
231 (with a space rather than an '='). The man page documented the '=' forms,
232 which are compatible with GNU grep; these now work.
233
234 6. No libpcreposix.pc file was created for pkg-config; there was just
235 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
236
237 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
238 when UCP support is not needed, by modifying the Python script that
239 generates it from Unicode data files. This should not matter if the module
240 is correctly used as a library, but I received one complaint about 50K of
241 unwanted data. My guess is that the person linked everything into his
242 program rather than using a library. Anyway, it does no harm.
243
244 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
245 was a minimum greater than 1 for a wide character in a possessive
246 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
247 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
248 character. Chaos in the form of incorrect output or a compiling loop could
249 result.
250
251 9. The restrictions on what a pattern can contain when partial matching is
252 requested for pcre_exec() have been removed. All patterns can now be
253 partially matched by this function. In addition, if there are at least two
254 slots in the offset vector, the offset of the earliest inspected character
255 for the match and the offset of the end of the subject are set in them when
256 PCRE_ERROR_PARTIAL is returned.
257
258 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
259 synonymous with PCRE_PARTIAL, for backwards compatibility, and
260 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
261 and may be more useful for multi-segment matching.
262
263 11. Partial matching with pcre_exec() is now more intuitive. A partial match
264 used to be given if ever the end of the subject was reached; now it is
265 given only if matching could not proceed because another character was
266 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
267 string "Z", which now yields "no match" instead of "partial match". In the
268 case of pcre_dfa_exec(), "no match" is given if every matching path for the
269 final character ended with (*FAIL).
270
271 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
272 if the pattern had a "must contain" character that was already found in the
273 earlier partial match, unless partial matching was again requested. For
274 example, with the pattern /dog.(body)?/, the "must contain" character is
275 "g". If the first part-match was for the string "dog", restarting with
276 "sbody" failed. This bug has been fixed.
277
278 13. The string returned by pcre_dfa_exec() after a partial match has been
279 changed so that it starts at the first inspected character rather than the
280 first character of the match. This makes a difference only if the pattern
281 starts with a lookbehind assertion or \b or \B (\K is not supported by
282 pcre_dfa_exec()). It's an incompatible change, but it makes the two
283 matching functions compatible, and I think it's the right thing to do.
284
285 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
286 so that the demonstration program is easily available in environments where
287 PCRE has not been installed from source.
288
289 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
290 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
291 library.
292
293 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
294 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
295 is not the first non-POSIX option to be added. Clearly some people find
296 these options useful.
297
298 17. If a caller to the POSIX matching function regexec() passes a non-zero
299 value for nmatch with a NULL value for pmatch, the value of
300 nmatch is forced to zero.
301
302 18. RunGrepTest did not have a test for the availability of the -u option of
303 the diff command, as RunTest does. It now checks in the same way as
304 RunTest, and also checks for the -b option.
305
306 19. If an odd number of negated classes containing just a single character
307 interposed, within parentheses, between a forward reference to a named
308 subpattern and the definition of the subpattern, compilation crashed with
309 an internal error, complaining that it could not find the referenced
310 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
311 [The bug was that it was starting one character too far in when skipping
312 over the character class, thus treating the ] as data rather than
313 terminating the class. This meant it could skip too much.]
314
315 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
316 /g option in pcretest when the pattern contains \K, which makes it possible
317 to have an empty string match not at the start, even when the pattern is
318 anchored. Updated pcretest and pcredemo to use this option.
319
320 21. If the maximum number of capturing subpatterns in a recursion was greater
321 than the maximum at the outer level, the higher number was returned, but
322 with unset values at the outer level. The correct (outer level) value is
323 now given.
324
325 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
326 PCRE did not set those parentheses (unlike Perl). I have now found a way to
327 make it do so. The string so far is captured, making this feature
328 compatible with Perl.
329
330 23. The tests have been re-organized, adding tests 11 and 12, to make it
331 possible to check the Perl 5.10 features against Perl 5.10.
332
333 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
334 pattern matches a fixed length string. PCRE did not allow this; now it
335 does. Neither allows recursion.
336
337 25. I finally figured out how to implement a request to provide the minimum
338 length of subject string that was needed in order to match a given pattern.
339 (It was back references and recursion that I had previously got hung up
340 on.) This code has now been added to pcre_study(); it finds a lower bound
341 to the length of subject needed. It is not necessarily the greatest lower
342 bound, but using it to avoid searching strings that are too short does give
343 some useful speed-ups. The value is available to calling programs via
344 pcre_fullinfo().
345
346 26. While implementing 25, I discovered to my embarrassment that pcretest had
347 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
348 study optimizations had never been tested with that matching function.
349 Oops. What is worse, even when it was passed study data, there was a bug in
350 pcre_dfa_exec() that meant it never actually used it. Double oops. There
351 were also very few tests of studied patterns with pcre_dfa_exec().
352
353 27. If (?| is used to create subpatterns with duplicate numbers, they are now
354 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
355 on the other side of the coin, they are no longer allowed to have different
356 names, because these cannot be distinguished in PCRE, and this has caused
357 confusion. (This is a difference from Perl.)
358
359 28. When duplicate subpattern names are present (necessarily with different
360 numbers, as required by 27 above), and a test is made by name in a
361 conditional pattern, either for a subpattern having been matched, or for
362 recursion in such a pattern, all the associated numbered subpatterns are
363 tested, and the overall condition is true if the condition is true for any
364 one of them. This is the way Perl works, and is also more like the way
365 testing by number works.
366
367
368 Version 7.9 11-Apr-09
369 ---------------------
370
371 1. When building with support for bzlib/zlib (pcregrep) and/or readline
372 (pcretest), all targets were linked against these libraries. This included
373 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
374 libraries. This caused unwanted dependencies to be created. This problem
375 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
376 pcretest is linked with readline.
377
378 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
379 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
380 moved outside it again, because FALSE and TRUE are already defined in AIX,
381 but BOOL is not.
382
383 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
384 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
385
386 4. The pcregrep documentation said spaces were inserted as well as colons (or
387 hyphens) following file names and line numbers when outputting matching
388 lines. This is not true; no spaces are inserted. I have also clarified the
389 wording for the --colour (or --color) option.
390
391 5. In pcregrep, when --colour was used with -o, the list of matching strings
392 was not coloured; this is different to GNU grep, so I have changed it to be
393 the same.
394
395 6. When --colo(u)r was used in pcregrep, only the first matching substring in
396 each matching line was coloured. Now it goes on to look for further matches
397 of any of the test patterns, which is the same behaviour as GNU grep.
398
399 7. A pattern that could match an empty string could cause pcregrep to loop; it
400 doesn't make sense to accept an empty string match in pcregrep, so I have
401 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
402 seems to be how GNU grep behaves.
403
404 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
405 start or after a newline", because the conditional assertion was not being
406 correctly handled. The rule now is that both the assertion and what follows
407 in the first alternative must satisfy the test.
408
409 9. If auto-callout was enabled in a pattern with a conditional group whose
410 condition was an assertion, PCRE could crash during matching, both with
411 pcre_exec() and pcre_dfa_exec().
412
413 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
414 used for matching.
415
416 11. Unicode property support in character classes was not working for
417 characters (bytes) greater than 127 when not in UTF-8 mode.
418
419 12. Added the -M command line option to pcretest.
420
421 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
422
423 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
424
425 16. Added comments and documentation about mis-use of no_arg in the C++
426 wrapper.
427
428 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
429 from Martin Jerabek that uses macro names for all relevant character and
430 string constants.
431
432 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
433 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
434 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
435 these, but not everybody uses configure.
436
437 19. A conditional group that had only one branch was not being correctly
438 recognized as an item that could match an empty string. This meant that an
439 enclosing group might also not be so recognized, causing infinite looping
440 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
441 with the subject "ab", where knowledge that the repeated group can match
442 nothing is needed in order to break the loop.
443
444 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
445 exec(), but without supplying a callout function, matching went wrong.
446
447 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
448 leak if the size of the offset vector was greater than 30. When the vector
449 is smaller, the saved offsets during recursion go onto a local stack
450 vector, but for larger vectors malloc() is used. It was failing to free
451 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
452 error, in fact).
453
454 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
455 heapframe that is used only when UTF-8 support is enabled. This caused no
456 problem, but was untidy.
457
458 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
459 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
460 included within another project.
461
462 24. Steven Van Ingelgem's patches to add more options to the CMake support,
463 slightly modified by me:
464
465 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
466 not building pcregrep.
467
468 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
469 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
470
471 25. Forward references, both numeric and by name, in patterns that made use of
472 duplicate group numbers, could behave incorrectly or give incorrect errors,
473 because when scanning forward to find the reference group, PCRE was not
474 taking into account the duplicate group numbers. A pattern such as
475 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
476
477 26. Changed a few more instances of "const unsigned char *" to USPTR, making
478 the feature of a custom pointer more persuasive (as requested by a user).
479
480 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
481 pcretest.c, inside #ifndefs, because it seems they are sometimes already
482 pre-defined.
483
484 28. Added support for (*UTF8) at the start of a pattern.
485
486 29. Arrange for flags added by the "release type" setting in CMake to be shown
487 in the configuration summary.
488
489
490 Version 7.8 05-Sep-08
491 ---------------------
492
493 1. Replaced UCP searching code with optimized version as implemented for Ad
494 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
495 stage table and inline lookup instead of a function, giving speed ups of 2
496 to 5 times on some simple patterns that I tested. Permission was given to
497 distribute the MultiStage2.py script that generates the tables (it's not in
498 the tarball, but is in the Subversion repository).
499
500 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
501 scripts.
502
503 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
504 a group with a zero qualifier. The result of the study could be incorrect,
505 or the function might crash, depending on the pattern.
506
507 4. Caseless matching was not working for non-ASCII characters in back
508 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
509 It now works when Unicode Property Support is available.
510
511 5. In pcretest, an escape such as \x{de} in the data was always generating
512 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
513 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
514 truncation.
515
516 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
517
518 7. Added two (int) casts to pcregrep when printing the difference of two
519 pointers, in case they are 64-bit values.
520
521 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
522 test 2 if it fails.
523
524 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
525 and a #define of that name to empty if it is not externally set. This is to
526 allow users of MSVC to set it if necessary.
527
528 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
529 the convenience functions in the pcre_get.c source file.
530
531 11. An option change at the start of a pattern that had top-level alternatives
532 could cause overwriting and/or a crash. This command provoked a crash in
533 some environments:
534
535 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
536
537 This potential security problem was recorded as CVE-2008-2371.
538
539 12. For a pattern where the match had to start at the beginning or immediately
540 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
541 pcre_dfa_exec() could read past the end of the passed subject if there was
542 no match. To help with detecting such bugs (e.g. with valgrind), I modified
543 pcretest so that it places the subject at the end of its malloc-ed buffer.
544
545 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
546 exec() might read past the end of the data buffer in UTF-8 mode.
547
548 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
549 the data contained the byte 0x85 as part of a UTF-8 character within its
550 first line. This applied both to normal and DFA matching.
551
552 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
553 /^[^d]*?$/8 failed to match "abc".
554
555 16. Added a missing copyright notice to pcrecpp_internal.h.
556
557 17. Make it more clear in the documentation that values returned from
558 pcre_exec() in ovector are byte offsets, not character counts.
559
560 18. Tidied a few places to stop certain compilers from issuing warnings.
561
562 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
563 supplied by Stefan Weber. I made a further small update for 7.8 because
564 there is a change of source arrangements: the pcre_searchfuncs.c module is
565 replaced by pcre_ucd.c.
566
567
568 Version 7.7 07-May-08
569 ---------------------
570
571 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
572 a string to a long long, pretend we don't even have a long long." This is
573 done by checking for the strtoq, strtoll, and _strtoi64 functions.
574
575 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
576 pre-7.6 versions, which defined a global no_arg variable instead of putting
577 it in the RE class. (See also #8 below.)
578
579 3. Remove a line of dead code, identified by coverity and reported by Nuno
580 Lopes.
581
582 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
583
584 (1) The include/exclude patterns were being applied to the whole pathnames
585 of files, instead of just to the final components.
586
587 (2) If there was more than one level of directory, the subdirectories were
588 skipped unless they satisfied the include/exclude conditions. This is
589 inconsistent with GNU grep (and could even be seen as contrary to the
590 pcregrep specification - which I improved to make it absolutely clear).
591 The action now is always to scan all levels of directory, and just
592 apply the include/exclude patterns to regular files.
593
594 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
595 --exclude_dir in the tests to avoid scanning .svn directories.
596
597 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
598 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
599 doesn't support NULs in patterns.
600
601 7. Added some missing "const"s to declarations of static tables in
602 pcre_compile.c and pcre_dfa_exec.c.
603
604 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
605 caused by fix #2 above. (Subsequently also a second patch to fix the
606 first patch. And a third patch - this was a messy problem.)
607
608 9. Applied Craig's patch to remove the use of push_back().
609
610 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
611 matching function regexec().
612
613 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
614 which, however, unlike Perl's \g{...}, are subroutine calls, not back
615 references. PCRE supports relative numbers with this syntax (I don't think
616 Oniguruma does).
617
618 12. Previously, a group with a zero repeat such as (...){0} was completely
619 omitted from the compiled regex. However, this means that if the group
620 was called as a subroutine from elsewhere in the pattern, things went wrong
621 (an internal error was given). Such groups are now left in the compiled
622 pattern, with a new opcode that causes them to be skipped at execution
623 time.
624
625 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
626 to the way PCRE behaves:
627
628 (a) A lone ] character is dis-allowed (Perl treats it as data).
629
630 (b) A back reference to an unmatched subpattern matches an empty string
631 (Perl fails the current match path).
632
633 (c) A data ] in a character class must be notated as \] because if the
634 first data character in a class is ], it defines an empty class. (In
635 Perl it is not possible to have an empty class.) The empty class []
636 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
637 The negative empty class [^] matches any one character, independently
638 of the DOTALL setting.
639
640 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
641 non-existent subpattern following a character class starting with ']' and
642 containing () gave an internal compiling error instead of "reference to
643 non-existent subpattern". Fortunately, when the pattern did exist, the
644 compiled code was correct. (When scanning forwards to check for the
645 existencd of the subpattern, it was treating the data ']' as terminating
646 the class, so got the count wrong. When actually compiling, the reference
647 was subsequently set up correctly.)
648
649 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
650 it was being rejected as not supported by pcre_dfa_exec(), even though
651 other assertions are supported. I have made pcre_dfa_exec() support
652 (*FAIL).
653
654 16. The implementation of 13c above involved the invention of a new opcode,
655 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
656 cannot be changed at match time, I realized I could make a small
657 improvement to matching performance by compiling OP_ALLANY instead of
658 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
659 on the OP_ANY path.
660
661 17. Compiling pcretest on Windows with readline support failed without the
662 following two fixes: (1) Make the unistd.h include conditional on
663 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
664
665 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
666 ncurses library to be included for pcretest when ReadLine support is
667 requested, but also to allow for it to be overridden. This patch came from
668 Daniel Bergström.
669
670 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
671 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
672 any errors with the current Unicode tables. Thanks to Peter Kankowski for
673 spotting this.
674
675
676 Version 7.6 28-Jan-08
677 ---------------------
678
679 1. A character class containing a very large number of characters with
680 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
681 overflow.
682
683 2. Patch to cut out the "long long" test in pcrecpp_unittest when
684 HAVE_LONG_LONG is not defined.
685
686 3. Applied Christian Ehrlicher's patch to update the CMake build files to
687 bring them up to date and include new features. This patch includes:
688
689 - Fixed PH's badly added libz and libbz2 support.
690 - Fixed a problem with static linking.
691 - Added pcredemo. [But later removed - see 7 below.]
692 - Fixed dftables problem and added an option.
693 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
694 HAVE_LONG_LONG.
695 - Added readline support for pcretest.
696 - Added an listing of the option settings after cmake has run.
697
698 4. A user submitted a patch to Makefile that makes it easy to create
699 "pcre.dll" under mingw when using Configure/Make. I added stuff to
700 Makefile.am that cause it to include this special target, without
701 affecting anything else. Note that the same mingw target plus all
702 the other distribution libraries and programs are now supported
703 when configuring with CMake (see 6 below) instead of with
704 Configure/Make.
705
706 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
707 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
708 exported in the Windows port". It has not yet been confirmed that the patch
709 solves the problem, but it does no harm.
710
711 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
712 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
713 with CMake, and also correct the comment about stack recursion.
714
715 7. Remove the automatic building of pcredemo from the ./configure system and
716 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
717 of a program that users should build themselves after PCRE is installed, so
718 building it automatically is not really right. What is more, it gave
719 trouble in some build environments.
720
721 8. Further tidies to CMakeLists.txt from Sheri and Christian.
722
723
724 Version 7.5 10-Jan-08
725 ---------------------
726
727 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
728 values in parens when parsing an RE using the C++ wrapper."
729
730 2. Negative specials like \S did not work in character classes in UTF-8 mode.
731 Characters greater than 255 were excluded from the class instead of being
732 included.
733
734 3. The same bug as (2) above applied to negated POSIX classes such as
735 [:^space:].
736
737 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
738 defined or documented. It seems to have been a typo for PCRE_STATIC, so
739 I have changed it.
740
741 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
742 first named subpattern) and a construct such as (?&a) would reference the
743 first named subpattern whose name started with "a" (in other words, the
744 length check was missing). Both these problems are fixed. "Subpattern name
745 expected" is now given for (?&) (a zero-length name), and this patch also
746 makes it give the same error for \k'' (previously it complained that that
747 was a reference to a non-existent subpattern).
748
749 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
750 this is right because (?- can be followed by option settings as well as by
751 digits. I have, however, made the messages clearer.
752
753 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
754 than the number used in the conditional) now cause a compile-time error.
755 This is actually not compatible with Perl, which accepts such patterns, but
756 treats the conditional as always being FALSE (as PCRE used to), but it
757 seems to me that giving a diagnostic is better.
758
759 8. Change "alphameric" to the more common word "alphanumeric" in comments
760 and messages.
761
762 9. Fix two occurrences of "backslash" in comments that should have been
763 "backspace".
764
765 10. Remove two redundant lines of code that can never be obeyed (their function
766 was moved elsewhere).
767
768 11. The program that makes PCRE's Unicode character property table had a bug
769 which caused it to generate incorrect table entries for sequences of
770 characters that have the same character type, but are in different scripts.
771 It amalgamated them into a single range, with the script of the first of
772 them. In other words, some characters were in the wrong script. There were
773 thirteen such cases, affecting characters in the following ranges:
774
775 U+002b0 - U+002c1
776 U+0060c - U+0060d
777 U+0061e - U+00612
778 U+0064b - U+0065e
779 U+0074d - U+0076d
780 U+01800 - U+01805
781 U+01d00 - U+01d77
782 U+01d9b - U+01dbf
783 U+0200b - U+0200f
784 U+030fc - U+030fe
785 U+03260 - U+0327f
786 U+0fb46 - U+0fbb1
787 U+10450 - U+1049d
788
789 12. The -o option (show only the matching part of a line) for pcregrep was not
790 compatible with GNU grep in that, if there was more than one match in a
791 line, it showed only the first of them. It now behaves in the same way as
792 GNU grep.
793
794 13. If the -o and -v options were combined for pcregrep, it printed a blank
795 line for every non-matching line. GNU grep prints nothing, and pcregrep now
796 does the same. The return code can be used to tell if there were any
797 non-matching lines.
798
799 14. Added --file-offsets and --line-offsets to pcregrep.
800
801 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
802 infinitely looping recursion. The bug was that positive lookaheads were not
803 being skipped when checking for a possible empty match (negative lookaheads
804 and both kinds of lookbehind were skipped).
805
806 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
807 inclusion of <windows.h> to before rather than after the definition of
808 INVALID_FILE_ATTRIBUTES (patch from David Byron).
809
810 17. Specifying a possessive quantifier with a specific limit for a Unicode
811 character property caused pcre_compile() to compile bad code, which led at
812 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
813 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
814 caused the error; without that there was no problem.
815
816 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
817
818 19. Added --enable-pcretest-libreadline.
819
820 20. In pcrecpp.cc, the variable 'count' was incremented twice in
821 RE::GlobalReplace(). As a result, the number of replacements returned was
822 double what it should be. I removed one of the increments, but Craig sent a
823 later patch that removed the other one (the right fix) and added unit tests
824 that check the return values (which was not done before).
825
826 21. Several CMake things:
827
828 (1) Arranged that, when cmake is used on Unix, the libraries end up with
829 the names libpcre and libpcreposix, not just pcre and pcreposix.
830
831 (2) The above change means that pcretest and pcregrep are now correctly
832 linked with the newly-built libraries, not previously installed ones.
833
834 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
835
836 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
837 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
838 UTF-8 newline character). The key issue is that the pattern starts .*;
839 this means that the match must be either at the beginning, or after a
840 newline. The bug was in the code for advancing after a failed match and
841 checking that the new position followed a newline. It was not taking
842 account of UTF-8 characters correctly.
843
844 23. PCRE was behaving differently from Perl in the way it recognized POSIX
845 character classes. PCRE was not treating the sequence [:...:] as a
846 character class unless the ... were all letters. Perl, however, seems to
847 allow any characters between [: and :], though of course it rejects as
848 unknown any "names" that contain non-letters, because all the known class
849 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
850 for example, whereas PCRE did not - it did not recognize a POSIX character
851 class. This seemed a bit dangerous, so the code has been changed to be
852 closer to Perl. The behaviour is not identical to Perl, because PCRE will
853 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
854 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
855 Perl does, and where it didn't before.
856
857 24. Rewrite so as to remove the single use of %n from pcregrep because in some
858 Windows environments %n is disabled by default.
859
860
861 Version 7.4 21-Sep-07
862 ---------------------
863
864 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
865 means that a class such as [\s] counted as "explicit reference to CR or
866 LF". That isn't really right - the whole point of the change was to try to
867 help when there was an actual mention of one of the two characters. So now
868 the change happens only if \r or \n (or a literal CR or LF) character is
869 encountered.
870
871 2. The 32-bit options word was also used for 6 internal flags, but the numbers
872 of both had grown to the point where there were only 3 bits left.
873 Fortunately, there was spare space in the data structure, and so I have
874 moved the internal flags into a new 16-bit field to free up more option
875 bits.
876
877 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
878 but did not set the internal JCHANGED flag - either of these is enough to
879 control the way the "get" function works - but the PCRE_INFO_JCHANGED
880 facility is supposed to tell if (?J) was ever used, so now (?J) at the
881 start sets both bits.
882
883 4. Added options (at build time, compile time, exec time) to change \R from
884 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
885
886 5. doc/pcresyntax.html was missing from the distribution.
887
888 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
889 compatibility, even though it is no longer used.
890
891 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
892 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
893 windows.h file is present (where different names are used). [This was
894 reversed later after testing - see 16 below.]
895
896 8. Changed all #include <config.h> to #include "config.h". There were also
897 some further <pcre.h> cases that I changed to "pcre.h".
898
899 9. When pcregrep was used with the --colour option, it missed the line ending
900 sequence off the lines that it output.
901
902 10. It was pointed out to me that arrays of string pointers cause lots of
903 relocations when a shared library is dynamically loaded. A technique of
904 using a single long string with a table of offsets can drastically reduce
905 these. I have refactored PCRE in four places to do this. The result is
906 dramatic:
907
908 Originally: 290
909 After changing UCP table: 187
910 After changing error message table: 43
911 After changing table of "verbs" 36
912 After changing table of Posix names 22
913
914 Thanks to the folks working on Gregex for glib for this insight.
915
916 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
917 unicode-properties was also set.
918
919 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
920
921 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
922 checked only for CRLF.
923
924 14. Added casts to pcretest.c to avoid compiler warnings.
925
926 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
927
928 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
929 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
930 entirely. This removes changes made in 7 above.
931
932 17. The CMake files have been updated, and there is now more information about
933 building with CMake in the NON-UNIX-USE document.
934
935
936 Version 7.3 28-Aug-07
937 ---------------------
938
939 1. In the rejigging of the build system that eventually resulted in 7.1, the
940 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
941 brackets there is not right, since it causes compilers to look for an
942 installed pcre.h, not the version that is in the source that is being
943 compiled (which of course may be different). I have changed it back to:
944
945 #include "pcre.h"
946
947 I have a vague recollection that the change was concerned with compiling in
948 different directories, but in the new build system, that is taken care of
949 by the VPATH setting the Makefile.
950
951 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
952 when the subject happened to end in the byte 0x85 (e.g. if the last
953 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
954 characters but of course it shouldn't be taken as a newline when it is part
955 of another character. The bug was that, for an unlimited repeat of . in
956 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
957 characters when looking for a newline.
958
959 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
960
961 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
962 in debug output.
963
964 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
965 long printing in the pcrecpp unittest when running under MinGW.
966
967 6. ESC_K was left out of the EBCDIC table.
968
969 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
970 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
971 limit also applies to "virtual nesting" when a pattern is recursive, and in
972 this case 1000 isn't so big. I have been able to remove this limit at the
973 expense of backing off one optimization in certain circumstances. Normally,
974 when pcre_exec() would call its internal match() function recursively and
975 immediately return the result unconditionally, it uses a "tail recursion"
976 feature to save stack. However, when a subpattern that can match an empty
977 string has an unlimited repetition quantifier, it no longer makes this
978 optimization. That gives it a stack frame in which to save the data for
979 checking that an empty string has been matched. Previously this was taken
980 from the 1000-entry workspace that had been reserved. So now there is no
981 explicit limit, but more stack is used.
982
983 8. Applied Daniel's patches to solve problems with the import/export magic
984 syntax that is required for Windows, and which was going wrong for the
985 pcreposix and pcrecpp parts of the library. These were overlooked when this
986 problem was solved for the main library.
987
988 9. There were some crude static tests to avoid integer overflow when computing
989 the size of patterns that contain repeated groups with explicit upper
990 limits. As the maximum quantifier is 65535, the maximum group length was
991 set at 30,000 so that the product of these two numbers did not overflow a
992 32-bit integer. However, it turns out that people want to use groups that
993 are longer than 30,000 bytes (though not repeat them that many times).
994 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
995 made it possible to implement the integer overflow checks in a much more
996 dynamic way, which I have now done. The artificial limitation on group
997 length has been removed - we now have only the limit on the total length of
998 the compiled pattern, which depends on the LINK_SIZE setting.
999
1000 10. Fixed a bug in the documentation for get/copy named substring when
1001 duplicate names are permitted. If none of the named substrings are set, the
1002 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1003 empty string.
1004
1005 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1006 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1007 because the ] is interpreted as the first data character and the
1008 terminating ] is not found. PCRE has been made compatible with Perl in this
1009 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1010 cause memory overwriting.
1011
1012 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1013 string has been matched (to stop an infinite loop). It was not recognizing
1014 a conditional subpattern that could match an empty string if that
1015 subpattern was within another subpattern. For example, it looped when
1016 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1017 condition was not nested. This bug has been fixed.
1018
1019 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1020 past the start of the subject in the presence of bytes with the top bit
1021 set, for example "\x8aBCD".
1022
1023 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1024 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1025
1026 14. Optimized (?!) to (*FAIL).
1027
1028 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1029 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1030 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1031 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1032 does: it's just the validity check that is more restrictive.
1033
1034 16. Inserted checks for integer overflows during escape sequence (backslash)
1035 processing, and also fixed erroneous offset values for syntax errors during
1036 backslash processing.
1037
1038 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1039 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1040
1041 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1042 caused an overrun.
1043
1044 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1045 something other than just ASCII characters) inside a group that had an
1046 unlimited repeat caused a loop at compile time (while checking to see
1047 whether the group could match an empty string).
1048
1049 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1050 [\P{Any}] did so. (Error in the code for printing property names.)
1051
1052 21. An orphan \E inside a character class could cause a crash.
1053
1054 22. A repeated capturing bracket such as (A)? could cause a wild memory
1055 reference during compilation.
1056
1057 23. There are several functions in pcre_compile() that scan along a compiled
1058 expression for various reasons (e.g. to see if it's fixed length for look
1059 behind). There were bugs in these functions when a repeated \p or \P was
1060 present in the pattern. These operators have additional parameters compared
1061 with \d, etc, and these were not being taken into account when moving along
1062 the compiled data. Specifically:
1063
1064 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1065 length.
1066
1067 (b) An item such as \pL+ within a repeated group could cause crashes or
1068 loops.
1069
1070 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1071 "reference to non-existent subpattern" error.
1072
1073 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1074
1075 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1076 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1077
1078 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1079
1080 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1081 character were causing crashes (broken optimization).
1082
1083 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1084 \p or \P) caused a compile-time loop.
1085
1086 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1087 break. For example, the unstudied pattern [\r\n]A does not match the string
1088 "\r\nA" because change 7.0/46 below moves the current point on by two
1089 characters after failing to match at the start. However, the pattern \nA
1090 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1091 the same is true. There doesn't seem any very clean way out of this, but
1092 what I have chosen to do makes the common cases work: PCRE now takes note
1093 of whether there can be an explicit match for \r or \n anywhere in the
1094 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1095 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1096 pattern has explicit CR or LF references.
1097
1098 29. Added (*CR) etc for changing newline setting at start of pattern.
1099
1100
1101 Version 7.2 19-Jun-07
1102 ---------------------
1103
1104 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1105 which is apparently normally available under Windows.
1106
1107 2. Re-jig the pcregrep tests with different newline settings in an attempt
1108 to make them independent of the local environment's newline setting.
1109
1110 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1111
1112 4. Some of the "internals" tests were previously cut out when the link size
1113 was not 2, because the output contained actual offsets. The recent new
1114 "Z" feature of pcretest means that these can be cut out, making the tests
1115 usable with all link sizes.
1116
1117 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1118 stack recursion. This gives a massive performance boost under BSD, but just
1119 a small improvement under Linux. However, it saves one field in the frame
1120 in all cases.
1121
1122 6. Added more features from the forthcoming Perl 5.10:
1123
1124 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1125 recursion call. It refers to the nth most recently opened parentheses.
1126
1127 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1128 to be opened parentheses.
1129
1130 (c) Conditions that refer to capturing parentheses can be specified
1131 relatively, for example, (?(-2)... or (?(+3)...
1132
1133 (d) \K resets the start of the current match so that everything before
1134 is not part of it.
1135
1136 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1137
1138 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1139 reference syntax.
1140
1141 (g) (?| introduces a group in which the numbering of parentheses in each
1142 alternative starts with the same number.
1143
1144 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1145
1146 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1147 PCRE_INFO_JCHANGED.
1148
1149 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1150 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1151 for detecting groups that can match an empty string.
1152
1153 9. A pattern with a very large number of alternatives (more than several
1154 hundred) was running out of internal workspace during the pre-compile
1155 phase, where pcre_compile() figures out how much memory will be needed. A
1156 bit of new cunning has reduced the workspace needed for groups with
1157 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1158 workspace instead of running out of the 4096 that are available.
1159
1160 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1161
1162 11. Applied patch from Google to remove an optimization that didn't quite work.
1163 The report of the bug said:
1164
1165 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1166 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1167 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1168
1169 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1170 it matched the wrong number of bytes.
1171
1172
1173 Version 7.1 24-Apr-07
1174 ---------------------
1175
1176 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1177 that is more "standard", making use of automake and other Autotools. There
1178 is some re-arrangement of the files and adjustment of comments consequent
1179 on this.
1180
1181 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1182 for recursive directory scanning broke on some systems because the files
1183 are not scanned in any specific order and on different systems the order
1184 was different. A call to "sort" has been inserted into RunGrepTest for the
1185 approprate test as a short-term fix. In the longer term there may be an
1186 alternative.
1187
1188 3. I had an email from Eric Raymond about problems translating some of PCRE's
1189 man pages to HTML (despite the fact that I distribute HTML pages, some
1190 people do their own conversions for various reasons). The problems
1191 concerned the use of low-level troff macros .br and .in. I have therefore
1192 removed all such uses from the man pages (some were redundant, some could
1193 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1194 HTML has been updated to handle .nf/.fi and to complain if it encounters
1195 .br or .in.
1196
1197 4. Updated comments in configure.ac that get placed in config.h.in and also
1198 arranged for config.h to be included in the distribution, with the name
1199 config.h.generic, for the benefit of those who have to compile without
1200 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1201
1202 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1203 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1204 makevp.bat for the current PCRE, using the additional files
1205 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1206
1207 6. A Windows user reported a minor discrepancy with test 2, which turned out
1208 to be caused by a trailing space on an input line that had got lost in his
1209 copy. The trailing space was an accident, so I've just removed it.
1210
1211 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1212 that is needed.
1213
1214 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1215 as "const" (a) because they are and (b) because it helps the PHP
1216 maintainers who have recently made a script to detect big data structures
1217 in the php code that should be moved to the .rodata section. I remembered
1218 to update Builducptable as well, so it won't revert if ucptable.h is ever
1219 re-created.
1220
1221 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1222 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1223 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1224 support is not required. This saves 1.5-2K of code, which is important in
1225 some applications.
1226
1227 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1228 so as not to refer to the tables, even though these functions will never be
1229 called when UTF-8 support is disabled. Otherwise there are problems with a
1230 shared library.
1231
1232 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1233
1234 (a) It was defining its arguments as char * instead of void *.
1235
1236 (b) It was assuming that all moves were upwards in memory; this was true
1237 a long time ago when I wrote it, but is no longer the case.
1238
1239 The emulated memove() is provided for those environments that have neither
1240 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1241 is clearly not the case, as these two bugs were recently reported.
1242
1243 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1244 and Detrail to create the HTML documentation, the .txt form of the man
1245 pages, and it removes trailing spaces from listed files. It also creates
1246 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1247 case, it wraps all the #defines with #ifndefs. This script should be run
1248 before "make dist".
1249
1250 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1251 with Unicode property support.
1252
1253 (a) For a maximizing quantifier, if the two different cases of the
1254 character were of different lengths in their UTF-8 codings (there are
1255 some cases like this - I found 11), and the matching function had to
1256 back up over a mixture of the two cases, it incorrectly assumed they
1257 were both the same length.
1258
1259 (b) When PCRE was configured to use the heap rather than the stack for
1260 recursion during matching, it was not correctly preserving the data for
1261 the other case of a UTF-8 character when checking ahead for a match
1262 while processing a minimizing repeat. If the check also involved
1263 matching a wide character, but failed, corruption could cause an
1264 erroneous result when trying to check for a repeat of the original
1265 character.
1266
1267 13. Some tidying changes to the testing mechanism:
1268
1269 (a) The RunTest script now detects the internal link size and whether there
1270 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1271 values substituted by "configure". (The RunGrepTest script already did
1272 this for UTF-8.) The configure.ac script no longer substitutes the
1273 relevant variables.
1274
1275 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1276 with length and offset values. This means that the output is different
1277 for different internal link sizes. Test 2 is skipped for link sizes
1278 other than 2 because of this, bypassing the problem. Unfortunately,
1279 there was also a test in test 3 (the locale tests) that used /B and
1280 failed for link sizes other than 2. Rather than cut the whole test out,
1281 I have added a new /Z option to pcretest that replaces the length and
1282 offset values with spaces. This is now used to make test 3 independent
1283 of link size. (Test 2 will be tidied up later.)
1284
1285 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1286 segmentation fault instead of returning the appropriate error message.
1287
1288 15. In multiline mode when the newline sequence was set to "any", the pattern
1289 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1290 This doesn't seem right; it now treats the CRLF combination as the line
1291 ending, and so does not match in that case. It's only a pattern such as ^$
1292 that would hit this one: something like ^ABC$ would have failed after \r
1293 and then tried again after \r\n.
1294
1295 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1296 in an attempt to make files that differ only in their line terminators
1297 compare equal. This works on Linux.
1298
1299 17. Under certain error circumstances pcregrep might try to free random memory
1300 as it exited. This is now fixed, thanks to valgrind.
1301
1302 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1303 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1304 was because its rules for how to advance for /g after matching an empty
1305 string at the end of a line did not allow for this case. They now check for
1306 it specially.
1307
1308 20. pcretest is supposed to handle patterns and data of any length, by
1309 extending its buffers when necessary. It was getting this wrong when the
1310 buffer for a data line had to be extended.
1311
1312 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1313 CRLF as a newline sequence.
1314
1315 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1316 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1317 I have nevertheless tidied it up.
1318
1319 23. Added some casts to kill warnings from HP-UX ia64 compiler.
1320
1321 24. Added a man page for pcre-config.
1322
1323
1324 Version 7.0 19-Dec-06
1325 ---------------------
1326
1327 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1328 moving to gcc 4.1.1.
1329
1330 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1331 sys/time.h, which is documented as needed for this function. It doesn't
1332 seem to matter on Linux, but it showed up on some releases of OS X.
1333
1334 3. It seems that there are systems where bytes whose values are greater than
1335 127 match isprint() in the "C" locale. The "C" locale should be the
1336 default when a C program starts up. In most systems, only ASCII printing
1337 characters match isprint(). This difference caused the output from pcretest
1338 to vary, making some of the tests fail. I have changed pcretest so that:
1339
1340 (a) When it is outputting text in the compiled version of a pattern, bytes
1341 other than 32-126 are always shown as hex escapes.
1342
1343 (b) When it is outputting text that is a matched part of a subject string,
1344 it does the same, unless a different locale has been set for the match
1345 (using the /L modifier). In this case, it uses isprint() to decide.
1346
1347 4. Fixed a major bug that caused incorrect computation of the amount of memory
1348 required for a compiled pattern when options that changed within the
1349 pattern affected the logic of the preliminary scan that determines the
1350 length. The relevant options are -x, and -i in UTF-8 mode. The result was
1351 that the computed length was too small. The symptoms of this bug were
1352 either the PCRE error "internal error: code overflow" from pcre_compile(),
1353 or a glibc crash with a message such as "pcretest: free(): invalid next
1354 size (fast)". Examples of patterns that provoked this bug (shown in
1355 pcretest format) are:
1356
1357 /(?-x: )/x
1358 /(?x)(?-x: \s*#\s*)/
1359 /((?i)[\x{c0}])/8
1360 /(?i:[\x{c0}])/8
1361
1362 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1363 is now done differently.
1364
1365 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1366 wrapper classes; (b) implement a new function in the C++ scanner that is
1367 more efficient than the old way of doing things because it avoids levels of
1368 recursion in the regex matching; (c) add a paragraph to the documentation
1369 for the FullMatch() function.
1370
1371 6. The escape sequence \n was being treated as whatever was defined as
1372 "newline". Not only was this contrary to the documentation, which states
1373 that \n is character 10 (hex 0A), but it also went horribly wrong when
1374 "newline" was defined as CRLF. This has been fixed.
1375
1376 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
1377 was being set to -1 for the "end of line" case (supposedly a value that no
1378 character can have). Though this value is never used (the check for end of
1379 line is "zero bytes in current character"), it caused compiler complaints.
1380 I've changed it to 0xffffffff.
1381
1382 8. In pcre_version.c, the version string was being built by a sequence of
1383 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
1384 string (as it is for production releases) called a macro with an empty
1385 argument. The C standard says the result of this is undefined. The gcc
1386 compiler treats it as an empty string (which was what was wanted) but it is
1387 reported that Visual C gives an error. The source has been hacked around to
1388 avoid this problem.
1389
1390 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
1391 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
1392 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
1393 of them did).
1394
1395 10. Originally, pcretest opened its input and output without "b"; then I was
1396 told that "b" was needed in some environments, so it was added for release
1397 5.0 to both the input and output. (It makes no difference on Unix-like
1398 systems.) Later I was told that it is wrong for the input on Windows. I've
1399 now abstracted the modes into two macros, to make it easier to fiddle with
1400 them, and removed "b" from the input mode under Windows.
1401
1402 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
1403
1404 12. Added -help and --help to pcretest as an official way of being reminded
1405 of the options.
1406
1407 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
1408 and pcrecpp.cc because they annoy compilers at high warning levels.
1409
1410 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
1411
1412 15. Fixed an occurrence of == in configure.ac that should have been = (shell
1413 scripts are not C programs :-) and which was not noticed because it works
1414 on Linux.
1415
1416 16. pcretest is supposed to handle any length of pattern and data line (as one
1417 line or as a continued sequence of lines) by extending its input buffer if
1418 necessary. This feature was broken for very long pattern lines, leading to
1419 a string of junk being passed to pcre_compile() if the pattern was longer
1420 than about 50K.
1421
1422 17. I have done a major re-factoring of the way pcre_compile() computes the
1423 amount of memory needed for a compiled pattern. Previously, there was code
1424 that made a preliminary scan of the pattern in order to do this. That was
1425 OK when PCRE was new, but as the facilities have expanded, it has become
1426 harder and harder to keep it in step with the real compile phase, and there
1427 have been a number of bugs (see for example, 4 above). I have now found a
1428 cunning way of running the real compile function in a "fake" mode that
1429 enables it to compute how much memory it would need, while actually only
1430 ever using a few hundred bytes of working memory and without too many
1431 tests of the mode. This should make future maintenance and development
1432 easier. A side effect of this work is that the limit of 200 on the nesting
1433 depth of parentheses has been removed (though this was never a serious
1434 limitation, I suspect). However, there is a downside: pcre_compile() now
1435 runs more slowly than before (30% or more, depending on the pattern). I
1436 hope this isn't a big issue. There is no effect on runtime performance.
1437
1438 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
1439 newline (only possible for the last line of a file) and it was a
1440 pattern that set a locale (followed by /Lsomething), pcretest crashed.
1441
1442 19. Added additional timing features to pcretest. (1) The -tm option now times
1443 matching only, not compiling. (2) Both -t and -tm can be followed, as a
1444 separate command line item, by a number that specifies the number of
1445 repeats to use when timing. The default is 50000; this gives better
1446 precision, but takes uncomfortably long for very large patterns.
1447
1448 20. Extended pcre_study() to be more clever in cases where a branch of a
1449 subpattern has no definite first character. For example, (a*|b*)[cd] would
1450 previously give no result from pcre_study(). Now it recognizes that the
1451 first character must be a, b, c, or d.
1452
1453 21. There was an incorrect error "recursive call could loop indefinitely" if
1454 a subpattern (or the entire pattern) that was being tested for matching an
1455 empty string contained only one non-empty item after a nested subpattern.
1456 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
1457 incorrectly, because the \d was being skipped in the check.
1458
1459 22. The pcretest program now has a new pattern option /B and a command line
1460 option -b, which is equivalent to adding /B to every pattern. This causes
1461 it to show the compiled bytecode, without the additional information that
1462 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
1463 is the same as /B/I).
1464
1465 23. A new optimization is now able automatically to treat some sequences such
1466 as a*b as a*+b. More specifically, if something simple (such as a character
1467 or a simple class like \d) has an unlimited quantifier, and is followed by
1468 something that cannot possibly match the quantified thing, the quantifier
1469 is automatically "possessified".
1470
1471 24. A recursive reference to a subpattern whose number was greater than 39
1472 went wrong under certain circumstances in UTF-8 mode. This bug could also
1473 have affected the operation of pcre_study().
1474
1475 25. Realized that a little bit of performance could be had by replacing
1476 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
1477
1478 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
1479
1480 27. Possessive quantifiers such as a++ were previously implemented by turning
1481 them into atomic groups such as ($>a+). Now they have their own opcodes,
1482 which improves performance. This includes the automatically created ones
1483 from 23 above.
1484
1485 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
1486 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
1487 the first matched character to be a colon. This applied both to named and
1488 numbered groups.
1489
1490 29. The ucpinternal.h header file was missing its idempotency #ifdef.
1491
1492 30. I was sent a "project" file called libpcre.a.dev which I understand makes
1493 building PCRE on Windows easier, so I have included it in the distribution.
1494
1495 31. There is now a check in pcretest against a ridiculously large number being
1496 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
1497 loop, the loop is abandoned.
1498
1499 32. Forward references to subpatterns in conditions such as (?(2)...) where
1500 subpattern 2 is defined later cause pcre_compile() to search forwards in
1501 the pattern for the relevant set of parentheses. This search went wrong
1502 when there were unescaped parentheses in a character class, parentheses
1503 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
1504
1505 33. "Subroutine" calls and backreferences were previously restricted to
1506 referencing subpatterns earlier in the regex. This restriction has now
1507 been removed.
1508
1509 34. Added a number of extra features that are going to be in Perl 5.10. On the
1510 whole, these are just syntactic alternatives for features that PCRE had
1511 previously implemented using the Python syntax or my own invention. The
1512 other formats are all retained for compatibility.
1513
1514 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
1515 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
1516 also .NET compatible.
1517
1518 (b) A recursion or subroutine call to a named group can now be defined as
1519 (?&name) as well as (?P>name).
1520
1521 (c) A backreference to a named group can now be defined as \k<name> or
1522 \k'name' as well as (?P=name). The new forms, as well as being in Perl
1523 5.10, are also .NET compatible.
1524
1525 (d) A conditional reference to a named group can now use the syntax
1526 (?(<name>) or (?('name') as well as (?(name).
1527
1528 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
1529 groups (named and numbered) that are never evaluated inline, but can be
1530 called as "subroutines" from elsewhere. In effect, the DEFINE condition
1531 is always false. There may be only one alternative in such a group.
1532
1533 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
1534 as the simple (?(R). The condition is true only if the most recent
1535 recursion is that of the given number or name. It does not search out
1536 through the entire recursion stack.
1537
1538 (g) The escape \gN or \g{N} has been added, where N is a positive or
1539 negative number, specifying an absolute or relative reference.
1540
1541 35. Tidied to get rid of some further signed/unsigned compiler warnings and
1542 some "unreachable code" warnings.
1543
1544 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
1545 things, this adds five new scripts.
1546
1547 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
1548 There were also incompatibilities regarding the handling of \Q..\E inside
1549 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
1550 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
1551
1552 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
1553 matches an empty string, and forcibly breaks the loop. There were bugs in
1554 this code in non-simple cases. For a pattern such as ^(a()*)* matched
1555 against aaaa the result was just "a" rather than "aaaa", for example. Two
1556 separate and independent bugs (that affected different cases) have been
1557 fixed.
1558
1559 39. Refactored the code to abolish the use of different opcodes for small
1560 capturing bracket numbers. This is a tidy that I avoided doing when I
1561 removed the limit on the number of capturing brackets for 3.5 back in 2001.
1562 The new approach is not only tidier, it makes it possible to reduce the
1563 memory needed to fix the previous bug (38).
1564
1565 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
1566 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
1567 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
1568 mode.
1569
1570 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
1571 report.
1572
1573 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
1574 copy construction and assignment in the C++ wrapper.
1575
1576 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
1577 couple of bugs that could have given wrong results in the "--newline=crlf"
1578 case.
1579
1580 44. Added a number of casts and did some reorganization of signed/unsigned int
1581 variables following suggestions from Dair Grant. Also renamed the variable
1582 "this" as "item" because it is a C++ keyword.
1583
1584 45. Arranged for dftables to add
1585
1586 #include "pcre_internal.h"
1587
1588 to pcre_chartables.c because without it, gcc 4.x may remove the array
1589 definition from the final binary if PCRE is built into a static library and
1590 dead code stripping is activated.
1591
1592 46. For an unanchored pattern, if a match attempt fails at the start of a
1593 newline sequence, and the newline setting is CRLF or ANY, and the next two
1594 characters are CRLF, advance by two characters instead of one.
1595
1596
1597 Version 6.7 04-Jul-06
1598 ---------------------
1599
1600 1. In order to handle tests when input lines are enormously long, pcretest has
1601 been re-factored so that it automatically extends its buffers when
1602 necessary. The code is crude, but this _is_ just a test program. The
1603 default size has been increased from 32K to 50K.
1604
1605 2. The code in pcre_study() was using the value of the re argument before
1606 testing it for NULL. (Of course, in any sensible call of the function, it
1607 won't be NULL.)
1608
1609 3. The memmove() emulation function in pcre_internal.h, which is used on
1610 systems that lack both memmove() and bcopy() - that is, hardly ever -
1611 was missing a "static" storage class specifier.
1612
1613 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
1614 containing an extended class (one that cannot be represented by a bitmap
1615 because it contains high-valued characters or Unicode property items, e.g.
1616 [\pZ]). Almost always one would set UTF-8 mode when processing such a
1617 pattern, but PCRE should not loop if you do not (it no longer does).
1618 [Detail: two cases were found: (a) a repeated subpattern containing an
1619 extended class; (b) a recursive reference to a subpattern that followed a
1620 previous extended class. It wasn't skipping over the extended class
1621 correctly when UTF-8 mode was not set.]
1622
1623 5. A negated single-character class was not being recognized as fixed-length
1624 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
1625 compile error "lookbehind assertion is not fixed length".
1626
1627 6. The RunPerlTest auxiliary script was showing an unexpected difference
1628 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
1629 write a Perl script that can interpret lines of an input file either as
1630 byte characters or as UTF-8, which is what "perltest" was being required to
1631 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
1632 can't do is switch easily at run time between having the "use utf8;" pragma
1633 or not. In the end, I fudged it by using the RunPerlTest script to insert
1634 "use utf8;" explicitly for the UTF-8 tests.
1635
1636 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
1637 the end of the subject string, contrary to the documentation and to what
1638 Perl does. This was true of both matching functions. Now it matches only at
1639 the start of the subject and immediately after *internal* newlines.
1640
1641 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
1642 a pointer to an int instead of a pointer to an unsigned long int. This
1643 caused problems on 64-bit systems.
1644
1645 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
1646 instance of the 'standard' template library not being so standard".
1647
1648 10. There was no check on the number of named subpatterns nor the maximum
1649 length of a subpattern name. The product of these values is used to compute
1650 the size of the memory block for a compiled pattern. By supplying a very
1651 long subpattern name and a large number of named subpatterns, the size
1652 computation could be caused to overflow. This is now prevented by limiting
1653 the length of names to 32 characters, and the number of named subpatterns
1654 to 10,000.
1655
1656 11. Subpatterns that are repeated with specific counts have to be replicated in
1657 the compiled pattern. The size of memory for this was computed from the
1658 length of the subpattern and the repeat count. The latter is limited to
1659 65535, but there was no limit on the former, meaning that integer overflow
1660 could in principle occur. The compiled length of a repeated subpattern is
1661 now limited to 30,000 bytes in order to prevent this.
1662
1663 12. Added the optional facility to have named substrings with the same name.
1664
1665 13. Added the ability to use a named substring as a condition, using the
1666 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
1667 are numbers (not recommended). Forward references are permitted.
1668
1669 14. Added forward references in named backreferences (if you see what I mean).
1670
1671 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
1672 pattern could run off the end of the subject. For example, the pattern
1673 "(?s)(.{1,5})"8 did this with the subject "ab".
1674
1675 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
1676 PCRE_CASELESS was set when matching characters that were quantified with ?
1677 or *.
1678
1679 17. A character class other than a single negated character that had a minimum
1680 but no maximum quantifier - for example [ab]{6,} - was not handled
1681 correctly by pce_dfa_exec(). It would match only one character.
1682
1683 18. A valid (though odd) pattern that looked like a POSIX character
1684 class but used an invalid character after [ (for example [[,abc,]]) caused
1685 pcre_compile() to give the error "Failed: internal error: code overflow" or
1686 in some cases to crash with a glibc free() error. This could even happen if
1687 the pattern terminated after [[ but there just happened to be a sequence of
1688 letters, a binary zero, and a closing ] in the memory that followed.
1689
1690 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
1691 over the years. Originally (before any Unicode support), just the bottom 8
1692 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
1693 output from "man perlunicode" includes this:
1694
1695 The regular expression compiler produces polymorphic opcodes. That
1696 is, the pattern adapts to the data and automatically switches to
1697 the Unicode character scheme when presented with Unicode data--or
1698 instead uses a traditional byte scheme when presented with byte
1699 data.
1700
1701 Sadly, a wide octal escape does not cause a switch, and in a string with
1702 no other multibyte characters, these octal escapes are treated as before.
1703 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
1704 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
1705 Unicode string.
1706
1707 I have not perpetrated such confusion in PCRE. Up till now, it took just
1708 the bottom 8 bits, as in old Perl. I have now made octal escapes with
1709 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
1710 translate to the appropriate multibyte character.
1711
1712 29. Applied some refactoring to reduce the number of warnings from Microsoft
1713 and Borland compilers. This has included removing the fudge introduced
1714 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
1715 a warning about an unused variable.
1716
1717 21. PCRE has not included VT (character 0x0b) in the set of whitespace
1718 characters since release 4.0, because Perl (from release 5.004) does not.
1719 [Or at least, is documented not to: some releases seem to be in conflict
1720 with the documentation.] However, when a pattern was studied with
1721 pcre_study() and all its branches started with \s, PCRE still included VT
1722 as a possible starting character. Of course, this did no harm; it just
1723 caused an unnecessary match attempt.
1724
1725 22. Removed a now-redundant internal flag bit that recorded the fact that case
1726 dependency changed within the pattern. This was once needed for "required
1727 byte" processing, but is no longer used. This recovers a now-scarce options
1728 bit. Also moved the least significant internal flag bit to the most-
1729 significant bit of the word, which was not previously used (hangover from
1730 the days when it was an int rather than a uint) to free up another bit for
1731 the future.
1732
1733 23. Added support for CRLF line endings as well as CR and LF. As well as the
1734 default being selectable at build time, it can now be changed at runtime
1735 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
1736 specify that it is scanning data with non-default line endings.
1737
1738 24. Changed the definition of CXXLINK to make it agree with the definition of
1739 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
1740
1741 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
1742 recursions. This makes a big different to stack usage for some patterns.
1743
1744 26. If a subpattern containing a named recursion or subroutine reference such
1745 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
1746 the space required for the compiled pattern went wrong and gave too small a
1747 value. Depending on the environment, this could lead to "Failed: internal
1748 error: code overflow at offset 49" or "glibc detected double free or
1749 corruption" errors.
1750
1751 27. Applied patches from Google (a) to support the new newline modes and (b) to
1752 advance over multibyte UTF-8 characters in GlobalReplace.
1753
1754 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
1755 difference for some implementation of PCRE in some Windows version.
1756
1757 29. Added some extra testing facilities to pcretest:
1758
1759 \q<number> in a data line sets the "match limit" value
1760 \Q<number> in a data line sets the "match recursion limt" value
1761 -S <number> sets the stack size, where <number> is in megabytes
1762
1763 The -S option isn't available for Windows.
1764
1765
1766 Version 6.6 06-Feb-06
1767 ---------------------
1768
1769 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
1770 in pcreposix.h. I have copied the definition from pcre.h.
1771
1772 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
1773 because pcre.h is no longer a built file.
1774
1775 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
1776 not normally included in the compiled code.
1777
1778
1779 Version 6.5 01-Feb-06
1780 ---------------------
1781
1782 1. When using the partial match feature with pcre_dfa_exec(), it was not
1783 anchoring the second and subsequent partial matches at the new starting
1784 point. This could lead to incorrect results. For example, with the pattern
1785 /1234/, partially matching against "123" and then "a4" gave a match.
1786
1787 2. Changes to pcregrep:
1788
1789 (a) All non-match returns from pcre_exec() were being treated as failures
1790 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
1791 error message is output. Some extra information is given for the
1792 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
1793 probably the only errors that are likely to be caused by users (by
1794 specifying a regex that has nested indefinite repeats, for instance).
1795 If there are more than 20 of these errors, pcregrep is abandoned.
1796
1797 (b) A binary zero was treated as data while matching, but terminated the
1798 output line if it was written out. This has been fixed: binary zeroes
1799 are now no different to any other data bytes.
1800
1801 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
1802 used to set a locale for matching. The --locale=xxxx long option has
1803 been added (no short equivalent) to specify a locale explicitly on the
1804 pcregrep command, overriding the environment variables.
1805
1806 (d) When -B was used with -n, some line numbers in the output were one less
1807 than they should have been.
1808
1809 (e) Added the -o (--only-matching) option.
1810
1811 (f) If -A or -C was used with -c (count only), some lines of context were
1812 accidentally printed for the final match.
1813
1814 (g) Added the -H (--with-filename) option.
1815
1816 (h) The combination of options -rh failed to suppress file names for files
1817 that were found from directory arguments.
1818
1819 (i) Added the -D (--devices) and -d (--directories) options.
1820
1821 (j) Added the -F (--fixed-strings) option.
1822
1823 (k) Allow "-" to be used as a file name for -f as well as for a data file.
1824
1825 (l) Added the --colo(u)r option.
1826
1827 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
1828 is not present by default.
1829
1830 3. A nasty bug was discovered in the handling of recursive patterns, that is,
1831 items such as (?R) or (?1), when the recursion could match a number of
1832 alternatives. If it matched one of the alternatives, but subsequently,
1833 outside the recursion, there was a failure, the code tried to back up into
1834 the recursion. However, because of the way PCRE is implemented, this is not
1835 possible, and the result was an incorrect result from the match.
1836
1837 In order to prevent this happening, the specification of recursion has
1838 been changed so that all such subpatterns are automatically treated as
1839 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
1840
1841 4. I had overlooked the fact that, in some locales, there are characters for
1842 which isalpha() is true but neither isupper() nor islower() are true. In
1843 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
1844 and ordfeminine) are like this. This affected the treatment of \w and \W
1845 when they appeared in character classes, but not when they appeared outside
1846 a character class. The bit map for "word" characters is now created
1847 separately from the results of isalnum() instead of just taking it from the
1848 upper, lower, and digit maps. (Plus the underscore character, of course.)
1849
1850 5. The above bug also affected the handling of POSIX character classes such as
1851 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
1852 permanent tables. Instead, the bit maps for such a class were previously
1853 created as the appropriate unions of the upper, lower, and digit bitmaps.
1854 Now they are created by subtraction from the [[:word:]] class, which has
1855 its own bitmap.
1856
1857 6. The [[:blank:]] character class matches horizontal, but not vertical space.
1858 It is created by subtracting the vertical space characters (\x09, \x0a,
1859 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
1860 subtraction was done in the overall bitmap for a character class, meaning
1861 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
1862 be recognized. This bug has been fixed.
1863
1864 7. Patches from the folks at Google:
1865
1866 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
1867 real life, but is still worth protecting against".
1868
1869 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
1870 regular expressions".
1871
1872 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
1873 have it.
1874
1875 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
1876 "configure" and the latter not, in order to fix a problem somebody had
1877 with compiling the Arg class on HP-UX.
1878
1879 (e) Improve the error-handling of the C++ wrapper a little bit.
1880
1881 (f) New tests for checking recursion limiting.
1882
1883 8. The pcre_memmove() function, which is used only if the environment does not
1884 have a standard memmove() function (and is therefore rarely compiled),
1885 contained two bugs: (a) use of int instead of size_t, and (b) it was not
1886 returning a result (though PCRE never actually uses the result).
1887
1888 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
1889 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
1890 returned instead of calling malloc() with an overflowing number that would
1891 most likely cause subsequent chaos.
1892
1893 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
1894
1895 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
1896 with this option is matched, the nmatch and pmatch options of regexec() are
1897 ignored.
1898
1899 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
1900 provided in case anyone wants to the the POSIX interface with UTF-8
1901 strings.
1902
1903 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
1904 C++ linking (needed for some HP-UX environments).
1905
1906 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
1907 (unused parameter) and in the pcre_printint() function (omitted "default"
1908 switch label when the default is to do nothing).
1909
1910 15. Added some code to make it possible, when PCRE is compiled as a C++
1911 library, to replace subject pointers for pcre_exec() with a smart pointer
1912 class, thus making it possible to process discontinuous strings.
1913
1914 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
1915 much the same function. They were added by different people who were trying
1916 to make PCRE easy to compile on non-Unix systems. It has been suggested
1917 that PCRE_EXPORT be abolished now that there is more automatic apparatus
1918 for compiling on Windows systems. I have therefore replaced it with
1919 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
1920 defaults to "extern" for C or "extern C" for C++, which works fine on
1921 Unix-like systems. It is now possible to override the value of PCRE_DATA_
1922 SCOPE with something explicit in config.h. In addition:
1923
1924 (a) pcreposix.h still had just "extern" instead of either of these macros;
1925 I have replaced it with PCRE_DATA_SCOPE.
1926
1927 (b) Functions such as _pcre_xclass(), which are internal to the library,
1928 but external in the C sense, all had PCRE_EXPORT in their definitions.
1929 This is apparently wrong for the Windows case, so I have removed it.
1930 (It makes no difference on Unix-like systems.)
1931
1932 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
1933 of recursive calls to match(). This is different to MATCH_LIMIT because
1934 that limits the total number of calls to match(), not all of which increase
1935 the depth of recursion. Limiting the recursion depth limits the amount of
1936 stack (or heap if NO_RECURSE is set) that is used. The default can be set
1937 when PCRE is compiled, and changed at run time. A patch from Google adds
1938 this functionality to the C++ interface.
1939
1940 18. Changes to the handling of Unicode character properties:
1941
1942 (a) Updated the table to Unicode 4.1.0.
1943
1944 (b) Recognize characters that are not in the table as "Cn" (undefined).
1945
1946 (c) I revised the way the table is implemented to a much improved format
1947 which includes recognition of ranges. It now supports the ranges that
1948 are defined in UnicodeData.txt, and it also amalgamates other
1949 characters into ranges. This has reduced the number of entries in the
1950 table from around 16,000 to around 3,000, thus reducing its size
1951 considerably. I realized I did not need to use a tree structure after
1952 all - a binary chop search is just as efficient. Having reduced the
1953 number of entries, I extended their size from 6 bytes to 8 bytes to
1954 allow for more data.
1955
1956 (d) Added support for Unicode script names via properties such as \p{Han}.
1957
1958 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
1959 matching that character.
1960
1961 20. When matching a repeated Unicode property with a minimum greater than zero,
1962 (for example \pL{2,}), PCRE could look past the end of the subject if it
1963 reached it while seeking the minimum number of characters. This could
1964 happen only if some of the characters were more than one byte long, because
1965 there is a check for at least the minimum number of bytes.
1966
1967 21. Refactored the implementation of \p and \P so as to be more general, to
1968 allow for more different types of property in future. This has changed the
1969 compiled form incompatibly. Anybody with saved compiled patterns that use
1970 \p or \P will have to recompile them.
1971
1972 22. Added "Any" and "L&" to the supported property types.
1973
1974 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
1975 but give a compile time error if the value is greater than 0xff.
1976
1977 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
1978 accidentally not being installed or uninstalled.
1979
1980 25. The pcre.h file was built from pcre.h.in, but the only changes that were
1981 made were to insert the current release number. This seemed silly, because
1982 it made things harder for people building PCRE on systems that don't run
1983 "configure". I have turned pcre.h into a distributed file, no longer built
1984 by "configure", with the version identification directly included. There is
1985 no longer a pcre.h.in file.
1986
1987 However, this change necessitated a change to the pcre-config script as
1988 well. It is built from pcre-config.in, and one of the substitutions was the
1989 release number. I have updated configure.ac so that ./configure now finds
1990 the release number by grepping pcre.h.
1991
1992 26. Added the ability to run the tests under valgrind.
1993
1994
1995 Version 6.4 05-Sep-05
1996 ---------------------
1997
1998 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
1999 "--" to be printed when multiple files were scanned, even when none of the
2000 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2001 consider it to be a bug, and have restored the previous behaviour.
2002
2003 2. A couple of code tidies to get rid of compiler warnings.
2004
2005 3. The pcretest program used to cheat by referring to symbols in the library
2006 whose names begin with _pcre_. These are internal symbols that are not
2007 really supposed to be visible externally, and in some environments it is
2008 possible to suppress them. The cheating is now confined to including
2009 certain files from the library's source, which is a bit cleaner.
2010
2011 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2012 file's purpose clearer.
2013
2014 5. Reorganized pcre_ucp_findchar().
2015
2016
2017 Version 6.3 15-Aug-05
2018 ---------------------
2019
2020 1. The file libpcre.pc.in did not have general read permission in the tarball.
2021
2022 2. There were some problems when building without C++ support:
2023
2024 (a) If C++ support was not built, "make install" and "make test" still
2025 tried to test it.
2026
2027 (b) There were problems when the value of CXX was explicitly set. Some
2028 changes have been made to try to fix these, and ...
2029
2030 (c) --disable-cpp can now be used to explicitly disable C++ support.
2031
2032 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2033 backslash in a target when C++ was disabled. This confuses some
2034 versions of "make", apparently. Using an intermediate variable solves
2035 this. (Same for CPP_LOBJ.)
2036
2037 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2038 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2039 necessary on certain architectures.
2040
2041 4. Added a setting of -export-symbols-regex to the link command to remove
2042 those symbols that are exported in the C sense, but actually are local
2043 within the library, and not documented. Their names all begin with
2044 "_pcre_". This is not a perfect job, because (a) we have to except some
2045 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2046 available (and never for static libraries). I have made a note to try to
2047 find a way round (a) in the future.
2048
2049
2050 Version 6.2 01-Aug-05
2051 ---------------------
2052
2053 1. There was no test for integer overflow of quantifier values. A construction
2054 such as {1111111111111111} would give undefined results. What is worse, if
2055 a minimum quantifier for a parenthesized subpattern overflowed and became
2056 negative, the calculation of the memory size went wrong. This could have
2057 led to memory overwriting.
2058
2059 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2060
2061 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2062 operating environments where this matters.
2063
2064 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2065 PCRE options from within the C++ wrapper.
2066
2067 5. Named capturing subpatterns were not being correctly counted when a pattern
2068 was compiled. This caused two problems: (a) If there were more than 100
2069 such subpatterns, the calculation of the memory needed for the whole
2070 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2071 back references of the form \12, where the number was greater than 9, were
2072 not recognized as back references, even though there were sufficient
2073 previous subpatterns.
2074
2075 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2076 versions of gcc, e.g. 2.95.4.
2077
2078
2079 Version 6.1 21-Jun-05
2080 ---------------------
2081
2082 1. There was one reference to the variable "posix" in pcretest.c that was not
2083 surrounded by "#if !defined NOPOSIX".
2084
2085 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2086 the cross-check on the old pcre_info() function, for the benefit of the
2087 cut-down version of PCRE that is currently imported into Exim.
2088
2089 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2090 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2091 patterns that start (?i) but I don't think that matters. The use of (?i) is
2092 just an example; this all applies to the other options as well.
2093
2094 4. Since libtool seems to echo the compile commands it is issuing, the output
2095 from "make" can be reduced a bit by putting "@" in front of each libtool
2096 compile command.
2097
2098 5. Patch from the folks at Google for configure.in to be a bit more thorough
2099 in checking for a suitable C++ installation before trying to compile the
2100 C++ stuff. This should fix a reported problem when a compiler was present,
2101 but no suitable headers.
2102
2103 6. The man pages all had just "PCRE" as their title. I have changed them to
2104 be the relevant file name. I have also arranged that these names are
2105 retained in the file doc/pcre.txt, which is a concatenation in text format
2106 of all the man pages except the little individual ones for each function.
2107
2108 7. The NON-UNIX-USE file had not been updated for the different set of source
2109 files that come with release 6. I also added a few comments about the C++
2110 wrapper.
2111
2112
2113 Version 6.0 07-Jun-05
2114 ---------------------
2115
2116 1. Some minor internal re-organization to help with my DFA experiments.
2117
2118 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2119 didn't matter for the library itself when fully configured, but did matter
2120 when compiling without UCP support, or within Exim, where the ucp files are
2121 not imported.
2122
2123 3. Refactoring of the library code to split up the various functions into
2124 different source modules. The addition of the new DFA matching code (see
2125 below) to a single monolithic source would have made it really too
2126 unwieldy, quite apart from causing all the code to be include in a
2127 statically linked application, when only some functions are used. This is
2128 relevant even without the DFA addition now that patterns can be compiled in
2129 one application and matched in another.
2130
2131 The downside of splitting up is that there have to be some external
2132 functions and data tables that are used internally in different modules of
2133 the library but which are not part of the API. These have all had their
2134 names changed to start with "_pcre_" so that they are unlikely to clash
2135 with other external names.
2136
2137 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2138 a different (DFA) algorithm. Although it is slower than the original
2139 function, it does have some advantages for certain types of matching
2140 problem.
2141
2142 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2143 including restarting after a partial match.
2144
2145 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2146 defined when compiling for Windows was sent to me. I have put it into the
2147 code, though I have no means of testing or verifying it.
2148
2149 7. Added the pcre_refcount() auxiliary function.
2150
2151 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2152 match before or at the first newline in the subject string. In pcretest,
2153 the /f option on a pattern can be used to set this.
2154
2155 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2156 would behave wrongly. This has been present in PCRE since release 4.0.
2157
2158 10. A number of changes to the pcregrep command:
2159
2160 (a) Refactored how -x works; insert ^(...)$ instead of setting
2161 PCRE_ANCHORED and checking the length, in preparation for adding
2162 something similar for -w.
2163
2164 (b) Added the -w (match as a word) option.
2165
2166 (c) Refactored the way lines are read and buffered so as to have more
2167 than one at a time available.
2168
2169 (d) Implemented a pcregrep test script.
2170
2171 (e) Added the -M (multiline match) option. This allows patterns to match
2172 over several lines of the subject. The buffering ensures that at least
2173 8K, or the rest of the document (whichever is the shorter) is available
2174 for matching (and similarly the previous 8K for lookbehind assertions).
2175
2176 (f) Changed the --help output so that it now says
2177
2178 -w, --word-regex(p)
2179
2180 instead of two lines, one with "regex" and the other with "regexp"
2181 because that confused at least one person since the short forms are the
2182 same. (This required a bit of code, as the output is generated
2183 automatically from a table. It wasn't just a text change.)
2184
2185 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2186 option but starts with a hyphen. Could be a pattern or a path name
2187 starting with a hyphen, for instance.
2188
2189 (h) "-" can be given as a file name to represent stdin.
2190
2191 (i) When file names are being printed, "(standard input)" is used for
2192 the standard input, for compatibility with GNU grep. Previously
2193 "<stdin>" was used.
2194
2195 (j) The option --label=xxx can be used to supply a name to be used for
2196 stdin when file names are being printed. There is no short form.
2197
2198 (k) Re-factored the options decoding logic because we are going to add
2199 two more options that take data. Such options can now be given in four
2200 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2201
2202 (l) Added the -A, -B, and -C options for requesting that lines of context
2203 around matches be printed.
2204
2205 (m) Added the -L option to print the names of files that do not contain
2206 any matching lines, that is, the complement of -l.
2207
2208 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2209 continue to scan other files.
2210
2211 (o) The -s option was incorrectly implemented. For compatibility with other
2212 greps, it now suppresses the error message for a non-existent or non-
2213 accessible file (but not the return code). There is a new option called
2214 -q that suppresses the output of matching lines, which was what -s was
2215 previously doing.
2216
2217 (p) Added --include and --exclude options to specify files for inclusion
2218 and exclusion when recursing.
2219
2220 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2221 Hopefully, it now does.
2222
2223 12. Missing cast in pcre_study().
2224
2225 13. Added an "uninstall" target to the makefile.
2226
2227 14. Replaced "extern" in the function prototypes in Makefile.in with
2228 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2229 world, but is set differently for Windows.
2230
2231 15. Added a second compiling function called pcre_compile2(). The only
2232 difference is that it has an extra argument, which is a pointer to an
2233 integer error code. When there is a compile-time failure, this is set
2234 non-zero, in addition to the error test pointer being set to point to an
2235 error message. The new argument may be NULL if no error number is required
2236 (but then you may as well call pcre_compile(), which is now just a
2237 wrapper). This facility is provided because some applications need a
2238 numeric error indication, but it has also enabled me to tidy up the way
2239 compile-time errors are handled in the POSIX wrapper.
2240
2241 16. Added VPATH=.libs to the makefile; this should help when building with one
2242 prefix path and installing with another. (Or so I'm told by someone who
2243 knows more about this stuff than I do.)
2244
2245 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2246 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2247 match everything, including newlines. This is not POSIX-compatible, but
2248 somebody wanted the feature. From pcretest it can be activated by using
2249 both the P and the s flags.
2250
2251 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2252
2253 19. libpcre.pc was being incorrectly installed as executable.
2254
2255 20. A couple of places in pcretest check for end-of-line by looking for '\n';
2256 it now also looks for '\r' so that it will work unmodified on Windows.
2257
2258 21. Added Google's contributed C++ wrapper to the distribution.
2259
2260 22. Added some untidy missing memory free() calls in pcretest, to keep
2261 Electric Fence happy when testing.
2262
2263
2264
2265 Version 5.0 13-Sep-04
2266 ---------------------
2267
2268 1. Internal change: literal characters are no longer packed up into items
2269 containing multiple characters in a single byte-string. Each character
2270 is now matched using a separate opcode. However, there may be more than one
2271 byte in the character in UTF-8 mode.
2272
2273 2. The pcre_callout_block structure has two new fields: pattern_position and
2274 next_item_length. These contain the offset in the pattern to the next match
2275 item, and its length, respectively.
2276
2277 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2278 insertion of callouts before each pattern item. Added the /C option to
2279 pcretest to make use of this.
2280
2281 4. On the advice of a Windows user, the lines
2282
2283 #if defined(_WIN32) || defined(WIN32)
2284 _setmode( _fileno( stdout ), 0x8000 );
2285 #endif /* defined(_WIN32) || defined(WIN32) */
2286
2287 have been added to the source of pcretest. This apparently does useful
2288 magic in relation to line terminators.
2289
2290 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
2291 for the benefit of those environments where the "b" makes a difference.
2292
2293 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
2294 to know about it. I have put a hack into configure.in that adds in code
2295 to set GCC=yes if CC=icc. This seems to end up at a point in the
2296 generated configure script that is early enough to affect the setting of
2297 compiler options, which is what is needed, but I have no means of testing
2298 whether it really works. (The user who reported this had patched the
2299 generated configure script, which of course I cannot do.)
2300
2301 LATER: After change 22 below (new libtool files), the configure script
2302 seems to know about icc (and also ecc). Therefore, I have commented out
2303 this hack in configure.in.
2304
2305 7. Added support for pkg-config (2 patches were sent in).
2306
2307 8. Negated POSIX character classes that used a combination of internal tables
2308 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
2309 [[:^ascii]]. Typically, they would match almost any characters. The other
2310 POSIX classes were not broken in this way.
2311
2312 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
2313 to find the match, as PCRE was deluded into thinking that the match had to
2314 start at the start point or following a newline. The same bug applied to
2315 patterns with negative forward assertions or any backward assertions
2316 preceding ".*" at the start, unless the pattern required a fixed first
2317 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
2318
2319 10. In UTF-8 mode, when moving forwards in the subject after a failed match
2320 starting at the last subject character, bytes beyond the end of the subject
2321 string were read.
2322
2323 11. Renamed the variable "class" as "classbits" to make life easier for C++
2324 users. (Previously there was a macro definition, but it apparently wasn't
2325 enough.)
2326
2327 12. Added the new field "tables" to the extra data so that tables can be passed
2328 in at exec time, or the internal tables can be re-selected. This allows
2329 a compiled regex to be saved and re-used at a later time by a different
2330 program that might have everything at different addresses.
2331
2332 13. Modified the pcre-config script so that, when run on Solaris, it shows a
2333 -R library as well as a -L library.
2334
2335 14. The debugging options of pcretest (-d on the command line or D on a
2336 pattern) showed incorrect output for anything following an extended class
2337 that contained multibyte characters and which was followed by a quantifier.
2338
2339 15. Added optional support for general category Unicode character properties
2340 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
2341 support. It adds about 90K to the size of the library. The meanings of the
2342 inbuilt class escapes such as \d and \s have NOT been changed.
2343
2344 16. Updated pcredemo.c to include calls to free() to release the memory for the
2345 compiled pattern.
2346
2347 17. The generated file chartables.c was being created in the source directory
2348 instead of in the building directory. This caused the build to fail if the
2349 source directory was different from the building directory, and was
2350 read-only.
2351
2352 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
2353 file. No doubt somebody will tell me if they don't make sense... Also added
2354 Dan Mooney's comments about building on OpenVMS.
2355
2356 19. Added support for partial matching via the PCRE_PARTIAL option for
2357 pcre_exec() and the \P data escape in pcretest.
2358
2359 20. Extended pcretest with 3 new pattern features:
2360
2361 (i) A pattern option of the form ">rest-of-line" causes pcretest to
2362 write the compiled pattern to the file whose name is "rest-of-line".
2363 This is a straight binary dump of the data, with the saved pointer to
2364 the character tables forced to be NULL. The study data, if any, is
2365 written too. After writing, pcretest reads a new pattern.
2366
2367 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
2368 compiled pattern from the given file. There must not be any
2369 occurrences of "<" in the file name (pretty unlikely); if there are,
2370 pcretest will instead treat the initial "<" as a pattern delimiter.
2371 After reading in the pattern, pcretest goes on to read data lines as
2372 usual.
2373
2374 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
2375 and 16-bit fields in a compiled pattern, to simulate a pattern that
2376 was compiled on a host of opposite endianness.
2377
2378 21. The pcre-exec() function can now cope with patterns that were compiled on
2379 hosts of opposite endianness, with this restriction:
2380
2381 As for any compiled expression that is saved and used later, the tables
2382 pointer field cannot be preserved; the extra_data field in the arguments
2383 to pcre_exec() should be used to pass in a tables address if a value
2384 other than the default internal tables were used at compile time.
2385
2386 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
2387 now diagnosed as an error. Previously, most of the time, a negative number
2388 would have been treated as zero, but if in addition "ovector" was passed as
2389 NULL, a crash could occur.
2390
2391 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
2392 new versions from the libtool 1.5 distribution (the last one is a copy of
2393 a file called libtool.m4). This seems to have fixed the need to patch
2394 "configure" to support Darwin 1.3 (which I used to do). However, I still
2395 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
2396 workstation).
2397
2398 24. Changed the PCRE licence to be the more standard "BSD" licence.
2399
2400
2401 Version 4.5 01-Dec-03
2402 ---------------------
2403
2404 1. There has been some re-arrangement of the code for the match() function so
2405 that it can be compiled in a version that does not call itself recursively.
2406 Instead, it keeps those local variables that need separate instances for
2407 each "recursion" in a frame on the heap, and gets/frees frames whenever it
2408 needs to "recurse". Keeping track of where control must go is done by means
2409 of setjmp/longjmp. The whole thing is implemented by a set of macros that
2410 hide most of the details from the main code, and operates only if
2411 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
2412 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
2413 operating.
2414
2415 To make it easier for callers to provide specially tailored get/free
2416 functions for this usage, two new functions, pcre_stack_malloc, and
2417 pcre_stack_free, are used. They are always called in strict stacking order,
2418 and the size of block requested is always the same.
2419
2420 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
2421 PCRE has been compiled to use the stack or the heap for recursion. The
2422 -C option of pcretest uses this to show which version is compiled.
2423
2424 A new data escape \S, is added to pcretest; it causes the amounts of store
2425 obtained and freed by both kinds of malloc/free at match time to be added
2426 to the output.
2427
2428 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
2429 what's available on my current Linux desktop machine.
2430
2431 3. When matching a UTF-8 string, the test for a valid string at the start has
2432 been extended. If start_offset is not zero, PCRE now checks that it points
2433 to a byte that is the start of a UTF-8 character. If not, it returns
2434 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
2435 this is necessary because there may be backward assertions in the pattern.
2436 When matching the same subject several times, it may save resources to use
2437 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
2438
2439 4. The code for checking the validity of UTF-8 strings has been tightened so
2440 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
2441 containing "overlong sequences".
2442
2443 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
2444 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
2445 should have been "&", but it just so happened that all the cases this let
2446 through by mistake were picked up later in the function.
2447
2448 6. I had used a variable called "isblank" - this is a C99 function, causing
2449 some compilers to warn. To avoid this, I renamed it (as "blankclass").
2450
2451 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
2452 prompting; (b) run "./pcretest /dev/null" at the start of the test script
2453 so the version is shown; (c) stop "make test" echoing "./RunTest".
2454
2455 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
2456
2457 9. The prototype for memmove() for systems that don't have it was using
2458 size_t, but the inclusion of the header that defines size_t was later. I've
2459 moved the #includes for the C headers earlier to avoid this.
2460
2461 10. Added some adjustments to the code to make it easier to compiler on certain
2462 special systems:
2463
2464 (a) Some "const" qualifiers were missing.
2465 (b) Added the macro EXPORT before all exported functions; by default this
2466 is defined to be empty.
2467 (c) Changed the dftables auxiliary program (that builds chartables.c) so
2468 that it reads its output file name as an argument instead of writing
2469 to the standard output and assuming this can be redirected.
2470
2471 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
2472 class containing characters with values greater than 255, PCRE compilation
2473 went into a loop.
2474
2475 12. A recursive reference to a subpattern that was within another subpattern
2476 that had a minimum quantifier of zero caused PCRE to crash. For example,
2477 (x(y(?2))z)? provoked this bug with a subject that got as far as the
2478 recursion. If the recursively-called subpattern itself had a zero repeat,
2479 that was OK.
2480
2481 13. In pcretest, the buffer for reading a data line was set at 30K, but the
2482 buffer into which it was copied (for escape processing) was still set at
2483 1024, so long lines caused crashes.
2484
2485 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
2486 "internal error: code overflow...". This applied to any character class
2487 that was followed by a possessive quantifier.
2488
2489 15. Modified the Makefile to add libpcre.la as a prerequisite for
2490 libpcreposix.la because I was told this is needed for a parallel build to
2491 work.
2492
2493 16. If a pattern that contained .* following optional items at the start was
2494 studied, the wrong optimizing data was generated, leading to matching
2495 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
2496 matching string must start with a or b or c. The correct conclusion for
2497 this pattern is that a match can start with any character.
2498
2499
2500 Version 4.4 13-Aug-03
2501 ---------------------
2502
2503 1. In UTF-8 mode, a character class containing characters with values between
2504 127 and 255 was not handled correctly if the compiled pattern was studied.
2505 In fixing this, I have also improved the studying algorithm for such
2506 classes (slightly).
2507
2508 2. Three internal functions had redundant arguments passed to them. Removal
2509 might give a very teeny performance improvement.
2510
2511 3. Documentation bug: the value of the capture_top field in a callout is *one
2512 more than* the number of the hightest numbered captured substring.
2513
2514 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
2515 in incorrectly linking with a previously installed version. They now link
2516 explicitly with libpcre.la.
2517
2518 5. configure.in no longer needs to recognize Cygwin specially.
2519
2520 6. A problem in pcre.in for Windows platforms is fixed.
2521
2522 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
2523 pcretest, it used to include the size of the study block as part of its
2524 output. Unfortunately, the structure contains a field that has a different
2525 size on different hardware architectures. This meant that the tests that
2526 showed this size failed. As the block is currently always of a fixed size,
2527 this information isn't actually particularly useful in pcretest output, so
2528 I have just removed it.
2529
2530 8. Three pre-processor statements accidentally did not start in column 1.
2531 Sadly, there are *still* compilers around that complain, even though
2532 standard C has not required this for well over a decade. Sigh.
2533
2534 9. In pcretest, the code for checking callouts passed small integers in the
2535 callout_data field, which is a void * field. However, some picky compilers
2536 complained about the casts involved for this on 64-bit systems. Now
2537 pcretest passes the address of the small integer instead, which should get
2538 rid of the warnings.
2539
2540 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
2541 both compile and run time, and gives an error if an invalid UTF-8 sequence
2542 is found. There is a option for disabling this check in cases where the
2543 string is known to be correct and/or the maximum performance is wanted.
2544
2545 11. In response to a bug report, I changed one line in Makefile.in from
2546
2547 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
2548 to
2549 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
2550
2551 to look similar to other lines, but I have no way of telling whether this
2552 is the right thing to do, as I do not use Windows. No doubt I'll get told
2553 if it's wrong...
2554
2555
2556 Version 4.3 21-May-03
2557 ---------------------
2558
2559 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
2560 Makefile.
2561
2562 2. Some refactoring to improve the quality of the code:
2563
2564 (i) The utf8_table... variables are now declared "const".
2565
2566 (ii) The code for \cx, which used the "case flipping" table to upper case
2567 lower case letters, now just substracts 32. This is ASCII-specific,
2568 but the whole concept of \cx is ASCII-specific, so it seems
2569 reasonable.
2570
2571 (iii) PCRE was using its character types table to recognize decimal and
2572 hexadecimal digits in the pattern. This is silly, because it handles
2573 only 0-9, a-f, and A-F, but the character types table is locale-
2574 specific, which means strange things might happen. A private
2575 table is now used for this - though it costs 256 bytes, a table is
2576 much faster than multiple explicit tests. Of course, the standard
2577 character types table is still used for matching digits in subject
2578 strings against \d.
2579
2580 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
2581 ending in _t are). So I've renamed it as ESC_tee.
2582
2583 3. The first argument for regexec() in the POSIX wrapper should have been
2584 defined as "const".
2585
2586 4. Changed pcretest to use malloc() for its buffers so that they can be
2587 Electric Fenced for debugging.
2588
2589 5. There were several places in the code where, in UTF-8 mode, PCRE would try
2590 to read one or more bytes before the start of the subject string. Often this
2591 had no effect on PCRE's behaviour, but in some circumstances it could
2592 provoke a segmentation fault.
2593
2594 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
2595 to try to read one or more bytes before the start of the subject string.
2596
2597 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
2598 UTF-8 support could misbehave in various ways if the subject string
2599 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
2600 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
2601 back over UTF-8 characters.)
2602
2603
2604 Version 4.2 14-Apr-03
2605 ---------------------
2606
2607 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
2608
2609 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
2610 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
2611 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
2612 [WIN_PREFIX]: new variable, "cyg" for Cygwin
2613 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
2614 and BUILD_EXEEXT
2615 Note: automatic setting of the BUILD variables is not yet working
2616 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
2617 compile-time but not at link-time
2618 [LINK]: use for linking executables only
2619 make different versions for Windows and non-Windows
2620 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
2621 libraries
2622 [LINK_FOR_BUILD]: new variable
2623 [OBJEXT]: use throughout
2624 [EXEEXT]: use throughout
2625 <winshared>: new target
2626 <wininstall>: new target
2627 <dftables.o>: use native compiler
2628 <dftables>: use native linker
2629 <install>: handle Windows platform correctly
2630 <clean>: ditto
2631 <check>: ditto
2632 copy DLL to top builddir before testing
2633
2634 As part of these changes, -no-undefined was removed again. This was reported
2635 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
2636 in any case.
2637
2638 3. Some tidies to get rid of compiler warnings:
2639
2640 . In the match_data structure, match_limit was an unsigned long int, whereas
2641 match_call_count was an int. I've made them both unsigned long ints.
2642
2643 . In pcretest the fact that a const uschar * doesn't automatically cast to
2644 a void * provoked a warning.
2645
2646 . Turning on some more compiler warnings threw up some "shadow" variables
2647 and a few more missing casts.
2648
2649 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
2650 option, a class that contained a single character with a value between 128
2651 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
2652
2653 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
2654 option, a class that contained several characters, but with at least one
2655 whose value was between 128 and 255 caused PCRE to crash.
2656
2657
2658 Version 4.1 12-Mar-03
2659 ---------------------
2660
2661 1. Compiling with gcc -pedantic found a couple of places where casts were
2662 needed, and a string in dftables.c that was longer than standard compilers are
2663 required to support.
2664
2665 2. Compiling with Sun's compiler found a few more places where the code could
2666 be tidied up in order to avoid warnings.
2667
2668 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
2669 first of these names is deprecated in the latest Autoconf in favour of the name
2670 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
2671 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
2672 analogy I have changed it to CFLAGS_FOR_BUILD.
2673
2674 4. Added -no-undefined to the linking command in the Makefile, because this is
2675 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
2676 linking step for the pcreposix library.
2677
2678 5. PCRE was failing to diagnose the case of two named groups with the same
2679 name.
2680
2681 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
2682 literal character that is needed in the subject for a match, and scans along to
2683 ensure that it is present before embarking on the full matching process. This
2684 saves time in cases of nested unlimited repeats that are never going to match.
2685 Problem: the scan can take a lot of time if the subject is very long (e.g.
2686 megabytes), thus penalizing straightforward matches. It is now done only if the
2687 amount of subject to be scanned is less than 1000 bytes.
2688
2689 7. A lesser problem with the same optimization is that it was recording the
2690 first character of an anchored pattern as "needed", thus provoking a search
2691 right along the subject, even when the first match of the pattern was going to
2692 fail. The "needed" character is now not set for anchored patterns, unless it
2693 follows something in the pattern that is of non-fixed length. Thus, it still
2694 fulfils its original purpose of finding quick non-matches in cases of nested
2695 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
2696
2697
2698 Version 4.0 17-Feb-03
2699 ---------------------
2700
2701 1. If a comment in an extended regex that started immediately after a meta-item
2702 extended to the end of string, PCRE compiled incorrect data. This could lead to
2703 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
2704
2705 2. Moved to autoconf 2.53 and libtool 1.4.2.
2706
2707 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
2708 the special perltest8 script is no longer needed - all the tests can be run
2709 from a single perltest script.
2710
2711 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
2712 by \s. It has now been removed in PCRE. This means it isn't recognized as
2713 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
2714 class [:space:] *does* include VT, thereby creating a mess.
2715
2716 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
2717 space and tab.
2718
2719 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
2720 its new features into the main test script, reducing the number of scripts.
2721
2722 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
2723 were backward compatible, and made the (?i) apply to the whole pattern, as if
2724 /i were given. Now it behaves more logically, and applies the option setting
2725 only to what follows. PCRE has been changed to follow suit. However, if it
2726 finds options settings right at the start of the pattern, it extracts them into
2727 the global options, as before. Thus, they show up in the info data.
2728
2729 8. Added support for the \Q...\E escape sequence. Characters in between are
2730 treated as literals. This is slightly different from Perl in that $ and @ are
2731 also handled as literals inside the quotes. In Perl, they will cause variable
2732 interpolation. Note the following examples:
2733
2734 Pattern PCRE matches Perl matches
2735
2736 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
2737 \Qabc\$xyz\E abc\$xyz abc\$xyz
2738 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2739
2740 For compatibility with Perl, \Q...\E sequences are recognized inside character
2741 classes as well as outside them.
2742
2743 9. Re-organized 3 code statements in pcretest to avoid "overflow in
2744 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
2745 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
2746 signed/unsigned warnings.
2747
2748 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
2749 option for pcretest, so I've replaced it by a simple function that does just
2750 that job.
2751
2752 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
2753 "pcregrep -".
2754
2755 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
2756 Java package. This provides some syntactic sugar for simple cases of what my
2757 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
2758 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
2759 item, you can use this simplified notation. Note that only makes sense with
2760 greedy quantifiers. Consequently, the use of the possessive quantifier forces
2761 greediness, whatever the setting of the PCRE_UNGREEDY option.
2762
2763 13. A change of greediness default within a pattern was not taking effect at
2764 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
2765 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
2766 was abstracted outside.
2767
2768 14. PCRE now supports the \G assertion. It is true when the current matching
2769 position is at the start point of the match. This differs from \A when the
2770 starting offset is non-zero. Used with the /g option of pcretest (or similar
2771 code), it works in the same way as it does for Perl's /g option. If all
2772 alternatives of a regex begin with \G, the expression is anchored to the start
2773 match position, and the "anchored" flag is set in the compiled expression.
2774
2775 15. Some bugs concerning the handling of certain option changes within patterns
2776 have been fixed. These applied to options other than (?ims). For example,
2777 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
2778 been the other way round. Some of this was related to change 7 above.
2779
2780 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
2781 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
2782 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
2783 POSIX classes only within a class (e.g. /[[:alpha:]]/).
2784
2785 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
2786 mode. Unlike ".", it always matches newline, whatever the setting of
2787 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
2788 assertions. Perl allows it, but it doesn't (in general) work because it can't
2789 calculate the length of the lookbehind. At least, that's the case for Perl
2790 5.8.0 - I've been told they are going to document that it doesn't work in
2791 future.
2792
2793 18. Added an error diagnosis for escapes that PCRE does not support: these are
2794 \L, \l, \N, \P, \p, \U, \u, and \X.
2795
2796 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
2797 reading past the end of the pattern in cases such as /[abcd/.
2798
2799 20. PCRE was getting more memory than necessary for patterns with classes that
2800 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
2801
2802 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
2803 compiling PCRE for use with Virtual Pascal.
2804
2805 22. Small fix to the Makefile to make it work properly if the build is done
2806 outside the source tree.
2807
2808 23. Added a new extension: a condition to go with recursion. If a conditional
2809 subpattern starts with (?(R) the "true" branch is used if recursion has
2810 happened, whereas the "false" branch is used only at the top level.
2811
2812 24. When there was a very long string of literal characters (over 255 bytes
2813 without UTF support, over 250 bytes with UTF support), the computation of how
2814 much memory was required could be incorrect, leading to segfaults or other
2815 strange effects.
2816
2817 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
2818 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
2819 there was a subsequent back reference to those brackets. This meant that, for
2820 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
2821 possible to check for precisely this case. All we can do is abandon the
2822 optimization if .* occurs inside capturing brackets when there are any back
2823 references whatsoever. (See below for a better fix that came later.)
2824
2825 26. The handling of the optimization for finding the first character of a
2826 non-anchored pattern, and for finding a character that is required later in the
2827 match were failing in some cases. This didn't break the matching; it just
2828 failed to optimize when it could. The way this is done has been re-implemented.
2829
2830 27. Fixed typo in error message for invalid (?R item (it said "(?p").
2831
2832 28. Added a new feature that provides some of the functionality that Perl
2833 provides with (?{...}). The facility is termed a "callout". The way it is done
2834 in PCRE is for the caller to provide an optional function, by setting
2835 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
2836 global variable. By default it is unset, which disables all calling out. To get
2837 the function called, the regex must include (?C) at appropriate points. This
2838 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
2839 This provides a means of identifying different callout points. When PCRE
2840 reaches such a point in the regex, if pcre_callout has been set, the external
2841 function is called. It is provided with data in a structure called
2842 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
2843 matching continues; if it returns a non-zero value, the match at the current
2844 point fails. However, backtracking will occur if possible. [This was changed
2845 later and other features added - see item 49 below.]
2846
2847 29. pcretest is upgraded to test the callout functionality. It provides a
2848 callout function that displays information. By default, it shows the start of
2849 the match and the current position in the text. There are some new data escapes
2850 to vary what happens:
2851
2852 \C+ in addition, show current contents of captured substrings
2853 \C- do not supply a callout function
2854 \C!n return 1 when callout number n is reached
2855 \C!n!m return 1 when callout number n is reached for the mth time
2856
2857 30. If pcregrep was called with the -l option and just a single file name, it
2858 output "<stdin>" if a match was found, instead of the file name.
2859
2860 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
2861 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
2862 pcre_exec(). This saves a malloc/free per call. The default value of
2863 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
2864 when configuring.
2865
2866 32. The default maximum size of a compiled pattern is 64K. There have been a
2867 few cases of people hitting this limit. The code now uses macros to handle the
2868 storing of links as offsets within the compiled pattern. It defaults to 2-byte
2869 links, but this can be changed to 3 or 4 bytes by --with-link-size when
2870 configuring. Tests 2 and 5 work only with 2-byte links because they output
2871 debugging information about compiled patterns.
2872
2873 33. Internal code re-arrangements:
2874
2875 (a) Moved the debugging function for printing out a compiled regex into
2876 its own source file (printint.c) and used #include to pull it into
2877 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
2878 separate copies.
2879
2880 (b) Defined the list of op-code names for debugging as a macro in
2881 internal.h so that it is next to the definition of the opcodes.
2882
2883 (c) Defined a table of op-code lengths for simpler skipping along compiled
2884 code. This is again a macro in internal.h so that it is next to the
2885 definition of the opcodes.
2886
2887 34. Added support for recursive calls to individual subpatterns, along the
2888 lines of Robin Houston's patch (but implemented somewhat differently).
2889
2890 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
2891 allow it to read and process whole directories in Win32. This code was
2892 contributed by Lionel Fourquaux; it has not been tested by me.
2893
2894 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
2895 used to name a group. Names consist of alphanumerics and underscores, and must
2896 be unique. Back references use the syntax (?P=name) and recursive calls use
2897 (?P>name) which is a PCRE extension to the Python extension. Groups still have
2898 numbers. The function pcre_fullinfo() can be used after compilation to extract
2899 a name/number map. There are three relevant calls:
2900
2901 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
2902 PCRE_INFO_NAMECOUNT yields the number of entries
2903 PCRE_INFO_NAMETABLE yields a pointer to the map.
2904
2905 The map is a vector of fixed-size entries. The size of each entry depends on
2906 the length of the longest name used. The first two bytes of each entry are the
2907 group number, most significant byte first. There follows the corresponding
2908 name, zero terminated. The names are in alphabetical order.
2909
2910 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
2911 case instead of 255. Making it the same both with and without UTF-8 support
2912 means that the same test output works with both.
2913
2914 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
2915 calling malloc() with a zero argument.
2916
2917 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
2918 optimization. I've improved things by keeping a bitmap of backreferences with
2919 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
2920 fact referenced, the optimization can be applied. It is unlikely that a
2921 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
2922 the match to follow \n) will appear inside brackets with a number greater than
2923 31, but if it does, any back reference > 31 suppresses the optimization.
2924
2925 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
2926 of disabling numbered capturing parentheses. Any opening parenthesis that is
2927 not followed by ? behaves as if it were followed by ?: but named parentheses
2928 can still be used for capturing (and they will acquire numbers in the usual
2929 way).
2930
2931 41. Redesigned the return codes from the match() function into yes/no/error so
2932 that errors can be passed back from deep inside the nested calls. A malloc
2933 failure while inside a recursive subpattern call now causes the
2934 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
2935
2936 42. It is now possible to set a limit on the number of times the match()
2937 function is called in a call to pcre_exec(). This facility makes it possible to
2938 limit the amount of recursion and backtracking, though not in a directly
2939 obvious way, because the match() function is used in a number of different
2940 circumstances. The count starts from zero for each position in the subject
2941 string (for non-anchored patterns). The default limit is, for compatibility, a
2942 large number, namely 10 000 000. You can change this in two ways:
2943
2944 (a) When configuring PCRE before making, you can use --with-match-limit=n
2945 to set a default value for the compiled library.
2946
2947 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
2948 a different value is set. See 45 below.
2949
2950 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
2951
2952 43. Added a new function pcre_config(int, void *) to enable run-time extraction
2953 of things that can be changed at compile time. The first argument specifies
2954 what is wanted and the second points to where the information is to be placed.
2955 The current list of available information is:
2956
2957 PCRE_CONFIG_UTF8
2958
2959 The output is an integer that is set to one if UTF-8 support is available;
2960 otherwise it is set to zero.
2961
2962 PCRE_CONFIG_NEWLINE
2963
2964 The output is an integer that it set to the value of the code that is used for
2965 newline. It is either LF (10) or CR (13).
2966
2967 PCRE_CONFIG_LINK_SIZE
2968
2969 The output is an integer that contains the number of bytes used for internal
2970 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
2971
2972 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
2973
2974 The output is an integer that contains the threshold above which the POSIX
2975 interface uses malloc() for output vectors. See item 31 above.
2976
2977 PCRE_CONFIG_MATCH_LIMIT
2978
2979 The output is an unsigned integer that contains the default limit of the number
2980 of match() calls in a pcre_exec() execution. See 42 above.
2981
2982 44. pcretest has been upgraded by the addition of the -C option. This causes it
2983 to extract all the available output from the new pcre_config() function, and to
2984 output it. The program then exits immediately.
2985
2986 45. A need has arisen to pass over additional data with calls to pcre_exec() in
2987 order to support additional features. One way would have been to define
2988 pcre_exec2() (for example) with extra arguments, but this would not have been
2989 extensible, and would also have required all calls to the original function to
2990 be mapped to the new one. Instead, I have chosen to extend the mechanism that
2991 is used for passing in "extra" data from pcre_study().
2992
2993 The pcre_extra structure is now exposed and defined in pcre.h. It currently
2994 contains the following fields:
2995
2996 flags a bitmap indicating which of the following fields are set
2997 study_data opaque data from pcre_study()
2998 match_limit a way of specifying a limit on match() calls for a specific
2999 call to pcre_exec()
3000 callout_data data for callouts (see 49 below)
3001
3002 The flag bits are also defined in pcre.h, and are
3003
3004 PCRE_EXTRA_STUDY_DATA
3005 PCRE_EXTRA_MATCH_LIMIT
3006 PCRE_EXTRA_CALLOUT_DATA
3007
3008 The pcre_study() function now returns one of these new pcre_extra blocks, with
3009 the actual study data pointed to by the study_data field, and the
3010 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3011 before. That is, this change is entirely upwards-compatible and requires no
3012 change to existing code.
3013
3014 If you want to pass in additional data to pcre_exec(), you can either place it
3015 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3016 block.
3017
3018 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3019 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3020 times with different match limits, until it finds the minimum value needed for
3021 pcre_exec() to complete. The value is then output. This can be instructive; for
3022 most simple matches the number is quite small, but for pathological cases it
3023 gets very large very quickly.
3024
3025 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3026 returns the size of the data block pointed to by the study_data field in a
3027 pcre_extra block, that is, the value that was passed as the argument to
3028 pcre_malloc() when PCRE was getting memory in which to place the information
3029 created by pcre_study(). The fourth argument should point to a size_t variable.
3030 pcretest has been extended so that this information is shown after a successful
3031 pcre_study() call when information about the compiled regex is being displayed.
3032
3033 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3034 because what follows is always an absolute path. (Later: it turns out that this
3035 is more than cosmetic for MinGW, because it doesn't like empty path
3036 components.)
3037
3038 49. Some changes have been made to the callout feature (see 28 above):
3039
3040 (i) A callout function now has three choices for what it returns:
3041
3042 0 => success, carry on matching
3043 > 0 => failure at this point, but backtrack if possible
3044 < 0 => serious error, return this value from pcre_exec()
3045
3046 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3047 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3048 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3049 use by callout functions. It will never be used by PCRE itself.
3050
3051 (ii) The pcre_extra structure (see 45 above) has a void * field called
3052 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3053 pcre_callout_block structure has a field of the same name. The contents of
3054 the field passed in the pcre_extra structure are passed to the callout
3055 function in the corresponding field in the callout block. This makes it
3056 easier to use the same callout-containing regex from multiple threads. For
3057 testing, the pcretest program has a new data escape
3058
3059 \C*n pass the number n (may be negative) as callout_data
3060
3061 If the callout function in pcretest receives a non-zero value as
3062 callout_data, it returns that value.
3063
3064 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3065 there were some redundant $(CFLAGS) in commands that are now specified as
3066 $(LINK), which already includes $(CFLAGS).
3067
3068 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3069 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3070 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3071 one-byte characters throughout. Note that case-insensitive matching applies
3072 only to characters whose values are less than 256. PCRE doesn't support the
3073 notion of cases for higher-valued characters.
3074
3075 (i) A character class whose characters are all within 0-255 is handled as
3076 a bit map, and the map is inverted for negative classes. Previously, a
3077 character > 255 always failed to match such a class; however it should
3078 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3079
3080 (ii) A negated character class with a single character < 255 is coded as
3081 "not this character" (OP_NOT). This wasn't working properly when the test
3082 character was multibyte, either singly or repeated.
3083
3084 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3085 mode, for example: \x{100}{2,3}.
3086
3087 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3088 singly or repeated) now correctly test multibyte characters. However,
3089 PCRE doesn't recognize any characters with values greater than 255 as
3090 digits, spaces, or word characters. Such characters always match \D, \S,
3091 and \W, and never match \d, \s, or \w.
3092
3093 (v) Classes may now contain characters and character ranges with values
3094 greater than 255. For example: [ab\x{100}-\x{400}].
3095
3096 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3097 PCRE in UTF-8 mode.
3098
3099 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3100 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3101 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3102 value.)
3103
3104 53. The single man page has become too large. I have therefore split it up into
3105 a number of separate man pages. These also give rise to individual HTML pages;
3106 these are now put in a separate directory, and there is an index.html page that
3107 lists them all. Some hyperlinking between the pages has been installed.
3108
3109 54. Added convenience functions for handling named capturing parentheses.
3110
3111 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3112 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3113 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3114 are faulted.
3115
3116 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3117 calling configure. These values are used when compiling the dftables.c program
3118 which is run to generate the source of the default character tables. They
3119 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3120 you will need to set these values.
3121
3122 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3123
3124
3125 Version 3.9 02-Jan-02
3126 ---------------------
3127
3128 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3129
3130 2. If --disable-static was given, the building process failed when trying to
3131 build pcretest and pcregrep. (For some reason it was using libtool to compile
3132 them, which is not right, as they aren't part of the library.)
3133
3134
3135 Version 3.8 18-Dec-01
3136 ---------------------
3137
3138 1. The experimental UTF-8 code was completely screwed up. It was packing the
3139 bytes in the wrong order. How dumb can you get?
3140
3141
3142 Version 3.7 29-Oct-01
3143 ---------------------
3144
3145 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3146 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3147 this didn't happen under Solaris 8, where I normally test things.
3148
3149 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3150 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3151 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3152 this fix an hour or so after the initial 3.7 release.)
3153
3154
3155 Version 3.6 23-Oct-01
3156 ---------------------
3157
3158 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3159 offsets passed as NULL with zero offset count.
3160
3161 2. The config.guess and config.sub files had not been updated when I moved to
3162 the latest autoconf.
3163
3164
3165 Version 3.5 15-Aug-01
3166 ---------------------
3167
3168 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3169 had been forgotten.
3170
3171 2. By using declared but undefined structures, we can avoid using "void"
3172 definitions in pcre.h while keeping the internal definitions of the structures
3173 private.
3174
3175 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3176 user point of view, this means that both static and shared libraries are built
3177 by default, but this can be individually controlled. More of the work of
3178 handling this static/shared cases is now inside libtool instead of PCRE's make
3179 file.
3180
3181 4. The pcretest utility is now installed along with pcregrep because it is
3182 useful for users (to test regexs) and by doing this, it automatically gets
3183 relinked by libtool. The documentation has been turned into a man page, so
3184 there are now .1, .txt, and .html versions in /doc.
3185
3186 5. Upgrades to pcregrep:
3187 (i) Added long-form option names like gnu grep.
3188 (ii) Added --help to list all options with an explanatory phrase.
3189 (iii) Added -r, --recursive to recurse into sub-directories.
3190 (iv) Added -f, --file to read patterns from a file.
3191
3192 6. pcre_exec() was referring to its "code" argument before testing that
3193 argument for NULL (and giving an error if it was NULL).
3194
3195 7. Upgraded Makefile.in to allow for compiling in a different directory from
3196 the source directory.
3197
3198 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3199 options bits, the pointer it was passed was to an int instead of to an unsigned
3200 long int. This mattered only on 64-bit systems.
3201
3202 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3203 generated) instead of pcre.in, which it its source. Also made the same change
3204 in several of the .c files.
3205
3206 10. A new release of gcc defines printf() as a macro, which broke pcretest
3207 because it had an ifdef in the middle of a string argument for printf(). Fixed
3208 by using separate calls to printf().
3209
3210 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3211 script, to force use of CR or LF instead of \n in the source. On non-Unix
3212 systems, the value can be set in config.h.
3213
3214 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3215 absolute limit. Changed the text of the error message to make this clear, and
3216 likewise updated the man page.
3217
3218 13. The limit of 99 on the number of capturing subpatterns has been removed.
3219 The new limit is 65535, which I hope will not be a "real" limit.
3220
3221
3222 Version 3.4 22-Aug-00
3223 ---------------------
3224
3225 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3226
3227 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3228
3229
3230 Version 3.3 01-Aug-00
3231 ---------------------
3232
3233 1. If an octal character was given, but the value was greater than \377, it
3234 was not getting masked to the least significant bits, as documented. This could
3235 lead to crashes in some systems.
3236
3237 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3238 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3239
3240 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3241 These just pass their arguments on to (pcre_free)(), but they are provided
3242 because some uses of PCRE bind it to non-C systems that can call its functions,
3243 but cannot call free() or pcre_free() directly.
3244
3245 4. Add "make test" as a synonym for "make check". Corrected some comments in
3246 the Makefile.
3247
3248 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3249 Makefile.
3250
3251 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3252 command called pgrep for grepping around the active processes.
3253
3254 7. Added the beginnings of support for UTF-8 character strings.
3255
3256 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
3257 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
3258 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
3259 out for the ar command.)
3260
3261
3262 Version 3.2 12-May-00
3263 ---------------------
3264
3265 This is purely a bug fixing release.
3266
3267 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3268 of ZA. This was just one example of several cases that could provoke this bug,
3269 which was introduced by change 9 of version 2.00. The code for breaking
3270 infinite loops after an iteration that matches an empty string was't working
3271 correctly.
3272
3273 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3274 when matched against abbab (for example). After matching an empty string, it
3275 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3276 caused it to match further down the string than it should.
3277
3278 3. The code contained an inclusion of sys/types.h. It isn't clear why this
3279 was there because it doesn't seem to be needed, and it causes trouble on some
3280 systems, as it is not a Standard C header. It has been removed.
3281
3282 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
3283 were reported on the Macintosh. The changes were from
3284
3285 while ((c = *(++ptr)) != 0 && c != '\n');
3286 to
3287 while ((c = *(++ptr)) != 0 && c != '\n') ;
3288
3289 Totally extraordinary, but if that's what it takes...
3290
3291 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
3292 available. Added HAVE_BCOPY and an autoconf test for it; if neither
3293 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
3294 assumes the way PCRE uses memmove() (always moving upwards).
3295
3296 6. PCRE is being used in one environment where strchr() is not available. There
3297 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
3298 faster code anyway.
3299
3300
3301 Version 3.1 09-Feb-00
3302 ---------------------
3303
3304 The only change in this release is the fixing of some bugs in Makefile.in for
3305 the "install" target:
3306
3307 (1) It was failing to install pcreposix.h.
3308
3309 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
3310
3311
3312 Version 3.0 01-Feb-00
3313 ---------------------
3314
3315 1. Add support for the /+ modifier to perltest (to output $` like it does in
3316 pcretest).
3317
3318 2. Add support for the /g modifier to perltest.
3319
3320 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
3321 matches null strings.
3322
3323 4. Fix perltest so that it doesn't do unwanted things when fed an empty
3324 pattern. Perl treats empty patterns specially - it reuses the most recent
3325 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
3326 effect.
3327
3328 5. The POSIX interface was broken in that it was just handing over the POSIX
3329 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
3330 required a bigger vector, with some working space on the end. This means that
3331 the POSIX wrapper now has to get and free some memory, and copy the results.
3332
3333 6. Added some simple autoconf support, placing the test data and the
3334 documentation in separate directories, re-organizing some of the
3335 information files, and making it build pcre-config (a GNU standard). Also added
3336 libtool support for building PCRE as a shared library, which is now the
3337 default.
3338
3339 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
3340 09 are not valid octal constants. Single digits will be used for minor values
3341 less than 10.
3342
3343 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
3344 existing programs that set these in the POSIX interface can use PCRE without
3345 modification.
3346
3347 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
3348 return all that pcre_info() returns, plus additional data. The pcre_info()
3349 function is retained for compatibility, but is considered to be obsolete.
3350
3351 10. Added experimental recursion feature (?R) to handle one common case that
3352 Perl 5.6 will be able to do with (?p{...}).
3353
3354 11. Added support for POSIX character classes like [:alpha:], which Perl is
3355 adopting.
3356
3357
3358 Version 2.08 31-Aug-99
3359 ----------------------
3360
3361 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
3362 trying to match at the startoffset position, but instead was moving forward to
3363 the next newline as if a previous match had failed.
3364
3365 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
3366 and could get into a loop if a null string was matched other than at the start
3367 of the subject.
3368
3369 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
3370 be distinguished at compile time, and for completeness also added PCRE_DATE.
3371
3372 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
3373 in GnuWin32 environments.
3374
3375
3376 Version 2.07 29-Jul-99
3377 ----------------------
3378
3379 1. The documentation is now supplied in plain text form and HTML as well as in
3380 the form of man page sources.
3381
3382 2. C++ compilers don't like assigning (void *) values to other pointer types.
3383 In particular this affects malloc(). Although there is no problem in Standard
3384 C, I've put in casts to keep C++ compilers happy.
3385
3386 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
3387 should be (const char *).
3388
3389 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
3390 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
3391 However, I haven't made this a standard facility. The documentation doesn't
3392 mention it, and the Makefile doesn't support it.
3393
3394 5. The Makefile now contains an "install" target, with editable destinations at
3395 the top of the file. The pcretest program is not installed.
3396
3397 6. pgrep -V now gives the PCRE version number and date.
3398
3399 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
3400 causing the entire string to be ignored, instead of just the last character.
3401
3402 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
3403 non-matching string, it can take a very, very long time, even for strings of
3404 quite modest length, because of the nested recursion. PCRE now does better in
3405 some of these cases. It does this by remembering the last required literal
3406 character in the pattern, and pre-searching the subject to ensure it is present
3407 before running the real match. In other words, it applies a heuristic to detect
3408 some types of certain failure quickly, and in the above example, if presented
3409 with a string that has no trailing " it gives "no match" very quickly.
3410
3411 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
3412 other alternatives are tried instead.
3413
3414
3415 Version 2.06 09-Jun-99
3416 ----------------------
3417
3418 1. Change pcretest's output for amount of store used to show just the code
3419 space, because the remainder (the data block) varies in size between 32-bit and
3420 64-bit systems.
3421
3422 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
3423 start matching at. This allows lookbehinds to work when searching for multiple
3424 occurrences in a string.
3425
3426 3. Added additional options to pcretest for testing multiple occurrences:
3427
3428 /+ outputs the rest of the string that follows a match
3429 /g loops for multiple occurrences, using the new startoffset argument
3430 /G loops for multiple occurrences by passing an incremented pointer
3431
3432 4. PCRE wasn't doing the "first character" optimization for patterns starting
3433 with \b or \B, though it was doing it for other lookbehind assertions. That is,
3434 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
3435 the letter 'x'. On long subject strings, this gives a significant speed-up.
3436
3437
3438 Version 2.05 21-Apr-99
3439 ----------------------
3440
3441 1. Changed the type of magic_number from int to long int so that it works
3442 properly on 16-bit systems.
3443
3444 2. Fixed a bug which caused patterns starting with .* not to work correctly
3445 when the subject string contained newline characters. PCRE was assuming
3446 anchoring for such patterns in all cases, which is not correct because .* will
3447 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
3448 DOTALL is set at top level; otherwise it knows that patterns starting with .*
3449 must be retried after every newline in the subject.
3450
3451
3452 Version 2.04 18-Feb-99
3453 ----------------------
3454
3455 1. For parenthesized subpatterns with repeats whose minimum was zero, the
3456 computation of the store needed to hold the pattern was incorrect (too large).
3457 If such patterns were nested a few deep, this could multiply and become a real
3458 problem.
3459
3460 2. Added /M option to pcretest to show the memory requirement of a specific
3461 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
3462
3463 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
3464 compiled in such a way that the backtracking after subsequent failure was
3465 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
3466 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
3467
3468
3469 Version 2.03 02-Feb-99
3470 ----------------------
3471
3472 1. Fixed typo and small mistake in man page.
3473
3474 2. Added 4th condition (GPL supersedes if conflict) and created separate
3475 LICENCE file containing the conditions.
3476
3477 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
3478 Perl, that is the internal \ allows the delimiter to be included in the
3479 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
3480 the final delimiter, add \ to the end of the pattern (to test the error).
3481
3482 4. Added the convenience functions for extracting substrings after a successful
3483 match. Updated pcretest to make it able to test these functions.
3484
3485
3486 Version 2.02 14-Jan-99
3487 ----------------------
3488
3489 1. Initialized the working variables associated with each extraction so that
3490 their saving and restoring doesn't refer to uninitialized store.
3491
3492 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
3493 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
3494 fix the problem.
3495
3496 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
3497 calls, and wasn't printing the correct value for compiling calls. Increased the
3498 default value of LOOPREPEAT, and the number of significant figures in the
3499 times.
3500
3501 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
3502
3503 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
3504 a building problem on Windows NT with a FAT file system.
3505
3506
3507 Version 2.01 21-Oct-98
3508 ----------------------
3509
3510 1. Changed the API for pcre_compile() to allow for the provision of a pointer
3511 to character tables built by pcre_maketables() in the current locale. If NULL
3512 is passed, the default tables are used.
3513
3514
3515 Version 2.00 24-Sep-98
3516 ----------------------
3517
3518 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
3519 it any more.
3520
3521 2. Allow quantification of (?>) groups, and make it work correctly.
3522
3523 3. The first character computation wasn't working for (?>) groups.
3524
3525 4. Correct the implementation of \Z (it is permitted to match on the \n at the
3526 end of the subject) and add 5.005's \z, which really does match only at the
3527 very end of the subject.
3528
3529 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
3530
3531 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
3532 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
3533 localized options. All options to pcre_study() were also removed.
3534
3535 7. Add other new features from 5.005:
3536
3537 $(?<= positive lookbehind
3538 $(?<! negative lookbehind
3539 (?imsx-imsx) added the unsetting capability
3540 such a setting is global if at outer level; local otherwise
3541 (?imsx-imsx:) non-capturing groups with option setting
3542 (?(cond)re|re) conditional pattern matching
3543
3544 A backreference to itself in a repeated group matches the previous
3545 captured string.
3546
3547 8. General tidying up of studying (both automatic and via "study")
3548 consequential on the addition of new assertions.
3549
3550 9. As in 5.005, unlimited repeated groups that could match an empty substring
3551 are no longer faulted at compile time. Instead, the loop is forcibly broken at
3552 runtime if any iteration does actually match an empty substring.
3553
3554 10. Include the RunTest script in the distribution.
3555
3556 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
3557 discrepancies, some of which were old and were also with respect to 5.004. They
3558 have now been fixed.
3559
3560
3561 Version 1.09 28-Apr-98
3562 ----------------------
3563
3564 1. A negated single character class followed by a quantifier with a minimum
3565 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
3566 program crashes, or just wrong answers. This did not apply to negated classes
3567 containing more than one character, or to minima other than one.
3568
3569
3570 Version 1.08 27-Mar-98
3571 ----------------------
3572
3573 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
3574
3575 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
3576 latter must appear before anything that relies on it in the pattern.
3577
3578
3579 Version 1.07 16-Feb-98
3580 ----------------------
3581
3582 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
3583 repeat of a potentially empty string).
3584
3585
3586 Version 1.06 23-Jan-98
3587 ----------------------
3588
3589 1. Added Markus Oberhumer's little patches for C++.
3590
3591 2. Literal strings longer than 255 characters were broken.
3592
3593
3594 Version 1.05 23-Dec-97
3595 ----------------------
3596
3597 1. Negated character classes containing more than one character were failing if
3598 PCRE_CASELESS was set at run time.
3599
3600
3601 Version 1.04 19-Dec-97
3602 ----------------------
3603
3604 1. Corrected the man page, where some "const" qualifiers had been omitted.
3605
3606 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
3607 input syntax.
3608
3609 3. Fixed memory leak which occurred when a regex with back references was
3610 matched with an offsets vector that wasn't big enough. The temporary memory
3611 that is used in this case wasn't being freed if the match failed.
3612
3613 4. Tidied pcretest to ensure it frees memory that it gets.
3614
3615 5. Temporary memory was being obtained in the case where the passed offsets
3616 vector was exactly big enough.
3617
3618 6. Corrected definition of offsetof() from change 5 below.
3619
3620 7. I had screwed up change 6 below and broken the rules for the use of
3621 setjmp(). Now fixed.
3622
3623
3624 Version 1.03 18-Dec-97
3625 ----------------------
3626
3627 1. A erroneous regex with a missing opening parenthesis was correctly
3628 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
3629 on some systems.
3630
3631 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
3632 it was reported that one broken compiler failed on the former because "code" is
3633 also an independent variable.
3634
3635 3. The erroneous regex a[]b caused an array overrun reference.
3636
3637 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
3638 fail on data ending with that character. (It was going on too far, and checking
3639 the next character, typically a binary zero.) This was specific to the
3640 optimized code for single-character negative classes.
3641
3642 5. Added a contributed patch from the TIN world which does the following:
3643
3644 + Add an undef for memmove, in case the the system defines a macro for it.
3645
3646 + Add a definition of offsetof(), in case there isn't one. (I don't know
3647 the reason behind this - offsetof() is part of the ANSI standard - but
3648 it does no harm).
3649
3650 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
3651 most of the places where whitespace preceded '#'. I have given up and
3652 allowed the remaining 2 cases to be at the margin.
3653
3654 + Rename some variables in pcre to eliminate shadowing. This seems very
3655 pedantic, but does no harm, of course.
3656
3657 6. Moved the call to setjmp() into its own function, to get rid of warnings
3658 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
3659
3660 7. Constructs such as \d{8,} were compiling into the equivalent of
3661 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
3662 outcome, but in this particular case used more store than had been allocated,
3663 which caused the bug to be discovered because it threw up an internal error.
3664
3665 8. The debugging code in both pcre and pcretest for outputting the compiled
3666 form of a regex was going wrong in the case of back references followed by
3667 curly-bracketed repeats.
3668
3669
3670 Version 1.02 12-Dec-97
3671 ----------------------
3672
3673 1. Typos in pcre.3 and comments in the source fixed.
3674
3675 2. Applied a contributed patch to get rid of places where it used to remove
3676 'const' from variables, and fixed some signed/unsigned and uninitialized
3677 variable warnings.
3678
3679 3. Added the "runtest" target to Makefile.
3680
3681 4. Set default compiler flag to -O2 rather than just -O.
3682
3683
3684 Version 1.01 19-Nov-97
3685 ----------------------
3686
3687 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
3688 like /([ab]*)*/, that is, for classes with more than one character in them.
3689
3690 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
3691 as /((?>a*))*/ (a PCRE_EXTRA facility).
3692
3693
3694 Version 1.00 18-Nov-97
3695 ----------------------
3696
3697 1. Added compile-time macros to support systems such as SunOS4 which don't have
3698 memmove() or strerror() but have other things that can be used instead.
3699
3700 2. Arranged that "make clean" removes the executables.
3701
3702
3703 Version 0.99 27-Oct-97
3704 ----------------------
3705
3706 1. Fixed bug in code for optimizing classes with only one character. It was
3707 initializing a 32-byte map regardless, which could cause it to run off the end
3708 of the memory it had got.
3709
3710 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
3711
3712
3713 Version 0.98 22-Oct-97
3714 ----------------------
3715
3716 1. Fixed bug in code for handling temporary memory usage when there are more
3717 back references than supplied space in the ovector. This could cause segfaults.
3718
3719
3720 Version 0.97 21-Oct-97
3721 ----------------------
3722
3723 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
3724
3725 2. Optimized negated single characters not to use a bit map.
3726
3727 3. Brought error texts together as macro definitions; clarified some of them;
3728 fixed one that was wrong - it said "range out of order" when it meant "invalid
3729 escape sequence".
3730
3731 4. Changed some char * arguments to const char *.
3732
3733 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
3734
3735 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
3736 pcretest.
3737
3738
3739 Version 0.96 16-Oct-97
3740 ----------------------
3741
3742 1. Added a simple "pgrep" utility to the distribution.
3743
3744 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
3745 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
3746 where "ddd" means "one or more decimal digits".
3747
3748 3. Fixed serious bug. If a pattern had a back reference, but the call to
3749 pcre_exec() didn't supply a large enough ovector to record the related
3750 identifying subpattern, the match always failed. PCRE now remembers the number
3751 of the largest back reference, and gets some temporary memory in which to save
3752 the offsets during matching if necessary, in order to ensure that
3753 backreferences always work.
3754
3755 4. Increased the compatibility with Perl in a number of ways:
3756
3757 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
3758 to request this handling. The option can be set at compile or exec time.
3759
3760 (b) $ matches before a terminating newline by default; an option
3761 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
3762 mode). The option can be set at compile or exec time.
3763
3764 (c) The handling of \ followed by a digit other than 0 is now supposed to be
3765 the same as Perl's. If the decimal number it represents is less than 10
3766 or there aren't that many previous left capturing parentheses, an octal
3767 escape is read. Inside a character class, it's always an octal escape,
3768 even if it is a single digit.
3769
3770 (d) An escaped but undefined alphabetic character is taken as a literal,
3771 unless PCRE_EXTRA is set. Currently this just reserves the remaining
3772 escapes.
3773
3774 (e) {0} is now permitted. (The previous item is removed from the compiled
3775 pattern).
3776
3777 5. Changed all the names of code files so that the basic parts are no longer
3778 than 10 characters, and abolished the teeny "globals.c" file.
3779
3780 6. Changed the handling of character classes; they are now done with a 32-byte
3781 bit map always.
3782
3783 7. Added the -d and /D options to pcretest to make it possible to look at the
3784 internals of compilation without having to recompile pcre.
3785
3786
3787 Version 0.95 23-Sep-97
3788 ----------------------
3789
3790 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
3791 \x20 at the start of a run of normal characters. These were being treated as
3792 real characters, instead of the source characters being re-checked.
3793
3794
3795 Version 0.94 18-Sep-97
3796 ----------------------
3797
3798 1. The functions are now thread-safe, with the caveat that the global variables
3799 containing pointers to malloc() and free() or alternative functions are the
3800 same for all threads.
3801
3802 2. Get pcre_study() to generate a bitmap of initial characters for non-
3803 anchored patterns when this is possible, and use it if passed to pcre_exec().
3804
3805
3806 Version 0.93 15-Sep-97
3807 ----------------------
3808
3809 1. /(b)|(:+)/ was computing an incorrect first character.
3810
3811 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
3812 but not actually doing anything yet.
3813
3814 3. Treat "-" characters in classes that cannot be part of ranges as literals,
3815 as Perl does (e.g. [-az] or [az-]).
3816
3817 4. Set the anchored flag if a branch starts with .* or .*? because that tests
3818 all possible positions.
3819
3820 5. Split up into different modules to avoid including unneeded functions in a
3821 compiled binary. However, compile and exec are still in one module. The "study"
3822 function is split off.
3823
3824 6. The character tables are now in a separate module whose source is generated
3825 by an auxiliary program - but can then be edited by hand if required. There are
3826 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
3827 toupper() in the code.
3828
3829 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
3830 make them global. Abolish the function for setting them, as the caller can now
3831 set them directly.
3832
3833
3834 Version 0.92 11-Sep-97
3835 ----------------------
3836
3837 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
3838 (e.g. /a{1,3}/) was broken (I mis-optimized it).
3839
3840 2. Caseless matching was not working in character classes if the characters in
3841 the pattern were in upper case.
3842
3843 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
3844
3845 4. Make PCRE_ANCHORED public and accept as a compile option.
3846
3847 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
3848 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
3849 pass them.
3850
3851 6. Give an error if bad option bits passed at compile or run time.
3852
3853 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
3854 pcretest to cause it to pass that flag.
3855
3856 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
3857 options, and the first character, if set.
3858
3859 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
3860
3861
3862 Version 0.91 10-Sep-97
3863 ----------------------
3864
3865 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
3866 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
3867
3868 2. PCRE was looping on encountering an indefinitely repeated back reference to
3869 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
3870 Perl does - treats the match as successful.
3871
3872 ****

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