/[pcre]/code/branches/pcre16/ChangeLog
ViewVC logotype

Contents of /code/branches/pcre16/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 818 - (show annotations)
Thu Dec 22 20:11:31 2011 UTC (7 years, 10 months ago) by ph10
File size: 223659 byte(s)
More 16-bit fixes, and removal of pcre_info().
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.22
5 ------------
6
7 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
8 name is defined in ctype.h.
9
10 2. Fixed a bug in the code for calculating the fixed length of lookbehind
11 assertions.
12
13 3. Removed the function pcre_info(), which has been obsolete and deprecated
14 since it was replaced by pcre_fullinfo() in February 2000.
15
16
17 Version 8.21 12-Dec-2011
18 ------------------------
19
20 1. Updating the JIT compiler.
21
22 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
23 are added as well.
24
25 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
26 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
27 calling _pcre_jit_exec. Some extra comments are added.
28
29 4. (*MARK) settings inside atomic groups that do not contain any capturing
30 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
31 was introduced by change 18 for 8.20.
32
33 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
34 ECMA-262 standard.
35
36 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
37 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
38 This bug was probably introduced by change 9 of 8.13.
39
40 7. While fixing 6 above, I noticed that a number of other items were being
41 incorrectly rejected as "not fixed length". This arose partly because newer
42 opcodes had not been added to the fixed-length checking code. I have (a)
43 corrected the bug and added tests for these items, and (b) arranged for an
44 error to occur if an unknown opcode is encountered while checking for fixed
45 length instead of just assuming "not fixed length". The items that were
46 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
47 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
48 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
49
50 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
51 being incorrectly compiled and would have given unpredicatble results.
52
53 9. A possessively repeated subpattern with minimum repeat count greater than
54 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
55 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
56 the first (A) could occur when it should not.
57
58 10. Add a cast and remove a redundant test from the code.
59
60 11. JIT should use pcre_malloc/pcre_free for allocation.
61
62 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
63 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
64 is anything other than /usr, -L is still shown).
65
66 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
67
68 14. Perl does not support \N without a following name in a [] class; PCRE now
69 also gives an error.
70
71 15. If a forward reference was repeated with an upper limit of around 2000,
72 it caused the error "internal error: overran compiling workspace". The
73 maximum number of forward references (including repeats) was limited by the
74 internal workspace, and dependent on the LINK_SIZE. The code has been
75 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
76 the default depends on LINK_SIZE. There is a new upper limit (for safety)
77 of around 200,000 forward references. While doing this, I also speeded up
78 the filling in of repeated forward references.
79
80 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
81 incorrectly expecting the subject to contain another "a" after the start.
82
83 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
84 in the match, the SKIP should be ignored. This was not happening; instead
85 the SKIP was being treated as NOMATCH. For patterns such as
86 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
87 tested.
88
89 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
90 now much more compatible with Perl, in particular in cases where the result
91 is a non-match for a non-anchored pattern. For example, if
92 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
93 "m", where previously it did not return a name. A side effect of this
94 change is that for partial matches, the last encountered mark name is
95 returned, as for non matches. A number of tests that were previously not
96 Perl-compatible have been moved into the Perl-compatible test files. The
97 refactoring has had the pleasing side effect of removing one argument from
98 the match() function, thus reducing its stack requirements.
99
100 19. If the /S+ option was used in pcretest to study a pattern using JIT,
101 subsequent uses of /S (without +) incorrectly behaved like /S+.
102
103 21. Retrieve executable code size support for the JIT compiler and fixing
104 some warnings.
105
106 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
107 not work when the shorter character appeared right at the end of the
108 subject string.
109
110 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
111 systems.
112
113 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
114 output it when the /M option is used in pcretest.
115
116 25. The CheckMan script was not being included in the distribution. Also, added
117 an explicit "perl" to run Perl scripts from the PrepareRelease script
118 because this is reportedly needed in Windows.
119
120 26. If study data was being save in a file and studying had not found a set of
121 "starts with" bytes for the pattern, the data written to the file (though
122 never used) was taken from uninitialized memory and so caused valgrind to
123 complain.
124
125 27. Updated RunTest.bat as provided by Sheri Pierce.
126
127 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
128
129 29. Computation of memory usage for the table of capturing group names was
130 giving an unnecessarily large value.
131
132
133 Version 8.20 21-Oct-2011
134 ------------------------
135
136 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
137 a POSIX class. After further experiments with Perl, which convinced me that
138 Perl has bugs and confusions, a closing square bracket is no longer allowed
139 in a POSIX name. This bug also affected patterns with classes that started
140 with full stops.
141
142 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
143 captured substring, but while checking the failing first alternative,
144 substring 1 is temporarily captured. If the output vector supplied to
145 pcre_exec() was not big enough for this capture, the yield of the function
146 was still zero ("insufficient space for captured substrings"). This cannot
147 be totally fixed without adding another stack variable, which seems a lot
148 of expense for a edge case. However, I have improved the situation in cases
149 such as /(a)(b)x|abc/ matched against "abc", where the return code
150 indicates that fewer than the maximum number of slots in the ovector have
151 been set.
152
153 3. Related to (2) above: when there are more back references in a pattern than
154 slots in the output vector, pcre_exec() uses temporary memory during
155 matching, and copies in the captures as far as possible afterwards. It was
156 using the entire output vector, but this conflicts with the specification
157 that only 2/3 is used for passing back captured substrings. Now it uses
158 only the first 2/3, for compatibility. This is, of course, another edge
159 case.
160
161 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
162 main code base, and can be used by building with --enable-jit. When this is
163 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
164 runtime --no-jit option is given.
165
166 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
167 ovector, the return from the function was zero, implying that there were
168 other matches that did not fit. The correct "exactly full" value is now
169 returned.
170
171 6. If a subpattern that was called recursively or as a subroutine contained
172 (*PRUNE) or any other control that caused it to give a non-standard return,
173 invalid errors such as "Error -26 (nested recursion at the same subject
174 position)" or even infinite loops could occur.
175
176 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
177 computing the minimum length on reaching *ACCEPT, and so ended up with the
178 wrong value of 1 rather than 0. Further investigation indicates that
179 computing a minimum subject length in the presence of *ACCEPT is difficult
180 (think back references, subroutine calls), and so I have changed the code
181 so that no minimum is registered for a pattern that contains *ACCEPT.
182
183 8. If (*THEN) was present in the first (true) branch of a conditional group,
184 it was not handled as intended. [But see 16 below.]
185
186 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
187 Sheri Pierce.
188
189 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
190 the first byte in a match must be "a".
191
192 11. Change 17 for 8.13 increased the recursion depth for patterns like
193 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
194 pattern contains any instances of (*THEN). If it does not, the old
195 optimizations are restored. It would be nice to do this on a per-group
196 basis, but at the moment that is not feasible.
197
198 12. In some environments, the output of pcretest -C is CRLF terminated. This
199 broke RunTest's code that checks for the link size. A single white space
200 character after the value is now allowed for.
201
202 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
203 For "fr", it uses the Windows-specific input and output files.
204
205 14. If (*THEN) appeared in a group that was called recursively or as a
206 subroutine, it did not work as intended. [But see next item.]
207
208 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
209 pattern fragments (but not containing any | characters). If A and B are
210 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
211 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
212 D. In other words, Perl considers parentheses that do not contain any |
213 characters to be part of a surrounding alternative, whereas PCRE was
214 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
215 differently. PCRE now behaves in the same way as Perl, except in the case
216 of subroutine/recursion calls such as (?1) which have in any case always
217 been different (but PCRE had them first :-).
218
219 16. Related to 15 above: Perl does not treat the | in a conditional group as
220 creating alternatives. Such a group is treated in the same way as an
221 ordinary group without any | characters when processing (*THEN). PCRE has
222 been changed to match Perl's behaviour.
223
224 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
225 RunGrepTest script failed.
226
227 18. Change 22 for version 13 caused atomic groups to use more stack. This is
228 inevitable for groups that contain captures, but it can lead to a lot of
229 stack use in large patterns. The old behaviour has been restored for atomic
230 groups that do not contain any capturing parentheses.
231
232 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
233 suppress the check for a minimum subject length at run time. (If it was
234 given to pcre_exec() or pcre_dfa_exec() it did work.)
235
236 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
237 fail to work when decoding hex characters in data strings in EBCDIC
238 environments.
239
240 21. It appears that in at least one Mac OS environment, the isxdigit() function
241 is implemented as a macro that evaluates to its argument more than once,
242 contravening the C 90 Standard (I haven't checked a later standard). There
243 was an instance in pcretest which caused it to go wrong when processing
244 \x{...} escapes in subject strings. The has been rewritten to avoid using
245 things like p++ in the argument of isxdigit().
246
247
248 Version 8.13 16-Aug-2011
249 ------------------------
250
251 1. The Unicode data tables have been updated to Unicode 6.0.0.
252
253 2. Two minor typos in pcre_internal.h have been fixed.
254
255 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
256 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
257 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
258
259 4. There were a number of related bugs in the code for matching backrefences
260 caselessly in UTF-8 mode when codes for the characters concerned were
261 different numbers of bytes. For example, U+023A and U+2C65 are an upper
262 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
263 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
264 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
265 2-byte code at the end of the subject (it thought there wasn't enough data
266 left).
267
268 5. Comprehensive information about what went wrong is now returned by
269 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
270 as the output vector has at least 2 elements. The offset of the start of
271 the failing character and a reason code are placed in the vector.
272
273 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
274 now returned is for the first byte of the failing character, instead of the
275 last byte inspected. This is an incompatible change, but I hope it is small
276 enough not to be a problem. It makes the returned offset consistent with
277 pcre_exec() and pcre_dfa_exec().
278
279 7. pcretest now gives a text phrase as well as the error number when
280 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
281 failure, the offset and reason code are output.
282
283 8. When \R was used with a maximizing quantifier it failed to skip backwards
284 over a \r\n pair if the subsequent match failed. Instead, it just skipped
285 back over a single character (\n). This seems wrong (because it treated the
286 two characters as a single entity when going forwards), conflicts with the
287 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
288 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
289 has been changed.
290
291 9. Some internal refactoring has changed the processing so that the handling
292 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
293 time (the PCRE_DOTALL option was changed this way some time ago: version
294 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
295 which was always a bit of a fudge. It also means that there is one less
296 argument for the match() function, which reduces its stack requirements
297 slightly. This change also fixes an incompatibility with Perl: the pattern
298 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
299
300 10. More internal refactoring has drastically reduced the number of recursive
301 calls to match() for possessively repeated groups such as (abc)++ when
302 using pcre_exec().
303
304 11. While implementing 10, a number of bugs in the handling of groups were
305 discovered and fixed:
306
307 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
308 (a|)*(?1) gave a compile-time internal error.
309 ((a|)+)+ did not notice that the outer group could match an empty string.
310 (^a|^)+ was not marked as anchored.
311 (.*a|.*)+ was not marked as matching at start or after a newline.
312
313 12. Yet more internal refactoring has removed another argument from the match()
314 function. Special calls to this function are now indicated by setting a
315 value in a variable in the "match data" data block.
316
317 13. Be more explicit in pcre_study() instead of relying on "default" for
318 opcodes that mean there is no starting character; this means that when new
319 ones are added and accidentally left out of pcre_study(), testing should
320 pick them up.
321
322 14. The -s option of pcretest has been documented for ages as being an old
323 synonym of -m (show memory usage). I have changed it to mean "force study
324 for every regex", that is, assume /S for every regex. This is similar to -i
325 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
326 using it. It makes it easier to run collections of tests with and without
327 study enabled, and thereby test pcre_study() more easily. All the standard
328 tests are now run with and without -s (but some patterns can be marked as
329 "never study" - see 20 below).
330
331 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
332 restoration of the capturing data to the outer values was not happening
333 correctly.
334
335 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
336 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
337 pattern had matched an empty string, and so incorrectly returned a no
338 match.
339
340 17. There was optimizing code for the last branch of non-capturing parentheses,
341 and also for the obeyed branch of a conditional subexpression, which used
342 tail recursion to cut down on stack usage. Unfortunately, now that there is
343 the possibility of (*THEN) occurring in these branches, tail recursion is
344 no longer possible because the return has to be checked for (*THEN). These
345 two optimizations have therefore been removed. [But see 8.20/11 above.]
346
347 18. If a pattern containing \R was studied, it was assumed that \R always
348 matched two bytes, thus causing the minimum subject length to be
349 incorrectly computed because \R can also match just one byte.
350
351 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
352 was incorrectly computed.
353
354 20. If /S is present twice on a test pattern in pcretest input, it now
355 *disables* studying, thereby overriding the use of -s on the command line
356 (see 14 above). This is necessary for one or two tests to keep the output
357 identical in both cases.
358
359 21. When (*ACCEPT) was used in an assertion that matched an empty string and
360 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
361
362 22. When an atomic group that contained a capturing parenthesis was
363 successfully matched, but the branch in which it appeared failed, the
364 capturing was not being forgotten if a higher numbered group was later
365 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
366 group 1 to "a", when in fact it should be unset. This applied to multi-
367 branched capturing and non-capturing groups, repeated or not, and also to
368 positive assertions (capturing in negative assertions does not happen
369 in PCRE) and also to nested atomic groups.
370
371 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
372 subject after a captured substring, to make it easier to tell which of a
373 number of identical substrings has been captured.
374
375 24. The way atomic groups are processed by pcre_exec() has been changed so that
376 if they are repeated, backtracking one repetition now resets captured
377 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
378 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
379 "aaa". Previously, it would have been "a". As part of this code
380 refactoring, the way recursive calls are handled has also been changed.
381
382 25. If an assertion condition captured any substrings, they were not passed
383 back unless some other capturing happened later. For example, if
384 (?(?=(a))a) was matched against "a", no capturing was returned.
385
386 26. When studying a pattern that contained subroutine calls or assertions,
387 the code for finding the minimum length of a possible match was handling
388 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
389 group 1 called group 2 while simultaneously a separate group 2 called group
390 1). A stack overflow occurred in this case. I have fixed this by limiting
391 the recursion depth to 10.
392
393 27. Updated RunTest.bat in the distribution to the version supplied by Tom
394 Fortmann. This supports explicit test numbers on the command line, and has
395 argument validation and error reporting.
396
397 28. An instance of \X with an unlimited repeat could fail if at any point the
398 first character it looked at was a mark character.
399
400 29. Some minor code refactoring concerning Unicode properties and scripts
401 should reduce the stack requirement of match() slightly.
402
403 30. Added the '=' option to pcretest to check the setting of unused capturing
404 slots at the end of the pattern, which are documented as being -1, but are
405 not included in the return count.
406
407 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
408 compiled something random. Now it gives a compile-time error (as does
409 Perl).
410
411 32. A *MARK encountered during the processing of a positive assertion is now
412 recorded and passed back (compatible with Perl).
413
414 33. If --only-matching or --colour was set on a pcregrep call whose pattern
415 had alternative anchored branches, the search for a second match in a line
416 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
417 matched the line "0102" twice. The same bug affected patterns that started
418 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
419 twice.
420
421 34. Previously, PCRE did not allow quantification of assertions. However, Perl
422 does, and because of capturing effects, quantifying parenthesized
423 assertions may at times be useful. Quantifiers are now allowed for
424 parenthesized assertions.
425
426 35. A minor code tidy in pcre_compile() when checking options for \R usage.
427
428 36. \g was being checked for fancy things in a character class, when it should
429 just be a literal "g".
430
431 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
432 appearance of a nested POSIX class supersedes an apparent external class.
433 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
434 unescaped square brackets may also appear as part of class names. For
435 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
436 more like Perl. (But see 8.20/1 above.)
437
438 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
439 was because it thought it was \N{name}, which is not supported).
440
441 39. Add minix to OS list not supporting the -S option in pcretest.
442
443 40. PCRE tries to detect cases of infinite recursion at compile time, but it
444 cannot analyze patterns in sufficient detail to catch mutual recursions
445 such as ((?1))((?2)). There is now a runtime test that gives an error if a
446 subgroup is called recursively as a subpattern for a second time at the
447 same position in the subject string. In previous releases this might have
448 been caught by the recursion limit, or it might have run out of stack.
449
450 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
451 happen only once. PCRE was, however incorrectly giving a compile time error
452 "recursive call could loop indefinitely" because it cannot analyze the
453 pattern in sufficient detail. The compile time test no longer happens when
454 PCRE is compiling a conditional subpattern, but actual runaway loops are
455 now caught at runtime (see 40 above).
456
457 42. It seems that Perl allows any characters other than a closing parenthesis
458 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
459 has been changed to be the same.
460
461 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
462 as not to get warnings when autogen.sh is called. Also changed
463 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
464
465 44. To help people who use pcregrep to scan files containing exceedingly long
466 lines, the following changes have been made:
467
468 (a) The default value of the buffer size parameter has been increased from
469 8K to 20K. (The actual buffer used is three times this size.)
470
471 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
472 PCRE is built.
473
474 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
475 to be set at run time.
476
477 (d) Numerical values in pcregrep options can be followed by K or M, for
478 example --buffer-size=50K.
479
480 (e) If a line being scanned overflows pcregrep's buffer, an error is now
481 given and the return code is set to 2.
482
483 45. Add a pointer to the latest mark to the callout data block.
484
485 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
486 partial match of an empty string instead of no match. This was specific to
487 the use of ".".
488
489 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
490 complete match instead of a partial match. This bug was dependent on both
491 the PCRE_UTF8 and PCRE_DOTALL options being set.
492
493 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
494 starting byte set, because \b was not being ignored.
495
496
497 Version 8.12 15-Jan-2011
498 ------------------------
499
500 1. Fixed some typos in the markup of the man pages, and wrote a script that
501 checks for such things as part of the documentation building process.
502
503 2. On a big-endian 64-bit system, pcregrep did not correctly process the
504 --match-limit and --recursion-limit options (added for 8.11). In
505 particular, this made one of the standard tests fail. (The integer value
506 went into the wrong half of a long int.)
507
508 3. If the --colour option was given to pcregrep with -v (invert match), it
509 did strange things, either producing crazy output, or crashing. It should,
510 of course, ignore a request for colour when reporting lines that do not
511 match.
512
513 4. Another pcregrep bug caused similar problems if --colour was specified with
514 -M (multiline) and the pattern match finished with a line ending.
515
516 5. In pcregrep, when a pattern that ended with a literal newline sequence was
517 matched in multiline mode, the following line was shown as part of the
518 match. This seems wrong, so I have changed it.
519
520 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
521 the check for further matches in the same line (so they could be coloured)
522 to overrun the end of the current line. If another match was found, it was
523 incorrectly shown (and then shown again when found in the next line).
524
525 7. If pcregrep was compiled under Windows, there was a reference to the
526 function pcregrep_exit() before it was defined. I am assuming this was
527 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
528 reported by a user. I've moved the definition above the reference.
529
530
531 Version 8.11 10-Dec-2010
532 ------------------------
533
534 1. (*THEN) was not working properly if there were untried alternatives prior
535 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
536 backtracked to try for "b" instead of moving to the next alternative branch
537 at the same level (in this case, to look for "c"). The Perl documentation
538 is clear that when (*THEN) is backtracked onto, it goes to the "next
539 alternative in the innermost enclosing group".
540
541 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
542 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
543 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
544 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
545 (*THEN).
546
547 3. If \s appeared in a character class, it removed the VT character from
548 the class, even if it had been included by some previous item, for example
549 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
550 of \s, but is part of the POSIX "space" class.)
551
552 4. A partial match never returns an empty string (because you can always
553 match an empty string at the end of the subject); however the checking for
554 an empty string was starting at the "start of match" point. This has been
555 changed to the "earliest inspected character" point, because the returned
556 data for a partial match starts at this character. This means that, for
557 example, /(?<=abc)def/ gives a partial match for the subject "abc"
558 (previously it gave "no match").
559
560 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
561 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
562 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
563 has an implication that the given string is incomplete (because a partial
564 match is preferred over a full match). For this reason, these items now
565 give a partial match in this situation. [Aside: previously, the one case
566 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
567 match rather than a full match, which was wrong by the old rules, but is
568 now correct.]
569
570 6. There was a bug in the handling of #-introduced comments, recognized when
571 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
572 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
573 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
574 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
575 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
576 places in pcre_compile().
577
578 7. Related to (6) above, when pcre_compile() was skipping #-introduced
579 comments when looking ahead for named forward references to subpatterns,
580 the only newline sequence it recognized was NL. It now handles newlines
581 according to the set newline convention.
582
583 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
584 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
585 cater for a lack of strerror(). These oversights have been fixed.
586
587 9. Added --match-limit and --recursion-limit to pcregrep.
588
589 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
590
591 11. When the -o option was used, pcregrep was setting a return code of 1, even
592 when matches were found, and --line-buffered was not being honoured.
593
594 12. Added an optional parentheses number to the -o and --only-matching options
595 of pcregrep.
596
597 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
598 can match an empty string. The code to do it in pcretest and pcredemo
599 needed fixing:
600
601 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
602 only one byte after an empty string match just before CRLF (this case
603 just got forgotten; "any" and "anycrlf" were OK).
604
605 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
606 mode when an empty string match preceded an ASCII character followed by
607 a non-ASCII character. (The code for advancing by one character rather
608 than one byte was nonsense.)
609
610 (c) The pcredemo.c sample program did not have any code at all to handle
611 the cases when CRLF is a valid newline sequence.
612
613 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
614 as a starting offset was within the subject string. There is now a new
615 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
616 negative or greater than the length of the string. In order to test this,
617 pcretest is extended to allow the setting of negative starting offsets.
618
619 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
620 starting offset points to the beginning of a UTF-8 character was
621 unnecessarily clumsy. I tidied it up.
622
623 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
624 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
625
626 17. Nobody had reported that the --include_dir option, which was added in
627 release 7.7 should have been called --include-dir (hyphen, not underscore)
628 for compatibility with GNU grep. I have changed it to --include-dir, but
629 left --include_dir as an undocumented synonym, and the same for
630 --exclude-dir, though that is not available in GNU grep, at least as of
631 release 2.5.4.
632
633 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
634 characters from a string of bytes) have been redefined so as not to use
635 loops, in order to improve performance in some environments. At the same
636 time, I abstracted some of the common code into auxiliary macros to save
637 repetition (this should not affect the compiled code).
638
639 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
640 compile-time error is now given if \c is not followed by an ASCII
641 character, that is, a byte less than 128. (In EBCDIC mode, the code is
642 different, and any byte value is allowed.)
643
644 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
645 START_OPTIMIZE option, which is now allowed at compile time - but just
646 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
647 to pcregrep and other applications that have no direct access to PCRE
648 options. The new /Y option in pcretest sets this option when calling
649 pcre_compile().
650
651 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
652 back references. Groups containing recursive back references were forced to
653 be atomic by that change, but in the case of named groups, the amount of
654 memory required was incorrectly computed, leading to "Failed: internal
655 error: code overflow". This has been fixed.
656
657 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
658 pcretest.c, to avoid build problems in some Borland environments.
659
660
661 Version 8.10 25-Jun-2010
662 ------------------------
663
664 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
665 THEN.
666
667 2. (*ACCEPT) was not working when inside an atomic group.
668
669 3. Inside a character class, \B is treated as a literal by default, but
670 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
671 causes the error). The code is unchanged, but I tidied the documentation.
672
673 4. Inside a character class, PCRE always treated \R and \X as literals,
674 whereas Perl faults them if its -w option is set. I have changed PCRE so
675 that it faults them when PCRE_EXTRA is set.
676
677 5. Added support for \N, which always matches any character other than
678 newline. (It is the same as "." when PCRE_DOTALL is not set.)
679
680 6. When compiling pcregrep with newer versions of gcc which may have
681 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
682 declared with attribute warn_unused_result" were given. Just casting the
683 result to (void) does not stop the warnings; a more elaborate fudge is
684 needed. I've used a macro to implement this.
685
686 7. Minor change to pcretest.c to avoid a compiler warning.
687
688 8. Added four artifical Unicode properties to help with an option to make
689 \s etc use properties (see next item). The new properties are: Xan
690 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
691
692 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
693 use Unicode properties. (*UCP) at the start of a pattern can be used to set
694 this option. Modified pcretest to add /W to test this facility. Added
695 REG_UCP to make it available via the POSIX interface.
696
697 10. Added --line-buffered to pcregrep.
698
699 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
700 studied, and the match started with a letter with a code point greater than
701 127 whose first byte was different to the first byte of the other case of
702 the letter, the other case of this starting letter was not recognized
703 (#976).
704
705 12. If a pattern that was studied started with a repeated Unicode property
706 test, for example, \p{Nd}+, there was the theoretical possibility of
707 setting up an incorrect bitmap of starting bytes, but fortunately it could
708 not have actually happened in practice until change 8 above was made (it
709 added property types that matched character-matching opcodes).
710
711 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
712 possible starting bytes for non-anchored patterns.
713
714 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
715 \R, and also a number of cases that involve Unicode properties, both
716 explicit and implicit when PCRE_UCP is set.
717
718 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
719 input, it could crash or give wrong results if characters with values
720 greater than 0xc0 were present in the subject string. (Detail: it assumed
721 UTF-8 input when processing these items.)
722
723 16. Added a lot of (int) casts to avoid compiler warnings in systems where
724 size_t is 64-bit (#991).
725
726 17. Added a check for running out of memory when PCRE is compiled with
727 --disable-stack-for-recursion (#990).
728
729 18. If the last data line in a file for pcretest does not have a newline on
730 the end, a newline was missing in the output.
731
732 19. The default pcre_chartables.c file recognizes only ASCII characters (values
733 less than 128) in its various bitmaps. However, there is a facility for
734 generating tables according to the current locale when PCRE is compiled. It
735 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
736 characters, are recognized by isspace() and therefore were getting set in
737 these tables, and indeed these tables seem to approximate to ISO 8859. This
738 caused a problem in UTF-8 mode when pcre_study() was used to create a list
739 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
740 which of course cannot start UTF-8 characters. I have changed the code so
741 that only real ASCII characters (less than 128) and the correct starting
742 bytes for UTF-8 encodings are set for characters greater than 127 when in
743 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
744 altogether.)
745
746 20. Added the /T option to pcretest so as to be able to run tests with non-
747 standard character tables, thus making it possible to include the tests
748 used for 19 above in the standard set of tests.
749
750 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
751 reference to a subpattern the other side of a comment that contains an
752 opening parenthesis caused either an internal compiling error, or a
753 reference to the wrong subpattern.
754
755
756 Version 8.02 19-Mar-2010
757 ------------------------
758
759 1. The Unicode data tables have been updated to Unicode 5.2.0.
760
761 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
762 configured.
763
764 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
765 original author of that file, following a query about its status.
766
767 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
768 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
769
770 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
771 quantifier applied to a forward-referencing subroutine call, could compile
772 incorrect code or give the error "internal error: previously-checked
773 referenced subpattern not found".
774
775 6. Both MS Visual Studio and Symbian OS have problems with initializing
776 variables to point to external functions. For these systems, therefore,
777 pcre_malloc etc. are now initialized to local functions that call the
778 relevant global functions.
779
780 7. There were two entries missing in the vectors called coptable and poptable
781 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
782 I've fixed the data, and added a kludgy way of testing at compile time that
783 the lengths are correct (equal to the number of opcodes).
784
785 8. Following on from 7, I added a similar kludge to check the length of the
786 eint vector in pcreposix.c.
787
788 9. Error texts for pcre_compile() are held as one long string to avoid too
789 much relocation at load time. To find a text, the string is searched,
790 counting zeros. There was no check for running off the end of the string,
791 which could happen if a new error number was added without updating the
792 string.
793
794 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
795
796 11. \K was not working if it appeared in an atomic group or in a group that
797 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
798 \K is "not well defined" if used in an assertion. PCRE now accepts it if
799 the assertion is positive, but not if it is negative.
800
801 12. Change 11 fortuitously reduced the size of the stack frame used in the
802 "match()" function of pcre_exec.c by one pointer. Forthcoming
803 implementation of support for (*MARK) will need an extra pointer on the
804 stack; I have reserved it now, so that the stack frame size does not
805 decrease.
806
807 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
808 item in branch that calls a recursion is a subroutine call - as in the
809 second branch in the above example - was incorrectly given the compile-
810 time error "recursive call could loop indefinitely" because pcre_compile()
811 was not correctly checking the subroutine for matching a non-empty string.
812
813 14. The checks for overrunning compiling workspace could trigger after an
814 overrun had occurred. This is a "should never occur" error, but it can be
815 triggered by pathological patterns such as hundreds of nested parentheses.
816 The checks now trigger 100 bytes before the end of the workspace.
817
818 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
819
820
821 Version 8.01 19-Jan-2010
822 ------------------------
823
824 1. If a pattern contained a conditional subpattern with only one branch (in
825 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
826 computed the wrong minimum data length (which is of course zero for such
827 subpatterns). This could cause incorrect "no match" results.
828
829 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
830 the pattern is reset in the first branch, pcre_compile() failed with
831 "internal error: code overflow at offset...". This happened only when
832 the reset was to the original external option setting. (An optimization
833 abstracts leading options settings into an external setting, which was the
834 cause of this.)
835
836 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
837 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
838 assertion pattern did not match (meaning that the assertion was true), it
839 was incorrectly treated as false if the SKIP had been reached during the
840 matching. This also applied to assertions used as conditions.
841
842 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
843 assertion subpattern, including such a pattern used as a condition,
844 unpredictable results occurred, instead of the error return
845 PCRE_ERROR_DFA_UITEM.
846
847 5. The C++ GlobalReplace function was not working like Perl for the special
848 situation when an empty string is matched. It now does the fancy magic
849 stuff that is necessary.
850
851 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
852 removed. (These were left over from very, very early versions of PCRE.)
853
854 7. Some cosmetic changes to the code to make life easier when compiling it
855 as part of something else:
856
857 (a) Change DEBUG to PCRE_DEBUG.
858
859 (b) In pcre_compile(), rename the member of the "branch_chain" structure
860 called "current" as "current_branch", to prevent a collision with the
861 Linux macro when compiled as a kernel module.
862
863 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
864 prevent a collision with the Linux macro when compiled as a kernel
865 module.
866
867 8. In pcre_compile() there are some checks for integer overflows that used to
868 cast potentially large values to (double). This has been changed to that
869 when building, a check for int64_t is made, and if it is found, it is used
870 instead, thus avoiding the use of floating point arithmetic. (There is no
871 other use of FP in PCRE.) If int64_t is not found, the fallback is to
872 double.
873
874 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
875 2005 (difference between two addresses compared to an unsigned value).
876
877 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
878 custom one, because of the following reported problem in Windows:
879
880 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
881 under Win32.
882 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
883 therefore missing the function definition.
884 - The compiler thus generates a "C" signature for the test function.
885 - The linker fails to find the "C" function.
886 - PCRE fails to configure if asked to do so against libbz2.
887
888 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
889 messages were output:
890
891 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
892 rerunning libtoolize, to keep the correct libtool macros in-tree.
893 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
894
895 I have done both of these things.
896
897 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
898 most of the time, it *can* run out if it is given a pattern that contains a
899 runaway infinite recursion. I updated the discussion in the pcrestack man
900 page.
901
902 13. Now that we have gone to the x.xx style of version numbers, the minor
903 version may start with zero. Using 08 or 09 is a bad idea because users
904 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
905 interpreted as invalid octal numbers. I've updated the previous comment in
906 configure.ac, and also added a check that gives an error if 08 or 09 are
907 used.
908
909 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
910 causing partial matching to fail when the end of the subject matched \W
911 in a UTF-8 pattern where \W was quantified with a minimum of 3.
912
913 15. There were some discrepancies between the declarations in pcre_internal.h
914 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
915 their definitions. The declarations used "const uschar *" and the
916 definitions used USPTR. Even though USPTR is normally defined as "const
917 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
918 reported that: "This difference in casting confuses some C++ compilers, for
919 example, SunCC recognizes above declarations as different functions and
920 generates broken code for hbpcre." I have changed the declarations to use
921 USPTR.
922
923 16. GNU libtool is named differently on some systems. The autogen.sh script now
924 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
925 (FreeBSD).
926
927 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
928 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
929 comment: "Figure out how to create a longlong from a string: strtoll and
930 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
931 instance, but it only takes 2 args instead of 3!"
932
933 18. A subtle bug concerned with back references has been fixed by a change of
934 specification, with a corresponding code fix. A pattern such as
935 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
936 refers, was giving matches when it shouldn't. For example, xa=xaaa would
937 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
938 same bug. Such groups have to be quantified to be useful, or contained
939 inside another quantified group. (If there's no repetition, the reference
940 can never match.) The problem arises because, having left the group and
941 moved on to the rest of the pattern, a later failure that backtracks into
942 the group uses the captured value from the final iteration of the group
943 rather than the correct earlier one. I have fixed this in PCRE by forcing
944 any group that contains a reference to itself to be an atomic group; that
945 is, there cannot be any backtracking into it once it has completed. This is
946 similar to recursive and subroutine calls.
947
948
949 Version 8.00 19-Oct-09
950 ----------------------
951
952 1. The table for translating pcre_compile() error codes into POSIX error codes
953 was out-of-date, and there was no check on the pcre_compile() error code
954 being within the table. This could lead to an OK return being given in
955 error.
956
957 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
958 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
959 in a Windows environment.
960
961 3. The pcregrep --count option prints the count for each file even when it is
962 zero, as does GNU grep. However, pcregrep was also printing all files when
963 --files-with-matches was added. Now, when both options are given, it prints
964 counts only for those files that have at least one match. (GNU grep just
965 prints the file name in this circumstance, but including the count seems
966 more useful - otherwise, why use --count?) Also ensured that the
967 combination -clh just lists non-zero counts, with no names.
968
969 4. The long form of the pcregrep -F option was incorrectly implemented as
970 --fixed_strings instead of --fixed-strings. This is an incompatible change,
971 but it seems right to fix it, and I didn't think it was worth preserving
972 the old behaviour.
973
974 5. The command line items --regex=pattern and --regexp=pattern were not
975 recognized by pcregrep, which required --regex pattern or --regexp pattern
976 (with a space rather than an '='). The man page documented the '=' forms,
977 which are compatible with GNU grep; these now work.
978
979 6. No libpcreposix.pc file was created for pkg-config; there was just
980 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
981
982 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
983 when UCP support is not needed, by modifying the Python script that
984 generates it from Unicode data files. This should not matter if the module
985 is correctly used as a library, but I received one complaint about 50K of
986 unwanted data. My guess is that the person linked everything into his
987 program rather than using a library. Anyway, it does no harm.
988
989 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
990 was a minimum greater than 1 for a wide character in a possessive
991 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
992 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
993 character. Chaos in the form of incorrect output or a compiling loop could
994 result.
995
996 9. The restrictions on what a pattern can contain when partial matching is
997 requested for pcre_exec() have been removed. All patterns can now be
998 partially matched by this function. In addition, if there are at least two
999 slots in the offset vector, the offset of the earliest inspected character
1000 for the match and the offset of the end of the subject are set in them when
1001 PCRE_ERROR_PARTIAL is returned.
1002
1003 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1004 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1005 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1006 and may be more useful for multi-segment matching.
1007
1008 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1009 used to be given if ever the end of the subject was reached; now it is
1010 given only if matching could not proceed because another character was
1011 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1012 string "Z", which now yields "no match" instead of "partial match". In the
1013 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1014 final character ended with (*FAIL).
1015
1016 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1017 if the pattern had a "must contain" character that was already found in the
1018 earlier partial match, unless partial matching was again requested. For
1019 example, with the pattern /dog.(body)?/, the "must contain" character is
1020 "g". If the first part-match was for the string "dog", restarting with
1021 "sbody" failed. This bug has been fixed.
1022
1023 13. The string returned by pcre_dfa_exec() after a partial match has been
1024 changed so that it starts at the first inspected character rather than the
1025 first character of the match. This makes a difference only if the pattern
1026 starts with a lookbehind assertion or \b or \B (\K is not supported by
1027 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1028 matching functions compatible, and I think it's the right thing to do.
1029
1030 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1031 so that the demonstration program is easily available in environments where
1032 PCRE has not been installed from source.
1033
1034 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1035 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1036 library.
1037
1038 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1039 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1040 is not the first non-POSIX option to be added. Clearly some people find
1041 these options useful.
1042
1043 17. If a caller to the POSIX matching function regexec() passes a non-zero
1044 value for nmatch with a NULL value for pmatch, the value of
1045 nmatch is forced to zero.
1046
1047 18. RunGrepTest did not have a test for the availability of the -u option of
1048 the diff command, as RunTest does. It now checks in the same way as
1049 RunTest, and also checks for the -b option.
1050
1051 19. If an odd number of negated classes containing just a single character
1052 interposed, within parentheses, between a forward reference to a named
1053 subpattern and the definition of the subpattern, compilation crashed with
1054 an internal error, complaining that it could not find the referenced
1055 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1056 [The bug was that it was starting one character too far in when skipping
1057 over the character class, thus treating the ] as data rather than
1058 terminating the class. This meant it could skip too much.]
1059
1060 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1061 /g option in pcretest when the pattern contains \K, which makes it possible
1062 to have an empty string match not at the start, even when the pattern is
1063 anchored. Updated pcretest and pcredemo to use this option.
1064
1065 21. If the maximum number of capturing subpatterns in a recursion was greater
1066 than the maximum at the outer level, the higher number was returned, but
1067 with unset values at the outer level. The correct (outer level) value is
1068 now given.
1069
1070 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1071 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1072 make it do so. The string so far is captured, making this feature
1073 compatible with Perl.
1074
1075 23. The tests have been re-organized, adding tests 11 and 12, to make it
1076 possible to check the Perl 5.10 features against Perl 5.10.
1077
1078 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1079 pattern matches a fixed length string. PCRE did not allow this; now it
1080 does. Neither allows recursion.
1081
1082 25. I finally figured out how to implement a request to provide the minimum
1083 length of subject string that was needed in order to match a given pattern.
1084 (It was back references and recursion that I had previously got hung up
1085 on.) This code has now been added to pcre_study(); it finds a lower bound
1086 to the length of subject needed. It is not necessarily the greatest lower
1087 bound, but using it to avoid searching strings that are too short does give
1088 some useful speed-ups. The value is available to calling programs via
1089 pcre_fullinfo().
1090
1091 26. While implementing 25, I discovered to my embarrassment that pcretest had
1092 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1093 study optimizations had never been tested with that matching function.
1094 Oops. What is worse, even when it was passed study data, there was a bug in
1095 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1096 were also very few tests of studied patterns with pcre_dfa_exec().
1097
1098 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1099 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1100 on the other side of the coin, they are no longer allowed to have different
1101 names, because these cannot be distinguished in PCRE, and this has caused
1102 confusion. (This is a difference from Perl.)
1103
1104 28. When duplicate subpattern names are present (necessarily with different
1105 numbers, as required by 27 above), and a test is made by name in a
1106 conditional pattern, either for a subpattern having been matched, or for
1107 recursion in such a pattern, all the associated numbered subpatterns are
1108 tested, and the overall condition is true if the condition is true for any
1109 one of them. This is the way Perl works, and is also more like the way
1110 testing by number works.
1111
1112
1113 Version 7.9 11-Apr-09
1114 ---------------------
1115
1116 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1117 (pcretest), all targets were linked against these libraries. This included
1118 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1119 libraries. This caused unwanted dependencies to be created. This problem
1120 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1121 pcretest is linked with readline.
1122
1123 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1124 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1125 moved outside it again, because FALSE and TRUE are already defined in AIX,
1126 but BOOL is not.
1127
1128 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1129 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1130
1131 4. The pcregrep documentation said spaces were inserted as well as colons (or
1132 hyphens) following file names and line numbers when outputting matching
1133 lines. This is not true; no spaces are inserted. I have also clarified the
1134 wording for the --colour (or --color) option.
1135
1136 5. In pcregrep, when --colour was used with -o, the list of matching strings
1137 was not coloured; this is different to GNU grep, so I have changed it to be
1138 the same.
1139
1140 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1141 each matching line was coloured. Now it goes on to look for further matches
1142 of any of the test patterns, which is the same behaviour as GNU grep.
1143
1144 7. A pattern that could match an empty string could cause pcregrep to loop; it
1145 doesn't make sense to accept an empty string match in pcregrep, so I have
1146 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1147 seems to be how GNU grep behaves.
1148
1149 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1150 start or after a newline", because the conditional assertion was not being
1151 correctly handled. The rule now is that both the assertion and what follows
1152 in the first alternative must satisfy the test.
1153
1154 9. If auto-callout was enabled in a pattern with a conditional group whose
1155 condition was an assertion, PCRE could crash during matching, both with
1156 pcre_exec() and pcre_dfa_exec().
1157
1158 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1159 used for matching.
1160
1161 11. Unicode property support in character classes was not working for
1162 characters (bytes) greater than 127 when not in UTF-8 mode.
1163
1164 12. Added the -M command line option to pcretest.
1165
1166 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1167
1168 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1169
1170 16. Added comments and documentation about mis-use of no_arg in the C++
1171 wrapper.
1172
1173 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1174 from Martin Jerabek that uses macro names for all relevant character and
1175 string constants.
1176
1177 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1178 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1179 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1180 these, but not everybody uses configure.
1181
1182 19. A conditional group that had only one branch was not being correctly
1183 recognized as an item that could match an empty string. This meant that an
1184 enclosing group might also not be so recognized, causing infinite looping
1185 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1186 with the subject "ab", where knowledge that the repeated group can match
1187 nothing is needed in order to break the loop.
1188
1189 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1190 exec(), but without supplying a callout function, matching went wrong.
1191
1192 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1193 leak if the size of the offset vector was greater than 30. When the vector
1194 is smaller, the saved offsets during recursion go onto a local stack
1195 vector, but for larger vectors malloc() is used. It was failing to free
1196 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1197 error, in fact).
1198
1199 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1200 heapframe that is used only when UTF-8 support is enabled. This caused no
1201 problem, but was untidy.
1202
1203 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1204 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1205 included within another project.
1206
1207 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1208 slightly modified by me:
1209
1210 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1211 not building pcregrep.
1212
1213 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1214 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1215
1216 25. Forward references, both numeric and by name, in patterns that made use of
1217 duplicate group numbers, could behave incorrectly or give incorrect errors,
1218 because when scanning forward to find the reference group, PCRE was not
1219 taking into account the duplicate group numbers. A pattern such as
1220 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1221
1222 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1223 the feature of a custom pointer more persuasive (as requested by a user).
1224
1225 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1226 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1227 pre-defined.
1228
1229 28. Added support for (*UTF8) at the start of a pattern.
1230
1231 29. Arrange for flags added by the "release type" setting in CMake to be shown
1232 in the configuration summary.
1233
1234
1235 Version 7.8 05-Sep-08
1236 ---------------------
1237
1238 1. Replaced UCP searching code with optimized version as implemented for Ad
1239 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1240 stage table and inline lookup instead of a function, giving speed ups of 2
1241 to 5 times on some simple patterns that I tested. Permission was given to
1242 distribute the MultiStage2.py script that generates the tables (it's not in
1243 the tarball, but is in the Subversion repository).
1244
1245 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1246 scripts.
1247
1248 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1249 a group with a zero qualifier. The result of the study could be incorrect,
1250 or the function might crash, depending on the pattern.
1251
1252 4. Caseless matching was not working for non-ASCII characters in back
1253 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1254 It now works when Unicode Property Support is available.
1255
1256 5. In pcretest, an escape such as \x{de} in the data was always generating
1257 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1258 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1259 truncation.
1260
1261 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1262
1263 7. Added two (int) casts to pcregrep when printing the difference of two
1264 pointers, in case they are 64-bit values.
1265
1266 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1267 test 2 if it fails.
1268
1269 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1270 and a #define of that name to empty if it is not externally set. This is to
1271 allow users of MSVC to set it if necessary.
1272
1273 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1274 the convenience functions in the pcre_get.c source file.
1275
1276 11. An option change at the start of a pattern that had top-level alternatives
1277 could cause overwriting and/or a crash. This command provoked a crash in
1278 some environments:
1279
1280 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1281
1282 This potential security problem was recorded as CVE-2008-2371.
1283
1284 12. For a pattern where the match had to start at the beginning or immediately
1285 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1286 pcre_dfa_exec() could read past the end of the passed subject if there was
1287 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1288 pcretest so that it places the subject at the end of its malloc-ed buffer.
1289
1290 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1291 exec() might read past the end of the data buffer in UTF-8 mode.
1292
1293 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1294 the data contained the byte 0x85 as part of a UTF-8 character within its
1295 first line. This applied both to normal and DFA matching.
1296
1297 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1298 /^[^d]*?$/8 failed to match "abc".
1299
1300 16. Added a missing copyright notice to pcrecpp_internal.h.
1301
1302 17. Make it more clear in the documentation that values returned from
1303 pcre_exec() in ovector are byte offsets, not character counts.
1304
1305 18. Tidied a few places to stop certain compilers from issuing warnings.
1306
1307 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1308 supplied by Stefan Weber. I made a further small update for 7.8 because
1309 there is a change of source arrangements: the pcre_searchfuncs.c module is
1310 replaced by pcre_ucd.c.
1311
1312
1313 Version 7.7 07-May-08
1314 ---------------------
1315
1316 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1317 a string to a long long, pretend we don't even have a long long." This is
1318 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1319
1320 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1321 pre-7.6 versions, which defined a global no_arg variable instead of putting
1322 it in the RE class. (See also #8 below.)
1323
1324 3. Remove a line of dead code, identified by coverity and reported by Nuno
1325 Lopes.
1326
1327 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1328
1329 (1) The include/exclude patterns were being applied to the whole pathnames
1330 of files, instead of just to the final components.
1331
1332 (2) If there was more than one level of directory, the subdirectories were
1333 skipped unless they satisfied the include/exclude conditions. This is
1334 inconsistent with GNU grep (and could even be seen as contrary to the
1335 pcregrep specification - which I improved to make it absolutely clear).
1336 The action now is always to scan all levels of directory, and just
1337 apply the include/exclude patterns to regular files.
1338
1339 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1340 --exclude_dir in the tests to avoid scanning .svn directories.
1341
1342 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1343 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1344 doesn't support NULs in patterns.
1345
1346 7. Added some missing "const"s to declarations of static tables in
1347 pcre_compile.c and pcre_dfa_exec.c.
1348
1349 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1350 caused by fix #2 above. (Subsequently also a second patch to fix the
1351 first patch. And a third patch - this was a messy problem.)
1352
1353 9. Applied Craig's patch to remove the use of push_back().
1354
1355 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1356 matching function regexec().
1357
1358 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1359 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1360 references. PCRE supports relative numbers with this syntax (I don't think
1361 Oniguruma does).
1362
1363 12. Previously, a group with a zero repeat such as (...){0} was completely
1364 omitted from the compiled regex. However, this means that if the group
1365 was called as a subroutine from elsewhere in the pattern, things went wrong
1366 (an internal error was given). Such groups are now left in the compiled
1367 pattern, with a new opcode that causes them to be skipped at execution
1368 time.
1369
1370 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1371 to the way PCRE behaves:
1372
1373 (a) A lone ] character is dis-allowed (Perl treats it as data).
1374
1375 (b) A back reference to an unmatched subpattern matches an empty string
1376 (Perl fails the current match path).
1377
1378 (c) A data ] in a character class must be notated as \] because if the
1379 first data character in a class is ], it defines an empty class. (In
1380 Perl it is not possible to have an empty class.) The empty class []
1381 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1382 The negative empty class [^] matches any one character, independently
1383 of the DOTALL setting.
1384
1385 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1386 non-existent subpattern following a character class starting with ']' and
1387 containing () gave an internal compiling error instead of "reference to
1388 non-existent subpattern". Fortunately, when the pattern did exist, the
1389 compiled code was correct. (When scanning forwards to check for the
1390 existencd of the subpattern, it was treating the data ']' as terminating
1391 the class, so got the count wrong. When actually compiling, the reference
1392 was subsequently set up correctly.)
1393
1394 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1395 it was being rejected as not supported by pcre_dfa_exec(), even though
1396 other assertions are supported. I have made pcre_dfa_exec() support
1397 (*FAIL).
1398
1399 16. The implementation of 13c above involved the invention of a new opcode,
1400 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1401 cannot be changed at match time, I realized I could make a small
1402 improvement to matching performance by compiling OP_ALLANY instead of
1403 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1404 on the OP_ANY path.
1405
1406 17. Compiling pcretest on Windows with readline support failed without the
1407 following two fixes: (1) Make the unistd.h include conditional on
1408 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1409
1410 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1411 ncurses library to be included for pcretest when ReadLine support is
1412 requested, but also to allow for it to be overridden. This patch came from
1413 Daniel Bergström.
1414
1415 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1416 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1417 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1418 spotting this.
1419
1420
1421 Version 7.6 28-Jan-08
1422 ---------------------
1423
1424 1. A character class containing a very large number of characters with
1425 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1426 overflow.
1427
1428 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1429 HAVE_LONG_LONG is not defined.
1430
1431 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1432 bring them up to date and include new features. This patch includes:
1433
1434 - Fixed PH's badly added libz and libbz2 support.
1435 - Fixed a problem with static linking.
1436 - Added pcredemo. [But later removed - see 7 below.]
1437 - Fixed dftables problem and added an option.
1438 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1439 HAVE_LONG_LONG.
1440 - Added readline support for pcretest.
1441 - Added an listing of the option settings after cmake has run.
1442
1443 4. A user submitted a patch to Makefile that makes it easy to create
1444 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1445 Makefile.am that cause it to include this special target, without
1446 affecting anything else. Note that the same mingw target plus all
1447 the other distribution libraries and programs are now supported
1448 when configuring with CMake (see 6 below) instead of with
1449 Configure/Make.
1450
1451 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1452 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1453 exported in the Windows port". It has not yet been confirmed that the patch
1454 solves the problem, but it does no harm.
1455
1456 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1457 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1458 with CMake, and also correct the comment about stack recursion.
1459
1460 7. Remove the automatic building of pcredemo from the ./configure system and
1461 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1462 of a program that users should build themselves after PCRE is installed, so
1463 building it automatically is not really right. What is more, it gave
1464 trouble in some build environments.
1465
1466 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1467
1468
1469 Version 7.5 10-Jan-08
1470 ---------------------
1471
1472 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1473 values in parens when parsing an RE using the C++ wrapper."
1474
1475 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1476 Characters greater than 255 were excluded from the class instead of being
1477 included.
1478
1479 3. The same bug as (2) above applied to negated POSIX classes such as
1480 [:^space:].
1481
1482 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1483 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1484 I have changed it.
1485
1486 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1487 first named subpattern) and a construct such as (?&a) would reference the
1488 first named subpattern whose name started with "a" (in other words, the
1489 length check was missing). Both these problems are fixed. "Subpattern name
1490 expected" is now given for (?&) (a zero-length name), and this patch also
1491 makes it give the same error for \k'' (previously it complained that that
1492 was a reference to a non-existent subpattern).
1493
1494 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1495 this is right because (?- can be followed by option settings as well as by
1496 digits. I have, however, made the messages clearer.
1497
1498 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1499 than the number used in the conditional) now cause a compile-time error.
1500 This is actually not compatible with Perl, which accepts such patterns, but
1501 treats the conditional as always being FALSE (as PCRE used to), but it
1502 seems to me that giving a diagnostic is better.
1503
1504 8. Change "alphameric" to the more common word "alphanumeric" in comments
1505 and messages.
1506
1507 9. Fix two occurrences of "backslash" in comments that should have been
1508 "backspace".
1509
1510 10. Remove two redundant lines of code that can never be obeyed (their function
1511 was moved elsewhere).
1512
1513 11. The program that makes PCRE's Unicode character property table had a bug
1514 which caused it to generate incorrect table entries for sequences of
1515 characters that have the same character type, but are in different scripts.
1516 It amalgamated them into a single range, with the script of the first of
1517 them. In other words, some characters were in the wrong script. There were
1518 thirteen such cases, affecting characters in the following ranges:
1519
1520 U+002b0 - U+002c1
1521 U+0060c - U+0060d
1522 U+0061e - U+00612
1523 U+0064b - U+0065e
1524 U+0074d - U+0076d
1525 U+01800 - U+01805
1526 U+01d00 - U+01d77
1527 U+01d9b - U+01dbf
1528 U+0200b - U+0200f
1529 U+030fc - U+030fe
1530 U+03260 - U+0327f
1531 U+0fb46 - U+0fbb1
1532 U+10450 - U+1049d
1533
1534 12. The -o option (show only the matching part of a line) for pcregrep was not
1535 compatible with GNU grep in that, if there was more than one match in a
1536 line, it showed only the first of them. It now behaves in the same way as
1537 GNU grep.
1538
1539 13. If the -o and -v options were combined for pcregrep, it printed a blank
1540 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1541 does the same. The return code can be used to tell if there were any
1542 non-matching lines.
1543
1544 14. Added --file-offsets and --line-offsets to pcregrep.
1545
1546 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1547 infinitely looping recursion. The bug was that positive lookaheads were not
1548 being skipped when checking for a possible empty match (negative lookaheads
1549 and both kinds of lookbehind were skipped).
1550
1551 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1552 inclusion of <windows.h> to before rather than after the definition of
1553 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1554
1555 17. Specifying a possessive quantifier with a specific limit for a Unicode
1556 character property caused pcre_compile() to compile bad code, which led at
1557 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1558 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1559 caused the error; without that there was no problem.
1560
1561 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1562
1563 19. Added --enable-pcretest-libreadline.
1564
1565 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1566 RE::GlobalReplace(). As a result, the number of replacements returned was
1567 double what it should be. I removed one of the increments, but Craig sent a
1568 later patch that removed the other one (the right fix) and added unit tests
1569 that check the return values (which was not done before).
1570
1571 21. Several CMake things:
1572
1573 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1574 the names libpcre and libpcreposix, not just pcre and pcreposix.
1575
1576 (2) The above change means that pcretest and pcregrep are now correctly
1577 linked with the newly-built libraries, not previously installed ones.
1578
1579 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1580
1581 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1582 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1583 UTF-8 newline character). The key issue is that the pattern starts .*;
1584 this means that the match must be either at the beginning, or after a
1585 newline. The bug was in the code for advancing after a failed match and
1586 checking that the new position followed a newline. It was not taking
1587 account of UTF-8 characters correctly.
1588
1589 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1590 character classes. PCRE was not treating the sequence [:...:] as a
1591 character class unless the ... were all letters. Perl, however, seems to
1592 allow any characters between [: and :], though of course it rejects as
1593 unknown any "names" that contain non-letters, because all the known class
1594 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1595 for example, whereas PCRE did not - it did not recognize a POSIX character
1596 class. This seemed a bit dangerous, so the code has been changed to be
1597 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1598 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1599 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1600 Perl does, and where it didn't before.
1601
1602 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1603 Windows environments %n is disabled by default.
1604
1605
1606 Version 7.4 21-Sep-07
1607 ---------------------
1608
1609 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1610 means that a class such as [\s] counted as "explicit reference to CR or
1611 LF". That isn't really right - the whole point of the change was to try to
1612 help when there was an actual mention of one of the two characters. So now
1613 the change happens only if \r or \n (or a literal CR or LF) character is
1614 encountered.
1615
1616 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1617 of both had grown to the point where there were only 3 bits left.
1618 Fortunately, there was spare space in the data structure, and so I have
1619 moved the internal flags into a new 16-bit field to free up more option
1620 bits.
1621
1622 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1623 but did not set the internal JCHANGED flag - either of these is enough to
1624 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1625 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1626 start sets both bits.
1627
1628 4. Added options (at build time, compile time, exec time) to change \R from
1629 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1630
1631 5. doc/pcresyntax.html was missing from the distribution.
1632
1633 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1634 compatibility, even though it is no longer used.
1635
1636 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1637 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1638 windows.h file is present (where different names are used). [This was
1639 reversed later after testing - see 16 below.]
1640
1641 8. Changed all #include <config.h> to #include "config.h". There were also
1642 some further <pcre.h> cases that I changed to "pcre.h".
1643
1644 9. When pcregrep was used with the --colour option, it missed the line ending
1645 sequence off the lines that it output.
1646
1647 10. It was pointed out to me that arrays of string pointers cause lots of
1648 relocations when a shared library is dynamically loaded. A technique of
1649 using a single long string with a table of offsets can drastically reduce
1650 these. I have refactored PCRE in four places to do this. The result is
1651 dramatic:
1652
1653 Originally: 290
1654 After changing UCP table: 187
1655 After changing error message table: 43
1656 After changing table of "verbs" 36
1657 After changing table of Posix names 22
1658
1659 Thanks to the folks working on Gregex for glib for this insight.
1660
1661 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1662 unicode-properties was also set.
1663
1664 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1665
1666 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1667 checked only for CRLF.
1668
1669 14. Added casts to pcretest.c to avoid compiler warnings.
1670
1671 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1672
1673 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1674 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1675 entirely. This removes changes made in 7 above.
1676
1677 17. The CMake files have been updated, and there is now more information about
1678 building with CMake in the NON-UNIX-USE document.
1679
1680
1681 Version 7.3 28-Aug-07
1682 ---------------------
1683
1684 1. In the rejigging of the build system that eventually resulted in 7.1, the
1685 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1686 brackets there is not right, since it causes compilers to look for an
1687 installed pcre.h, not the version that is in the source that is being
1688 compiled (which of course may be different). I have changed it back to:
1689
1690 #include "pcre.h"
1691
1692 I have a vague recollection that the change was concerned with compiling in
1693 different directories, but in the new build system, that is taken care of
1694 by the VPATH setting the Makefile.
1695
1696 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1697 when the subject happened to end in the byte 0x85 (e.g. if the last
1698 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1699 characters but of course it shouldn't be taken as a newline when it is part
1700 of another character. The bug was that, for an unlimited repeat of . in
1701 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1702 characters when looking for a newline.
1703
1704 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1705
1706 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1707 in debug output.
1708
1709 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1710 long printing in the pcrecpp unittest when running under MinGW.
1711
1712 6. ESC_K was left out of the EBCDIC table.
1713
1714 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1715 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1716 limit also applies to "virtual nesting" when a pattern is recursive, and in
1717 this case 1000 isn't so big. I have been able to remove this limit at the
1718 expense of backing off one optimization in certain circumstances. Normally,
1719 when pcre_exec() would call its internal match() function recursively and
1720 immediately return the result unconditionally, it uses a "tail recursion"
1721 feature to save stack. However, when a subpattern that can match an empty
1722 string has an unlimited repetition quantifier, it no longer makes this
1723 optimization. That gives it a stack frame in which to save the data for
1724 checking that an empty string has been matched. Previously this was taken
1725 from the 1000-entry workspace that had been reserved. So now there is no
1726 explicit limit, but more stack is used.
1727
1728 8. Applied Daniel's patches to solve problems with the import/export magic
1729 syntax that is required for Windows, and which was going wrong for the
1730 pcreposix and pcrecpp parts of the library. These were overlooked when this
1731 problem was solved for the main library.
1732
1733 9. There were some crude static tests to avoid integer overflow when computing
1734 the size of patterns that contain repeated groups with explicit upper
1735 limits. As the maximum quantifier is 65535, the maximum group length was
1736 set at 30,000 so that the product of these two numbers did not overflow a
1737 32-bit integer. However, it turns out that people want to use groups that
1738 are longer than 30,000 bytes (though not repeat them that many times).
1739 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1740 made it possible to implement the integer overflow checks in a much more
1741 dynamic way, which I have now done. The artificial limitation on group
1742 length has been removed - we now have only the limit on the total length of
1743 the compiled pattern, which depends on the LINK_SIZE setting.
1744
1745 10. Fixed a bug in the documentation for get/copy named substring when
1746 duplicate names are permitted. If none of the named substrings are set, the
1747 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1748 empty string.
1749
1750 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1751 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1752 because the ] is interpreted as the first data character and the
1753 terminating ] is not found. PCRE has been made compatible with Perl in this
1754 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1755 cause memory overwriting.
1756
1757 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1758 string has been matched (to stop an infinite loop). It was not recognizing
1759 a conditional subpattern that could match an empty string if that
1760 subpattern was within another subpattern. For example, it looped when
1761 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1762 condition was not nested. This bug has been fixed.
1763
1764 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1765 past the start of the subject in the presence of bytes with the top bit
1766 set, for example "\x8aBCD".
1767
1768 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1769 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1770
1771 14. Optimized (?!) to (*FAIL).
1772
1773 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1774 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1775 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1776 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1777 does: it's just the validity check that is more restrictive.
1778
1779 16. Inserted checks for integer overflows during escape sequence (backslash)
1780 processing, and also fixed erroneous offset values for syntax errors during
1781 backslash processing.
1782
1783 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1784 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1785
1786 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1787 caused an overrun.
1788
1789 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1790 something other than just ASCII characters) inside a group that had an
1791 unlimited repeat caused a loop at compile time (while checking to see
1792 whether the group could match an empty string).
1793
1794 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1795 [\P{Any}] did so. (Error in the code for printing property names.)
1796
1797 21. An orphan \E inside a character class could cause a crash.
1798
1799 22. A repeated capturing bracket such as (A)? could cause a wild memory
1800 reference during compilation.
1801
1802 23. There are several functions in pcre_compile() that scan along a compiled
1803 expression for various reasons (e.g. to see if it's fixed length for look
1804 behind). There were bugs in these functions when a repeated \p or \P was
1805 present in the pattern. These operators have additional parameters compared
1806 with \d, etc, and these were not being taken into account when moving along
1807 the compiled data. Specifically:
1808
1809 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1810 length.
1811
1812 (b) An item such as \pL+ within a repeated group could cause crashes or
1813 loops.
1814
1815 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1816 "reference to non-existent subpattern" error.
1817
1818 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1819
1820 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1821 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1822
1823 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1824
1825 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1826 character were causing crashes (broken optimization).
1827
1828 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1829 \p or \P) caused a compile-time loop.
1830
1831 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1832 break. For example, the unstudied pattern [\r\n]A does not match the string
1833 "\r\nA" because change 7.0/46 below moves the current point on by two
1834 characters after failing to match at the start. However, the pattern \nA
1835 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1836 the same is true. There doesn't seem any very clean way out of this, but
1837 what I have chosen to do makes the common cases work: PCRE now takes note
1838 of whether there can be an explicit match for \r or \n anywhere in the
1839 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1840 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1841 pattern has explicit CR or LF references.
1842
1843 29. Added (*CR) etc for changing newline setting at start of pattern.
1844
1845
1846 Version 7.2 19-Jun-07
1847 ---------------------
1848
1849 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1850 which is apparently normally available under Windows.
1851
1852 2. Re-jig the pcregrep tests with different newline settings in an attempt
1853 to make them independent of the local environment's newline setting.
1854
1855 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1856
1857 4. Some of the "internals" tests were previously cut out when the link size
1858 was not 2, because the output contained actual offsets. The recent new
1859 "Z" feature of pcretest means that these can be cut out, making the tests
1860 usable with all link sizes.
1861
1862 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1863 stack recursion. This gives a massive performance boost under BSD, but just
1864 a small improvement under Linux. However, it saves one field in the frame
1865 in all cases.
1866
1867 6. Added more features from the forthcoming Perl 5.10:
1868
1869 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1870 recursion call. It refers to the nth most recently opened parentheses.
1871
1872 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1873 to be opened parentheses.
1874
1875 (c) Conditions that refer to capturing parentheses can be specified
1876 relatively, for example, (?(-2)... or (?(+3)...
1877
1878 (d) \K resets the start of the current match so that everything before
1879 is not part of it.
1880
1881 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1882
1883 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1884 reference syntax.
1885
1886 (g) (?| introduces a group in which the numbering of parentheses in each
1887 alternative starts with the same number.
1888
1889 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1890
1891 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1892 PCRE_INFO_JCHANGED.
1893
1894 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1895 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1896 for detecting groups that can match an empty string.
1897
1898 9. A pattern with a very large number of alternatives (more than several
1899 hundred) was running out of internal workspace during the pre-compile
1900 phase, where pcre_compile() figures out how much memory will be needed. A
1901 bit of new cunning has reduced the workspace needed for groups with
1902 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1903 workspace instead of running out of the 4096 that are available.
1904
1905 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1906
1907 11. Applied patch from Google to remove an optimization that didn't quite work.
1908 The report of the bug said:
1909
1910 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1911 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1912 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1913
1914 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1915 it matched the wrong number of bytes.
1916
1917
1918 Version 7.1 24-Apr-07
1919 ---------------------
1920
1921 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1922 that is more "standard", making use of automake and other Autotools. There
1923 is some re-arrangement of the files and adjustment of comments consequent
1924 on this.
1925
1926 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1927 for recursive directory scanning broke on some systems because the files
1928 are not scanned in any specific order and on different systems the order
1929 was different. A call to "sort" has been inserted into RunGrepTest for the
1930 approprate test as a short-term fix. In the longer term there may be an
1931 alternative.
1932
1933 3. I had an email from Eric Raymond about problems translating some of PCRE's
1934 man pages to HTML (despite the fact that I distribute HTML pages, some
1935 people do their own conversions for various reasons). The problems
1936 concerned the use of low-level troff macros .br and .in. I have therefore
1937 removed all such uses from the man pages (some were redundant, some could
1938 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1939 HTML has been updated to handle .nf/.fi and to complain if it encounters
1940 .br or .in.
1941
1942 4. Updated comments in configure.ac that get placed in config.h.in and also
1943 arranged for config.h to be included in the distribution, with the name
1944 config.h.generic, for the benefit of those who have to compile without
1945 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1946
1947 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1948 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1949 makevp.bat for the current PCRE, using the additional files
1950 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1951
1952 6. A Windows user reported a minor discrepancy with test 2, which turned out
1953 to be caused by a trailing space on an input line that had got lost in his
1954 copy. The trailing space was an accident, so I've just removed it.
1955
1956 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1957 that is needed.
1958
1959 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1960 as "const" (a) because they are and (b) because it helps the PHP
1961 maintainers who have recently made a script to detect big data structures
1962 in the php code that should be moved to the .rodata section. I remembered
1963 to update Builducptable as well, so it won't revert if ucptable.h is ever
1964 re-created.
1965
1966 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1967 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1968 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1969 support is not required. This saves 1.5-2K of code, which is important in
1970 some applications.
1971
1972 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1973 so as not to refer to the tables, even though these functions will never be
1974 called when UTF-8 support is disabled. Otherwise there are problems with a
1975 shared library.
1976
1977 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1978
1979 (a) It was defining its arguments as char * instead of void *.
1980
1981 (b) It was assuming that all moves were upwards in memory; this was true
1982 a long time ago when I wrote it, but is no longer the case.
1983
1984 The emulated memove() is provided for those environments that have neither
1985 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1986 is clearly not the case, as these two bugs were recently reported.
1987
1988 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1989 and Detrail to create the HTML documentation, the .txt form of the man
1990 pages, and it removes trailing spaces from listed files. It also creates
1991 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1992 case, it wraps all the #defines with #ifndefs. This script should be run
1993 before "make dist".
1994
1995 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1996 with Unicode property support.
1997
1998 (a) For a maximizing quantifier, if the two different cases of the
1999 character were of different lengths in their UTF-8 codings (there are
2000 some cases like this - I found 11), and the matching function had to
2001 back up over a mixture of the two cases, it incorrectly assumed they
2002 were both the same length.
2003
2004 (b) When PCRE was configured to use the heap rather than the stack for
2005 recursion during matching, it was not correctly preserving the data for
2006 the other case of a UTF-8 character when checking ahead for a match
2007 while processing a minimizing repeat. If the check also involved
2008 matching a wide character, but failed, corruption could cause an
2009 erroneous result when trying to check for a repeat of the original
2010 character.
2011
2012 13. Some tidying changes to the testing mechanism:
2013
2014 (a) The RunTest script now detects the internal link size and whether there
2015 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2016 values substituted by "configure". (The RunGrepTest script already did
2017 this for UTF-8.) The configure.ac script no longer substitutes the
2018 relevant variables.
2019
2020 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2021 with length and offset values. This means that the output is different
2022 for different internal link sizes. Test 2 is skipped for link sizes
2023 other than 2 because of this, bypassing the problem. Unfortunately,
2024 there was also a test in test 3 (the locale tests) that used /B and
2025 failed for link sizes other than 2. Rather than cut the whole test out,
2026 I have added a new /Z option to pcretest that replaces the length and
2027 offset values with spaces. This is now used to make test 3 independent
2028 of link size. (Test 2 will be tidied up later.)
2029
2030 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2031 segmentation fault instead of returning the appropriate error message.
2032
2033 15. In multiline mode when the newline sequence was set to "any", the pattern
2034 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2035 This doesn't seem right; it now treats the CRLF combination as the line
2036 ending, and so does not match in that case. It's only a pattern such as ^$
2037 that would hit this one: something like ^ABC$ would have failed after \r
2038 and then tried again after \r\n.
2039
2040 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2041 in an attempt to make files that differ only in their line terminators
2042 compare equal. This works on Linux.
2043
2044 17. Under certain error circumstances pcregrep might try to free random memory
2045 as it exited. This is now fixed, thanks to valgrind.
2046
2047 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2048 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2049 was because its rules for how to advance for /g after matching an empty
2050 string at the end of a line did not allow for this case. They now check for
2051 it specially.
2052
2053 20. pcretest is supposed to handle patterns and data of any length, by
2054 extending its buffers when necessary. It was getting this wrong when the
2055 buffer for a data line had to be extended.
2056
2057 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2058 CRLF as a newline sequence.
2059
2060 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2061 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2062 I have nevertheless tidied it up.
2063
2064 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2065
2066 24. Added a man page for pcre-config.
2067
2068
2069 Version 7.0 19-Dec-06
2070 ---------------------
2071
2072 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2073 moving to gcc 4.1.1.
2074
2075 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2076 sys/time.h, which is documented as needed for this function. It doesn't
2077 seem to matter on Linux, but it showed up on some releases of OS X.
2078
2079 3. It seems that there are systems where bytes whose values are greater than
2080 127 match isprint() in the "C" locale. The "C" locale should be the
2081 default when a C program starts up. In most systems, only ASCII printing
2082 characters match isprint(). This difference caused the output from pcretest
2083 to vary, making some of the tests fail. I have changed pcretest so that:
2084
2085 (a) When it is outputting text in the compiled version of a pattern, bytes
2086 other than 32-126 are always shown as hex escapes.
2087
2088 (b) When it is outputting text that is a matched part of a subject string,
2089 it does the same, unless a different locale has been set for the match
2090 (using the /L modifier). In this case, it uses isprint() to decide.
2091
2092 4. Fixed a major bug that caused incorrect computation of the amount of memory
2093 required for a compiled pattern when options that changed within the
2094 pattern affected the logic of the preliminary scan that determines the
2095 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2096 that the computed length was too small. The symptoms of this bug were
2097 either the PCRE error "internal error: code overflow" from pcre_compile(),
2098 or a glibc crash with a message such as "pcretest: free(): invalid next
2099 size (fast)". Examples of patterns that provoked this bug (shown in
2100 pcretest format) are:
2101
2102 /(?-x: )/x
2103 /(?x)(?-x: \s*#\s*)/
2104 /((?i)[\x{c0}])/8
2105 /(?i:[\x{c0}])/8
2106
2107 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2108 is now done differently.
2109
2110 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2111 wrapper classes; (b) implement a new function in the C++ scanner that is
2112 more efficient than the old way of doing things because it avoids levels of
2113 recursion in the regex matching; (c) add a paragraph to the documentation
2114 for the FullMatch() function.
2115
2116 6. The escape sequence \n was being treated as whatever was defined as
2117 "newline". Not only was this contrary to the documentation, which states
2118 that \n is character 10 (hex 0A), but it also went horribly wrong when
2119 "newline" was defined as CRLF. This has been fixed.
2120
2121 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2122 was being set to -1 for the "end of line" case (supposedly a value that no
2123 character can have). Though this value is never used (the check for end of
2124 line is "zero bytes in current character"), it caused compiler complaints.
2125 I've changed it to 0xffffffff.
2126
2127 8. In pcre_version.c, the version string was being built by a sequence of
2128 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2129 string (as it is for production releases) called a macro with an empty
2130 argument. The C standard says the result of this is undefined. The gcc
2131 compiler treats it as an empty string (which was what was wanted) but it is
2132 reported that Visual C gives an error. The source has been hacked around to
2133 avoid this problem.
2134
2135 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2136 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2137 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2138 of them did).
2139
2140 10. Originally, pcretest opened its input and output without "b"; then I was
2141 told that "b" was needed in some environments, so it was added for release
2142 5.0 to both the input and output. (It makes no difference on Unix-like
2143 systems.) Later I was told that it is wrong for the input on Windows. I've
2144 now abstracted the modes into two macros, to make it easier to fiddle with
2145 them, and removed "b" from the input mode under Windows.
2146
2147 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2148
2149 12. Added -help and --help to pcretest as an official way of being reminded
2150 of the options.
2151
2152 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2153 and pcrecpp.cc because they annoy compilers at high warning levels.
2154
2155 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2156
2157 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2158 scripts are not C programs :-) and which was not noticed because it works
2159 on Linux.
2160
2161 16. pcretest is supposed to handle any length of pattern and data line (as one
2162 line or as a continued sequence of lines) by extending its input buffer if
2163 necessary. This feature was broken for very long pattern lines, leading to
2164 a string of junk being passed to pcre_compile() if the pattern was longer
2165 than about 50K.
2166
2167 17. I have done a major re-factoring of the way pcre_compile() computes the
2168 amount of memory needed for a compiled pattern. Previously, there was code
2169 that made a preliminary scan of the pattern in order to do this. That was
2170 OK when PCRE was new, but as the facilities have expanded, it has become
2171 harder and harder to keep it in step with the real compile phase, and there
2172 have been a number of bugs (see for example, 4 above). I have now found a
2173 cunning way of running the real compile function in a "fake" mode that
2174 enables it to compute how much memory it would need, while actually only
2175 ever using a few hundred bytes of working memory and without too many
2176 tests of the mode. This should make future maintenance and development
2177 easier. A side effect of this work is that the limit of 200 on the nesting
2178 depth of parentheses has been removed (though this was never a serious
2179 limitation, I suspect). However, there is a downside: pcre_compile() now
2180 runs more slowly than before (30% or more, depending on the pattern). I
2181 hope this isn't a big issue. There is no effect on runtime performance.
2182
2183 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2184 newline (only possible for the last line of a file) and it was a
2185 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2186
2187 19. Added additional timing features to pcretest. (1) The -tm option now times
2188 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2189 separate command line item, by a number that specifies the number of
2190 repeats to use when timing. The default is 50000; this gives better
2191 precision, but takes uncomfortably long for very large patterns.
2192
2193 20. Extended pcre_study() to be more clever in cases where a branch of a
2194 subpattern has no definite first character. For example, (a*|b*)[cd] would
2195 previously give no result from pcre_study(). Now it recognizes that the
2196 first character must be a, b, c, or d.
2197
2198 21. There was an incorrect error "recursive call could loop indefinitely" if
2199 a subpattern (or the entire pattern) that was being tested for matching an
2200 empty string contained only one non-empty item after a nested subpattern.
2201 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2202 incorrectly, because the \d was being skipped in the check.
2203
2204 22. The pcretest program now has a new pattern option /B and a command line
2205 option -b, which is equivalent to adding /B to every pattern. This causes
2206 it to show the compiled bytecode, without the additional information that
2207 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2208 is the same as /B/I).
2209
2210 23. A new optimization is now able automatically to treat some sequences such
2211 as a*b as a*+b. More specifically, if something simple (such as a character
2212 or a simple class like \d) has an unlimited quantifier, and is followed by
2213 something that cannot possibly match the quantified thing, the quantifier
2214 is automatically "possessified".
2215
2216 24. A recursive reference to a subpattern whose number was greater than 39
2217 went wrong under certain circumstances in UTF-8 mode. This bug could also
2218 have affected the operation of pcre_study().
2219
2220 25. Realized that a little bit of performance could be had by replacing
2221 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2222
2223 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2224
2225 27. Possessive quantifiers such as a++ were previously implemented by turning
2226 them into atomic groups such as ($>a+). Now they have their own opcodes,
2227 which improves performance. This includes the automatically created ones
2228 from 23 above.
2229
2230 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2231 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2232 the first matched character to be a colon. This applied both to named and
2233 numbered groups.
2234
2235 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2236
2237 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2238 building PCRE on Windows easier, so I have included it in the distribution.
2239
2240 31. There is now a check in pcretest against a ridiculously large number being
2241 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2242 loop, the loop is abandoned.
2243
2244 32. Forward references to subpatterns in conditions such as (?(2)...) where
2245 subpattern 2 is defined later cause pcre_compile() to search forwards in
2246 the pattern for the relevant set of parentheses. This search went wrong
2247 when there were unescaped parentheses in a character class, parentheses
2248 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2249
2250 33. "Subroutine" calls and backreferences were previously restricted to
2251 referencing subpatterns earlier in the regex. This restriction has now
2252 been removed.
2253
2254 34. Added a number of extra features that are going to be in Perl 5.10. On the
2255 whole, these are just syntactic alternatives for features that PCRE had
2256 previously implemented using the Python syntax or my own invention. The
2257 other formats are all retained for compatibility.
2258
2259 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2260 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2261 also .NET compatible.
2262
2263 (b) A recursion or subroutine call to a named group can now be defined as
2264 (?&name) as well as (?P>name).
2265
2266 (c) A backreference to a named group can now be defined as \k<name> or
2267 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2268 5.10, are also .NET compatible.
2269
2270 (d) A conditional reference to a named group can now use the syntax
2271 (?(<name>) or (?('name') as well as (?(name).
2272
2273 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2274 groups (named and numbered) that are never evaluated inline, but can be
2275 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2276 is always false. There may be only one alternative in such a group.
2277
2278 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2279 as the simple (?(R). The condition is true only if the most recent
2280 recursion is that of the given number or name. It does not search out
2281 through the entire recursion stack.
2282
2283 (g) The escape \gN or \g{N} has been added, where N is a positive or
2284 negative number, specifying an absolute or relative reference.
2285
2286 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2287 some "unreachable code" warnings.
2288
2289 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2290 things, this adds five new scripts.
2291
2292 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2293 There were also incompatibilities regarding the handling of \Q..\E inside
2294 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2295 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2296
2297 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2298 matches an empty string, and forcibly breaks the loop. There were bugs in
2299 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2300 against aaaa the result was just "a" rather than "aaaa", for example. Two
2301 separate and independent bugs (that affected different cases) have been
2302 fixed.
2303
2304 39. Refactored the code to abolish the use of different opcodes for small
2305 capturing bracket numbers. This is a tidy that I avoided doing when I
2306 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2307 The new approach is not only tidier, it makes it possible to reduce the
2308 memory needed to fix the previous bug (38).
2309
2310 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2311 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2312 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2313 mode.
2314
2315 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2316 report.
2317
2318 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2319 copy construction and assignment in the C++ wrapper.
2320
2321 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2322 couple of bugs that could have given wrong results in the "--newline=crlf"
2323 case.
2324
2325 44. Added a number of casts and did some reorganization of signed/unsigned int
2326 variables following suggestions from Dair Grant. Also renamed the variable
2327 "this" as "item" because it is a C++ keyword.
2328
2329 45. Arranged for dftables to add
2330
2331 #include "pcre_internal.h"
2332
2333 to pcre_chartables.c because without it, gcc 4.x may remove the array
2334 definition from the final binary if PCRE is built into a static library and
2335 dead code stripping is activated.
2336
2337 46. For an unanchored pattern, if a match attempt fails at the start of a
2338 newline sequence, and the newline setting is CRLF or ANY, and the next two
2339 characters are CRLF, advance by two characters instead of one.
2340
2341
2342 Version 6.7 04-Jul-06
2343 ---------------------
2344
2345 1. In order to handle tests when input lines are enormously long, pcretest has
2346 been re-factored so that it automatically extends its buffers when
2347 necessary. The code is crude, but this _is_ just a test program. The
2348 default size has been increased from 32K to 50K.
2349
2350 2. The code in pcre_study() was using the value of the re argument before
2351 testing it for NULL. (Of course, in any sensible call of the function, it
2352 won't be NULL.)
2353
2354 3. The memmove() emulation function in pcre_internal.h, which is used on
2355 systems that lack both memmove() and bcopy() - that is, hardly ever -
2356 was missing a "static" storage class specifier.
2357
2358 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2359 containing an extended class (one that cannot be represented by a bitmap
2360 because it contains high-valued characters or Unicode property items, e.g.
2361 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2362 pattern, but PCRE should not loop if you do not (it no longer does).
2363 [Detail: two cases were found: (a) a repeated subpattern containing an
2364 extended class; (b) a recursive reference to a subpattern that followed a
2365 previous extended class. It wasn't skipping over the extended class
2366 correctly when UTF-8 mode was not set.]
2367
2368 5. A negated single-character class was not being recognized as fixed-length
2369 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2370 compile error "lookbehind assertion is not fixed length".
2371
2372 6. The RunPerlTest auxiliary script was showing an unexpected difference
2373 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2374 write a Perl script that can interpret lines of an input file either as
2375 byte characters or as UTF-8, which is what "perltest" was being required to
2376 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2377 can't do is switch easily at run time between having the "use utf8;" pragma
2378 or not. In the end, I fudged it by using the RunPerlTest script to insert
2379 "use utf8;" explicitly for the UTF-8 tests.
2380
2381 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2382 the end of the subject string, contrary to the documentation and to what
2383 Perl does. This was true of both matching functions. Now it matches only at
2384 the start of the subject and immediately after *internal* newlines.
2385
2386 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2387 a pointer to an int instead of a pointer to an unsigned long int. This
2388 caused problems on 64-bit systems.
2389
2390 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2391 instance of the 'standard' template library not being so standard".
2392
2393 10. There was no check on the number of named subpatterns nor the maximum
2394 length of a subpattern name. The product of these values is used to compute
2395 the size of the memory block for a compiled pattern. By supplying a very
2396 long subpattern name and a large number of named subpatterns, the size
2397 computation could be caused to overflow. This is now prevented by limiting
2398 the length of names to 32 characters, and the number of named subpatterns
2399 to 10,000.
2400
2401 11. Subpatterns that are repeated with specific counts have to be replicated in
2402 the compiled pattern. The size of memory for this was computed from the
2403 length of the subpattern and the repeat count. The latter is limited to
2404 65535, but there was no limit on the former, meaning that integer overflow
2405 could in principle occur. The compiled length of a repeated subpattern is
2406 now limited to 30,000 bytes in order to prevent this.
2407
2408 12. Added the optional facility to have named substrings with the same name.
2409
2410 13. Added the ability to use a named substring as a condition, using the
2411 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2412 are numbers (not recommended). Forward references are permitted.
2413
2414 14. Added forward references in named backreferences (if you see what I mean).
2415
2416 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2417 pattern could run off the end of the subject. For example, the pattern
2418 "(?s)(.{1,5})"8 did this with the subject "ab".
2419
2420 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2421 PCRE_CASELESS was set when matching characters that were quantified with ?
2422 or *.
2423
2424 17. A character class other than a single negated character that had a minimum
2425 but no maximum quantifier - for example [ab]{6,} - was not handled
2426 correctly by pce_dfa_exec(). It would match only one character.
2427
2428 18. A valid (though odd) pattern that looked like a POSIX character
2429 class but used an invalid character after [ (for example [[,abc,]]) caused
2430 pcre_compile() to give the error "Failed: internal error: code overflow" or
2431 in some cases to crash with a glibc free() error. This could even happen if
2432 the pattern terminated after [[ but there just happened to be a sequence of
2433 letters, a binary zero, and a closing ] in the memory that followed.
2434
2435 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2436 over the years. Originally (before any Unicode support), just the bottom 8
2437 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2438 output from "man perlunicode" includes this:
2439
2440 The regular expression compiler produces polymorphic opcodes. That
2441 is, the pattern adapts to the data and automatically switches to
2442 the Unicode character scheme when presented with Unicode data--or
2443 instead uses a traditional byte scheme when presented with byte
2444 data.
2445
2446 Sadly, a wide octal escape does not cause a switch, and in a string with
2447 no other multibyte characters, these octal escapes are treated as before.
2448 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2449 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2450 Unicode string.
2451
2452 I have not perpetrated such confusion in PCRE. Up till now, it took just
2453 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2454 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2455 translate to the appropriate multibyte character.
2456
2457 29. Applied some refactoring to reduce the number of warnings from Microsoft
2458 and Borland compilers. This has included removing the fudge introduced
2459 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2460 a warning about an unused variable.
2461
2462 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2463 characters since release 4.0, because Perl (from release 5.004) does not.
2464 [Or at least, is documented not to: some releases seem to be in conflict
2465 with the documentation.] However, when a pattern was studied with
2466 pcre_study() and all its branches started with \s, PCRE still included VT
2467 as a possible starting character. Of course, this did no harm; it just
2468 caused an unnecessary match attempt.
2469
2470 22. Removed a now-redundant internal flag bit that recorded the fact that case
2471 dependency changed within the pattern. This was once needed for "required
2472 byte" processing, but is no longer used. This recovers a now-scarce options
2473 bit. Also moved the least significant internal flag bit to the most-
2474 significant bit of the word, which was not previously used (hangover from
2475 the days when it was an int rather than a uint) to free up another bit for
2476 the future.
2477
2478 23. Added support for CRLF line endings as well as CR and LF. As well as the
2479 default being selectable at build time, it can now be changed at runtime
2480 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2481 specify that it is scanning data with non-default line endings.
2482
2483 24. Changed the definition of CXXLINK to make it agree with the definition of
2484 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2485
2486 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2487 recursions. This makes a big different to stack usage for some patterns.
2488
2489 26. If a subpattern containing a named recursion or subroutine reference such
2490 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2491 the space required for the compiled pattern went wrong and gave too small a
2492 value. Depending on the environment, this could lead to "Failed: internal
2493 error: code overflow at offset 49" or "glibc detected double free or
2494 corruption" errors.
2495
2496 27. Applied patches from Google (a) to support the new newline modes and (b) to
2497 advance over multibyte UTF-8 characters in GlobalReplace.
2498
2499 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2500 difference for some implementation of PCRE in some Windows version.
2501
2502 29. Added some extra testing facilities to pcretest:
2503
2504 \q<number> in a data line sets the "match limit" value
2505 \Q<number> in a data line sets the "match recursion limt" value
2506 -S <number> sets the stack size, where <number> is in megabytes
2507
2508 The -S option isn't available for Windows.
2509
2510
2511 Version 6.6 06-Feb-06
2512 ---------------------
2513
2514 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2515 in pcreposix.h. I have copied the definition from pcre.h.
2516
2517 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2518 because pcre.h is no longer a built file.
2519
2520 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2521 not normally included in the compiled code.
2522
2523
2524 Version 6.5 01-Feb-06
2525 ---------------------
2526
2527 1. When using the partial match feature with pcre_dfa_exec(), it was not
2528 anchoring the second and subsequent partial matches at the new starting
2529 point. This could lead to incorrect results. For example, with the pattern
2530 /1234/, partially matching against "123" and then "a4" gave a match.
2531
2532 2. Changes to pcregrep:
2533
2534 (a) All non-match returns from pcre_exec() were being treated as failures
2535 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2536 error message is output. Some extra information is given for the
2537 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2538 probably the only errors that are likely to be caused by users (by
2539 specifying a regex that has nested indefinite repeats, for instance).
2540 If there are more than 20 of these errors, pcregrep is abandoned.
2541
2542 (b) A binary zero was treated as data while matching, but terminated the
2543 output line if it was written out. This has been fixed: binary zeroes
2544 are now no different to any other data bytes.
2545
2546 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2547 used to set a locale for matching. The --locale=xxxx long option has
2548 been added (no short equivalent) to specify a locale explicitly on the
2549 pcregrep command, overriding the environment variables.
2550
2551 (d) When -B was used with -n, some line numbers in the output were one less
2552 than they should have been.
2553
2554 (e) Added the -o (--only-matching) option.
2555
2556 (f) If -A or -C was used with -c (count only), some lines of context were
2557 accidentally printed for the final match.
2558
2559 (g) Added the -H (--with-filename) option.
2560
2561 (h) The combination of options -rh failed to suppress file names for files
2562 that were found from directory arguments.
2563
2564 (i) Added the -D (--devices) and -d (--directories) options.
2565
2566 (j) Added the -F (--fixed-strings) option.
2567
2568 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2569
2570 (l) Added the --colo(u)r option.
2571
2572 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2573 is not present by default.
2574
2575 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2576 items such as (?R) or (?1), when the recursion could match a number of
2577 alternatives. If it matched one of the alternatives, but subsequently,
2578 outside the recursion, there was a failure, the code tried to back up into
2579 the recursion. However, because of the way PCRE is implemented, this is not
2580 possible, and the result was an incorrect result from the match.
2581
2582 In order to prevent this happening, the specification of recursion has
2583 been changed so that all such subpatterns are automatically treated as
2584 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2585
2586 4. I had overlooked the fact that, in some locales, there are characters for
2587 which isalpha() is true but neither isupper() nor islower() are true. In
2588 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2589 and ordfeminine) are like this. This affected the treatment of \w and \W
2590 when they appeared in character classes, but not when they appeared outside
2591 a character class. The bit map for "word" characters is now created
2592 separately from the results of isalnum() instead of just taking it from the
2593 upper, lower, and digit maps. (Plus the underscore character, of course.)
2594
2595 5. The above bug also affected the handling of POSIX character classes such as
2596 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2597 permanent tables. Instead, the bit maps for such a class were previously
2598 created as the appropriate unions of the upper, lower, and digit bitmaps.
2599 Now they are created by subtraction from the [[:word:]] class, which has
2600 its own bitmap.
2601
2602 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2603 It is created by subtracting the vertical space characters (\x09, \x0a,
2604 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2605 subtraction was done in the overall bitmap for a character class, meaning
2606 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2607 be recognized. This bug has been fixed.
2608
2609 7. Patches from the folks at Google:
2610
2611 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2612 real life, but is still worth protecting against".
2613
2614 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2615 regular expressions".
2616
2617 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2618 have it.
2619
2620 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2621 "configure" and the latter not, in order to fix a problem somebody had
2622 with compiling the Arg class on HP-UX.
2623
2624 (e) Improve the error-handling of the C++ wrapper a little bit.
2625
2626 (f) New tests for checking recursion limiting.
2627
2628 8. The pcre_memmove() function, which is used only if the environment does not
2629 have a standard memmove() function (and is therefore rarely compiled),
2630 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2631 returning a result (though PCRE never actually uses the result).
2632
2633 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2634 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2635 returned instead of calling malloc() with an overflowing number that would
2636 most likely cause subsequent chaos.
2637
2638 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2639
2640 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2641 with this option is matched, the nmatch and pmatch options of regexec() are
2642 ignored.
2643
2644 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2645 provided in case anyone wants to the the POSIX interface with UTF-8
2646 strings.
2647
2648 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2649 C++ linking (needed for some HP-UX environments).
2650
2651 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2652 (unused parameter) and in the pcre_printint() function (omitted "default"
2653 switch label when the default is to do nothing).
2654
2655 15. Added some code to make it possible, when PCRE is compiled as a C++
2656 library, to replace subject pointers for pcre_exec() with a smart pointer
2657 class, thus making it possible to process discontinuous strings.
2658
2659 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2660 much the same function. They were added by different people who were trying
2661 to make PCRE easy to compile on non-Unix systems. It has been suggested
2662 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2663 for compiling on Windows systems. I have therefore replaced it with
2664 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2665 defaults to "extern" for C or "extern C" for C++, which works fine on
2666 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2667 SCOPE with something explicit in config.h. In addition:
2668
2669 (a) pcreposix.h still had just "extern" instead of either of these macros;
2670 I have replaced it with PCRE_DATA_SCOPE.
2671
2672 (b) Functions such as _pcre_xclass(), which are internal to the library,
2673 but external in the C sense, all had PCRE_EXPORT in their definitions.
2674 This is apparently wrong for the Windows case, so I have removed it.
2675 (It makes no difference on Unix-like systems.)
2676
2677 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2678 of recursive calls to match(). This is different to MATCH_LIMIT because
2679 that limits the total number of calls to match(), not all of which increase
2680 the depth of recursion. Limiting the recursion depth limits the amount of
2681 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2682 when PCRE is compiled, and changed at run time. A patch from Google adds
2683 this functionality to the C++ interface.
2684
2685 18. Changes to the handling of Unicode character properties:
2686
2687 (a) Updated the table to Unicode 4.1.0.
2688
2689 (b) Recognize characters that are not in the table as "Cn" (undefined).
2690
2691 (c) I revised the way the table is implemented to a much improved format
2692 which includes recognition of ranges. It now supports the ranges that
2693 are defined in UnicodeData.txt, and it also amalgamates other
2694 characters into ranges. This has reduced the number of entries in the
2695 table from around 16,000 to around 3,000, thus reducing its size
2696 considerably. I realized I did not need to use a tree structure after
2697 all - a binary chop search is just as efficient. Having reduced the
2698 number of entries, I extended their size from 6 bytes to 8 bytes to
2699 allow for more data.
2700
2701 (d) Added support for Unicode script names via properties such as \p{Han}.
2702
2703 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2704 matching that character.
2705
2706 20. When matching a repeated Unicode property with a minimum greater than zero,
2707 (for example \pL{2,}), PCRE could look past the end of the subject if it
2708 reached it while seeking the minimum number of characters. This could
2709 happen only if some of the characters were more than one byte long, because
2710 there is a check for at least the minimum number of bytes.
2711
2712 21. Refactored the implementation of \p and \P so as to be more general, to
2713 allow for more different types of property in future. This has changed the
2714 compiled form incompatibly. Anybody with saved compiled patterns that use
2715 \p or \P will have to recompile them.
2716
2717 22. Added "Any" and "L&" to the supported property types.
2718
2719 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2720 but give a compile time error if the value is greater than 0xff.
2721
2722 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2723 accidentally not being installed or uninstalled.
2724
2725 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2726 made were to insert the current release number. This seemed silly, because
2727 it made things harder for people building PCRE on systems that don't run
2728 "configure". I have turned pcre.h into a distributed file, no longer built
2729 by "configure", with the version identification directly included. There is
2730 no longer a pcre.h.in file.
2731
2732 However, this change necessitated a change to the pcre-config script as
2733 well. It is built from pcre-config.in, and one of the substitutions was the
2734 release number. I have updated configure.ac so that ./configure now finds
2735 the release number by grepping pcre.h.
2736
2737 26. Added the ability to run the tests under valgrind.
2738
2739
2740 Version 6.4 05-Sep-05
2741 ---------------------
2742
2743 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2744 "--" to be printed when multiple files were scanned, even when none of the
2745 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2746 consider it to be a bug, and have restored the previous behaviour.
2747
2748 2. A couple of code tidies to get rid of compiler warnings.
2749
2750 3. The pcretest program used to cheat by referring to symbols in the library
2751 whose names begin with _pcre_. These are internal symbols that are not
2752 really supposed to be visible externally, and in some environments it is
2753 possible to suppress them. The cheating is now confined to including
2754 certain files from the library's source, which is a bit cleaner.
2755
2756 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2757 file's purpose clearer.
2758
2759 5. Reorganized pcre_ucp_findchar().
2760
2761
2762 Version 6.3 15-Aug-05
2763 ---------------------
2764
2765 1. The file libpcre.pc.in did not have general read permission in the tarball.
2766
2767 2. There were some problems when building without C++ support:
2768
2769 (a) If C++ support was not built, "make install" and "make test" still
2770 tried to test it.
2771
2772 (b) There were problems when the value of CXX was explicitly set. Some
2773 changes have been made to try to fix these, and ...
2774
2775 (c) --disable-cpp can now be used to explicitly disable C++ support.
2776
2777 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2778 backslash in a target when C++ was disabled. This confuses some
2779 versions of "make", apparently. Using an intermediate variable solves
2780 this. (Same for CPP_LOBJ.)
2781
2782 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2783 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2784 necessary on certain architectures.
2785
2786 4. Added a setting of -export-symbols-regex to the link command to remove
2787 those symbols that are exported in the C sense, but actually are local
2788 within the library, and not documented. Their names all begin with
2789 "_pcre_". This is not a perfect job, because (a) we have to except some
2790 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2791 available (and never for static libraries). I have made a note to try to
2792 find a way round (a) in the future.
2793
2794
2795 Version 6.2 01-Aug-05
2796 ---------------------
2797
2798 1. There was no test for integer overflow of quantifier values. A construction
2799 such as {1111111111111111} would give undefined results. What is worse, if
2800 a minimum quantifier for a parenthesized subpattern overflowed and became
2801 negative, the calculation of the memory size went wrong. This could have
2802 led to memory overwriting.
2803
2804 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2805
2806 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2807 operating environments where this matters.
2808
2809 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2810 PCRE options from within the C++ wrapper.
2811
2812 5. Named capturing subpatterns were not being correctly counted when a pattern
2813 was compiled. This caused two problems: (a) If there were more than 100
2814 such subpatterns, the calculation of the memory needed for the whole
2815 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2816 back references of the form \12, where the number was greater than 9, were
2817 not recognized as back references, even though there were sufficient
2818 previous subpatterns.
2819
2820 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2821 versions of gcc, e.g. 2.95.4.
2822
2823
2824 Version 6.1 21-Jun-05
2825 ---------------------
2826
2827 1. There was one reference to the variable "posix" in pcretest.c that was not
2828 surrounded by "#if !defined NOPOSIX".
2829
2830 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2831 the cross-check on the old pcre_info() function, for the benefit of the
2832 cut-down version of PCRE that is currently imported into Exim.
2833
2834 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2835 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2836 patterns that start (?i) but I don't think that matters. The use of (?i) is
2837 just an example; this all applies to the other options as well.
2838
2839 4. Since libtool seems to echo the compile commands it is issuing, the output
2840 from "make" can be reduced a bit by putting "@" in front of each libtool
2841 compile command.
2842
2843 5. Patch from the folks at Google for configure.in to be a bit more thorough
2844 in checking for a suitable C++ installation before trying to compile the
2845 C++ stuff. This should fix a reported problem when a compiler was present,
2846 but no suitable headers.
2847
2848 6. The man pages all had just "PCRE" as their title. I have changed them to
2849 be the relevant file name. I have also arranged that these names are
2850 retained in the file doc/pcre.txt, which is a concatenation in text format
2851 of all the man pages except the little individual ones for each function.
2852
2853 7. The NON-UNIX-USE file had not been updated for the different set of source
2854 files that come with release 6. I also added a few comments about the C++
2855 wrapper.
2856
2857
2858 Version 6.0 07-Jun-05
2859 ---------------------
2860
2861 1. Some minor internal re-organization to help with my DFA experiments.
2862
2863 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2864 didn't matter for the library itself when fully configured, but did matter
2865 when compiling without UCP support, or within Exim, where the ucp files are
2866 not imported.
2867
2868 3. Refactoring of the library code to split up the various functions into
2869 different source modules. The addition of the new DFA matching code (see
2870 below) to a single monolithic source would have made it really too
2871 unwieldy, quite apart from causing all the code to be include in a
2872 statically linked application, when only some functions are used. This is
2873 relevant even without the DFA addition now that patterns can be compiled in
2874 one application and matched in another.
2875
2876 The downside of splitting up is that there have to be some external
2877 functions and data tables that are used internally in different modules of
2878 the library but which are not part of the API. These have all had their
2879 names changed to start with "_pcre_" so that they are unlikely to clash
2880 with other external names.
2881
2882 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2883 a different (DFA) algorithm. Although it is slower than the original
2884 function, it does have some advantages for certain types of matching
2885 problem.
2886
2887 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2888 including restarting after a partial match.
2889
2890 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2891 defined when compiling for Windows was sent to me. I have put it into the
2892 code, though I have no means of testing or verifying it.
2893
2894 7. Added the pcre_refcount() auxiliary function.
2895
2896 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2897 match before or at the first newline in the subject string. In pcretest,
2898 the /f option on a pattern can be used to set this.
2899
2900 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2901 would behave wrongly. This has been present in PCRE since release 4.0.
2902
2903 10. A number of changes to the pcregrep command:
2904
2905 (a) Refactored how -x works; insert ^(...)$ instead of setting
2906 PCRE_ANCHORED and checking the length, in preparation for adding
2907 something similar for -w.
2908
2909 (b) Added the -w (match as a word) option.
2910
2911 (c) Refactored the way lines are read and buffered so as to have more
2912 than one at a time available.
2913
2914 (d) Implemented a pcregrep test script.
2915
2916 (e) Added the -M (multiline match) option. This allows patterns to match
2917 over several lines of the subject. The buffering ensures that at least
2918 8K, or the rest of the document (whichever is the shorter) is available
2919 for matching (and similarly the previous 8K for lookbehind assertions).
2920
2921 (f) Changed the --help output so that it now says
2922
2923 -w, --word-regex(p)
2924
2925 instead of two lines, one with "regex" and the other with "regexp"
2926 because that confused at least one person since the short forms are the
2927 same. (This required a bit of code, as the output is generated
2928 automatically from a table. It wasn't just a text change.)
2929
2930 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2931 option but starts with a hyphen. Could be a pattern or a path name
2932 starting with a hyphen, for instance.
2933
2934 (h) "-" can be given as a file name to represent stdin.
2935
2936 (i) When file names are being printed, "(standard input)" is used for
2937 the standard input, for compatibility with GNU grep. Previously
2938 "<stdin>" was used.
2939
2940 (j) The option --label=xxx can be used to supply a name to be used for
2941 stdin when file names are being printed. There is no short form.
2942
2943 (k) Re-factored the options decoding logic because we are going to add
2944 two more options that take data. Such options can now be given in four
2945 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2946
2947 (l) Added the -A, -B, and -C options for requesting that lines of context
2948 around matches be printed.
2949
2950 (m) Added the -L option to print the names of files that do not contain
2951 any matching lines, that is, the complement of -l.
2952
2953 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2954 continue to scan other files.
2955
2956 (o) The -s option was incorrectly implemented. For compatibility with other
2957 greps, it now suppresses the error message for a non-existent or non-
2958 accessible file (but not the return code). There is a new option called
2959 -q that suppresses the output of matching lines, which was what -s was
2960 previously doing.
2961
2962 (p) Added --include and --exclude options to specify files for inclusion
2963 and exclusion when recursing.
2964
2965 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2966 Hopefully, it now does.
2967
2968 12. Missing cast in pcre_study().
2969
2970 13. Added an "uninstall" target to the makefile.
2971
2972 14. Replaced "extern" in the function prototypes in Makefile.in with
2973 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2974 world, but is set differently for Windows.
2975
2976 15. Added a second compiling function called pcre_compile2(). The only
2977 difference is that it has an extra argument, which is a pointer to an
2978 integer error code. When there is a compile-time failure, this is set
2979 non-zero, in addition to the error test pointer being set to point to an
2980 error message. The new argument may be NULL if no error number is required
2981 (but then you may as well call pcre_compile(), which is now just a
2982 wrapper). This facility is provided because some applications need a
2983 numeric error indication, but it has also enabled me to tidy up the way
2984 compile-time errors are handled in the POSIX wrapper.
2985
2986 16. Added VPATH=.libs to the makefile; this should help when building with one
2987 prefix path and installing with another. (Or so I'm told by someone who
2988 knows more about this stuff than I do.)
2989
2990 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2991 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2992 match everything, including newlines. This is not POSIX-compatible, but
2993 somebody wanted the feature. From pcretest it can be activated by using
2994 both the P and the s flags.
2995
2996 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2997
2998 19. libpcre.pc was being incorrectly installed as executable.
2999
3000 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3001 it now also looks for '\r' so that it will work unmodified on Windows.
3002
3003 21. Added Google's contributed C++ wrapper to the distribution.
3004
3005 22. Added some untidy missing memory free() calls in pcretest, to keep
3006 Electric Fence happy when testing.
3007
3008
3009
3010 Version 5.0 13-Sep-04
3011 ---------------------
3012
3013 1. Internal change: literal characters are no longer packed up into items
3014 containing multiple characters in a single byte-string. Each character
3015 is now matched using a separate opcode. However, there may be more than one
3016 byte in the character in UTF-8 mode.
3017
3018 2. The pcre_callout_block structure has two new fields: pattern_position and
3019 next_item_length. These contain the offset in the pattern to the next match
3020 item, and its length, respectively.
3021
3022 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3023 insertion of callouts before each pattern item. Added the /C option to
3024 pcretest to make use of this.
3025
3026 4. On the advice of a Windows user, the lines
3027
3028 #if defined(_WIN32) || defined(WIN32)
3029 _setmode( _fileno( stdout ), 0x8000 );
3030 #endif /* defined(_WIN32) || defined(WIN32) */
3031
3032 have been added to the source of pcretest. This apparently does useful
3033 magic in relation to line terminators.
3034
3035 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3036 for the benefit of those environments where the "b" makes a difference.
3037
3038 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3039 to know about it. I have put a hack into configure.in that adds in code
3040 to set GCC=yes if CC=icc. This seems to end up at a point in the
3041 generated configure script that is early enough to affect the setting of
3042 compiler options, which is what is needed, but I have no means of testing
3043 whether it really works. (The user who reported this had patched the
3044 generated configure script, which of course I cannot do.)
3045
3046 LATER: After change 22 below (new libtool files), the configure script
3047 seems to know about icc (and also ecc). Therefore, I have commented out
3048 this hack in configure.in.
3049
3050 7. Added support for pkg-config (2 patches were sent in).
3051
3052 8. Negated POSIX character classes that used a combination of internal tables
3053 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3054 [[:^ascii]]. Typically, they would match almost any characters. The other
3055 POSIX classes were not broken in this way.
3056
3057 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3058 to find the match, as PCRE was deluded into thinking that the match had to
3059 start at the start point or following a newline. The same bug applied to
3060 patterns with negative forward assertions or any backward assertions
3061 preceding ".*" at the start, unless the pattern required a fixed first
3062 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3063
3064 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3065 starting at the last subject character, bytes beyond the end of the subject
3066 string were read.
3067
3068 11. Renamed the variable "class" as "classbits" to make life easier for C++
3069 users. (Previously there was a macro definition, but it apparently wasn't
3070 enough.)
3071
3072 12. Added the new field "tables" to the extra data so that tables can be passed
3073 in at exec time, or the internal tables can be re-selected. This allows
3074 a compiled regex to be saved and re-used at a later time by a different
3075 program that might have everything at different addresses.
3076
3077 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3078 -R library as well as a -L library.
3079
3080 14. The debugging options of pcretest (-d on the command line or D on a
3081 pattern) showed incorrect output for anything following an extended class
3082 that contained multibyte characters and which was followed by a quantifier.
3083
3084 15. Added optional support for general category Unicode character properties
3085 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3086 support. It adds about 90K to the size of the library. The meanings of the
3087 inbuilt class escapes such as \d and \s have NOT been changed.
3088
3089 16. Updated pcredemo.c to include calls to free() to release the memory for the
3090 compiled pattern.
3091
3092 17. The generated file chartables.c was being created in the source directory
3093 instead of in the building directory. This caused the build to fail if the
3094 source directory was different from the building directory, and was
3095 read-only.
3096
3097 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3098 file. No doubt somebody will tell me if they don't make sense... Also added
3099 Dan Mooney's comments about building on OpenVMS.
3100
3101 19. Added support for partial matching via the PCRE_PARTIAL option for
3102 pcre_exec() and the \P data escape in pcretest.
3103
3104 20. Extended pcretest with 3 new pattern features:
3105
3106 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3107 write the compiled pattern to the file whose name is "rest-of-line".
3108 This is a straight binary dump of the data, with the saved pointer to
3109 the character tables forced to be NULL. The study data, if any, is
3110 written too. After writing, pcretest reads a new pattern.
3111
3112 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3113 compiled pattern from the given file. There must not be any
3114 occurrences of "<" in the file name (pretty unlikely); if there are,
3115 pcretest will instead treat the initial "<" as a pattern delimiter.
3116 After reading in the pattern, pcretest goes on to read data lines as
3117 usual.
3118
3119 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3120 and 16-bit fields in a compiled pattern, to simulate a pattern that
3121 was compiled on a host of opposite endianness.
3122
3123 21. The pcre-exec() function can now cope with patterns that were compiled on
3124 hosts of opposite endianness, with this restriction:
3125
3126 As for any compiled expression that is saved and used later, the tables
3127 pointer field cannot be preserved; the extra_data field in the arguments
3128 to pcre_exec() should be used to pass in a tables address if a value
3129 other than the default internal tables were used at compile time.
3130
3131 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3132 now diagnosed as an error. Previously, most of the time, a negative number
3133 would have been treated as zero, but if in addition "ovector" was passed as
3134 NULL, a crash could occur.
3135
3136 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3137 new versions from the libtool 1.5 distribution (the last one is a copy of
3138 a file called libtool.m4). This seems to have fixed the need to patch
3139 "configure" to support Darwin 1.3 (which I used to do). However, I still
3140 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3141 workstation).
3142
3143 24. Changed the PCRE licence to be the more standard "BSD" licence.
3144
3145
3146 Version 4.5 01-Dec-03
3147 ---------------------
3148
3149 1. There has been some re-arrangement of the code for the match() function so
3150 that it can be compiled in a version that does not call itself recursively.
3151 Instead, it keeps those local variables that need separate instances for
3152 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3153 needs to "recurse". Keeping track of where control must go is done by means
3154 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3155 hide most of the details from the main code, and operates only if
3156 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3157 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3158 operating.
3159
3160 To make it easier for callers to provide specially tailored get/free
3161 functions for this usage, two new functions, pcre_stack_malloc, and
3162 pcre_stack_free, are used. They are always called in strict stacking order,
3163 and the size of block requested is always the same.
3164
3165 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3166 PCRE has been compiled to use the stack or the heap for recursion. The
3167 -C option of pcretest uses this to show which version is compiled.
3168
3169 A new data escape \S, is added to pcretest; it causes the amounts of store
3170 obtained and freed by both kinds of malloc/free at match time to be added
3171 to the output.
3172
3173 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3174 what's available on my current Linux desktop machine.
3175
3176 3. When matching a UTF-8 string, the test for a valid string at the start has
3177 been extended. If start_offset is not zero, PCRE now checks that it points
3178 to a byte that is the start of a UTF-8 character. If not, it returns
3179 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3180 this is necessary because there may be backward assertions in the pattern.
3181 When matching the same subject several times, it may save resources to use
3182 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3183
3184 4. The code for checking the validity of UTF-8 strings has been tightened so
3185 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3186 containing "overlong sequences".
3187
3188 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3189 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3190 should have been "&", but it just so happened that all the cases this let
3191 through by mistake were picked up later in the function.
3192
3193 6. I had used a variable called "isblank" - this is a C99 function, causing
3194 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3195
3196 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3197 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3198 so the version is shown; (c) stop "make test" echoing "./RunTest".
3199
3200 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3201
3202 9. The prototype for memmove() for systems that don't have it was using
3203 size_t, but the inclusion of the header that defines size_t was later. I've
3204 moved the #includes for the C headers earlier to avoid this.
3205
3206 10. Added some adjustments to the code to make it easier to compiler on certain
3207 special systems:
3208
3209 (a) Some "const" qualifiers were missing.
3210 (b) Added the macro EXPORT before all exported functions; by default this
3211 is defined to be empty.
3212 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3213 that it reads its output file name as an argument instead of writing
3214 to the standard output and assuming this can be redirected.
3215
3216 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3217 class containing characters with values greater than 255, PCRE compilation
3218 went into a loop.
3219
3220 12. A recursive reference to a subpattern that was within another subpattern
3221 that had a minimum quantifier of zero caused PCRE to crash. For example,
3222 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3223 recursion. If the recursively-called subpattern itself had a zero repeat,
3224 that was OK.
3225
3226 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3227 buffer into which it was copied (for escape processing) was still set at
3228 1024, so long lines caused crashes.
3229
3230 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3231 "internal error: code overflow...". This applied to any character class
3232 that was followed by a possessive quantifier.
3233
3234 15. Modified the Makefile to add libpcre.la as a prerequisite for
3235 libpcreposix.la because I was told this is needed for a parallel build to
3236 work.
3237
3238 16. If a pattern that contained .* following optional items at the start was
3239 studied, the wrong optimizing data was generated, leading to matching
3240 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3241 matching string must start with a or b or c. The correct conclusion for
3242 this pattern is that a match can start with any character.
3243
3244
3245 Version 4.4 13-Aug-03
3246 ---------------------
3247
3248 1. In UTF-8 mode, a character class containing characters with values between
3249 127 and 255 was not handled correctly if the compiled pattern was studied.
3250 In fixing this, I have also improved the studying algorithm for such
3251 classes (slightly).
3252
3253 2. Three internal functions had redundant arguments passed to them. Removal
3254 might give a very teeny performance improvement.
3255
3256 3. Documentation bug: the value of the capture_top field in a callout is *one
3257 more than* the number of the hightest numbered captured substring.
3258
3259 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3260 in incorrectly linking with a previously installed version. They now link
3261 explicitly with libpcre.la.
3262
3263 5. configure.in no longer needs to recognize Cygwin specially.
3264
3265 6. A problem in pcre.in for Windows platforms is fixed.
3266
3267 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3268 pcretest, it used to include the size of the study block as part of its
3269 output. Unfortunately, the structure contains a field that has a different
3270 size on different hardware architectures. This meant that the tests that
3271 showed this size failed. As the block is currently always of a fixed size,
3272 this information isn't actually particularly useful in pcretest output, so
3273 I have just removed it.
3274
3275 8. Three pre-processor statements accidentally did not start in column 1.
3276 Sadly, there are *still* compilers around that complain, even though
3277 standard C has not required this for well over a decade. Sigh.
3278
3279 9. In pcretest, the code for checking callouts passed small integers in the
3280 callout_data field, which is a void * field. However, some picky compilers
3281 complained about the casts involved for this on 64-bit systems. Now
3282 pcretest passes the address of the small integer instead, which should get
3283 rid of the warnings.
3284
3285 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3286 both compile and run time, and gives an error if an invalid UTF-8 sequence
3287 is found. There is a option for disabling this check in cases where the
3288 string is known to be correct and/or the maximum performance is wanted.
3289
3290 11. In response to a bug report, I changed one line in Makefile.in from
3291
3292 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3293 to
3294 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3295
3296 to look similar to other lines, but I have no way of telling whether this
3297 is the right thing to do, as I do not use Windows. No doubt I'll get told
3298 if it's wrong...
3299
3300
3301 Version 4.3 21-May-03
3302 ---------------------
3303
3304 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3305 Makefile.
3306
3307 2. Some refactoring to improve the quality of the code:
3308
3309 (i) The utf8_table... variables are now declared "const".
3310
3311 (ii) The code for \cx, which used the "case flipping" table to upper case
3312 lower case letters, now just substracts 32. This is ASCII-specific,
3313 but the whole concept of \cx is ASCII-specific, so it seems
3314 reasonable.
3315
3316 (iii) PCRE was using its character types table to recognize decimal and
3317 hexadecimal digits in the pattern. This is silly, because it handles
3318 only 0-9, a-f, and A-F, but the character types table is locale-
3319 specific, which means strange things might happen. A private
3320 table is now used for this - though it costs 256 bytes, a table is
3321 much faster than multiple explicit tests. Of course, the standard
3322 character types table is still used for matching digits in subject
3323 strings against \d.
3324
3325 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3326 ending in _t are). So I've renamed it as ESC_tee.
3327
3328 3. The first argument for regexec() in the POSIX wrapper should have been
3329 defined as "const".
3330
3331 4. Changed pcretest to use malloc() for its buffers so that they can be
3332 Electric Fenced for debugging.
3333
3334 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3335 to read one or more bytes before the start of the subject string. Often this
3336 had no effect on PCRE's behaviour, but in some circumstances it could
3337 provoke a segmentation fault.
3338
3339 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3340 to try to read one or more bytes before the start of the subject string.
3341
3342 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3343 UTF-8 support could misbehave in various ways if the subject string
3344 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3345 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3346 back over UTF-8 characters.)
3347
3348
3349 Version 4.2 14-Apr-03
3350 ---------------------
3351
3352 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3353
3354 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3355 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3356 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3357 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3358 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3359 and BUILD_EXEEXT
3360 Note: automatic setting of the BUILD variables is not yet working
3361 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3362 compile-time but not at link-time
3363 [LINK]: use for linking executables only
3364 make different versions for Windows and non-Windows
3365 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3366 libraries
3367 [LINK_FOR_BUILD]: new variable
3368 [OBJEXT]: use throughout
3369 [EXEEXT]: use throughout
3370 <winshared>: new target
3371 <wininstall>: new target
3372 <dftables.o>: use native compiler
3373 <dftables>: use native linker
3374 <install>: handle Windows platform correctly
3375 <clean>: ditto
3376 <check>: ditto
3377 copy DLL to top builddir before testing
3378
3379 As part of these changes, -no-undefined was removed again. This was reported
3380 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3381 in any case.
3382
3383 3. Some tidies to get rid of compiler warnings:
3384
3385 . In the match_data structure, match_limit was an unsigned long int, whereas
3386 match_call_count was an int. I've made them both unsigned long ints.
3387
3388 . In pcretest the fact that a const uschar * doesn't automatically cast to
3389 a void * provoked a warning.
3390
3391 . Turning on some more compiler warnings threw up some "shadow" variables
3392 and a few more missing casts.
3393
3394 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3395 option, a class that contained a single character with a value between 128
3396 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3397
3398 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3399 option, a class that contained several characters, but with at least one
3400 whose value was between 128 and 255 caused PCRE to crash.
3401
3402
3403 Version 4.1 12-Mar-03
3404 ---------------------
3405
3406 1. Compiling with gcc -pedantic found a couple of places where casts were
3407 needed, and a string in dftables.c that was longer than standard compilers are
3408 required to support.
3409
3410 2. Compiling with Sun's compiler found a few more places where the code could
3411 be tidied up in order to avoid warnings.
3412
3413 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3414 first of these names is deprecated in the latest Autoconf in favour of the name
3415 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3416 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3417 analogy I have changed it to CFLAGS_FOR_BUILD.
3418
3419 4. Added -no-undefined to the linking command in the Makefile, because this is
3420 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3421 linking step for the pcreposix library.
3422
3423 5. PCRE was failing to diagnose the case of two named groups with the same
3424 name.
3425
3426 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3427 literal character that is needed in the subject for a match, and scans along to
3428 ensure that it is present before embarking on the full matching process. This
3429 saves time in cases of nested unlimited repeats that are never going to match.
3430 Problem: the scan can take a lot of time if the subject is very long (e.g.
3431 megabytes), thus penalizing straightforward matches. It is now done only if the
3432 amount of subject to be scanned is less than 1000 bytes.
3433
3434 7. A lesser problem with the same optimization is that it was recording the
3435 first character of an anchored pattern as "needed", thus provoking a search
3436 right along the subject, even when the first match of the pattern was going to
3437 fail. The "needed" character is now not set for anchored patterns, unless it
3438 follows something in the pattern that is of non-fixed length. Thus, it still
3439 fulfils its original purpose of finding quick non-matches in cases of nested
3440 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3441
3442
3443 Version 4.0 17-Feb-03
3444 ---------------------
3445
3446 1. If a comment in an extended regex that started immediately after a meta-item
3447 extended to the end of string, PCRE compiled incorrect data. This could lead to
3448 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3449
3450 2. Moved to autoconf 2.53 and libtool 1.4.2.
3451
3452 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3453 the special perltest8 script is no longer needed - all the tests can be run
3454 from a single perltest script.
3455
3456 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3457 by \s. It has now been removed in PCRE. This means it isn't recognized as
3458 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3459 class [:space:] *does* include VT, thereby creating a mess.
3460
3461 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3462 space and tab.
3463
3464 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3465 its new features into the main test script, reducing the number of scripts.
3466
3467 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3468 were backward compatible, and made the (?i) apply to the whole pattern, as if
3469 /i were given. Now it behaves more logically, and applies the option setting
3470 only to what follows. PCRE has been changed to follow suit. However, if it
3471 finds options settings right at the start of the pattern, it extracts them into
3472 the global options, as before. Thus, they show up in the info data.
3473
3474 8. Added support for the \Q...\E escape sequence. Characters in between are
3475 treated as literals. This is slightly different from Perl in that $ and @ are
3476 also handled as literals inside the quotes. In Perl, they will cause variable
3477 interpolation. Note the following examples:
3478
3479 Pattern PCRE matches Perl matches
3480
3481 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3482 \Qabc\$xyz\E abc\$xyz abc\$xyz
3483 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3484
3485 For compatibility with Perl, \Q...\E sequences are recognized inside character
3486 classes as well as outside them.
3487
3488 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3489 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3490 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3491 signed/unsigned warnings.
3492
3493 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3494 option for pcretest, so I've replaced it by a simple function that does just
3495 that job.
3496
3497 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3498 "pcregrep -".
3499
3500 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3501 Java package. This provides some syntactic sugar for simple cases of what my
3502 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3503 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3504 item, you can use this simplified notation. Note that only makes sense with
3505 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3506 greediness, whatever the setting of the PCRE_UNGREEDY option.
3507
3508 13. A change of greediness default within a pattern was not taking effect at
3509 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3510 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3511 was abstracted outside.
3512
3513 14. PCRE now supports the \G assertion. It is true when the current matching
3514 position is at the start point of the match. This differs from \A when the
3515 starting offset is non-zero. Used with the /g option of pcretest (or similar
3516 code), it works in the same way as it does for Perl's /g option. If all
3517 alternatives of a regex begin with \G, the expression is anchored to the start
3518 match position, and the "anchored" flag is set in the compiled expression.
3519
3520 15. Some bugs concerning the handling of certain option changes within patterns
3521 have been fixed. These applied to options other than (?ims). For example,
3522 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3523 been the other way round. Some of this was related to change 7 above.
3524
3525 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3526 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3527 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3528 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3529
3530 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3531 mode. Unlike ".", it always matches newline, whatever the setting of
3532 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3533 assertions. Perl allows it, but it doesn't (in general) work because it can't
3534 calculate the length of the lookbehind. At least, that's the case for Perl
3535 5.8.0 - I've been told they are going to document that it doesn't work in
3536 future.
3537
3538 18. Added an error diagnosis for escapes that PCRE does not support: these are
3539 \L, \l, \N, \P, \p, \U, \u, and \X.
3540
3541 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3542 reading past the end of the pattern in cases such as /[abcd/.
3543
3544 20. PCRE was getting more memory than necessary for patterns with classes that
3545 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3546
3547 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3548 compiling PCRE for use with Virtual Pascal.
3549
3550 22. Small fix to the Makefile to make it work properly if the build is done
3551 outside the source tree.
3552
3553 23. Added a new extension: a condition to go with recursion. If a conditional
3554 subpattern starts with (?(R) the "true" branch is used if recursion has
3555 happened, whereas the "false" branch is used only at the top level.
3556
3557 24. When there was a very long string of literal characters (over 255 bytes
3558 without UTF support, over 250 bytes with UTF support), the computation of how
3559 much memory was required could be incorrect, leading to segfaults or other
3560 strange effects.
3561
3562 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3563 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3564 there was a subsequent back reference to those brackets. This meant that, for
3565 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3566 possible to check for precisely this case. All we can do is abandon the
3567 optimization if .* occurs inside capturing brackets when there are any back
3568 references whatsoever. (See below for a better fix that came later.)
3569
3570 26. The handling of the optimization for finding the first character of a
3571 non-anchored pattern, and for finding a character that is required later in the
3572 match were failing in some cases. This didn't break the matching; it just
3573 failed to optimize when it could. The way this is done has been re-implemented.
3574
3575 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3576
3577 28. Added a new feature that provides some of the functionality that Perl
3578 provides with (?{...}). The facility is termed a "callout". The way it is done
3579 in PCRE is for the caller to provide an optional function, by setting
3580 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3581 global variable. By default it is unset, which disables all calling out. To get
3582 the function called, the regex must include (?C) at appropriate points. This
3583 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3584 This provides a means of identifying different callout points. When PCRE
3585 reaches such a point in the regex, if pcre_callout has been set, the external
3586 function is called. It is provided with data in a structure called
3587 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3588 matching continues; if it returns a non-zero value, the match at the current
3589 point fails. However, backtracking will occur if possible. [This was changed
3590 later and other features added - see item 49 below.]
3591
3592 29. pcretest is upgraded to test the callout functionality. It provides a
3593 callout function that displays information. By default, it shows the start of
3594 the match and the current position in the text. There are some new data escapes
3595 to vary what happens:
3596
3597 \C+ in addition, show current contents of captured substrings
3598 \C- do not supply a callout function
3599 \C!n return 1 when callout number n is reached
3600 \C!n!m return 1 when callout number n is reached for the mth time
3601
3602 30. If pcregrep was called with the -l option and just a single file name, it
3603 output "<stdin>" if a match was found, instead of the file name.
3604
3605 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3606 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3607 pcre_exec(). This saves a malloc/free per call. The default value of
3608 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3609 when configuring.
3610
3611 32. The default maximum size of a compiled pattern is 64K. There have been a
3612 few cases of people hitting this limit. The code now uses macros to handle the
3613 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3614 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3615 configuring. Tests 2 and 5 work only with 2-byte links because they output
3616 debugging information about compiled patterns.
3617
3618 33. Internal code re-arrangements:
3619
3620 (a) Moved the debugging function for printing out a compiled regex into
3621 its own source file (printint.c) and used #include to pull it into
3622 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3623 separate copies.
3624
3625 (b) Defined the list of op-code names for debugging as a macro in
3626 internal.h so that it is next to the definition of the opcodes.
3627
3628 (c) Defined a table of op-code lengths for simpler skipping along compiled
3629 code. This is again a macro in internal.h so that it is next to the
3630 definition of the opcodes.
3631
3632 34. Added support for recursive calls to individual subpatterns, along the
3633 lines of Robin Houston's patch (but implemented somewhat differently).
3634
3635 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3636 allow it to read and process whole directories in Win32. This code was
3637 contributed by Lionel Fourquaux; it has not been tested by me.
3638
3639 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3640 used to name a group. Names consist of alphanumerics and underscores, and must
3641 be unique. Back references use the syntax (?P=name) and recursive calls use
3642 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3643 numbers. The function pcre_fullinfo() can be used after compilation to extract
3644 a name/number map. There are three relevant calls:
3645
3646 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3647 PCRE_INFO_NAMECOUNT yields the number of entries
3648 PCRE_INFO_NAMETABLE yields a pointer to the map.
3649
3650 The map is a vector of fixed-size entries. The size of each entry depends on
3651 the length of the longest name used. The first two bytes of each entry are the
3652 group number, most significant byte first. There follows the corresponding
3653 name, zero terminated. The names are in alphabetical order.
3654
3655 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3656 case instead of 255. Making it the same both with and without UTF-8 support
3657 means that the same test output works with both.
3658
3659 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3660 calling malloc() with a zero argument.
3661
3662 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3663 optimization. I've improved things by keeping a bitmap of backreferences with
3664 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3665 fact referenced, the optimization can be applied. It is unlikely that a
3666 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3667 the match to follow \n) will appear inside brackets with a number greater than
3668 31, but if it does, any back reference > 31 suppresses the optimization.
3669
3670 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3671 of disabling numbered capturing parentheses. Any opening parenthesis that is
3672 not followed by ? behaves as if it were followed by ?: but named parentheses
3673 can still be used for capturing (and they will acquire numbers in the usual
3674 way).
3675
3676 41. Redesigned the return codes from the match() function into yes/no/error so
3677 that errors can be passed back from deep inside the nested calls. A malloc
3678 failure while inside a recursive subpattern call now causes the
3679 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3680
3681 42. It is now possible to set a limit on the number of times the match()
3682 function is called in a call to pcre_exec(). This facility makes it possible to
3683 limit the amount of recursion and backtracking, though not in a directly
3684 obvious way, because the match() function is used in a number of different
3685 circumstances. The count starts from zero for each position in the subject
3686 string (for non-anchored patterns). The default limit is, for compatibility, a
3687 large number, namely 10 000 000. You can change this in two ways:
3688
3689 (a) When configuring PCRE before making, you can use --with-match-limit=n
3690 to set a default value for the compiled library.
3691
3692 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3693 a different value is set. See 45 below.
3694
3695 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3696
3697 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3698 of things that can be changed at compile time. The first argument specifies
3699 what is wanted and the second points to where the information is to be placed.
3700 The current list of available information is:
3701
3702 PCRE_CONFIG_UTF8
3703
3704 The output is an integer that is set to one if UTF-8 support is available;
3705 otherwise it is set to zero.
3706
3707 PCRE_CONFIG_NEWLINE
3708
3709 The output is an integer that it set to the value of the code that is used for
3710 newline. It is either LF (10) or CR (13).
3711
3712 PCRE_CONFIG_LINK_SIZE
3713
3714 The output is an integer that contains the number of bytes used for internal
3715 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3716
3717 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3718
3719 The output is an integer that contains the threshold above which the POSIX
3720 interface uses malloc() for output vectors. See item 31 above.
3721
3722 PCRE_CONFIG_MATCH_LIMIT
3723
3724 The output is an unsigned integer that contains the default limit of the number
3725 of match() calls in a pcre_exec() execution. See 42 above.
3726
3727 44. pcretest has been upgraded by the addition of the -C option. This causes it
3728 to extract all the available output from the new pcre_config() function, and to
3729 output it. The program then exits immediately.
3730
3731 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3732 order to support additional features. One way would have been to define
3733 pcre_exec2() (for example) with extra arguments, but this would not have been
3734 extensible, and would also have required all calls to the original function to
3735 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3736 is used for passing in "extra" data from pcre_study().
3737
3738 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3739 contains the following fields:
3740
3741 flags a bitmap indicating which of the following fields are set
3742 study_data opaque data from pcre_study()
3743 match_limit a way of specifying a limit on match() calls for a specific
3744 call to pcre_exec()
3745 callout_data data for callouts (see 49 below)
3746
3747 The flag bits are also defined in pcre.h, and are
3748
3749 PCRE_EXTRA_STUDY_DATA
3750 PCRE_EXTRA_MATCH_LIMIT
3751 PCRE_EXTRA_CALLOUT_DATA
3752
3753 The pcre_study() function now returns one of these new pcre_extra blocks, with
3754 the actual study data pointed to by the study_data field, and the
3755 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3756 before. That is, this change is entirely upwards-compatible and requires no
3757 change to existing code.
3758
3759 If you want to pass in additional data to pcre_exec(), you can either place it
3760 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3761 block.
3762
3763 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3764 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3765 times with different match limits, until it finds the minimum value needed for
3766 pcre_exec() to complete. The value is then output. This can be instructive; for
3767 most simple matches the number is quite small, but for pathological cases it
3768 gets very large very quickly.
3769
3770 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3771 returns the size of the data block pointed to by the study_data field in a
3772 pcre_extra block, that is, the value that was passed as the argument to
3773 pcre_malloc() when PCRE was getting memory in which to place the information
3774 created by pcre_study(). The fourth argument should point to a size_t variable.
3775 pcretest has been extended so that this information is shown after a successful
3776 pcre_study() call when information about the compiled regex is being displayed.
3777
3778 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3779 because what follows is always an absolute path. (Later: it turns out that this
3780 is more than cosmetic for MinGW, because it doesn't like empty path
3781 components.)
3782
3783 49. Some changes have been made to the callout feature (see 28 above):
3784
3785 (i) A callout function now has three choices for what it returns:
3786
3787 0 => success, carry on matching
3788 > 0 => failure at this point, but backtrack if possible
3789 < 0 => serious error, return this value from pcre_exec()
3790
3791 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3792 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3793 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3794 use by callout functions. It will never be used by PCRE itself.
3795
3796 (ii) The pcre_extra structure (see 45 above) has a void * field called
3797 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3798 pcre_callout_block structure has a field of the same name. The contents of
3799 the field passed in the pcre_extra structure are passed to the callout
3800 function in the corresponding field in the callout block. This makes it
3801 easier to use the same callout-containing regex from multiple threads. For
3802 testing, the pcretest program has a new data escape
3803
3804 \C*n pass the number n (may be negative) as callout_data
3805
3806 If the callout function in pcretest receives a non-zero value as
3807 callout_data, it returns that value.
3808
3809 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3810 there were some redundant $(CFLAGS) in commands that are now specified as
3811 $(LINK), which already includes $(CFLAGS).
3812
3813 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3814 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3815 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3816 one-byte characters throughout. Note that case-insensitive matching applies
3817 only to characters whose values are less than 256. PCRE doesn't support the
3818 notion of cases for higher-valued characters.
3819
3820 (i) A character class whose characters are all within 0-255 is handled as
3821 a bit map, and the map is inverted for negative classes. Previously, a
3822 character > 255 always failed to match such a class; however it should
3823 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3824
3825 (ii) A negated character class with a single character < 255 is coded as
3826 "not this character" (OP_NOT). This wasn't working properly when the test
3827 character was multibyte, either singly or repeated.
3828
3829 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3830 mode, for example: \x{100}{2,3}.
3831
3832 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3833 singly or repeated) now correctly test multibyte characters. However,
3834 PCRE doesn't recognize any characters with values greater than 255 as
3835 digits, spaces, or word characters. Such characters always match \D, \S,
3836 and \W, and never match \d, \s, or \w.
3837
3838 (v) Classes may now contain characters and character ranges with values
3839 greater than 255. For example: [ab\x{100}-\x{400}].
3840
3841 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3842 PCRE in UTF-8 mode.
3843
3844 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3845 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3846 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3847 value.)
3848
3849 53. The single man page has become too large. I have therefore split it up into
3850 a number of separate man pages. These also give rise to individual HTML pages;
3851 these are now put in a separate directory, and there is an index.html page that
3852 lists them all. Some hyperlinking between the pages has been installed.
3853
3854 54. Added convenience functions for handling named capturing parentheses.
3855
3856 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3857 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3858 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3859 are faulted.
3860
3861 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3862 calling configure. These values are used when compiling the dftables.c program
3863 which is run to generate the source of the default character tables. They
3864 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3865 you will need to set these values.
3866
3867 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3868
3869
3870 Version 3.9 02-Jan-02
3871 ---------------------
3872
3873 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3874
3875 2. If --disable-static was given, the building process failed when trying to
3876 build pcretest and pcregrep. (For some reason it was using libtool to compile
3877 them, which is not right, as they aren't part of the library.)
3878
3879
3880 Version 3.8 18-Dec-01
3881 ---------------------
3882
3883 1. The experimental UTF-8 code was completely screwed up. It was packing the
3884 bytes in the wrong order. How dumb can you get?
3885
3886
3887 Version 3.7 29-Oct-01
3888 ---------------------
3889
3890 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3891 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3892 this didn't happen under Solaris 8, where I normally test things.
3893
3894 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3895 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3896 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3897 this fix an hour or so after the initial 3.7 release.)
3898
3899
3900 Version 3.6 23-Oct-01
3901 ---------------------
3902
3903 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3904 offsets passed as NULL with zero offset count.
3905
3906 2. The config.guess and config.sub files had not been updated when I moved to
3907 the latest autoconf.
3908
3909
3910 Version 3.5 15-Aug-01
3911 ---------------------
3912
3913 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3914 had been forgotten.
3915
3916 2. By using declared but undefined structures, we can avoid using "void"
3917 definitions in pcre.h while keeping the internal definitions of the structures
3918 private.
3919
3920 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3921 user point of view, this means that both static and shared libraries are built
3922 by default, but this can be individually controlled. More of the work of
3923 handling this static/shared cases is now inside libtool instead of PCRE's make
3924 file.
3925
3926 4. The pcretest utility is now installed along with pcregrep because it is
3927 useful for users (to test regexs) and by doing this, it automatically gets
3928 relinked by libtool. The documentation has been turned into a man page, so
3929 there are now .1, .txt, and .html versions in /doc.
3930
3931 5. Upgrades to pcregrep:
3932 (i) Added long-form option names like gnu grep.
3933 (ii) Added --help to list all options with an explanatory phrase.
3934 (iii) Added -r, --recursive to recurse into sub-directories.
3935 (iv) Added -f, --file to read patterns from a file.
3936
3937 6. pcre_exec() was referring to its "code" argument before testing that
3938 argument for NULL (and giving an error if it was NULL).
3939
3940 7. Upgraded Makefile.in to allow for compiling in a different directory from
3941 the source directory.
3942
3943 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3944 options bits, the pointer it was passed was to an int instead of to an unsigned
3945 long int. This mattered only on 64-bit systems.
3946
3947 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3948 generated) instead of pcre.in, which it its source. Also made the same change
3949 in several of the .c files.
3950
3951 10. A new release of gcc defines printf() as a macro, which broke pcretest
3952 because it had an ifdef in the middle of a string argument for printf(). Fixed
3953 by using separate calls to printf().
3954
3955 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3956 script, to force use of CR or LF instead of \n in the source. On non-Unix
3957 systems, the value can be set in config.h.
3958
3959 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3960 absolute limit. Changed the text of the error message to make this clear, and
3961 likewise updated the man page.
3962
3963 13. The limit of 99 on the number of capturing subpatterns has been removed.
3964 The new limit is 65535, which I hope will not be a "real" limit.
3965
3966
3967 Version 3.4 22-Aug-00
3968 ---------------------
3969
3970 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3971
3972 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3973
3974
3975 Version 3.3 01-Aug-00
3976 ---------------------
3977
3978 1. If an octal character was given, but the value was greater than \377, it
3979 was not getting masked to the least significant bits, as documented. This could
3980 lead to crashes in some systems.
3981
3982 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3983 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3984
3985 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3986 These just pass their arguments on to (pcre_free)(), but they are provided
3987 because some uses of PCRE bind it to non-C systems that can call its functions,
3988 but cannot call free() or pcre_free() directly.
3989
3990 4. Add "make test" as a synonym for "make check". Corrected some comments in
3991 the Makefile.
3992
3993 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3994 Makefile.
3995
3996 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3997 command called pgrep for grepping around the active processes.
3998
3999 7. Added the beginnings of support for UTF-8 character strings.
4000
4001 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4002 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4003 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4004 out for the ar command.)
4005
4006
4007 Version 3.2 12-May-00
4008 ---------------------
4009
4010 This is purely a bug fixing release.
4011
4012 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4013 of ZA. This was just one example of several cases that could provoke this bug,
4014 which was introduced by change 9 of version 2.00. The code for breaking
4015 infinite loops after an iteration that matches an empty string was't working
4016 correctly.
4017
4018 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4019 when matched against abbab (for example). After matching an empty string, it
4020 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4021 caused it to match further down the string than it should.
4022
4023 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4024 was there because it doesn't seem to be needed, and it causes trouble on some
4025 systems, as it is not a Standard C header. It has been removed.
4026
4027 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4028 were reported on the Macintosh. The changes were from
4029
4030 while ((c = *(++ptr)) != 0 && c != '\n');
4031 to
4032 while ((c = *(++ptr)) != 0 && c != '\n') ;
4033
4034 Totally extraordinary, but if that's what it takes...
4035
4036 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4037 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4038 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4039 assumes the way PCRE uses memmove() (always moving upwards).
4040
4041 6. PCRE is being used in one environment where strchr() is not available. There
4042 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4043 faster code anyway.
4044
4045
4046 Version 3.1 09-Feb-00
4047 ---------------------
4048
4049 The only change in this release is the fixing of some bugs in Makefile.in for
4050 the "install" target:
4051
4052 (1) It was failing to install pcreposix.h.
4053
4054 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4055
4056
4057 Version 3.0 01-Feb-00
4058 ---------------------
4059
4060 1. Add support for the /+ modifier to perltest (to output $` like it does in
4061 pcretest).
4062
4063 2. Add support for the /g modifier to perltest.
4064
4065 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4066 matches null strings.
4067
4068 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4069 pattern. Perl treats empty patterns specially - it reuses the most recent
4070 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4071 effect.
4072
4073 5. The POSIX interface was broken in that it was just handing over the POSIX
4074 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4075 required a bigger vector, with some working space on the end. This means that
4076 the POSIX wrapper now has to get and free some memory, and copy the results.
4077
4078 6. Added some simple autoconf support, placing the test data and the
4079 documentation in separate directories, re-organizing some of the
4080 information files, and making it build pcre-config (a GNU standard). Also added
4081 libtool support for building PCRE as a shared library, which is now the
4082 default.
4083
4084 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4085 09 are not valid octal constants. Single digits will be used for minor values
4086 less than 10.
4087
4088 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4089 existing programs that set these in the POSIX interface can use PCRE without
4090 modification.
4091
4092 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4093 return all that pcre_info() returns, plus additional data. The pcre_info()
4094 function is retained for compatibility, but is considered to be obsolete.
4095
4096 10. Added experimental recursion feature (?R) to handle one common case that
4097 Perl 5.6 will be able to do with (?p{...}).
4098
4099 11. Added support for POSIX character classes like [:alpha:], which Perl is
4100 adopting.
4101
4102
4103 Version 2.08 31-Aug-99
4104 ----------------------
4105
4106 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4107 trying to match at the startoffset position, but instead was moving forward to
4108 the next newline as if a previous match had failed.
4109
4110 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4111 and could get into a loop if a null string was matched other than at the start
4112 of the subject.
4113
4114 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4115 be distinguished at compile time, and for completeness also added PCRE_DATE.
4116
4117 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4118 in GnuWin32 environments.
4119
4120
4121 Version 2.07 29-Jul-99
4122 ----------------------
4123
4124 1. The documentation is now supplied in plain text form and HTML as well as in
4125 the form of man page sources.
4126
4127 2. C++ compilers don't like assigning (void *) values to other pointer types.
4128 In particular this affects malloc(). Although there is no problem in Standard
4129 C, I've put in casts to keep C++ compilers happy.
4130
4131 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4132 should be (const char *).
4133
4134 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4135 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4136 However, I haven't made this a standard facility. The documentation doesn't
4137 mention it, and the Makefile doesn't support it.
4138
4139 5. The Makefile now contains an "install" target, with editable destinations at
4140 the top of the file. The pcretest program is not installed.
4141
4142 6. pgrep -V now gives the PCRE version number and date.
4143
4144 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4145 causing the entire string to be ignored, instead of just the last character.
4146
4147 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4148 non-matching string, it can take a very, very long time, even for strings of
4149 quite modest length, because of the nested recursion. PCRE now does better in
4150 some of these cases. It does this by remembering the last required literal
4151 character in the pattern, and pre-searching the subject to ensure it is present
4152 before running the real match. In other words, it applies a heuristic to detect
4153 some types of certain failure quickly, and in the above example, if presented
4154 with a string that has no trailing " it gives "no match" very quickly.
4155
4156 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4157 other alternatives are tried instead.
4158
4159
4160 Version 2.06 09-Jun-99
4161 ----------------------
4162
4163 1. Change pcretest's output for amount of store used to show just the code
4164 space, because the remainder (the data block) varies in size between 32-bit and
4165 64-bit systems.
4166
4167 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4168 start matching at. This allows lookbehinds to work when searching for multiple
4169 occurrences in a string.
4170
4171 3. Added additional options to pcretest for testing multiple occurrences:
4172
4173 /+ outputs the rest of the string that follows a match
4174 /g loops for multiple occurrences, using the new startoffset argument
4175 /G loops for multiple occurrences by passing an incremented pointer
4176
4177 4. PCRE wasn't doing the "first character" optimization for patterns starting
4178 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4179 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4180 the letter 'x'. On long subject strings, this gives a significant speed-up.
4181
4182
4183 Version 2.05 21-Apr-99
4184 ----------------------
4185
4186 1. Changed the type of magic_number from int to long int so that it works
4187 properly on 16-bit systems.
4188
4189 2. Fixed a bug which caused patterns starting with .* not to work correctly
4190 when the subject string contained newline characters. PCRE was assuming
4191 anchoring for such patterns in all cases, which is not correct because .* will
4192 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4193 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4194 must be retried after every newline in the subject.
4195
4196
4197 Version 2.04 18-Feb-99
4198 ----------------------
4199
4200 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4201 computation of the store needed to hold the pattern was incorrect (too large).
4202 If such patterns were nested a few deep, this could multiply and become a real
4203 problem.
4204
4205 2. Added /M option to pcretest to show the memory requirement of a specific
4206 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4207
4208 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4209 compiled in such a way that the backtracking after subsequent failure was
4210 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4211 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4212
4213
4214 Version 2.03 02-Feb-99
4215 ----------------------
4216
4217 1. Fixed typo and small mistake in man page.
4218
4219 2. Added 4th condition (GPL supersedes if conflict) and created separate
4220 LICENCE file containing the conditions.
4221
4222 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4223 Perl, that is the internal \ allows the delimiter to be included in the
4224 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4225 the final delimiter, add \ to the end of the pattern (to test the error).
4226
4227 4. Added the convenience functions for extracting substrings after a successful
4228 match. Updated pcretest to make it able to test these functions.
4229
4230
4231 Version 2.02 14-Jan-99
4232 ----------------------
4233
4234 1. Initialized the working variables associated with each extraction so that
4235 their saving and restoring doesn't refer to uninitialized store.
4236
4237 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4238 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4239 fix the problem.
4240
4241 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4242 calls, and wasn't printing the correct value for compiling calls. Increased the
4243 default value of LOOPREPEAT, and the number of significant figures in the
4244 times.
4245
4246 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4247
4248 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4249 a building problem on Windows NT with a FAT file system.
4250
4251
4252 Version 2.01 21-Oct-98
4253 ----------------------
4254
4255 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4256 to character tables built by pcre_maketables() in the current locale. If NULL
4257 is passed, the default tables are used.
4258
4259
4260 Version 2.00 24-Sep-98
4261 ----------------------
4262
4263 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4264 it any more.
4265
4266 2. Allow quantification of (?>) groups, and make it work correctly.
4267
4268 3. The first character computation wasn't working for (?>) groups.
4269
4270 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4271 end of the subject) and add 5.005's \z, which really does match only at the
4272 very end of the subject.
4273
4274 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4275
4276 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4277 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4278 localized options. All options to pcre_study() were also removed.
4279
4280 7. Add other new features from 5.005:
4281
4282 $(?<= positive lookbehind
4283 $(?<! negative lookbehind
4284 (?imsx-imsx) added the unsetting capability
4285 such a setting is global if at outer level; local otherwise
4286 (?imsx-imsx:) non-capturing groups with option setting
4287 (?(cond)re|re) conditional pattern matching
4288
4289 A backreference to itself in a repeated group matches the previous
4290 captured string.
4291
4292 8. General tidying up of studying (both automatic and via "study")
4293 consequential on the addition of new assertions.
4294
4295 9. As in 5.005, unlimited repeated groups that could match an empty substring
4296 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4297 runtime if any iteration does actually match an empty substring.
4298
4299 10. Include the RunTest script in the distribution.
4300
4301 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4302 discrepancies, some of which were old and were also with respect to 5.004. They
4303 have now been fixed.
4304
4305
4306 Version 1.09 28-Apr-98
4307 ----------------------
4308
4309 1. A negated single character class followed by a quantifier with a minimum
4310 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4311 program crashes, or just wrong answers. This did not apply to negated classes
4312 containing more than one character, or to minima other than one.
4313
4314
4315 Version 1.08 27-Mar-98
4316 ----------------------
4317
4318 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4319
4320 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4321 latter must appear before anything that relies on it in the pattern.
4322
4323
4324 Version 1.07 16-Feb-98
4325 ----------------------
4326
4327 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4328 repeat of a potentially empty string).
4329
4330
4331 Version 1.06 23-Jan-98
4332 ----------------------
4333
4334 1. Added Markus Oberhumer's little patches for C++.
4335
4336 2. Literal strings longer than 255 characters were broken.
4337
4338
4339 Version 1.05 23-Dec-97
4340 ----------------------
4341
4342 1. Negated character classes containing more than one character were failing if
4343 PCRE_CASELESS was set at run time.
4344
4345
4346 Version 1.04 19-Dec-97
4347 ----------------------
4348
4349 1. Corrected the man page, where some "const" qualifiers had been omitted.
4350
4351 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4352 input syntax.
4353
4354 3. Fixed memory leak which occurred when a regex with back references was
4355 matched with an offsets vector that wasn't big enough. The temporary memory
4356 that is used in this case wasn't being freed if the match failed.
4357
4358 4. Tidied pcretest to ensure it frees memory that it gets.
4359
4360 5. Temporary memory was being obtained in the case where the passed offsets
4361 vector was exactly big enough.
4362
4363 6. Corrected definition of offsetof() from change 5 below.
4364
4365 7. I had screwed up change 6 below and broken the rules for the use of
4366 setjmp(). Now fixed.
4367
4368
4369 Version 1.03 18-Dec-97
4370 ----------------------
4371
4372 1. A erroneous regex with a missing opening parenthesis was correctly
4373 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4374 on some systems.
4375
4376 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4377 it was reported that one broken compiler failed on the former because "code" is
4378 also an independent variable.
4379
4380 3. The erroneous regex a[]b caused an array overrun reference.
4381
4382 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4383 fail on data ending with that character. (It was going on too far, and checking
4384 the next character, typically a binary zero.) This was specific to the
4385 optimized code for single-character negative classes.
4386
4387 5. Added a contributed patch from the TIN world which does the following:
4388
4389 + Add an undef for memmove, in case the the system defines a macro for it.
4390
4391 + Add a definition of offsetof(), in case there isn't one. (I don't know
4392 the reason behind this - offsetof() is part of the ANSI standard - but
4393 it does no harm).
4394
4395 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4396 most of the places where whitespace preceded '#'. I have given up and
4397 allowed the remaining 2 cases to be at the margin.
4398
4399 + Rename some variables in pcre to eliminate shadowing. This seems very
4400 pedantic, but does no harm, of course.
4401
4402 6. Moved the call to setjmp() into its own function, to get rid of warnings
4403 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4404
4405 7. Constructs such as \d{8,} were compiling into the equivalent of
4406 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4407 outcome, but in this particular case used more store than had been allocated,
4408 which caused the bug to be discovered because it threw up an internal error.
4409
4410 8. The debugging code in both pcre and pcretest for outputting the compiled
4411 form of a regex was going wrong in the case of back references followed by
4412 curly-bracketed repeats.
4413
4414
4415 Version 1.02 12-Dec-97
4416 ----------------------
4417
4418 1. Typos in pcre.3 and comments in the source fixed.
4419
4420 2. Applied a contributed patch to get rid of places where it used to remove
4421 'const' from variables, and fixed some signed/unsigned and uninitialized
4422 variable warnings.
4423
4424 3. Added the "runtest" target to Makefile.
4425
4426 4. Set default compiler flag to -O2 rather than just -O.
4427
4428
4429 Version 1.01 19-Nov-97
4430 ----------------------
4431
4432 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4433 like /([ab]*)*/, that is, for classes with more than one character in them.
4434
4435 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4436 as /((?>a*))*/ (a PCRE_EXTRA facility).
4437
4438
4439 Version 1.00 18-Nov-97
4440 ----------------------
4441
4442 1. Added compile-time macros to support systems such as SunOS4 which don't have
4443 memmove() or strerror() but have other things that can be used instead.
4444
4445 2. Arranged that "make clean" removes the executables.
4446
4447
4448 Version 0.99 27-Oct-97
4449 ----------------------
4450
4451 1. Fixed bug in code for optimizing classes with only one character. It was
4452 initializing a 32-byte map regardless, which could cause it to run off the end
4453 of the memory it had got.
4454
4455 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4456
4457
4458 Version 0.98 22-Oct-97
4459 ----------------------
4460
4461 1. Fixed bug in code for handling temporary memory usage when there are more
4462 back references than supplied space in the ovector. This could cause segfaults.
4463
4464
4465 Version 0.97 21-Oct-97
4466 ----------------------
4467
4468 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4469
4470 2. Optimized negated single characters not to use a bit map.
4471
4472 3. Brought error texts together as macro definitions; clarified some of them;
4473 fixed one that was wrong - it said "range out of order" when it meant "invalid
4474 escape sequence".
4475
4476 4. Changed some char * arguments to const char *.
4477
4478 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4479
4480 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4481 pcretest.
4482
4483
4484 Version 0.96 16-Oct-97
4485 ----------------------
4486
4487 1. Added a simple "pgrep" utility to the distribution.
4488
4489 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4490 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4491 where "ddd" means "one or more decimal digits".
4492
4493 3. Fixed serious bug. If a pattern had a back reference, but the call to
4494 pcre_exec() didn't supply a large enough ovector to record the related
4495 identifying subpattern, the match always failed. PCRE now remembers the number
4496 of the largest back reference, and gets some temporary memory in which to save
4497 the offsets during matching if necessary, in order to ensure that
4498 backreferences always work.
4499
4500 4. Increased the compatibility with Perl in a number of ways:
4501
4502 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4503 to request this handling. The option can be set at compile or exec time.
4504
4505 (b) $ matches before a terminating newline by default; an option
4506 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4507 mode). The option can be set at compile or exec time.
4508
4509 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4510 the same as Perl's. If the decimal number it represents is less than 10
4511 or there aren't that many previous left capturing parentheses, an octal
4512 escape is read. Inside a character class, it's always an octal escape,
4513 even if it is a single digit.
4514
4515 (d) An escaped but undefined alphabetic character is taken as a literal,
4516 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4517 escapes.
4518
4519 (e) {0} is now permitted. (The previous item is removed from the compiled
4520 pattern).
4521
4522 5. Changed all the names of code files so that the basic parts are no longer
4523 than 10 characters, and abolished the teeny "globals.c" file.
4524
4525 6. Changed the handling of character classes; they are now done with a 32-byte
4526 bit map always.
4527
4528 7. Added the -d and /D options to pcretest to make it possible to look at the
4529 internals of compilation without having to recompile pcre.
4530
4531
4532 Version 0.95 23-Sep-97
4533 ----------------------
4534
4535 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4536 \x20 at the start of a run of normal characters. These were being treated as
4537 real characters, instead of the source characters being re-checked.
4538
4539
4540 Version 0.94 18-Sep-97
4541 ----------------------
4542
4543 1. The functions are now thread-safe, with the caveat that the global variables
4544 containing pointers to malloc() and free() or alternative functions are the
4545 same for all threads.
4546
4547 2. Get pcre_study() to generate a bitmap of initial characters for non-
4548 anchored patterns when this is possible, and use it if passed to pcre_exec().
4549
4550
4551 Version 0.93 15-Sep-97
4552 ----------------------
4553
4554 1. /(b)|(:+)/ was computing an incorrect first character.
4555
4556 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4557 but not actually doing anything yet.
4558
4559 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4560 as Perl does (e.g. [-az] or [az-]).
4561
4562 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4563 all possible positions.
4564
4565 5. Split up into different modules to avoid including unneeded functions in a
4566 compiled binary. However, compile and exec are still in one module. The "study"
4567 function is split off.
4568
4569 6. The character tables are now in a separate module whose source is generated
4570 by an auxiliary program - but can then be edited by hand if required. There are
4571 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4572 toupper() in the code.
4573
4574 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4575 make them global. Abolish the function for setting them, as the caller can now
4576 set them directly.
4577
4578
4579 Version 0.92 11-Sep-97
4580 ----------------------
4581
4582 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4583 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4584
4585 2. Caseless matching was not working in character classes if the characters in
4586 the pattern were in upper case.
4587
4588 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4589
4590 4. Make PCRE_ANCHORED public and accept as a compile option.
4591
4592 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4593 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4594 pass them.
4595
4596 6. Give an error if bad option bits passed at compile or run time.
4597
4598 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4599 pcretest to cause it to pass that flag.
4600
4601 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4602 options, and the first character, if set.
4603
4604 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4605
4606
4607 Version 0.91 10-Sep-97
4608 ----------------------
4609
4610 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4611 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4612
4613 2. PCRE was looping on encountering an indefinitely repeated back reference to
4614 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4615 Perl does - treats the match as successful.
4616
4617 ****

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

  ViewVC Help
Powered by ViewVC 1.1.5