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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 333 - (show annotations) (download)
Thu Apr 10 19:55:57 2008 UTC (6 years ago) by ph10
File size: 153110 byte(s)
Add Oniguruma syntax \g<...> and \g'...' for subroutine calls.

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

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