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

Contents of /code/branches/pcre16/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 821 - (show annotations)
Fri Dec 23 16:38:13 2011 UTC (7 years, 7 months ago) by ph10
File size: 223941 byte(s)
More pcretest 16-bit updates; also a bug fix in pcre_exec.
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 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
17 match a (*MARK), and the match failed at the start of the subject, a
18 reference to memory before the start of the subject could occur. This bug
19 was introduced by fix 17 of release 8.21.
20
21
22 Version 8.21 12-Dec-2011
23 ------------------------
24
25 1. Updating the JIT compiler.
26
27 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
28 are added as well.
29
30 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
31 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
32 calling _pcre_jit_exec. Some extra comments are added.
33
34 4. (*MARK) settings inside atomic groups that do not contain any capturing
35 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
36 was introduced by change 18 for 8.20.
37
38 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
39 ECMA-262 standard.
40
41 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
42 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
43 This bug was probably introduced by change 9 of 8.13.
44
45 7. While fixing 6 above, I noticed that a number of other items were being
46 incorrectly rejected as "not fixed length". This arose partly because newer
47 opcodes had not been added to the fixed-length checking code. I have (a)
48 corrected the bug and added tests for these items, and (b) arranged for an
49 error to occur if an unknown opcode is encountered while checking for fixed
50 length instead of just assuming "not fixed length". The items that were
51 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
52 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
53 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
54
55 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
56 being incorrectly compiled and would have given unpredicatble results.
57
58 9. A possessively repeated subpattern with minimum repeat count greater than
59 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
60 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
61 the first (A) could occur when it should not.
62
63 10. Add a cast and remove a redundant test from the code.
64
65 11. JIT should use pcre_malloc/pcre_free for allocation.
66
67 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
68 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
69 is anything other than /usr, -L is still shown).
70
71 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
72
73 14. Perl does not support \N without a following name in a [] class; PCRE now
74 also gives an error.
75
76 15. If a forward reference was repeated with an upper limit of around 2000,
77 it caused the error "internal error: overran compiling workspace". The
78 maximum number of forward references (including repeats) was limited by the
79 internal workspace, and dependent on the LINK_SIZE. The code has been
80 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
81 the default depends on LINK_SIZE. There is a new upper limit (for safety)
82 of around 200,000 forward references. While doing this, I also speeded up
83 the filling in of repeated forward references.
84
85 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
86 incorrectly expecting the subject to contain another "a" after the start.
87
88 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
89 in the match, the SKIP should be ignored. This was not happening; instead
90 the SKIP was being treated as NOMATCH. For patterns such as
91 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
92 tested.
93
94 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
95 now much more compatible with Perl, in particular in cases where the result
96 is a non-match for a non-anchored pattern. For example, if
97 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
98 "m", where previously it did not return a name. A side effect of this
99 change is that for partial matches, the last encountered mark name is
100 returned, as for non matches. A number of tests that were previously not
101 Perl-compatible have been moved into the Perl-compatible test files. The
102 refactoring has had the pleasing side effect of removing one argument from
103 the match() function, thus reducing its stack requirements.
104
105 19. If the /S+ option was used in pcretest to study a pattern using JIT,
106 subsequent uses of /S (without +) incorrectly behaved like /S+.
107
108 21. Retrieve executable code size support for the JIT compiler and fixing
109 some warnings.
110
111 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
112 not work when the shorter character appeared right at the end of the
113 subject string.
114
115 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
116 systems.
117
118 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
119 output it when the /M option is used in pcretest.
120
121 25. The CheckMan script was not being included in the distribution. Also, added
122 an explicit "perl" to run Perl scripts from the PrepareRelease script
123 because this is reportedly needed in Windows.
124
125 26. If study data was being save in a file and studying had not found a set of
126 "starts with" bytes for the pattern, the data written to the file (though
127 never used) was taken from uninitialized memory and so caused valgrind to
128 complain.
129
130 27. Updated RunTest.bat as provided by Sheri Pierce.
131
132 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
133
134 29. Computation of memory usage for the table of capturing group names was
135 giving an unnecessarily large value.
136
137
138 Version 8.20 21-Oct-2011
139 ------------------------
140
141 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
142 a POSIX class. After further experiments with Perl, which convinced me that
143 Perl has bugs and confusions, a closing square bracket is no longer allowed
144 in a POSIX name. This bug also affected patterns with classes that started
145 with full stops.
146
147 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
148 captured substring, but while checking the failing first alternative,
149 substring 1 is temporarily captured. If the output vector supplied to
150 pcre_exec() was not big enough for this capture, the yield of the function
151 was still zero ("insufficient space for captured substrings"). This cannot
152 be totally fixed without adding another stack variable, which seems a lot
153 of expense for a edge case. However, I have improved the situation in cases
154 such as /(a)(b)x|abc/ matched against "abc", where the return code
155 indicates that fewer than the maximum number of slots in the ovector have
156 been set.
157
158 3. Related to (2) above: when there are more back references in a pattern than
159 slots in the output vector, pcre_exec() uses temporary memory during
160 matching, and copies in the captures as far as possible afterwards. It was
161 using the entire output vector, but this conflicts with the specification
162 that only 2/3 is used for passing back captured substrings. Now it uses
163 only the first 2/3, for compatibility. This is, of course, another edge
164 case.
165
166 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
167 main code base, and can be used by building with --enable-jit. When this is
168 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
169 runtime --no-jit option is given.
170
171 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
172 ovector, the return from the function was zero, implying that there were
173 other matches that did not fit. The correct "exactly full" value is now
174 returned.
175
176 6. If a subpattern that was called recursively or as a subroutine contained
177 (*PRUNE) or any other control that caused it to give a non-standard return,
178 invalid errors such as "Error -26 (nested recursion at the same subject
179 position)" or even infinite loops could occur.
180
181 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
182 computing the minimum length on reaching *ACCEPT, and so ended up with the
183 wrong value of 1 rather than 0. Further investigation indicates that
184 computing a minimum subject length in the presence of *ACCEPT is difficult
185 (think back references, subroutine calls), and so I have changed the code
186 so that no minimum is registered for a pattern that contains *ACCEPT.
187
188 8. If (*THEN) was present in the first (true) branch of a conditional group,
189 it was not handled as intended. [But see 16 below.]
190
191 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
192 Sheri Pierce.
193
194 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
195 the first byte in a match must be "a".
196
197 11. Change 17 for 8.13 increased the recursion depth for patterns like
198 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
199 pattern contains any instances of (*THEN). If it does not, the old
200 optimizations are restored. It would be nice to do this on a per-group
201 basis, but at the moment that is not feasible.
202
203 12. In some environments, the output of pcretest -C is CRLF terminated. This
204 broke RunTest's code that checks for the link size. A single white space
205 character after the value is now allowed for.
206
207 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
208 For "fr", it uses the Windows-specific input and output files.
209
210 14. If (*THEN) appeared in a group that was called recursively or as a
211 subroutine, it did not work as intended. [But see next item.]
212
213 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
214 pattern fragments (but not containing any | characters). If A and B are
215 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
216 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
217 D. In other words, Perl considers parentheses that do not contain any |
218 characters to be part of a surrounding alternative, whereas PCRE was
219 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
220 differently. PCRE now behaves in the same way as Perl, except in the case
221 of subroutine/recursion calls such as (?1) which have in any case always
222 been different (but PCRE had them first :-).
223
224 16. Related to 15 above: Perl does not treat the | in a conditional group as
225 creating alternatives. Such a group is treated in the same way as an
226 ordinary group without any | characters when processing (*THEN). PCRE has
227 been changed to match Perl's behaviour.
228
229 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
230 RunGrepTest script failed.
231
232 18. Change 22 for version 13 caused atomic groups to use more stack. This is
233 inevitable for groups that contain captures, but it can lead to a lot of
234 stack use in large patterns. The old behaviour has been restored for atomic
235 groups that do not contain any capturing parentheses.
236
237 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
238 suppress the check for a minimum subject length at run time. (If it was
239 given to pcre_exec() or pcre_dfa_exec() it did work.)
240
241 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
242 fail to work when decoding hex characters in data strings in EBCDIC
243 environments.
244
245 21. It appears that in at least one Mac OS environment, the isxdigit() function
246 is implemented as a macro that evaluates to its argument more than once,
247 contravening the C 90 Standard (I haven't checked a later standard). There
248 was an instance in pcretest which caused it to go wrong when processing
249 \x{...} escapes in subject strings. The has been rewritten to avoid using
250 things like p++ in the argument of isxdigit().
251
252
253 Version 8.13 16-Aug-2011
254 ------------------------
255
256 1. The Unicode data tables have been updated to Unicode 6.0.0.
257
258 2. Two minor typos in pcre_internal.h have been fixed.
259
260 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
261 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
262 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
263
264 4. There were a number of related bugs in the code for matching backrefences
265 caselessly in UTF-8 mode when codes for the characters concerned were
266 different numbers of bytes. For example, U+023A and U+2C65 are an upper
267 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
268 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
269 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
270 2-byte code at the end of the subject (it thought there wasn't enough data
271 left).
272
273 5. Comprehensive information about what went wrong is now returned by
274 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
275 as the output vector has at least 2 elements. The offset of the start of
276 the failing character and a reason code are placed in the vector.
277
278 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
279 now returned is for the first byte of the failing character, instead of the
280 last byte inspected. This is an incompatible change, but I hope it is small
281 enough not to be a problem. It makes the returned offset consistent with
282 pcre_exec() and pcre_dfa_exec().
283
284 7. pcretest now gives a text phrase as well as the error number when
285 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
286 failure, the offset and reason code are output.
287
288 8. When \R was used with a maximizing quantifier it failed to skip backwards
289 over a \r\n pair if the subsequent match failed. Instead, it just skipped
290 back over a single character (\n). This seems wrong (because it treated the
291 two characters as a single entity when going forwards), conflicts with the
292 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
293 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
294 has been changed.
295
296 9. Some internal refactoring has changed the processing so that the handling
297 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
298 time (the PCRE_DOTALL option was changed this way some time ago: version
299 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
300 which was always a bit of a fudge. It also means that there is one less
301 argument for the match() function, which reduces its stack requirements
302 slightly. This change also fixes an incompatibility with Perl: the pattern
303 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
304
305 10. More internal refactoring has drastically reduced the number of recursive
306 calls to match() for possessively repeated groups such as (abc)++ when
307 using pcre_exec().
308
309 11. While implementing 10, a number of bugs in the handling of groups were
310 discovered and fixed:
311
312 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
313 (a|)*(?1) gave a compile-time internal error.
314 ((a|)+)+ did not notice that the outer group could match an empty string.
315 (^a|^)+ was not marked as anchored.
316 (.*a|.*)+ was not marked as matching at start or after a newline.
317
318 12. Yet more internal refactoring has removed another argument from the match()
319 function. Special calls to this function are now indicated by setting a
320 value in a variable in the "match data" data block.
321
322 13. Be more explicit in pcre_study() instead of relying on "default" for
323 opcodes that mean there is no starting character; this means that when new
324 ones are added and accidentally left out of pcre_study(), testing should
325 pick them up.
326
327 14. The -s option of pcretest has been documented for ages as being an old
328 synonym of -m (show memory usage). I have changed it to mean "force study
329 for every regex", that is, assume /S for every regex. This is similar to -i
330 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
331 using it. It makes it easier to run collections of tests with and without
332 study enabled, and thereby test pcre_study() more easily. All the standard
333 tests are now run with and without -s (but some patterns can be marked as
334 "never study" - see 20 below).
335
336 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
337 restoration of the capturing data to the outer values was not happening
338 correctly.
339
340 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
341 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
342 pattern had matched an empty string, and so incorrectly returned a no
343 match.
344
345 17. There was optimizing code for the last branch of non-capturing parentheses,
346 and also for the obeyed branch of a conditional subexpression, which used
347 tail recursion to cut down on stack usage. Unfortunately, now that there is
348 the possibility of (*THEN) occurring in these branches, tail recursion is
349 no longer possible because the return has to be checked for (*THEN). These
350 two optimizations have therefore been removed. [But see 8.20/11 above.]
351
352 18. If a pattern containing \R was studied, it was assumed that \R always
353 matched two bytes, thus causing the minimum subject length to be
354 incorrectly computed because \R can also match just one byte.
355
356 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
357 was incorrectly computed.
358
359 20. If /S is present twice on a test pattern in pcretest input, it now
360 *disables* studying, thereby overriding the use of -s on the command line
361 (see 14 above). This is necessary for one or two tests to keep the output
362 identical in both cases.
363
364 21. When (*ACCEPT) was used in an assertion that matched an empty string and
365 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
366
367 22. When an atomic group that contained a capturing parenthesis was
368 successfully matched, but the branch in which it appeared failed, the
369 capturing was not being forgotten if a higher numbered group was later
370 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
371 group 1 to "a", when in fact it should be unset. This applied to multi-
372 branched capturing and non-capturing groups, repeated or not, and also to
373 positive assertions (capturing in negative assertions does not happen
374 in PCRE) and also to nested atomic groups.
375
376 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
377 subject after a captured substring, to make it easier to tell which of a
378 number of identical substrings has been captured.
379
380 24. The way atomic groups are processed by pcre_exec() has been changed so that
381 if they are repeated, backtracking one repetition now resets captured
382 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
383 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
384 "aaa". Previously, it would have been "a". As part of this code
385 refactoring, the way recursive calls are handled has also been changed.
386
387 25. If an assertion condition captured any substrings, they were not passed
388 back unless some other capturing happened later. For example, if
389 (?(?=(a))a) was matched against "a", no capturing was returned.
390
391 26. When studying a pattern that contained subroutine calls or assertions,
392 the code for finding the minimum length of a possible match was handling
393 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
394 group 1 called group 2 while simultaneously a separate group 2 called group
395 1). A stack overflow occurred in this case. I have fixed this by limiting
396 the recursion depth to 10.
397
398 27. Updated RunTest.bat in the distribution to the version supplied by Tom
399 Fortmann. This supports explicit test numbers on the command line, and has
400 argument validation and error reporting.
401
402 28. An instance of \X with an unlimited repeat could fail if at any point the
403 first character it looked at was a mark character.
404
405 29. Some minor code refactoring concerning Unicode properties and scripts
406 should reduce the stack requirement of match() slightly.
407
408 30. Added the '=' option to pcretest to check the setting of unused capturing
409 slots at the end of the pattern, which are documented as being -1, but are
410 not included in the return count.
411
412 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
413 compiled something random. Now it gives a compile-time error (as does
414 Perl).
415
416 32. A *MARK encountered during the processing of a positive assertion is now
417 recorded and passed back (compatible with Perl).
418
419 33. If --only-matching or --colour was set on a pcregrep call whose pattern
420 had alternative anchored branches, the search for a second match in a line
421 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
422 matched the line "0102" twice. The same bug affected patterns that started
423 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
424 twice.
425
426 34. Previously, PCRE did not allow quantification of assertions. However, Perl
427 does, and because of capturing effects, quantifying parenthesized
428 assertions may at times be useful. Quantifiers are now allowed for
429 parenthesized assertions.
430
431 35. A minor code tidy in pcre_compile() when checking options for \R usage.
432
433 36. \g was being checked for fancy things in a character class, when it should
434 just be a literal "g".
435
436 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
437 appearance of a nested POSIX class supersedes an apparent external class.
438 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
439 unescaped square brackets may also appear as part of class names. For
440 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
441 more like Perl. (But see 8.20/1 above.)
442
443 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
444 was because it thought it was \N{name}, which is not supported).
445
446 39. Add minix to OS list not supporting the -S option in pcretest.
447
448 40. PCRE tries to detect cases of infinite recursion at compile time, but it
449 cannot analyze patterns in sufficient detail to catch mutual recursions
450 such as ((?1))((?2)). There is now a runtime test that gives an error if a
451 subgroup is called recursively as a subpattern for a second time at the
452 same position in the subject string. In previous releases this might have
453 been caught by the recursion limit, or it might have run out of stack.
454
455 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
456 happen only once. PCRE was, however incorrectly giving a compile time error
457 "recursive call could loop indefinitely" because it cannot analyze the
458 pattern in sufficient detail. The compile time test no longer happens when
459 PCRE is compiling a conditional subpattern, but actual runaway loops are
460 now caught at runtime (see 40 above).
461
462 42. It seems that Perl allows any characters other than a closing parenthesis
463 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
464 has been changed to be the same.
465
466 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
467 as not to get warnings when autogen.sh is called. Also changed
468 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
469
470 44. To help people who use pcregrep to scan files containing exceedingly long
471 lines, the following changes have been made:
472
473 (a) The default value of the buffer size parameter has been increased from
474 8K to 20K. (The actual buffer used is three times this size.)
475
476 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
477 PCRE is built.
478
479 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
480 to be set at run time.
481
482 (d) Numerical values in pcregrep options can be followed by K or M, for
483 example --buffer-size=50K.
484
485 (e) If a line being scanned overflows pcregrep's buffer, an error is now
486 given and the return code is set to 2.
487
488 45. Add a pointer to the latest mark to the callout data block.
489
490 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
491 partial match of an empty string instead of no match. This was specific to
492 the use of ".".
493
494 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
495 complete match instead of a partial match. This bug was dependent on both
496 the PCRE_UTF8 and PCRE_DOTALL options being set.
497
498 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
499 starting byte set, because \b was not being ignored.
500
501
502 Version 8.12 15-Jan-2011
503 ------------------------
504
505 1. Fixed some typos in the markup of the man pages, and wrote a script that
506 checks for such things as part of the documentation building process.
507
508 2. On a big-endian 64-bit system, pcregrep did not correctly process the
509 --match-limit and --recursion-limit options (added for 8.11). In
510 particular, this made one of the standard tests fail. (The integer value
511 went into the wrong half of a long int.)
512
513 3. If the --colour option was given to pcregrep with -v (invert match), it
514 did strange things, either producing crazy output, or crashing. It should,
515 of course, ignore a request for colour when reporting lines that do not
516 match.
517
518 4. Another pcregrep bug caused similar problems if --colour was specified with
519 -M (multiline) and the pattern match finished with a line ending.
520
521 5. In pcregrep, when a pattern that ended with a literal newline sequence was
522 matched in multiline mode, the following line was shown as part of the
523 match. This seems wrong, so I have changed it.
524
525 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
526 the check for further matches in the same line (so they could be coloured)
527 to overrun the end of the current line. If another match was found, it was
528 incorrectly shown (and then shown again when found in the next line).
529
530 7. If pcregrep was compiled under Windows, there was a reference to the
531 function pcregrep_exit() before it was defined. I am assuming this was
532 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
533 reported by a user. I've moved the definition above the reference.
534
535
536 Version 8.11 10-Dec-2010
537 ------------------------
538
539 1. (*THEN) was not working properly if there were untried alternatives prior
540 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
541 backtracked to try for "b" instead of moving to the next alternative branch
542 at the same level (in this case, to look for "c"). The Perl documentation
543 is clear that when (*THEN) is backtracked onto, it goes to the "next
544 alternative in the innermost enclosing group".
545
546 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
547 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
548 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
549 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
550 (*THEN).
551
552 3. If \s appeared in a character class, it removed the VT character from
553 the class, even if it had been included by some previous item, for example
554 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
555 of \s, but is part of the POSIX "space" class.)
556
557 4. A partial match never returns an empty string (because you can always
558 match an empty string at the end of the subject); however the checking for
559 an empty string was starting at the "start of match" point. This has been
560 changed to the "earliest inspected character" point, because the returned
561 data for a partial match starts at this character. This means that, for
562 example, /(?<=abc)def/ gives a partial match for the subject "abc"
563 (previously it gave "no match").
564
565 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
566 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
567 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
568 has an implication that the given string is incomplete (because a partial
569 match is preferred over a full match). For this reason, these items now
570 give a partial match in this situation. [Aside: previously, the one case
571 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
572 match rather than a full match, which was wrong by the old rules, but is
573 now correct.]
574
575 6. There was a bug in the handling of #-introduced comments, recognized when
576 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
577 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
578 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
579 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
580 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
581 places in pcre_compile().
582
583 7. Related to (6) above, when pcre_compile() was skipping #-introduced
584 comments when looking ahead for named forward references to subpatterns,
585 the only newline sequence it recognized was NL. It now handles newlines
586 according to the set newline convention.
587
588 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
589 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
590 cater for a lack of strerror(). These oversights have been fixed.
591
592 9. Added --match-limit and --recursion-limit to pcregrep.
593
594 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
595
596 11. When the -o option was used, pcregrep was setting a return code of 1, even
597 when matches were found, and --line-buffered was not being honoured.
598
599 12. Added an optional parentheses number to the -o and --only-matching options
600 of pcregrep.
601
602 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
603 can match an empty string. The code to do it in pcretest and pcredemo
604 needed fixing:
605
606 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
607 only one byte after an empty string match just before CRLF (this case
608 just got forgotten; "any" and "anycrlf" were OK).
609
610 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
611 mode when an empty string match preceded an ASCII character followed by
612 a non-ASCII character. (The code for advancing by one character rather
613 than one byte was nonsense.)
614
615 (c) The pcredemo.c sample program did not have any code at all to handle
616 the cases when CRLF is a valid newline sequence.
617
618 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
619 as a starting offset was within the subject string. There is now a new
620 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
621 negative or greater than the length of the string. In order to test this,
622 pcretest is extended to allow the setting of negative starting offsets.
623
624 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
625 starting offset points to the beginning of a UTF-8 character was
626 unnecessarily clumsy. I tidied it up.
627
628 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
629 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
630
631 17. Nobody had reported that the --include_dir option, which was added in
632 release 7.7 should have been called --include-dir (hyphen, not underscore)
633 for compatibility with GNU grep. I have changed it to --include-dir, but
634 left --include_dir as an undocumented synonym, and the same for
635 --exclude-dir, though that is not available in GNU grep, at least as of
636 release 2.5.4.
637
638 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
639 characters from a string of bytes) have been redefined so as not to use
640 loops, in order to improve performance in some environments. At the same
641 time, I abstracted some of the common code into auxiliary macros to save
642 repetition (this should not affect the compiled code).
643
644 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
645 compile-time error is now given if \c is not followed by an ASCII
646 character, that is, a byte less than 128. (In EBCDIC mode, the code is
647 different, and any byte value is allowed.)
648
649 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
650 START_OPTIMIZE option, which is now allowed at compile time - but just
651 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
652 to pcregrep and other applications that have no direct access to PCRE
653 options. The new /Y option in pcretest sets this option when calling
654 pcre_compile().
655
656 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
657 back references. Groups containing recursive back references were forced to
658 be atomic by that change, but in the case of named groups, the amount of
659 memory required was incorrectly computed, leading to "Failed: internal
660 error: code overflow". This has been fixed.
661
662 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
663 pcretest.c, to avoid build problems in some Borland environments.
664
665
666 Version 8.10 25-Jun-2010
667 ------------------------
668
669 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
670 THEN.
671
672 2. (*ACCEPT) was not working when inside an atomic group.
673
674 3. Inside a character class, \B is treated as a literal by default, but
675 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
676 causes the error). The code is unchanged, but I tidied the documentation.
677
678 4. Inside a character class, PCRE always treated \R and \X as literals,
679 whereas Perl faults them if its -w option is set. I have changed PCRE so
680 that it faults them when PCRE_EXTRA is set.
681
682 5. Added support for \N, which always matches any character other than
683 newline. (It is the same as "." when PCRE_DOTALL is not set.)
684
685 6. When compiling pcregrep with newer versions of gcc which may have
686 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
687 declared with attribute warn_unused_result" were given. Just casting the
688 result to (void) does not stop the warnings; a more elaborate fudge is
689 needed. I've used a macro to implement this.
690
691 7. Minor change to pcretest.c to avoid a compiler warning.
692
693 8. Added four artifical Unicode properties to help with an option to make
694 \s etc use properties (see next item). The new properties are: Xan
695 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
696
697 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
698 use Unicode properties. (*UCP) at the start of a pattern can be used to set
699 this option. Modified pcretest to add /W to test this facility. Added
700 REG_UCP to make it available via the POSIX interface.
701
702 10. Added --line-buffered to pcregrep.
703
704 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
705 studied, and the match started with a letter with a code point greater than
706 127 whose first byte was different to the first byte of the other case of
707 the letter, the other case of this starting letter was not recognized
708 (#976).
709
710 12. If a pattern that was studied started with a repeated Unicode property
711 test, for example, \p{Nd}+, there was the theoretical possibility of
712 setting up an incorrect bitmap of starting bytes, but fortunately it could
713 not have actually happened in practice until change 8 above was made (it
714 added property types that matched character-matching opcodes).
715
716 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
717 possible starting bytes for non-anchored patterns.
718
719 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
720 \R, and also a number of cases that involve Unicode properties, both
721 explicit and implicit when PCRE_UCP is set.
722
723 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
724 input, it could crash or give wrong results if characters with values
725 greater than 0xc0 were present in the subject string. (Detail: it assumed
726 UTF-8 input when processing these items.)
727
728 16. Added a lot of (int) casts to avoid compiler warnings in systems where
729 size_t is 64-bit (#991).
730
731 17. Added a check for running out of memory when PCRE is compiled with
732 --disable-stack-for-recursion (#990).
733
734 18. If the last data line in a file for pcretest does not have a newline on
735 the end, a newline was missing in the output.
736
737 19. The default pcre_chartables.c file recognizes only ASCII characters (values
738 less than 128) in its various bitmaps. However, there is a facility for
739 generating tables according to the current locale when PCRE is compiled. It
740 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
741 characters, are recognized by isspace() and therefore were getting set in
742 these tables, and indeed these tables seem to approximate to ISO 8859. This
743 caused a problem in UTF-8 mode when pcre_study() was used to create a list
744 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
745 which of course cannot start UTF-8 characters. I have changed the code so
746 that only real ASCII characters (less than 128) and the correct starting
747 bytes for UTF-8 encodings are set for characters greater than 127 when in
748 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
749 altogether.)
750
751 20. Added the /T option to pcretest so as to be able to run tests with non-
752 standard character tables, thus making it possible to include the tests
753 used for 19 above in the standard set of tests.
754
755 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
756 reference to a subpattern the other side of a comment that contains an
757 opening parenthesis caused either an internal compiling error, or a
758 reference to the wrong subpattern.
759
760
761 Version 8.02 19-Mar-2010
762 ------------------------
763
764 1. The Unicode data tables have been updated to Unicode 5.2.0.
765
766 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
767 configured.
768
769 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
770 original author of that file, following a query about its status.
771
772 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
773 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
774
775 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
776 quantifier applied to a forward-referencing subroutine call, could compile
777 incorrect code or give the error "internal error: previously-checked
778 referenced subpattern not found".
779
780 6. Both MS Visual Studio and Symbian OS have problems with initializing
781 variables to point to external functions. For these systems, therefore,
782 pcre_malloc etc. are now initialized to local functions that call the
783 relevant global functions.
784
785 7. There were two entries missing in the vectors called coptable and poptable
786 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
787 I've fixed the data, and added a kludgy way of testing at compile time that
788 the lengths are correct (equal to the number of opcodes).
789
790 8. Following on from 7, I added a similar kludge to check the length of the
791 eint vector in pcreposix.c.
792
793 9. Error texts for pcre_compile() are held as one long string to avoid too
794 much relocation at load time. To find a text, the string is searched,
795 counting zeros. There was no check for running off the end of the string,
796 which could happen if a new error number was added without updating the
797 string.
798
799 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
800
801 11. \K was not working if it appeared in an atomic group or in a group that
802 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
803 \K is "not well defined" if used in an assertion. PCRE now accepts it if
804 the assertion is positive, but not if it is negative.
805
806 12. Change 11 fortuitously reduced the size of the stack frame used in the
807 "match()" function of pcre_exec.c by one pointer. Forthcoming
808 implementation of support for (*MARK) will need an extra pointer on the
809 stack; I have reserved it now, so that the stack frame size does not
810 decrease.
811
812 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
813 item in branch that calls a recursion is a subroutine call - as in the
814 second branch in the above example - was incorrectly given the compile-
815 time error "recursive call could loop indefinitely" because pcre_compile()
816 was not correctly checking the subroutine for matching a non-empty string.
817
818 14. The checks for overrunning compiling workspace could trigger after an
819 overrun had occurred. This is a "should never occur" error, but it can be
820 triggered by pathological patterns such as hundreds of nested parentheses.
821 The checks now trigger 100 bytes before the end of the workspace.
822
823 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
824
825
826 Version 8.01 19-Jan-2010
827 ------------------------
828
829 1. If a pattern contained a conditional subpattern with only one branch (in
830 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
831 computed the wrong minimum data length (which is of course zero for such
832 subpatterns). This could cause incorrect "no match" results.
833
834 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
835 the pattern is reset in the first branch, pcre_compile() failed with
836 "internal error: code overflow at offset...". This happened only when
837 the reset was to the original external option setting. (An optimization
838 abstracts leading options settings into an external setting, which was the
839 cause of this.)
840
841 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
842 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
843 assertion pattern did not match (meaning that the assertion was true), it
844 was incorrectly treated as false if the SKIP had been reached during the
845 matching. This also applied to assertions used as conditions.
846
847 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
848 assertion subpattern, including such a pattern used as a condition,
849 unpredictable results occurred, instead of the error return
850 PCRE_ERROR_DFA_UITEM.
851
852 5. The C++ GlobalReplace function was not working like Perl for the special
853 situation when an empty string is matched. It now does the fancy magic
854 stuff that is necessary.
855
856 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
857 removed. (These were left over from very, very early versions of PCRE.)
858
859 7. Some cosmetic changes to the code to make life easier when compiling it
860 as part of something else:
861
862 (a) Change DEBUG to PCRE_DEBUG.
863
864 (b) In pcre_compile(), rename the member of the "branch_chain" structure
865 called "current" as "current_branch", to prevent a collision with the
866 Linux macro when compiled as a kernel module.
867
868 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
869 prevent a collision with the Linux macro when compiled as a kernel
870 module.
871
872 8. In pcre_compile() there are some checks for integer overflows that used to
873 cast potentially large values to (double). This has been changed to that
874 when building, a check for int64_t is made, and if it is found, it is used
875 instead, thus avoiding the use of floating point arithmetic. (There is no
876 other use of FP in PCRE.) If int64_t is not found, the fallback is to
877 double.
878
879 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
880 2005 (difference between two addresses compared to an unsigned value).
881
882 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
883 custom one, because of the following reported problem in Windows:
884
885 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
886 under Win32.
887 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
888 therefore missing the function definition.
889 - The compiler thus generates a "C" signature for the test function.
890 - The linker fails to find the "C" function.
891 - PCRE fails to configure if asked to do so against libbz2.
892
893 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
894 messages were output:
895
896 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
897 rerunning libtoolize, to keep the correct libtool macros in-tree.
898 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
899
900 I have done both of these things.
901
902 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
903 most of the time, it *can* run out if it is given a pattern that contains a
904 runaway infinite recursion. I updated the discussion in the pcrestack man
905 page.
906
907 13. Now that we have gone to the x.xx style of version numbers, the minor
908 version may start with zero. Using 08 or 09 is a bad idea because users
909 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
910 interpreted as invalid octal numbers. I've updated the previous comment in
911 configure.ac, and also added a check that gives an error if 08 or 09 are
912 used.
913
914 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
915 causing partial matching to fail when the end of the subject matched \W
916 in a UTF-8 pattern where \W was quantified with a minimum of 3.
917
918 15. There were some discrepancies between the declarations in pcre_internal.h
919 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
920 their definitions. The declarations used "const uschar *" and the
921 definitions used USPTR. Even though USPTR is normally defined as "const
922 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
923 reported that: "This difference in casting confuses some C++ compilers, for
924 example, SunCC recognizes above declarations as different functions and
925 generates broken code for hbpcre." I have changed the declarations to use
926 USPTR.
927
928 16. GNU libtool is named differently on some systems. The autogen.sh script now
929 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
930 (FreeBSD).
931
932 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
933 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
934 comment: "Figure out how to create a longlong from a string: strtoll and
935 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
936 instance, but it only takes 2 args instead of 3!"
937
938 18. A subtle bug concerned with back references has been fixed by a change of
939 specification, with a corresponding code fix. A pattern such as
940 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
941 refers, was giving matches when it shouldn't. For example, xa=xaaa would
942 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
943 same bug. Such groups have to be quantified to be useful, or contained
944 inside another quantified group. (If there's no repetition, the reference
945 can never match.) The problem arises because, having left the group and
946 moved on to the rest of the pattern, a later failure that backtracks into
947 the group uses the captured value from the final iteration of the group
948 rather than the correct earlier one. I have fixed this in PCRE by forcing
949 any group that contains a reference to itself to be an atomic group; that
950 is, there cannot be any backtracking into it once it has completed. This is
951 similar to recursive and subroutine calls.
952
953
954 Version 8.00 19-Oct-09
955 ----------------------
956
957 1. The table for translating pcre_compile() error codes into POSIX error codes
958 was out-of-date, and there was no check on the pcre_compile() error code
959 being within the table. This could lead to an OK return being given in
960 error.
961
962 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
963 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
964 in a Windows environment.
965
966 3. The pcregrep --count option prints the count for each file even when it is
967 zero, as does GNU grep. However, pcregrep was also printing all files when
968 --files-with-matches was added. Now, when both options are given, it prints
969 counts only for those files that have at least one match. (GNU grep just
970 prints the file name in this circumstance, but including the count seems
971 more useful - otherwise, why use --count?) Also ensured that the
972 combination -clh just lists non-zero counts, with no names.
973
974 4. The long form of the pcregrep -F option was incorrectly implemented as
975 --fixed_strings instead of --fixed-strings. This is an incompatible change,
976 but it seems right to fix it, and I didn't think it was worth preserving
977 the old behaviour.
978
979 5. The command line items --regex=pattern and --regexp=pattern were not
980 recognized by pcregrep, which required --regex pattern or --regexp pattern
981 (with a space rather than an '='). The man page documented the '=' forms,
982 which are compatible with GNU grep; these now work.
983
984 6. No libpcreposix.pc file was created for pkg-config; there was just
985 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
986
987 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
988 when UCP support is not needed, by modifying the Python script that
989 generates it from Unicode data files. This should not matter if the module
990 is correctly used as a library, but I received one complaint about 50K of
991 unwanted data. My guess is that the person linked everything into his
992 program rather than using a library. Anyway, it does no harm.
993
994 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
995 was a minimum greater than 1 for a wide character in a possessive
996 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
997 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
998 character. Chaos in the form of incorrect output or a compiling loop could
999 result.
1000
1001 9. The restrictions on what a pattern can contain when partial matching is
1002 requested for pcre_exec() have been removed. All patterns can now be
1003 partially matched by this function. In addition, if there are at least two
1004 slots in the offset vector, the offset of the earliest inspected character
1005 for the match and the offset of the end of the subject are set in them when
1006 PCRE_ERROR_PARTIAL is returned.
1007
1008 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1009 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1010 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1011 and may be more useful for multi-segment matching.
1012
1013 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1014 used to be given if ever the end of the subject was reached; now it is
1015 given only if matching could not proceed because another character was
1016 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1017 string "Z", which now yields "no match" instead of "partial match". In the
1018 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1019 final character ended with (*FAIL).
1020
1021 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1022 if the pattern had a "must contain" character that was already found in the
1023 earlier partial match, unless partial matching was again requested. For
1024 example, with the pattern /dog.(body)?/, the "must contain" character is
1025 "g". If the first part-match was for the string "dog", restarting with
1026 "sbody" failed. This bug has been fixed.
1027
1028 13. The string returned by pcre_dfa_exec() after a partial match has been
1029 changed so that it starts at the first inspected character rather than the
1030 first character of the match. This makes a difference only if the pattern
1031 starts with a lookbehind assertion or \b or \B (\K is not supported by
1032 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1033 matching functions compatible, and I think it's the right thing to do.
1034
1035 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1036 so that the demonstration program is easily available in environments where
1037 PCRE has not been installed from source.
1038
1039 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1040 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1041 library.
1042
1043 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1044 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1045 is not the first non-POSIX option to be added. Clearly some people find
1046 these options useful.
1047
1048 17. If a caller to the POSIX matching function regexec() passes a non-zero
1049 value for nmatch with a NULL value for pmatch, the value of
1050 nmatch is forced to zero.
1051
1052 18. RunGrepTest did not have a test for the availability of the -u option of
1053 the diff command, as RunTest does. It now checks in the same way as
1054 RunTest, and also checks for the -b option.
1055
1056 19. If an odd number of negated classes containing just a single character
1057 interposed, within parentheses, between a forward reference to a named
1058 subpattern and the definition of the subpattern, compilation crashed with
1059 an internal error, complaining that it could not find the referenced
1060 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1061 [The bug was that it was starting one character too far in when skipping
1062 over the character class, thus treating the ] as data rather than
1063 terminating the class. This meant it could skip too much.]
1064
1065 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1066 /g option in pcretest when the pattern contains \K, which makes it possible
1067 to have an empty string match not at the start, even when the pattern is
1068 anchored. Updated pcretest and pcredemo to use this option.
1069
1070 21. If the maximum number of capturing subpatterns in a recursion was greater
1071 than the maximum at the outer level, the higher number was returned, but
1072 with unset values at the outer level. The correct (outer level) value is
1073 now given.
1074
1075 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1076 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1077 make it do so. The string so far is captured, making this feature
1078 compatible with Perl.
1079
1080 23. The tests have been re-organized, adding tests 11 and 12, to make it
1081 possible to check the Perl 5.10 features against Perl 5.10.
1082
1083 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1084 pattern matches a fixed length string. PCRE did not allow this; now it
1085 does. Neither allows recursion.
1086
1087 25. I finally figured out how to implement a request to provide the minimum
1088 length of subject string that was needed in order to match a given pattern.
1089 (It was back references and recursion that I had previously got hung up
1090 on.) This code has now been added to pcre_study(); it finds a lower bound
1091 to the length of subject needed. It is not necessarily the greatest lower
1092 bound, but using it to avoid searching strings that are too short does give
1093 some useful speed-ups. The value is available to calling programs via
1094 pcre_fullinfo().
1095
1096 26. While implementing 25, I discovered to my embarrassment that pcretest had
1097 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1098 study optimizations had never been tested with that matching function.
1099 Oops. What is worse, even when it was passed study data, there was a bug in
1100 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1101 were also very few tests of studied patterns with pcre_dfa_exec().
1102
1103 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1104 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1105 on the other side of the coin, they are no longer allowed to have different
1106 names, because these cannot be distinguished in PCRE, and this has caused
1107 confusion. (This is a difference from Perl.)
1108
1109 28. When duplicate subpattern names are present (necessarily with different
1110 numbers, as required by 27 above), and a test is made by name in a
1111 conditional pattern, either for a subpattern having been matched, or for
1112 recursion in such a pattern, all the associated numbered subpatterns are
1113 tested, and the overall condition is true if the condition is true for any
1114 one of them. This is the way Perl works, and is also more like the way
1115 testing by number works.
1116
1117
1118 Version 7.9 11-Apr-09
1119 ---------------------
1120
1121 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1122 (pcretest), all targets were linked against these libraries. This included
1123 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1124 libraries. This caused unwanted dependencies to be created. This problem
1125 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1126 pcretest is linked with readline.
1127
1128 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1129 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1130 moved outside it again, because FALSE and TRUE are already defined in AIX,
1131 but BOOL is not.
1132
1133 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1134 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1135
1136 4. The pcregrep documentation said spaces were inserted as well as colons (or
1137 hyphens) following file names and line numbers when outputting matching
1138 lines. This is not true; no spaces are inserted. I have also clarified the
1139 wording for the --colour (or --color) option.
1140
1141 5. In pcregrep, when --colour was used with -o, the list of matching strings
1142 was not coloured; this is different to GNU grep, so I have changed it to be
1143 the same.
1144
1145 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1146 each matching line was coloured. Now it goes on to look for further matches
1147 of any of the test patterns, which is the same behaviour as GNU grep.
1148
1149 7. A pattern that could match an empty string could cause pcregrep to loop; it
1150 doesn't make sense to accept an empty string match in pcregrep, so I have
1151 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1152 seems to be how GNU grep behaves.
1153
1154 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1155 start or after a newline", because the conditional assertion was not being
1156 correctly handled. The rule now is that both the assertion and what follows
1157 in the first alternative must satisfy the test.
1158
1159 9. If auto-callout was enabled in a pattern with a conditional group whose
1160 condition was an assertion, PCRE could crash during matching, both with
1161 pcre_exec() and pcre_dfa_exec().
1162
1163 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1164 used for matching.
1165
1166 11. Unicode property support in character classes was not working for
1167 characters (bytes) greater than 127 when not in UTF-8 mode.
1168
1169 12. Added the -M command line option to pcretest.
1170
1171 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1172
1173 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1174
1175 16. Added comments and documentation about mis-use of no_arg in the C++
1176 wrapper.
1177
1178 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1179 from Martin Jerabek that uses macro names for all relevant character and
1180 string constants.
1181
1182 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1183 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1184 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1185 these, but not everybody uses configure.
1186
1187 19. A conditional group that had only one branch was not being correctly
1188 recognized as an item that could match an empty string. This meant that an
1189 enclosing group might also not be so recognized, causing infinite looping
1190 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1191 with the subject "ab", where knowledge that the repeated group can match
1192 nothing is needed in order to break the loop.
1193
1194 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1195 exec(), but without supplying a callout function, matching went wrong.
1196
1197 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1198 leak if the size of the offset vector was greater than 30. When the vector
1199 is smaller, the saved offsets during recursion go onto a local stack
1200 vector, but for larger vectors malloc() is used. It was failing to free
1201 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1202 error, in fact).
1203
1204 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1205 heapframe that is used only when UTF-8 support is enabled. This caused no
1206 problem, but was untidy.
1207
1208 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1209 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1210 included within another project.
1211
1212 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1213 slightly modified by me:
1214
1215 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1216 not building pcregrep.
1217
1218 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1219 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1220
1221 25. Forward references, both numeric and by name, in patterns that made use of
1222 duplicate group numbers, could behave incorrectly or give incorrect errors,
1223 because when scanning forward to find the reference group, PCRE was not
1224 taking into account the duplicate group numbers. A pattern such as
1225 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1226
1227 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1228 the feature of a custom pointer more persuasive (as requested by a user).
1229
1230 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1231 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1232 pre-defined.
1233
1234 28. Added support for (*UTF8) at the start of a pattern.
1235
1236 29. Arrange for flags added by the "release type" setting in CMake to be shown
1237 in the configuration summary.
1238
1239
1240 Version 7.8 05-Sep-08
1241 ---------------------
1242
1243 1. Replaced UCP searching code with optimized version as implemented for Ad
1244 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1245 stage table and inline lookup instead of a function, giving speed ups of 2
1246 to 5 times on some simple patterns that I tested. Permission was given to
1247 distribute the MultiStage2.py script that generates the tables (it's not in
1248 the tarball, but is in the Subversion repository).
1249
1250 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1251 scripts.
1252
1253 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1254 a group with a zero qualifier. The result of the study could be incorrect,
1255 or the function might crash, depending on the pattern.
1256
1257 4. Caseless matching was not working for non-ASCII characters in back
1258 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1259 It now works when Unicode Property Support is available.
1260
1261 5. In pcretest, an escape such as \x{de} in the data was always generating
1262 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1263 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1264 truncation.
1265
1266 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1267
1268 7. Added two (int) casts to pcregrep when printing the difference of two
1269 pointers, in case they are 64-bit values.
1270
1271 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1272 test 2 if it fails.
1273
1274 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1275 and a #define of that name to empty if it is not externally set. This is to
1276 allow users of MSVC to set it if necessary.
1277
1278 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1279 the convenience functions in the pcre_get.c source file.
1280
1281 11. An option change at the start of a pattern that had top-level alternatives
1282 could cause overwriting and/or a crash. This command provoked a crash in
1283 some environments:
1284
1285 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1286
1287 This potential security problem was recorded as CVE-2008-2371.
1288
1289 12. For a pattern where the match had to start at the beginning or immediately
1290 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1291 pcre_dfa_exec() could read past the end of the passed subject if there was
1292 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1293 pcretest so that it places the subject at the end of its malloc-ed buffer.
1294
1295 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1296 exec() might read past the end of the data buffer in UTF-8 mode.
1297
1298 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1299 the data contained the byte 0x85 as part of a UTF-8 character within its
1300 first line. This applied both to normal and DFA matching.
1301
1302 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1303 /^[^d]*?$/8 failed to match "abc".
1304
1305 16. Added a missing copyright notice to pcrecpp_internal.h.
1306
1307 17. Make it more clear in the documentation that values returned from
1308 pcre_exec() in ovector are byte offsets, not character counts.
1309
1310 18. Tidied a few places to stop certain compilers from issuing warnings.
1311
1312 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1313 supplied by Stefan Weber. I made a further small update for 7.8 because
1314 there is a change of source arrangements: the pcre_searchfuncs.c module is
1315 replaced by pcre_ucd.c.
1316
1317
1318 Version 7.7 07-May-08
1319 ---------------------
1320
1321 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1322 a string to a long long, pretend we don't even have a long long." This is
1323 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1324
1325 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1326 pre-7.6 versions, which defined a global no_arg variable instead of putting
1327 it in the RE class. (See also #8 below.)
1328
1329 3. Remove a line of dead code, identified by coverity and reported by Nuno
1330 Lopes.
1331
1332 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1333
1334 (1) The include/exclude patterns were being applied to the whole pathnames
1335 of files, instead of just to the final components.
1336
1337 (2) If there was more than one level of directory, the subdirectories were
1338 skipped unless they satisfied the include/exclude conditions. This is
1339 inconsistent with GNU grep (and could even be seen as contrary to the
1340 pcregrep specification - which I improved to make it absolutely clear).
1341 The action now is always to scan all levels of directory, and just
1342 apply the include/exclude patterns to regular files.
1343
1344 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1345 --exclude_dir in the tests to avoid scanning .svn directories.
1346
1347 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1348 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1349 doesn't support NULs in patterns.
1350
1351 7. Added some missing "const"s to declarations of static tables in
1352 pcre_compile.c and pcre_dfa_exec.c.
1353
1354 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1355 caused by fix #2 above. (Subsequently also a second patch to fix the
1356 first patch. And a third patch - this was a messy problem.)
1357
1358 9. Applied Craig's patch to remove the use of push_back().
1359
1360 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1361 matching function regexec().
1362
1363 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1364 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1365 references. PCRE supports relative numbers with this syntax (I don't think
1366 Oniguruma does).
1367
1368 12. Previously, a group with a zero repeat such as (...){0} was completely
1369 omitted from the compiled regex. However, this means that if the group
1370 was called as a subroutine from elsewhere in the pattern, things went wrong
1371 (an internal error was given). Such groups are now left in the compiled
1372 pattern, with a new opcode that causes them to be skipped at execution
1373 time.
1374
1375 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1376 to the way PCRE behaves:
1377
1378 (a) A lone ] character is dis-allowed (Perl treats it as data).
1379
1380 (b) A back reference to an unmatched subpattern matches an empty string
1381 (Perl fails the current match path).
1382
1383 (c) A data ] in a character class must be notated as \] because if the
1384 first data character in a class is ], it defines an empty class. (In
1385 Perl it is not possible to have an empty class.) The empty class []
1386 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1387 The negative empty class [^] matches any one character, independently
1388 of the DOTALL setting.
1389
1390 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1391 non-existent subpattern following a character class starting with ']' and
1392 containing () gave an internal compiling error instead of "reference to
1393 non-existent subpattern". Fortunately, when the pattern did exist, the
1394 compiled code was correct. (When scanning forwards to check for the
1395 existencd of the subpattern, it was treating the data ']' as terminating
1396 the class, so got the count wrong. When actually compiling, the reference
1397 was subsequently set up correctly.)
1398
1399 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1400 it was being rejected as not supported by pcre_dfa_exec(), even though
1401 other assertions are supported. I have made pcre_dfa_exec() support
1402 (*FAIL).
1403
1404 16. The implementation of 13c above involved the invention of a new opcode,
1405 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1406 cannot be changed at match time, I realized I could make a small
1407 improvement to matching performance by compiling OP_ALLANY instead of
1408 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1409 on the OP_ANY path.
1410
1411 17. Compiling pcretest on Windows with readline support failed without the
1412 following two fixes: (1) Make the unistd.h include conditional on
1413 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1414
1415 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1416 ncurses library to be included for pcretest when ReadLine support is
1417 requested, but also to allow for it to be overridden. This patch came from
1418 Daniel Bergström.
1419
1420 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1421 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1422 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1423 spotting this.
1424
1425
1426 Version 7.6 28-Jan-08
1427 ---------------------
1428
1429 1. A character class containing a very large number of characters with
1430 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1431 overflow.
1432
1433 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1434 HAVE_LONG_LONG is not defined.
1435
1436 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1437 bring them up to date and include new features. This patch includes:
1438
1439 - Fixed PH's badly added libz and libbz2 support.
1440 - Fixed a problem with static linking.
1441 - Added pcredemo. [But later removed - see 7 below.]
1442 - Fixed dftables problem and added an option.
1443 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1444 HAVE_LONG_LONG.
1445 - Added readline support for pcretest.
1446 - Added an listing of the option settings after cmake has run.
1447
1448 4. A user submitted a patch to Makefile that makes it easy to create
1449 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1450 Makefile.am that cause it to include this special target, without
1451 affecting anything else. Note that the same mingw target plus all
1452 the other distribution libraries and programs are now supported
1453 when configuring with CMake (see 6 below) instead of with
1454 Configure/Make.
1455
1456 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1457 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1458 exported in the Windows port". It has not yet been confirmed that the patch
1459 solves the problem, but it does no harm.
1460
1461 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1462 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1463 with CMake, and also correct the comment about stack recursion.
1464
1465 7. Remove the automatic building of pcredemo from the ./configure system and
1466 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1467 of a program that users should build themselves after PCRE is installed, so
1468 building it automatically is not really right. What is more, it gave
1469 trouble in some build environments.
1470
1471 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1472
1473
1474 Version 7.5 10-Jan-08
1475 ---------------------
1476
1477 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1478 values in parens when parsing an RE using the C++ wrapper."
1479
1480 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1481 Characters greater than 255 were excluded from the class instead of being
1482 included.
1483
1484 3. The same bug as (2) above applied to negated POSIX classes such as
1485 [:^space:].
1486
1487 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1488 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1489 I have changed it.
1490
1491 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1492 first named subpattern) and a construct such as (?&a) would reference the
1493 first named subpattern whose name started with "a" (in other words, the
1494 length check was missing). Both these problems are fixed. "Subpattern name
1495 expected" is now given for (?&) (a zero-length name), and this patch also
1496 makes it give the same error for \k'' (previously it complained that that
1497 was a reference to a non-existent subpattern).
1498
1499 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1500 this is right because (?- can be followed by option settings as well as by
1501 digits. I have, however, made the messages clearer.
1502
1503 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1504 than the number used in the conditional) now cause a compile-time error.
1505 This is actually not compatible with Perl, which accepts such patterns, but
1506 treats the conditional as always being FALSE (as PCRE used to), but it
1507 seems to me that giving a diagnostic is better.
1508
1509 8. Change "alphameric" to the more common word "alphanumeric" in comments
1510 and messages.
1511
1512 9. Fix two occurrences of "backslash" in comments that should have been
1513 "backspace".
1514
1515 10. Remove two redundant lines of code that can never be obeyed (their function
1516 was moved elsewhere).
1517
1518 11. The program that makes PCRE's Unicode character property table had a bug
1519 which caused it to generate incorrect table entries for sequences of
1520 characters that have the same character type, but are in different scripts.
1521 It amalgamated them into a single range, with the script of the first of
1522 them. In other words, some characters were in the wrong script. There were
1523 thirteen such cases, affecting characters in the following ranges:
1524
1525 U+002b0 - U+002c1
1526 U+0060c - U+0060d
1527 U+0061e - U+00612
1528 U+0064b - U+0065e
1529 U+0074d - U+0076d
1530 U+01800 - U+01805
1531 U+01d00 - U+01d77
1532 U+01d9b - U+01dbf
1533 U+0200b - U+0200f
1534 U+030fc - U+030fe
1535 U+03260 - U+0327f
1536 U+0fb46 - U+0fbb1
1537 U+10450 - U+1049d
1538
1539 12. The -o option (show only the matching part of a line) for pcregrep was not
1540 compatible with GNU grep in that, if there was more than one match in a
1541 line, it showed only the first of them. It now behaves in the same way as
1542 GNU grep.
1543
1544 13. If the -o and -v options were combined for pcregrep, it printed a blank
1545 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1546 does the same. The return code can be used to tell if there were any
1547 non-matching lines.
1548
1549 14. Added --file-offsets and --line-offsets to pcregrep.
1550
1551 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1552 infinitely looping recursion. The bug was that positive lookaheads were not
1553 being skipped when checking for a possible empty match (negative lookaheads
1554 and both kinds of lookbehind were skipped).
1555
1556 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1557 inclusion of <windows.h> to before rather than after the definition of
1558 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1559
1560 17. Specifying a possessive quantifier with a specific limit for a Unicode
1561 character property caused pcre_compile() to compile bad code, which led at
1562 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1563 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1564 caused the error; without that there was no problem.
1565
1566 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1567
1568 19. Added --enable-pcretest-libreadline.
1569
1570 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1571 RE::GlobalReplace(). As a result, the number of replacements returned was
1572 double what it should be. I removed one of the increments, but Craig sent a
1573 later patch that removed the other one (the right fix) and added unit tests
1574 that check the return values (which was not done before).
1575
1576 21. Several CMake things:
1577
1578 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1579 the names libpcre and libpcreposix, not just pcre and pcreposix.
1580
1581 (2) The above change means that pcretest and pcregrep are now correctly
1582 linked with the newly-built libraries, not previously installed ones.
1583
1584 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1585
1586 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1587 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1588 UTF-8 newline character). The key issue is that the pattern starts .*;
1589 this means that the match must be either at the beginning, or after a
1590 newline. The bug was in the code for advancing after a failed match and
1591 checking that the new position followed a newline. It was not taking
1592 account of UTF-8 characters correctly.
1593
1594 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1595 character classes. PCRE was not treating the sequence [:...:] as a
1596 character class unless the ... were all letters. Perl, however, seems to
1597 allow any characters between [: and :], though of course it rejects as
1598 unknown any "names" that contain non-letters, because all the known class
1599 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1600 for example, whereas PCRE did not - it did not recognize a POSIX character
1601 class. This seemed a bit dangerous, so the code has been changed to be
1602 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1603 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1604 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1605 Perl does, and where it didn't before.
1606
1607 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1608 Windows environments %n is disabled by default.
1609
1610
1611 Version 7.4 21-Sep-07
1612 ---------------------
1613
1614 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1615 means that a class such as [\s] counted as "explicit reference to CR or
1616 LF". That isn't really right - the whole point of the change was to try to
1617 help when there was an actual mention of one of the two characters. So now
1618 the change happens only if \r or \n (or a literal CR or LF) character is
1619 encountered.
1620
1621 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1622 of both had grown to the point where there were only 3 bits left.
1623 Fortunately, there was spare space in the data structure, and so I have
1624 moved the internal flags into a new 16-bit field to free up more option
1625 bits.
1626
1627 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1628 but did not set the internal JCHANGED flag - either of these is enough to
1629 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1630 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1631 start sets both bits.
1632
1633 4. Added options (at build time, compile time, exec time) to change \R from
1634 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1635
1636 5. doc/pcresyntax.html was missing from the distribution.
1637
1638 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1639 compatibility, even though it is no longer used.
1640
1641 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1642 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1643 windows.h file is present (where different names are used). [This was
1644 reversed later after testing - see 16 below.]
1645
1646 8. Changed all #include <config.h> to #include "config.h". There were also
1647 some further <pcre.h> cases that I changed to "pcre.h".
1648
1649 9. When pcregrep was used with the --colour option, it missed the line ending
1650 sequence off the lines that it output.
1651
1652 10. It was pointed out to me that arrays of string pointers cause lots of
1653 relocations when a shared library is dynamically loaded. A technique of
1654 using a single long string with a table of offsets can drastically reduce
1655 these. I have refactored PCRE in four places to do this. The result is
1656 dramatic:
1657
1658 Originally: 290
1659 After changing UCP table: 187
1660 After changing error message table: 43
1661 After changing table of "verbs" 36
1662 After changing table of Posix names 22
1663
1664 Thanks to the folks working on Gregex for glib for this insight.
1665
1666 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1667 unicode-properties was also set.
1668
1669 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1670
1671 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1672 checked only for CRLF.
1673
1674 14. Added casts to pcretest.c to avoid compiler warnings.
1675
1676 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1677
1678 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1679 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1680 entirely. This removes changes made in 7 above.
1681
1682 17. The CMake files have been updated, and there is now more information about
1683 building with CMake in the NON-UNIX-USE document.
1684
1685
1686 Version 7.3 28-Aug-07
1687 ---------------------
1688
1689 1. In the rejigging of the build system that eventually resulted in 7.1, the
1690 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1691 brackets there is not right, since it causes compilers to look for an
1692 installed pcre.h, not the version that is in the source that is being
1693 compiled (which of course may be different). I have changed it back to:
1694
1695 #include "pcre.h"
1696
1697 I have a vague recollection that the change was concerned with compiling in
1698 different directories, but in the new build system, that is taken care of
1699 by the VPATH setting the Makefile.
1700
1701 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1702 when the subject happened to end in the byte 0x85 (e.g. if the last
1703 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1704 characters but of course it shouldn't be taken as a newline when it is part
1705 of another character. The bug was that, for an unlimited repeat of . in
1706 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1707 characters when looking for a newline.
1708
1709 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1710
1711 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1712 in debug output.
1713
1714 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1715 long printing in the pcrecpp unittest when running under MinGW.
1716
1717 6. ESC_K was left out of the EBCDIC table.
1718
1719 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1720 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1721 limit also applies to "virtual nesting" when a pattern is recursive, and in
1722 this case 1000 isn't so big. I have been able to remove this limit at the
1723 expense of backing off one optimization in certain circumstances. Normally,
1724 when pcre_exec() would call its internal match() function recursively and
1725 immediately return the result unconditionally, it uses a "tail recursion"
1726 feature to save stack. However, when a subpattern that can match an empty
1727 string has an unlimited repetition quantifier, it no longer makes this
1728 optimization. That gives it a stack frame in which to save the data for
1729 checking that an empty string has been matched. Previously this was taken
1730 from the 1000-entry workspace that had been reserved. So now there is no
1731 explicit limit, but more stack is used.
1732
1733 8. Applied Daniel's patches to solve problems with the import/export magic
1734 syntax that is required for Windows, and which was going wrong for the
1735 pcreposix and pcrecpp parts of the library. These were overlooked when this
1736 problem was solved for the main library.
1737
1738 9. There were some crude static tests to avoid integer overflow when computing
1739 the size of patterns that contain repeated groups with explicit upper
1740 limits. As the maximum quantifier is 65535, the maximum group length was
1741 set at 30,000 so that the product of these two numbers did not overflow a
1742 32-bit integer. However, it turns out that people want to use groups that
1743 are longer than 30,000 bytes (though not repeat them that many times).
1744 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1745 made it possible to implement the integer overflow checks in a much more
1746 dynamic way, which I have now done. The artificial limitation on group
1747 length has been removed - we now have only the limit on the total length of
1748 the compiled pattern, which depends on the LINK_SIZE setting.
1749
1750 10. Fixed a bug in the documentation for get/copy named substring when
1751 duplicate names are permitted. If none of the named substrings are set, the
1752 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1753 empty string.
1754
1755 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1756 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1757 because the ] is interpreted as the first data character and the
1758 terminating ] is not found. PCRE has been made compatible with Perl in this
1759 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1760 cause memory overwriting.
1761
1762 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1763 string has been matched (to stop an infinite loop). It was not recognizing
1764 a conditional subpattern that could match an empty string if that
1765 subpattern was within another subpattern. For example, it looped when
1766 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1767 condition was not nested. This bug has been fixed.
1768
1769 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1770 past the start of the subject in the presence of bytes with the top bit
1771 set, for example "\x8aBCD".
1772
1773 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1774 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1775
1776 14. Optimized (?!) to (*FAIL).
1777
1778 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1779 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1780 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1781 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1782 does: it's just the validity check that is more restrictive.
1783
1784 16. Inserted checks for integer overflows during escape sequence (backslash)
1785 processing, and also fixed erroneous offset values for syntax errors during
1786 backslash processing.
1787
1788 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1789 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1790
1791 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1792 caused an overrun.
1793
1794 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1795 something other than just ASCII characters) inside a group that had an
1796 unlimited repeat caused a loop at compile time (while checking to see
1797 whether the group could match an empty string).
1798
1799 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1800 [\P{Any}] did so. (Error in the code for printing property names.)
1801
1802 21. An orphan \E inside a character class could cause a crash.
1803
1804 22. A repeated capturing bracket such as (A)? could cause a wild memory
1805 reference during compilation.
1806
1807 23. There are several functions in pcre_compile() that scan along a compiled
1808 expression for various reasons (e.g. to see if it's fixed length for look
1809 behind). There were bugs in these functions when a repeated \p or \P was
1810 present in the pattern. These operators have additional parameters compared
1811 with \d, etc, and these were not being taken into account when moving along
1812 the compiled data. Specifically:
1813
1814 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1815 length.
1816
1817 (b) An item such as \pL+ within a repeated group could cause crashes or
1818 loops.
1819
1820 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1821 "reference to non-existent subpattern" error.
1822
1823 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1824
1825 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1826 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1827
1828 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1829
1830 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1831 character were causing crashes (broken optimization).
1832
1833 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1834 \p or \P) caused a compile-time loop.
1835
1836 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1837 break. For example, the unstudied pattern [\r\n]A does not match the string
1838 "\r\nA" because change 7.0/46 below moves the current point on by two
1839 characters after failing to match at the start. However, the pattern \nA
1840 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1841 the same is true. There doesn't seem any very clean way out of this, but
1842 what I have chosen to do makes the common cases work: PCRE now takes note
1843 of whether there can be an explicit match for \r or \n anywhere in the
1844 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1845 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1846 pattern has explicit CR or LF references.
1847
1848 29. Added (*CR) etc for changing newline setting at start of pattern.
1849
1850
1851 Version 7.2 19-Jun-07
1852 ---------------------
1853
1854 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1855 which is apparently normally available under Windows.
1856
1857 2. Re-jig the pcregrep tests with different newline settings in an attempt
1858 to make them independent of the local environment's newline setting.
1859
1860 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1861
1862 4. Some of the "internals" tests were previously cut out when the link size
1863 was not 2, because the output contained actual offsets. The recent new
1864 "Z" feature of pcretest means that these can be cut out, making the tests
1865 usable with all link sizes.
1866
1867 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1868 stack recursion. This gives a massive performance boost under BSD, but just
1869 a small improvement under Linux. However, it saves one field in the frame
1870 in all cases.
1871
1872 6. Added more features from the forthcoming Perl 5.10:
1873
1874 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1875 recursion call. It refers to the nth most recently opened parentheses.
1876
1877 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1878 to be opened parentheses.
1879
1880 (c) Conditions that refer to capturing parentheses can be specified
1881 relatively, for example, (?(-2)... or (?(+3)...
1882
1883 (d) \K resets the start of the current match so that everything before
1884 is not part of it.
1885
1886 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1887
1888 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1889 reference syntax.
1890
1891 (g) (?| introduces a group in which the numbering of parentheses in each
1892 alternative starts with the same number.
1893
1894 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1895
1896 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1897 PCRE_INFO_JCHANGED.
1898
1899 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1900 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1901 for detecting groups that can match an empty string.
1902
1903 9. A pattern with a very large number of alternatives (more than several
1904 hundred) was running out of internal workspace during the pre-compile
1905 phase, where pcre_compile() figures out how much memory will be needed. A
1906 bit of new cunning has reduced the workspace needed for groups with
1907 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1908 workspace instead of running out of the 4096 that are available.
1909
1910 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1911
1912 11. Applied patch from Google to remove an optimization that didn't quite work.
1913 The report of the bug said:
1914
1915 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1916 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1917 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1918
1919 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1920 it matched the wrong number of bytes.
1921
1922
1923 Version 7.1 24-Apr-07
1924 ---------------------
1925
1926 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1927 that is more "standard", making use of automake and other Autotools. There
1928 is some re-arrangement of the files and adjustment of comments consequent
1929 on this.
1930
1931 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1932 for recursive directory scanning broke on some systems because the files
1933 are not scanned in any specific order and on different systems the order
1934 was different. A call to "sort" has been inserted into RunGrepTest for the
1935 approprate test as a short-term fix. In the longer term there may be an
1936 alternative.
1937
1938 3. I had an email from Eric Raymond about problems translating some of PCRE's
1939 man pages to HTML (despite the fact that I distribute HTML pages, some
1940 people do their own conversions for various reasons). The problems
1941 concerned the use of low-level troff macros .br and .in. I have therefore
1942 removed all such uses from the man pages (some were redundant, some could
1943 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1944 HTML has been updated to handle .nf/.fi and to complain if it encounters
1945 .br or .in.
1946
1947 4. Updated comments in configure.ac that get placed in config.h.in and also
1948 arranged for config.h to be included in the distribution, with the name
1949 config.h.generic, for the benefit of those who have to compile without
1950 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1951
1952 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1953 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1954 makevp.bat for the current PCRE, using the additional files
1955 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1956
1957 6. A Windows user reported a minor discrepancy with test 2, which turned out
1958 to be caused by a trailing space on an input line that had got lost in his
1959 copy. The trailing space was an accident, so I've just removed it.
1960
1961 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1962 that is needed.
1963
1964 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1965 as "const" (a) because they are and (b) because it helps the PHP
1966 maintainers who have recently made a script to detect big data structures
1967 in the php code that should be moved to the .rodata section. I remembered
1968 to update Builducptable as well, so it won't revert if ucptable.h is ever
1969 re-created.
1970
1971 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1972 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1973 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1974 support is not required. This saves 1.5-2K of code, which is important in
1975 some applications.
1976
1977 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1978 so as not to refer to the tables, even though these functions will never be
1979 called when UTF-8 support is disabled. Otherwise there are problems with a
1980 shared library.
1981
1982 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1983
1984 (a) It was defining its arguments as char * instead of void *.
1985
1986 (b) It was assuming that all moves were upwards in memory; this was true
1987 a long time ago when I wrote it, but is no longer the case.
1988
1989 The emulated memove() is provided for those environments that have neither
1990 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1991 is clearly not the case, as these two bugs were recently reported.
1992
1993 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1994 and Detrail to create the HTML documentation, the .txt form of the man
1995 pages, and it removes trailing spaces from listed files. It also creates
1996 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1997 case, it wraps all the #defines with #ifndefs. This script should be run
1998 before "make dist".
1999
2000 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2001 with Unicode property support.
2002
2003 (a) For a maximizing quantifier, if the two different cases of the
2004 character were of different lengths in their UTF-8 codings (there are
2005 some cases like this - I found 11), and the matching function had to
2006 back up over a mixture of the two cases, it incorrectly assumed they
2007 were both the same length.
2008
2009 (b) When PCRE was configured to use the heap rather than the stack for
2010 recursion during matching, it was not correctly preserving the data for
2011 the other case of a UTF-8 character when checking ahead for a match
2012 while processing a minimizing repeat. If the check also involved
2013 matching a wide character, but failed, corruption could cause an
2014 erroneous result when trying to check for a repeat of the original
2015 character.
2016
2017 13. Some tidying changes to the testing mechanism:
2018
2019 (a) The RunTest script now detects the internal link size and whether there
2020 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2021 values substituted by "configure". (The RunGrepTest script already did
2022 this for UTF-8.) The configure.ac script no longer substitutes the
2023 relevant variables.
2024
2025 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2026 with length and offset values. This means that the output is different
2027 for different internal link sizes. Test 2 is skipped for link sizes
2028 other than 2 because of this, bypassing the problem. Unfortunately,
2029 there was also a test in test 3 (the locale tests) that used /B and
2030 failed for link sizes other than 2. Rather than cut the whole test out,
2031 I have added a new /Z option to pcretest that replaces the length and
2032 offset values with spaces. This is now used to make test 3 independent
2033 of link size. (Test 2 will be tidied up later.)
2034
2035 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2036 segmentation fault instead of returning the appropriate error message.
2037
2038 15. In multiline mode when the newline sequence was set to "any", the pattern
2039 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2040 This doesn't seem right; it now treats the CRLF combination as the line
2041 ending, and so does not match in that case. It's only a pattern such as ^$
2042 that would hit this one: something like ^ABC$ would have failed after \r
2043 and then tried again after \r\n.
2044
2045 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2046 in an attempt to make files that differ only in their line terminators
2047 compare equal. This works on Linux.
2048
2049 17. Under certain error circumstances pcregrep might try to free random memory
2050 as it exited. This is now fixed, thanks to valgrind.
2051
2052 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2053 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2054 was because its rules for how to advance for /g after matching an empty
2055 string at the end of a line did not allow for this case. They now check for
2056 it specially.
2057
2058 20. pcretest is supposed to handle patterns and data of any length, by
2059 extending its buffers when necessary. It was getting this wrong when the
2060 buffer for a data line had to be extended.
2061
2062 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2063 CRLF as a newline sequence.
2064
2065 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2066 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2067 I have nevertheless tidied it up.
2068
2069 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2070
2071 24. Added a man page for pcre-config.
2072
2073
2074 Version 7.0 19-Dec-06
2075 ---------------------
2076
2077 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2078 moving to gcc 4.1.1.
2079
2080 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2081 sys/time.h, which is documented as needed for this function. It doesn't
2082 seem to matter on Linux, but it showed up on some releases of OS X.
2083
2084 3. It seems that there are systems where bytes whose values are greater than
2085 127 match isprint() in the "C" locale. The "C" locale should be the
2086 default when a C program starts up. In most systems, only ASCII printing
2087 characters match isprint(). This difference caused the output from pcretest
2088 to vary, making some of the tests fail. I have changed pcretest so that:
2089
2090 (a) When it is outputting text in the compiled version of a pattern, bytes
2091 other than 32-126 are always shown as hex escapes.
2092
2093 (b) When it is outputting text that is a matched part of a subject string,
2094 it does the same, unless a different locale has been set for the match
2095 (using the /L modifier). In this case, it uses isprint() to decide.
2096
2097 4. Fixed a major bug that caused incorrect computation of the amount of memory
2098 required for a compiled pattern when options that changed within the
2099 pattern affected the logic of the preliminary scan that determines the
2100 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2101 that the computed length was too small. The symptoms of this bug were
2102 either the PCRE error "internal error: code overflow" from pcre_compile(),
2103 or a glibc crash with a message such as "pcretest: free(): invalid next
2104 size (fast)". Examples of patterns that provoked this bug (shown in
2105 pcretest format) are:
2106
2107 /(?-x: )/x
2108 /(?x)(?-x: \s*#\s*)/
2109 /((?i)[\x{c0}])/8
2110 /(?i:[\x{c0}])/8
2111
2112 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2113 is now done differently.
2114
2115 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2116 wrapper classes; (b) implement a new function in the C++ scanner that is
2117 more efficient than the old way of doing things because it avoids levels of
2118 recursion in the regex matching; (c) add a paragraph to the documentation
2119 for the FullMatch() function.
2120
2121 6. The escape sequence \n was being treated as whatever was defined as
2122 "newline". Not only was this contrary to the documentation, which states
2123 that \n is character 10 (hex 0A), but it also went horribly wrong when
2124 "newline" was defined as CRLF. This has been fixed.
2125
2126 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2127 was being set to -1 for the "end of line" case (supposedly a value that no
2128 character can have). Though this value is never used (the check for end of
2129 line is "zero bytes in current character"), it caused compiler complaints.
2130 I've changed it to 0xffffffff.
2131
2132 8. In pcre_version.c, the version string was being built by a sequence of
2133 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2134 string (as it is for production releases) called a macro with an empty
2135 argument. The C standard says the result of this is undefined. The gcc
2136 compiler treats it as an empty string (which was what was wanted) but it is
2137 reported that Visual C gives an error. The source has been hacked around to
2138 avoid this problem.
2139
2140 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2141 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2142 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2143 of them did).
2144
2145 10. Originally, pcretest opened its input and output without "b"; then I was
2146 told that "b" was needed in some environments, so it was added for release
2147 5.0 to both the input and output. (It makes no difference on Unix-like
2148 systems.) Later I was told that it is wrong for the input on Windows. I've
2149 now abstracted the modes into two macros, to make it easier to fiddle with
2150 them, and removed "b" from the input mode under Windows.
2151
2152 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2153
2154 12. Added -help and --help to pcretest as an official way of being reminded
2155 of the options.
2156
2157 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2158 and pcrecpp.cc because they annoy compilers at high warning levels.
2159
2160 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2161
2162 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2163 scripts are not C programs :-) and which was not noticed because it works
2164 on Linux.
2165
2166 16. pcretest is supposed to handle any length of pattern and data line (as one
2167 line or as a continued sequence of lines) by extending its input buffer if
2168 necessary. This feature was broken for very long pattern lines, leading to
2169 a string of junk being passed to pcre_compile() if the pattern was longer
2170 than about 50K.
2171
2172 17. I have done a major re-factoring of the way pcre_compile() computes the
2173 amount of memory needed for a compiled pattern. Previously, there was code
2174 that made a preliminary scan of the pattern in order to do this. That was
2175 OK when PCRE was new, but as the facilities have expanded, it has become
2176 harder and harder to keep it in step with the real compile phase, and there
2177 have been a number of bugs (see for example, 4 above). I have now found a
2178 cunning way of running the real compile function in a "fake" mode that
2179 enables it to compute how much memory it would need, while actually only
2180 ever using a few hundred bytes of working memory and without too many
2181 tests of the mode. This should make future maintenance and development
2182 easier. A side effect of this work is that the limit of 200 on the nesting
2183 depth of parentheses has been removed (though this was never a serious
2184 limitation, I suspect). However, there is a downside: pcre_compile() now
2185 runs more slowly than before (30% or more, depending on the pattern). I
2186 hope this isn't a big issue. There is no effect on runtime performance.
2187
2188 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2189 newline (only possible for the last line of a file) and it was a
2190 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2191
2192 19. Added additional timing features to pcretest. (1) The -tm option now times
2193 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2194 separate command line item, by a number that specifies the number of
2195 repeats to use when timing. The default is 50000; this gives better
2196 precision, but takes uncomfortably long for very large patterns.
2197
2198 20. Extended pcre_study() to be more clever in cases where a branch of a
2199 subpattern has no definite first character. For example, (a*|b*)[cd] would
2200 previously give no result from pcre_study(). Now it recognizes that the
2201 first character must be a, b, c, or d.
2202
2203 21. There was an incorrect error "recursive call could loop indefinitely" if
2204 a subpattern (or the entire pattern) that was being tested for matching an
2205 empty string contained only one non-empty item after a nested subpattern.
2206 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2207 incorrectly, because the \d was being skipped in the check.
2208
2209 22. The pcretest program now has a new pattern option /B and a command line
2210 option -b, which is equivalent to adding /B to every pattern. This causes
2211 it to show the compiled bytecode, without the additional information that
2212 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2213 is the same as /B/I).
2214
2215 23. A new optimization is now able automatically to treat some sequences such
2216 as a*b as a*+b. More specifically, if something simple (such as a character
2217 or a simple class like \d) has an unlimited quantifier, and is followed by
2218 something that cannot possibly match the quantified thing, the quantifier
2219 is automatically "possessified".
2220
2221 24. A recursive reference to a subpattern whose number was greater than 39
2222 went wrong under certain circumstances in UTF-8 mode. This bug could also
2223 have affected the operation of pcre_study().
2224
2225 25. Realized that a little bit of performance could be had by replacing
2226 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2227
2228 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2229
2230 27. Possessive quantifiers such as a++ were previously implemented by turning
2231 them into atomic groups such as ($>a+). Now they have their own opcodes,
2232 which improves performance. This includes the automatically created ones
2233 from 23 above.
2234
2235 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2236 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2237 the first matched character to be a colon. This applied both to named and
2238 numbered groups.
2239
2240 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2241
2242 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2243 building PCRE on Windows easier, so I have included it in the distribution.
2244
2245 31. There is now a check in pcretest against a ridiculously large number being
2246 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2247 loop, the loop is abandoned.
2248
2249 32. Forward references to subpatterns in conditions such as (?(2)...) where
2250 subpattern 2 is defined later cause pcre_compile() to search forwards in
2251 the pattern for the relevant set of parentheses. This search went wrong
2252 when there were unescaped parentheses in a character class, parentheses
2253 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2254
2255 33. "Subroutine" calls and backreferences were previously restricted to
2256 referencing subpatterns earlier in the regex. This restriction has now
2257 been removed.
2258
2259 34. Added a number of extra features that are going to be in Perl 5.10. On the
2260 whole, these are just syntactic alternatives for features that PCRE had
2261 previously implemented using the Python syntax or my own invention. The
2262 other formats are all retained for compatibility.
2263
2264 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2265 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2266 also .NET compatible.
2267
2268 (b) A recursion or subroutine call to a named group can now be defined as
2269 (?&name) as well as (?P>name).
2270
2271 (c) A backreference to a named group can now be defined as \k<name> or
2272 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2273 5.10, are also .NET compatible.
2274
2275 (d) A conditional reference to a named group can now use the syntax
2276 (?(<name>) or (?('name') as well as (?(name).
2277
2278 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2279 groups (named and numbered) that are never evaluated inline, but can be
2280 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2281 is always false. There may be only one alternative in such a group.
2282
2283 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2284 as the simple (?(R). The condition is true only if the most recent
2285 recursion is that of the given number or name. It does not search out
2286 through the entire recursion stack.
2287
2288 (g) The escape \gN or \g{N} has been added, where N is a positive or
2289 negative number, specifying an absolute or relative reference.
2290
2291 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2292 some "unreachable code" warnings.
2293
2294 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2295 things, this adds five new scripts.
2296
2297 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2298 There were also incompatibilities regarding the handling of \Q..\E inside
2299 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2300 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2301
2302 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2303 matches an empty string, and forcibly breaks the loop. There were bugs in
2304 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2305 against aaaa the result was just "a" rather than "aaaa", for example. Two
2306 separate and independent bugs (that affected different cases) have been
2307 fixed.
2308
2309 39. Refactored the code to abolish the use of different opcodes for small
2310 capturing bracket numbers. This is a tidy that I avoided doing when I
2311 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2312 The new approach is not only tidier, it makes it possible to reduce the
2313 memory needed to fix the previous bug (38).
2314
2315 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2316 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2317 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2318 mode.
2319
2320 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2321 report.
2322
2323 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2324 copy construction and assignment in the C++ wrapper.
2325
2326 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2327 couple of bugs that could have given wrong results in the "--newline=crlf"
2328 case.
2329
2330 44. Added a number of casts and did some reorganization of signed/unsigned int
2331 variables following suggestions from Dair Grant. Also renamed the variable
2332 "this" as "item" because it is a C++ keyword.
2333
2334 45. Arranged for dftables to add
2335
2336 #include "pcre_internal.h"
2337
2338 to pcre_chartables.c because without it, gcc 4.x may remove the array
2339 definition from the final binary if PCRE is built into a static library and
2340 dead code stripping is activated.
2341
2342 46. For an unanchored pattern, if a match attempt fails at the start of a
2343 newline sequence, and the newline setting is CRLF or ANY, and the next two
2344 characters are CRLF, advance by two characters instead of one.
2345
2346
2347 Version 6.7 04-Jul-06
2348 ---------------------
2349
2350 1. In order to handle tests when input lines are enormously long, pcretest has
2351 been re-factored so that it automatically extends its buffers when
2352 necessary. The code is crude, but this _is_ just a test program. The
2353 default size has been increased from 32K to 50K.
2354
2355 2. The code in pcre_study() was using the value of the re argument before
2356 testing it for NULL. (Of course, in any sensible call of the function, it
2357 won't be NULL.)
2358
2359 3. The memmove() emulation function in pcre_internal.h, which is used on
2360 systems that lack both memmove() and bcopy() - that is, hardly ever -
2361 was missing a "static" storage class specifier.
2362
2363 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2364 containing an extended class (one that cannot be represented by a bitmap
2365 because it contains high-valued characters or Unicode property items, e.g.
2366 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2367 pattern, but PCRE should not loop if you do not (it no longer does).
2368 [Detail: two cases were found: (a) a repeated subpattern containing an
2369 extended class; (b) a recursive reference to a subpattern that followed a
2370 previous extended class. It wasn't skipping over the extended class
2371 correctly when UTF-8 mode was not set.]
2372
2373 5. A negated single-character class was not being recognized as fixed-length
2374 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2375 compile error "lookbehind assertion is not fixed length".
2376
2377 6. The RunPerlTest auxiliary script was showing an unexpected difference
2378 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2379 write a Perl script that can interpret lines of an input file either as
2380 byte characters or as UTF-8, which is what "perltest" was being required to
2381 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2382 can't do is switch easily at run time between having the "use utf8;" pragma
2383 or not. In the end, I fudged it by using the RunPerlTest script to insert
2384 "use utf8;" explicitly for the UTF-8 tests.
2385
2386 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2387 the end of the subject string, contrary to the documentation and to what
2388 Perl does. This was true of both matching functions. Now it matches only at
2389 the start of the subject and immediately after *internal* newlines.
2390
2391 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2392 a pointer to an int instead of a pointer to an unsigned long int. This
2393 caused problems on 64-bit systems.
2394
2395 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2396 instance of the 'standard' template library not being so standard".
2397
2398 10. There was no check on the number of named subpatterns nor the maximum
2399 length of a subpattern name. The product of these values is used to compute
2400 the size of the memory block for a compiled pattern. By supplying a very
2401 long subpattern name and a large number of named subpatterns, the size
2402 computation could be caused to overflow. This is now prevented by limiting
2403 the length of names to 32 characters, and the number of named subpatterns
2404 to 10,000.
2405
2406 11. Subpatterns that are repeated with specific counts have to be replicated in
2407 the compiled pattern. The size of memory for this was computed from the
2408 length of the subpattern and the repeat count. The latter is limited to
2409 65535, but there was no limit on the former, meaning that integer overflow
2410 could in principle occur. The compiled length of a repeated subpattern is
2411 now limited to 30,000 bytes in order to prevent this.
2412
2413 12. Added the optional facility to have named substrings with the same name.
2414
2415 13. Added the ability to use a named substring as a condition, using the
2416 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2417 are numbers (not recommended). Forward references are permitted.
2418
2419 14. Added forward references in named backreferences (if you see what I mean).
2420
2421 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2422 pattern could run off the end of the subject. For example, the pattern
2423 "(?s)(.{1,5})"8 did this with the subject "ab".
2424
2425 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2426 PCRE_CASELESS was set when matching characters that were quantified with ?
2427 or *.
2428
2429 17. A character class other than a single negated character that had a minimum
2430 but no maximum quantifier - for example [ab]{6,} - was not handled
2431 correctly by pce_dfa_exec(). It would match only one character.
2432
2433 18. A valid (though odd) pattern that looked like a POSIX character
2434 class but used an invalid character after [ (for example [[,abc,]]) caused
2435 pcre_compile() to give the error "Failed: internal error: code overflow" or
2436 in some cases to crash with a glibc free() error. This could even happen if
2437 the pattern terminated after [[ but there just happened to be a sequence of
2438 letters, a binary zero, and a closing ] in the memory that followed.
2439
2440 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2441 over the years. Originally (before any Unicode support), just the bottom 8
2442 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2443 output from "man perlunicode" includes this:
2444
2445 The regular expression compiler produces polymorphic opcodes. That
2446 is, the pattern adapts to the data and automatically switches to
2447 the Unicode character scheme when presented with Unicode data--or
2448 instead uses a traditional byte scheme when presented with byte
2449 data.
2450
2451 Sadly, a wide octal escape does not cause a switch, and in a string with
2452 no other multibyte characters, these octal escapes are treated as before.
2453 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2454 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2455 Unicode string.
2456
2457 I have not perpetrated such confusion in PCRE. Up till now, it took just
2458 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2459 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2460 translate to the appropriate multibyte character.
2461
2462 29. Applied some refactoring to reduce the number of warnings from Microsoft
2463 and Borland compilers. This has included removing the fudge introduced
2464 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2465 a warning about an unused variable.
2466
2467 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2468 characters since release 4.0, because Perl (from release 5.004) does not.
2469 [Or at least, is documented not to: some releases seem to be in conflict
2470 with the documentation.] However, when a pattern was studied with
2471 pcre_study() and all its branches started with \s, PCRE still included VT
2472 as a possible starting character. Of course, this did no harm; it just
2473 caused an unnecessary match attempt.
2474
2475 22. Removed a now-redundant internal flag bit that recorded the fact that case
2476 dependency changed within the pattern. This was once needed for "required
2477 byte" processing, but is no longer used. This recovers a now-scarce options
2478 bit. Also moved the least significant internal flag bit to the most-
2479 significant bit of the word, which was not previously used (hangover from
2480 the days when it was an int rather than a uint) to free up another bit for
2481 the future.
2482
2483 23. Added support for CRLF line endings as well as CR and LF. As well as the
2484 default being selectable at build time, it can now be changed at runtime
2485 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2486 specify that it is scanning data with non-default line endings.
2487
2488 24. Changed the definition of CXXLINK to make it agree with the definition of
2489 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2490
2491 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2492 recursions. This makes a big different to stack usage for some patterns.
2493
2494 26. If a subpattern containing a named recursion or subroutine reference such
2495 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2496 the space required for the compiled pattern went wrong and gave too small a
2497 value. Depending on the environment, this could lead to "Failed: internal
2498 error: code overflow at offset 49" or "glibc detected double free or
2499 corruption" errors.
2500
2501 27. Applied patches from Google (a) to support the new newline modes and (b) to
2502 advance over multibyte UTF-8 characters in GlobalReplace.
2503
2504 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2505 difference for some implementation of PCRE in some Windows version.
2506
2507 29. Added some extra testing facilities to pcretest:
2508
2509 \q<number> in a data line sets the "match limit" value
2510 \Q<number> in a data line sets the "match recursion limt" value
2511 -S <number> sets the stack size, where <number> is in megabytes
2512
2513 The -S option isn't available for Windows.
2514
2515
2516 Version 6.6 06-Feb-06
2517 ---------------------
2518
2519 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2520 in pcreposix.h. I have copied the definition from pcre.h.
2521
2522 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2523 because pcre.h is no longer a built file.
2524
2525 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2526 not normally included in the compiled code.
2527
2528
2529 Version 6.5 01-Feb-06
2530 ---------------------
2531
2532 1. When using the partial match feature with pcre_dfa_exec(), it was not
2533 anchoring the second and subsequent partial matches at the new starting
2534 point. This could lead to incorrect results. For example, with the pattern
2535 /1234/, partially matching against "123" and then "a4" gave a match.
2536
2537 2. Changes to pcregrep:
2538
2539 (a) All non-match returns from pcre_exec() were being treated as failures
2540 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2541 error message is output. Some extra information is given for the
2542 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2543 probably the only errors that are likely to be caused by users (by
2544 specifying a regex that has nested indefinite repeats, for instance).
2545 If there are more than 20 of these errors, pcregrep is abandoned.
2546
2547 (b) A binary zero was treated as data while matching, but terminated the
2548 output line if it was written out. This has been fixed: binary zeroes
2549 are now no different to any other data bytes.
2550
2551 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2552 used to set a locale for matching. The --locale=xxxx long option has
2553 been added (no short equivalent) to specify a locale explicitly on the
2554 pcregrep command, overriding the environment variables.
2555
2556 (d) When -B was used with -n, some line numbers in the output were one less
2557 than they should have been.
2558
2559 (e) Added the -o (--only-matching) option.
2560
2561 (f) If -A or -C was used with -c (count only), some lines of context were
2562 accidentally printed for the final match.
2563
2564 (g) Added the -H (--with-filename) option.
2565
2566 (h) The combination of options -rh failed to suppress file names for files
2567 that were found from directory arguments.
2568
2569 (i) Added the -D (--devices) and -d (--directories) options.
2570
2571 (j) Added the -F (--fixed-strings) option.
2572
2573 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2574
2575 (l) Added the --colo(u)r option.
2576
2577 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2578 is not present by default.
2579
2580 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2581 items such as (?R) or (?1), when the recursion could match a number of
2582 alternatives. If it matched one of the alternatives, but subsequently,
2583 outside the recursion, there was a failure, the code tried to back up into
2584 the recursion. However, because of the way PCRE is implemented, this is not
2585 possible, and the result was an incorrect result from the match.
2586
2587 In order to prevent this happening, the specification of recursion has
2588 been changed so that all such subpatterns are automatically treated as
2589 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2590
2591 4. I had overlooked the fact that, in some locales, there are characters for
2592 which isalpha() is true but neither isupper() nor islower() are true. In
2593 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2594 and ordfeminine) are like this. This affected the treatment of \w and \W
2595 when they appeared in character classes, but not when they appeared outside
2596 a character class. The bit map for "word" characters is now created
2597 separately from the results of isalnum() instead of just taking it from the
2598 upper, lower, and digit maps. (Plus the underscore character, of course.)
2599
2600 5. The above bug also affected the handling of POSIX character classes such as
2601 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2602 permanent tables. Instead, the bit maps for such a class were previously
2603 created as the appropriate unions of the upper, lower, and digit bitmaps.
2604 Now they are created by subtraction from the [[:word:]] class, which has
2605 its own bitmap.
2606
2607 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2608 It is created by subtracting the vertical space characters (\x09, \x0a,
2609 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2610 subtraction was done in the overall bitmap for a character class, meaning
2611 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2612 be recognized. This bug has been fixed.
2613
2614 7. Patches from the folks at Google:
2615
2616 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2617 real life, but is still worth protecting against".
2618
2619 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2620 regular expressions".
2621
2622 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2623 have it.
2624
2625 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2626 "configure" and the latter not, in order to fix a problem somebody had
2627 with compiling the Arg class on HP-UX.
2628
2629 (e) Improve the error-handling of the C++ wrapper a little bit.
2630
2631 (f) New tests for checking recursion limiting.
2632
2633 8. The pcre_memmove() function, which is used only if the environment does not
2634 have a standard memmove() function (and is therefore rarely compiled),
2635 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2636 returning a result (though PCRE never actually uses the result).
2637
2638 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2639 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2640 returned instead of calling malloc() with an overflowing number that would
2641 most likely cause subsequent chaos.
2642
2643 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2644
2645 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2646 with this option is matched, the nmatch and pmatch options of regexec() are
2647 ignored.
2648
2649 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2650 provided in case anyone wants to the the POSIX interface with UTF-8
2651 strings.
2652
2653 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2654 C++ linking (needed for some HP-UX environments).
2655
2656 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2657 (unused parameter) and in the pcre_printint() function (omitted "default"
2658 switch label when the default is to do nothing).
2659
2660 15. Added some code to make it possible, when PCRE is compiled as a C++
2661 library, to replace subject pointers for pcre_exec() with a smart pointer
2662 class, thus making it possible to process discontinuous strings.
2663
2664 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2665 much the same function. They were added by different people who were trying
2666 to make PCRE easy to compile on non-Unix systems. It has been suggested
2667 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2668 for compiling on Windows systems. I have therefore replaced it with
2669 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2670 defaults to "extern" for C or "extern C" for C++, which works fine on
2671 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2672 SCOPE with something explicit in config.h. In addition:
2673
2674 (a) pcreposix.h still had just "extern" instead of either of these macros;
2675 I have replaced it with PCRE_DATA_SCOPE.
2676
2677 (b) Functions such as _pcre_xclass(), which are internal to the library,
2678 but external in the C sense, all had PCRE_EXPORT in their definitions.
2679 This is apparently wrong for the Windows case, so I have removed it.
2680 (It makes no difference on Unix-like systems.)
2681
2682 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2683 of recursive calls to match(). This is different to MATCH_LIMIT because
2684 that limits the total number of calls to match(), not all of which increase
2685 the depth of recursion. Limiting the recursion depth limits the amount of
2686 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2687 when PCRE is compiled, and changed at run time. A patch from Google adds
2688 this functionality to the C++ interface.
2689
2690 18. Changes to the handling of Unicode character properties:
2691
2692 (a) Updated the table to Unicode 4.1.0.
2693
2694 (b) Recognize characters that are not in the table as "Cn" (undefined).
2695
2696 (c) I revised the way the table is implemented to a much improved format
2697 which includes recognition of ranges. It now supports the ranges that
2698 are defined in UnicodeData.txt, and it also amalgamates other
2699 characters into ranges. This has reduced the number of entries in the
2700 table from around 16,000 to around 3,000, thus reducing its size
2701 considerably. I realized I did not need to use a tree structure after
2702 all - a binary chop search is just as efficient. Having reduced the
2703 number of entries, I extended their size from 6 bytes to 8 bytes to
2704 allow for more data.
2705
2706 (d) Added support for Unicode script names via properties such as \p{Han}.
2707
2708 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2709 matching that character.
2710
2711 20. When matching a repeated Unicode property with a minimum greater than zero,
2712 (for example \pL{2,}), PCRE could look past the end of the subject if it
2713 reached it while seeking the minimum number of characters. This could
2714 happen only if some of the characters were more than one byte long, because
2715 there is a check for at least the minimum number of bytes.
2716
2717 21. Refactored the implementation of \p and \P so as to be more general, to
2718 allow for more different types of property in future. This has changed the
2719 compiled form incompatibly. Anybody with saved compiled patterns that use
2720 \p or \P will have to recompile them.
2721
2722 22. Added "Any" and "L&" to the supported property types.
2723
2724 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2725 but give a compile time error if the value is greater than 0xff.
2726
2727 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2728 accidentally not being installed or uninstalled.
2729
2730 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2731 made were to insert the current release number. This seemed silly, because
2732 it made things harder for people building PCRE on systems that don't run
2733 "configure". I have turned pcre.h into a distributed file, no longer built
2734 by "configure", with the version identification directly included. There is
2735 no longer a pcre.h.in file.
2736
2737 However, this change necessitated a change to the pcre-config script as
2738 well. It is built from pcre-config.in, and one of the substitutions was the
2739 release number. I have updated configure.ac so that ./configure now finds
2740 the release number by grepping pcre.h.
2741
2742 26. Added the ability to run the tests under valgrind.
2743
2744
2745 Version 6.4 05-Sep-05
2746 ---------------------
2747
2748 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2749 "--" to be printed when multiple files were scanned, even when none of the
2750 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2751 consider it to be a bug, and have restored the previous behaviour.
2752
2753 2. A couple of code tidies to get rid of compiler warnings.
2754
2755 3. The pcretest program used to cheat by referring to symbols in the library
2756 whose names begin with _pcre_. These are internal symbols that are not
2757 really supposed to be visible externally, and in some environments it is
2758 possible to suppress them. The cheating is now confined to including
2759 certain files from the library's source, which is a bit cleaner.
2760
2761 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2762 file's purpose clearer.
2763
2764 5. Reorganized pcre_ucp_findchar().
2765
2766
2767 Version 6.3 15-Aug-05
2768 ---------------------
2769
2770 1. The file libpcre.pc.in did not have general read permission in the tarball.
2771
2772 2. There were some problems when building without C++ support:
2773
2774 (a) If C++ support was not built, "make install" and "make test" still
2775 tried to test it.
2776
2777 (b) There were problems when the value of CXX was explicitly set. Some
2778 changes have been made to try to fix these, and ...
2779
2780 (c) --disable-cpp can now be used to explicitly disable C++ support.
2781
2782 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2783 backslash in a target when C++ was disabled. This confuses some
2784 versions of "make", apparently. Using an intermediate variable solves
2785 this. (Same for CPP_LOBJ.)
2786
2787 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2788 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2789 necessary on certain architectures.
2790
2791 4. Added a setting of -export-symbols-regex to the link command to remove
2792 those symbols that are exported in the C sense, but actually are local
2793 within the library, and not documented. Their names all begin with
2794 "_pcre_". This is not a perfect job, because (a) we have to except some
2795 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2796 available (and never for static libraries). I have made a note to try to
2797 find a way round (a) in the future.
2798
2799
2800 Version 6.2 01-Aug-05
2801 ---------------------
2802
2803 1. There was no test for integer overflow of quantifier values. A construction
2804 such as {1111111111111111} would give undefined results. What is worse, if
2805 a minimum quantifier for a parenthesized subpattern overflowed and became
2806 negative, the calculation of the memory size went wrong. This could have
2807 led to memory overwriting.
2808
2809 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2810
2811 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2812 operating environments where this matters.
2813
2814 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2815 PCRE options from within the C++ wrapper.
2816
2817 5. Named capturing subpatterns were not being correctly counted when a pattern
2818 was compiled. This caused two problems: (a) If there were more than 100
2819 such subpatterns, the calculation of the memory needed for the whole
2820 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2821 back references of the form \12, where the number was greater than 9, were
2822 not recognized as back references, even though there were sufficient
2823 previous subpatterns.
2824
2825 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2826 versions of gcc, e.g. 2.95.4.
2827
2828
2829 Version 6.1 21-Jun-05
2830 ---------------------
2831
2832 1. There was one reference to the variable "posix" in pcretest.c that was not
2833 surrounded by "#if !defined NOPOSIX".
2834
2835 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2836 the cross-check on the old pcre_info() function, for the benefit of the
2837 cut-down version of PCRE that is currently imported into Exim.
2838
2839 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2840 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2841 patterns that start (?i) but I don't think that matters. The use of (?i) is
2842 just an example; this all applies to the other options as well.
2843
2844 4. Since libtool seems to echo the compile commands it is issuing, the output
2845 from "make" can be reduced a bit by putting "@" in front of each libtool
2846 compile command.
2847
2848 5. Patch from the folks at Google for configure.in to be a bit more thorough
2849 in checking for a suitable C++ installation before trying to compile the
2850 C++ stuff. This should fix a reported problem when a compiler was present,
2851 but no suitable headers.
2852
2853 6. The man pages all had just "PCRE" as their title. I have changed them to
2854 be the relevant file name. I have also arranged that these names are
2855 retained in the file doc/pcre.txt, which is a concatenation in text format
2856 of all the man pages except the little individual ones for each function.
2857
2858 7. The NON-UNIX-USE file had not been updated for the different set of source
2859 files that come with release 6. I also added a few comments about the C++
2860 wrapper.
2861
2862
2863 Version 6.0 07-Jun-05
2864 ---------------------
2865
2866 1. Some minor internal re-organization to help with my DFA experiments.
2867
2868 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2869 didn't matter for the library itself when fully configured, but did matter
2870 when compiling without UCP support, or within Exim, where the ucp files are
2871 not imported.
2872
2873 3. Refactoring of the library code to split up the various functions into
2874 different source modules. The addition of the new DFA matching code (see
2875 below) to a single monolithic source would have made it really too
2876 unwieldy, quite apart from causing all the code to be include in a
2877 statically linked application, when only some functions are used. This is
2878 relevant even without the DFA addition now that patterns can be compiled in
2879 one application and matched in another.
2880
2881 The downside of splitting up is that there have to be some external
2882 functions and data tables that are used internally in different modules of
2883 the library but which are not part of the API. These have all had their
2884 names changed to start with "_pcre_" so that they are unlikely to clash
2885 with other external names.
2886
2887 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2888 a different (DFA) algorithm. Although it is slower than the original
2889 function, it does have some advantages for certain types of matching
2890 problem.
2891
2892 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2893 including restarting after a partial match.
2894
2895 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2896 defined when compiling for Windows was sent to me. I have put it into the
2897 code, though I have no means of testing or verifying it.
2898
2899 7. Added the pcre_refcount() auxiliary function.
2900
2901 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2902 match before or at the first newline in the subject string. In pcretest,
2903 the /f option on a pattern can be used to set this.
2904
2905 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2906 would behave wrongly. This has been present in PCRE since release 4.0.
2907
2908 10. A number of changes to the pcregrep command:
2909
2910 (a) Refactored how -x works; insert ^(...)$ instead of setting
2911 PCRE_ANCHORED and checking the length, in preparation for adding
2912 something similar for -w.
2913
2914 (b) Added the -w (match as a word) option.
2915
2916 (c) Refactored the way lines are read and buffered so as to have more
2917 than one at a time available.
2918
2919 (d) Implemented a pcregrep test script.
2920
2921 (e) Added the -M (multiline match) option. This allows patterns to match
2922 over several lines of the subject. The buffering ensures that at least
2923 8K, or the rest of the document (whichever is the shorter) is available
2924 for matching (and similarly the previous 8K for lookbehind assertions).
2925
2926 (f) Changed the --help output so that it now says
2927
2928 -w, --word-regex(p)
2929
2930 instead of two lines, one with "regex" and the other with "regexp"
2931 because that confused at least one person since the short forms are the
2932 same. (This required a bit of code, as the output is generated
2933 automatically from a table. It wasn't just a text change.)
2934
2935 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2936 option but starts with a hyphen. Could be a pattern or a path name
2937 starting with a hyphen, for instance.
2938
2939 (h) "-" can be given as a file name to represent stdin.
2940
2941 (i) When file names are being printed, "(standard input)" is used for
2942 the standard input, for compatibility with GNU grep. Previously
2943 "<stdin>" was used.
2944
2945 (j) The option --label=xxx can be used to supply a name to be used for
2946 stdin when file names are being printed. There is no short form.
2947
2948 (k) Re-factored the options decoding logic because we are going to add
2949 two more options that take data. Such options can now be given in four
2950 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2951
2952 (l) Added the -A, -B, and -C options for requesting that lines of context
2953 around matches be printed.
2954
2955 (m) Added the -L option to print the names of files that do not contain
2956 any matching lines, that is, the complement of -l.
2957
2958 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2959 continue to scan other files.
2960
2961 (o) The -s option was incorrectly implemented. For compatibility with other
2962 greps, it now suppresses the error message for a non-existent or non-
2963 accessible file (but not the return code). There is a new option called
2964 -q that suppresses the output of matching lines, which was what -s was
2965 previously doing.
2966
2967 (p) Added --include and --exclude options to specify files for inclusion
2968 and exclusion when recursing.
2969
2970 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2971 Hopefully, it now does.
2972
2973 12. Missing cast in pcre_study().
2974
2975 13. Added an "uninstall" target to the makefile.
2976
2977 14. Replaced "extern" in the function prototypes in Makefile.in with
2978 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2979 world, but is set differently for Windows.
2980
2981 15. Added a second compiling function called pcre_compile2(). The only
2982 difference is that it has an extra argument, which is a pointer to an
2983 integer error code. When there is a compile-time failure, this is set
2984 non-zero, in addition to the error test pointer being set to point to an
2985 error message. The new argument may be NULL if no error number is required
2986 (but then you may as well call pcre_compile(), which is now just a
2987 wrapper). This facility is provided because some applications need a
2988 numeric error indication, but it has also enabled me to tidy up the way
2989 compile-time errors are handled in the POSIX wrapper.
2990
2991 16. Added VPATH=.libs to the makefile; this should help when building with one
2992 prefix path and installing with another. (Or so I'm told by someone who
2993 knows more about this stuff than I do.)
2994
2995 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2996 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2997 match everything, including newlines. This is not POSIX-compatible, but
2998 somebody wanted the feature. From pcretest it can be activated by using
2999 both the P and the s flags.
3000
3001 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3002
3003 19. libpcre.pc was being incorrectly installed as executable.
3004
3005 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3006 it now also looks for '\r' so that it will work unmodified on Windows.
3007
3008 21. Added Google's contributed C++ wrapper to the distribution.
3009
3010 22. Added some untidy missing memory free() calls in pcretest, to keep
3011 Electric Fence happy when testing.
3012
3013
3014
3015 Version 5.0 13-Sep-04
3016 ---------------------
3017
3018 1. Internal change: literal characters are no longer packed up into items
3019 containing multiple characters in a single byte-string. Each character
3020 is now matched using a separate opcode. However, there may be more than one
3021 byte in the character in UTF-8 mode.
3022
3023 2. The pcre_callout_block structure has two new fields: pattern_position and
3024 next_item_length. These contain the offset in the pattern to the next match
3025 item, and its length, respectively.
3026
3027 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3028 insertion of callouts before each pattern item. Added the /C option to
3029 pcretest to make use of this.
3030
3031 4. On the advice of a Windows user, the lines
3032
3033 #if defined(_WIN32) || defined(WIN32)
3034 _setmode( _fileno( stdout ), 0x8000 );
3035 #endif /* defined(_WIN32) || defined(WIN32) */
3036
3037 have been added to the source of pcretest. This apparently does useful
3038 magic in relation to line terminators.
3039
3040 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3041 for the benefit of those environments where the "b" makes a difference.
3042
3043 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3044 to know about it. I have put a hack into configure.in that adds in code
3045 to set GCC=yes if CC=icc. This seems to end up at a point in the
3046 generated configure script that is early enough to affect the setting of
3047 compiler options, which is what is needed, but I have no means of testing
3048 whether it really works. (The user who reported this had patched the
3049 generated configure script, which of course I cannot do.)
3050
3051 LATER: After change 22 below (new libtool files), the configure script
3052 seems to know about icc (and also ecc). Therefore, I have commented out
3053 this hack in configure.in.
3054
3055 7. Added support for pkg-config (2 patches were sent in).
3056
3057 8. Negated POSIX character classes that used a combination of internal tables
3058 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3059 [[:^ascii]]. Typically, they would match almost any characters. The other
3060 POSIX classes were not broken in this way.
3061
3062 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3063 to find the match, as PCRE was deluded into thinking that the match had to
3064 start at the start point or following a newline. The same bug applied to
3065 patterns with negative forward assertions or any backward assertions
3066 preceding ".*" at the start, unless the pattern required a fixed first
3067 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3068
3069 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3070 starting at the last subject character, bytes beyond the end of the subject
3071 string were read.
3072
3073 11. Renamed the variable "class" as "classbits" to make life easier for C++
3074 users. (Previously there was a macro definition, but it apparently wasn't
3075 enough.)
3076
3077 12. Added the new field "tables" to the extra data so that tables can be passed
3078 in at exec time, or the internal tables can be re-selected. This allows
3079 a compiled regex to be saved and re-used at a later time by a different
3080 program that might have everything at different addresses.
3081
3082 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3083 -R library as well as a -L library.
3084
3085 14. The debugging options of pcretest (-d on the command line or D on a
3086 pattern) showed incorrect output for anything following an extended class
3087 that contained multibyte characters and which was followed by a quantifier.
3088
3089 15. Added optional support for general category Unicode character properties
3090 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3091 support. It adds about 90K to the size of the library. The meanings of the
3092 inbuilt class escapes such as \d and \s have NOT been changed.
3093
3094 16. Updated pcredemo.c to include calls to free() to release the memory for the
3095 compiled pattern.
3096
3097 17. The generated file chartables.c was being created in the source directory
3098 instead of in the building directory. This caused the build to fail if the
3099 source directory was different from the building directory, and was
3100 read-only.
3101
3102 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3103 file. No doubt somebody will tell me if they don't make sense... Also added
3104 Dan Mooney's comments about building on OpenVMS.
3105
3106 19. Added support for partial matching via the PCRE_PARTIAL option for
3107 pcre_exec() and the \P data escape in pcretest.
3108
3109 20. Extended pcretest with 3 new pattern features:
3110
3111 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3112 write the compiled pattern to the file whose name is "rest-of-line".
3113 This is a straight binary dump of the data, with the saved pointer to
3114 the character tables forced to be NULL. The study data, if any, is
3115 written too. After writing, pcretest reads a new pattern.
3116
3117 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3118 compiled pattern from the given file. There must not be any
3119 occurrences of "<" in the file name (pretty unlikely); if there are,
3120 pcretest will instead treat the initial "<" as a pattern delimiter.
3121 After reading in the pattern, pcretest goes on to read data lines as
3122 usual.
3123
3124 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3125 and 16-bit fields in a compiled pattern, to simulate a pattern that
3126 was compiled on a host of opposite endianness.
3127
3128 21. The pcre-exec() function can now cope with patterns that were compiled on
3129 hosts of opposite endianness, with this restriction:
3130
3131 As for any compiled expression that is saved and used later, the tables
3132 pointer field cannot be preserved; the extra_data field in the arguments
3133 to pcre_exec() should be used to pass in a tables address if a value
3134 other than the default internal tables were used at compile time.
3135
3136 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3137 now diagnosed as an error. Previously, most of the time, a negative number
3138 would have been treated as zero, but if in addition "ovector" was passed as
3139 NULL, a crash could occur.
3140
3141 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3142 new versions from the libtool 1.5 distribution (the last one is a copy of
3143 a file called libtool.m4). This seems to have fixed the need to patch
3144 "configure" to support Darwin 1.3 (which I used to do). However, I still
3145 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3146 workstation).
3147
3148 24. Changed the PCRE licence to be the more standard "BSD" licence.
3149
3150
3151 Version 4.5 01-Dec-03
3152 ---------------------
3153
3154 1. There has been some re-arrangement of the code for the match() function so
3155 that it can be compiled in a version that does not call itself recursively.
3156 Instead, it keeps those local variables that need separate instances for
3157 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3158 needs to "recurse". Keeping track of where control must go is done by means
3159 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3160 hide most of the details from the main code, and operates only if
3161 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3162 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3163 operating.
3164
3165 To make it easier for callers to provide specially tailored get/free
3166 functions for this usage, two new functions, pcre_stack_malloc, and
3167 pcre_stack_free, are used. They are always called in strict stacking order,
3168 and the size of block requested is always the same.
3169
3170 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3171 PCRE has been compiled to use the stack or the heap for recursion. The
3172 -C option of pcretest uses this to show which version is compiled.
3173
3174 A new data escape \S, is added to pcretest; it causes the amounts of store
3175 obtained and freed by both kinds of malloc/free at match time to be added
3176 to the output.
3177
3178 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3179 what's available on my current Linux desktop machine.
3180
3181 3. When matching a UTF-8 string, the test for a valid string at the start has
3182 been extended. If start_offset is not zero, PCRE now checks that it points
3183 to a byte that is the start of a UTF-8 character. If not, it returns
3184 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3185 this is necessary because there may be backward assertions in the pattern.
3186 When matching the same subject several times, it may save resources to use
3187 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3188
3189 4. The code for checking the validity of UTF-8 strings has been tightened so
3190 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3191 containing "overlong sequences".
3192
3193 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3194 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3195 should have been "&", but it just so happened that all the cases this let
3196 through by mistake were picked up later in the function.
3197
3198 6. I had used a variable called "isblank" - this is a C99 function, causing
3199 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3200
3201 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3202 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3203 so the version is shown; (c) stop "make test" echoing "./RunTest".
3204
3205 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3206
3207 9. The prototype for memmove() for systems that don't have it was using
3208 size_t, but the inclusion of the header that defines size_t was later. I've
3209 moved the #includes for the C headers earlier to avoid this.
3210
3211 10. Added some adjustments to the code to make it easier to compiler on certain
3212 special systems:
3213
3214 (a) Some "const" qualifiers were missing.
3215 (b) Added the macro EXPORT before all exported functions; by default this
3216 is defined to be empty.
3217 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3218 that it reads its output file name as an argument instead of writing
3219 to the standard output and assuming this can be redirected.
3220
3221 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3222 class containing characters with values greater than 255, PCRE compilation
3223 went into a loop.
3224
3225 12. A recursive reference to a subpattern that was within another subpattern
3226 that had a minimum quantifier of zero caused PCRE to crash. For example,
3227 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3228 recursion. If the recursively-called subpattern itself had a zero repeat,
3229 that was OK.
3230
3231 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3232 buffer into which it was copied (for escape processing) was still set at
3233 1024, so long lines caused crashes.
3234
3235 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3236 "internal error: code overflow...". This applied to any character class
3237 that was followed by a possessive quantifier.
3238
3239 15. Modified the Makefile to add libpcre.la as a prerequisite for
3240 libpcreposix.la because I was told this is needed for a parallel build to
3241 work.
3242
3243 16. If a pattern that contained .* following optional items at the start was
3244 studied, the wrong optimizing data was generated, leading to matching
3245 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3246 matching string must start with a or b or c. The correct conclusion for
3247 this pattern is that a match can start with any character.
3248
3249
3250 Version 4.4 13-Aug-03
3251 ---------------------
3252
3253 1. In UTF-8 mode, a character class containing characters with values between
3254 127 and 255 was not handled correctly if the compiled pattern was studied.
3255 In fixing this, I have also improved the studying algorithm for such
3256 classes (slightly).
3257
3258 2. Three internal functions had redundant arguments passed to them. Removal
3259 might give a very teeny performance improvement.
3260
3261 3. Documentation bug: the value of the capture_top field in a callout is *one
3262 more than* the number of the hightest numbered captured substring.
3263
3264 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3265 in incorrectly linking with a previously installed version. They now link
3266 explicitly with libpcre.la.
3267
3268 5. configure.in no longer needs to recognize Cygwin specially.
3269
3270 6. A problem in pcre.in for Windows platforms is fixed.
3271
3272 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3273 pcretest, it used to include the size of the study block as part of its
3274 output. Unfortunately, the structure contains a field that has a different
3275 size on different hardware architectures. This meant that the tests that
3276 showed this size failed. As the block is currently always of a fixed size,
3277 this information isn't actually particularly useful in pcretest output, so
3278 I have just removed it.
3279
3280 8. Three pre-processor statements accidentally did not start in column 1.
3281 Sadly, there are *still* compilers around that complain, even though
3282 standard C has not required this for well over a decade. Sigh.
3283
3284 9. In pcretest, the code for checking callouts passed small integers in the
3285 callout_data field, which is a void * field. However, some picky compilers
3286 complained about the casts involved for this on 64-bit systems. Now
3287 pcretest passes the address of the small integer instead, which should get
3288 rid of the warnings.
3289
3290 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3291 both compile and run time, and gives an error if an invalid UTF-8 sequence
3292 is found. There is a option for disabling this check in cases where the
3293 string is known to be correct and/or the maximum performance is wanted.
3294
3295 11. In response to a bug report, I changed one line in Makefile.in from
3296
3297 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3298 to
3299 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3300
3301 to look similar to other lines, but I have no way of telling whether this
3302 is the right thing to do, as I do not use Windows. No doubt I'll get told
3303 if it's wrong...
3304
3305
3306 Version 4.3 21-May-03
3307 ---------------------
3308
3309 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3310 Makefile.
3311
3312 2. Some refactoring to improve the quality of the code:
3313
3314 (i) The utf8_table... variables are now declared "const".
3315
3316 (ii) The code for \cx, which used the "case flipping" table to upper case
3317 lower case letters, now just substracts 32. This is ASCII-specific,
3318 but the whole concept of \cx is ASCII-specific, so it seems
3319 reasonable.
3320
3321 (iii) PCRE was using its character types table to recognize decimal and
3322 hexadecimal digits in the pattern. This is silly, because it handles
3323 only 0-9, a-f, and A-F, but the character types table is locale-
3324 specific, which means strange things might happen. A private
3325 table is now used for this - though it costs 256 bytes, a table is
3326 much faster than multiple explicit tests. Of course, the standard
3327 character types table is still used for matching digits in subject
3328 strings against \d.
3329
3330 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3331 ending in _t are). So I've renamed it as ESC_tee.
3332
3333 3. The first argument for regexec() in the POSIX wrapper should have been
3334 defined as "const".
3335
3336 4. Changed pcretest to use malloc() for its buffers so that they can be
3337 Electric Fenced for debugging.
3338
3339 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3340 to read one or more bytes before the start of the subject string. Often this
3341 had no effect on PCRE's behaviour, but in some circumstances it could
3342 provoke a segmentation fault.
3343
3344 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3345 to try to read one or more bytes before the start of the subject string.
3346
3347 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3348 UTF-8 support could misbehave in various ways if the subject string
3349 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3350 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3351 back over UTF-8 characters.)
3352
3353
3354 Version 4.2 14-Apr-03
3355 ---------------------
3356
3357 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3358
3359 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3360 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3361 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3362 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3363 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3364 and BUILD_EXEEXT
3365 Note: automatic setting of the BUILD variables is not yet working
3366 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3367 compile-time but not at link-time
3368 [LINK]: use for linking executables only
3369 make different versions for Windows and non-Windows
3370 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3371 libraries
3372 [LINK_FOR_BUILD]: new variable
3373 [OBJEXT]: use throughout
3374 [EXEEXT]: use throughout
3375 <winshared>: new target
3376 <wininstall>: new target
3377 <dftables.o>: use native compiler
3378 <dftables>: use native linker
3379 <install>: handle Windows platform correctly
3380 <clean>: ditto
3381 <check>: ditto
3382 copy DLL to top builddir before testing
3383
3384 As part of these changes, -no-undefined was removed again. This was reported
3385 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3386 in any case.
3387
3388 3. Some tidies to get rid of compiler warnings:
3389
3390 . In the match_data structure, match_limit was an unsigned long int, whereas
3391 match_call_count was an int. I've made them both unsigned long ints.
3392
3393 . In pcretest the fact that a const uschar * doesn't automatically cast to
3394 a void * provoked a warning.
3395
3396 . Turning on some more compiler warnings threw up some "shadow" variables
3397 and a few more missing casts.
3398
3399 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3400 option, a class that contained a single character with a value between 128
3401 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3402
3403 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3404 option, a class that contained several characters, but with at least one
3405 whose value was between 128 and 255 caused PCRE to crash.
3406
3407
3408 Version 4.1 12-Mar-03
3409 ---------------------
3410
3411 1. Compiling with gcc -pedantic found a couple of places where casts were
3412 needed, and a string in dftables.c that was longer than standard compilers are
3413 required to support.
3414
3415 2. Compiling with Sun's compiler found a few more places where the code could
3416 be tidied up in order to avoid warnings.
3417
3418 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3419 first of these names is deprecated in the latest Autoconf in favour of the name
3420 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3421 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3422 analogy I have changed it to CFLAGS_FOR_BUILD.
3423
3424 4. Added -no-undefined to the linking command in the Makefile, because this is
3425 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3426 linking step for the pcreposix library.
3427
3428 5. PCRE was failing to diagnose the case of two named groups with the same
3429 name.
3430
3431 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3432 literal character that is needed in the subject for a match, and scans along to
3433 ensure that it is present before embarking on the full matching process. This
3434 saves time in cases of nested unlimited repeats that are never going to match.
3435 Problem: the scan can take a lot of time if the subject is very long (e.g.
3436 megabytes), thus penalizing straightforward matches. It is now done only if the
3437 amount of subject to be scanned is less than 1000 bytes.
3438
3439 7. A lesser problem with the same optimization is that it was recording the
3440 first character of an anchored pattern as "needed", thus provoking a search
3441 right along the subject, even when the first match of the pattern was going to
3442 fail. The "needed" character is now not set for anchored patterns, unless it
3443 follows something in the pattern that is of non-fixed length. Thus, it still
3444 fulfils its original purpose of finding quick non-matches in cases of nested
3445 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3446
3447
3448 Version 4.0 17-Feb-03
3449 ---------------------
3450
3451 1. If a comment in an extended regex that started immediately after a meta-item
3452 extended to the end of string, PCRE compiled incorrect data. This could lead to
3453 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3454
3455 2. Moved to autoconf 2.53 and libtool 1.4.2.
3456
3457 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3458 the special perltest8 script is no longer needed - all the tests can be run
3459 from a single perltest script.
3460
3461 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3462 by \s. It has now been removed in PCRE. This means it isn't recognized as
3463 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3464 class [:space:] *does* include VT, thereby creating a mess.
3465
3466 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3467 space and tab.
3468
3469 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3470 its new features into the main test script, reducing the number of scripts.
3471
3472 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3473 were backward compatible, and made the (?i) apply to the whole pattern, as if
3474 /i were given. Now it behaves more logically, and applies the option setting
3475 only to what follows. PCRE has been changed to follow suit. However, if it
3476 finds options settings right at the start of the pattern, it extracts them into
3477 the global options, as before. Thus, they show up in the info data.
3478
3479 8. Added support for the \Q...\E escape sequence. Characters in between are
3480 treated as literals. This is slightly different from Perl in that $ and @ are
3481 also handled as literals inside the quotes. In Perl, they will cause variable
3482 interpolation. Note the following examples:
3483
3484 Pattern PCRE matches Perl matches
3485
3486 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3487 \Qabc\$xyz\E abc\$xyz abc\$xyz
3488 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3489
3490 For compatibility with Perl, \Q...\E sequences are recognized inside character
3491 classes as well as outside them.
3492
3493 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3494 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3495 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3496 signed/unsigned warnings.
3497
3498 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3499 option for pcretest, so I've replaced it by a simple function that does just
3500 that job.
3501
3502 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3503 "pcregrep -".
3504
3505 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3506 Java package. This provides some syntactic sugar for simple cases of what my
3507 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3508 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3509 item, you can use this simplified notation. Note that only makes sense with
3510 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3511 greediness, whatever the setting of the PCRE_UNGREEDY option.
3512
3513 13. A change of greediness default within a pattern was not taking effect at
3514 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3515 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3516 was abstracted outside.
3517
3518 14. PCRE now supports the \G assertion. It is true when the current matching
3519 position is at the start point of the match. This differs from \A when the
3520 starting offset is non-zero. Used with the /g option of pcretest (or similar
3521 code), it works in the same way as it does for Perl's /g option. If all
3522 alternatives of a regex begin with \G, the expression is anchored to the start
3523 match position, and the "anchored" flag is set in the compiled expression.
3524
3525 15. Some bugs concerning the handling of certain option changes within patterns
3526 have been fixed. These applied to options other than (?ims). For example,
3527 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3528 been the other way round. Some of this was related to change 7 above.
3529
3530 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3531 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3532 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3533 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3534
3535 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3536 mode. Unlike ".", it always matches newline, whatever the setting of
3537 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3538 assertions. Perl allows it, but it doesn't (in general) work because it can't
3539 calculate the length of the lookbehind. At least, that's the case for Perl
3540 5.8.0 - I've been told they are going to document that it doesn't work in
3541 future.
3542
3543 18. Added an error diagnosis for escapes that PCRE does not support: these are
3544 \L, \l, \N, \P, \p, \U, \u, and \X.
3545
3546 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3547 reading past the end of the pattern in cases such as /[abcd/.
3548
3549 20. PCRE was getting more memory than necessary for patterns with classes that
3550 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3551
3552 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3553 compiling PCRE for use with Virtual Pascal.
3554
3555 22. Small fix to the Makefile to make it work properly if the build is done
3556 outside the source tree.
3557
3558 23. Added a new extension: a condition to go with recursion. If a conditional
3559 subpattern starts with (?(R) the "true" branch is used if recursion has
3560 happened, whereas the "false" branch is used only at the top level.
3561
3562 24. When there was a very long string of literal characters (over 255 bytes
3563 without UTF support, over 250 bytes with UTF support), the computation of how
3564 much memory was required could be incorrect, leading to segfaults or other
3565 strange effects.
3566
3567 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3568 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3569 there was a subsequent back reference to those brackets. This meant that, for
3570 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3571 possible to check for precisely this case. All we can do is abandon the
3572 optimization if .* occurs inside capturing brackets when there are any back
3573 references whatsoever. (See below for a better fix that came later.)
3574
3575 26. The handling of the optimization for finding the first character of a
3576 non-anchored pattern, and for finding a character that is required later in the
3577 match were failing in some cases. This didn't break the matching; it just
3578 failed to optimize when it could. The way this is done has been re-implemented.
3579
3580 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3581
3582 28. Added a new feature that provides some of the functionality that Perl
3583 provides with (?{...}). The facility is termed a "callout". The way it is done
3584 in PCRE is for the caller to provide an optional function, by setting
3585 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3586 global variable. By default it is unset, which disables all calling out. To get
3587 the function called, the regex must include (?C) at appropriate points. This
3588 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3589 This provides a means of identifying different callout points. When PCRE
3590 reaches such a point in the regex, if pcre_callout has been set, the external
3591 function is called. It is provided with data in a structure called
3592 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3593 matching continues; if it returns a non-zero value, the match at the current
3594 point fails. However, backtracking will occur if possible. [This was changed
3595 later and other features added - see item 49 below.]
3596
3597 29. pcretest is upgraded to test the callout functionality. It provides a
3598 callout function that displays information. By default, it shows the start of
3599 the match and the current position in the text. There are some new data escapes
3600 to vary what happens:
3601
3602 \C+ in addition, show current contents of captured substrings
3603 \C- do not supply a callout function
3604 \C!n return 1 when callout number n is reached
3605 \C!n!m return 1 when callout number n is reached for the mth time
3606
3607 30. If pcregrep was called with the -l option and just a single file name, it
3608 output "<stdin>" if a match was found, instead of the file name.
3609
3610 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3611 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3612 pcre_exec(). This saves a malloc/free per call. The default value of
3613 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3614 when configuring.
3615
3616 32. The default maximum size of a compiled pattern is 64K. There have been a
3617 few cases of people hitting this limit. The code now uses macros to handle the
3618 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3619 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3620 configuring. Tests 2 and 5 work only with 2-byte links because they output
3621 debugging information about compiled patterns.
3622
3623 33. Internal code re-arrangements:
3624
3625 (a) Moved the debugging function for printing out a compiled regex into
3626 its own source file (printint.c) and used #include to pull it into
3627 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3628 separate copies.
3629
3630 (b) Defined the list of op-code names for debugging as a macro in
3631 internal.h so that it is next to the definition of the opcodes.
3632
3633 (c) Defined a table of op-code lengths for simpler skipping along compiled
3634 code. This is again a macro in internal.h so that it is next to the
3635 definition of the opcodes.
3636
3637 34. Added support for recursive calls to individual subpatterns, along the
3638 lines of Robin Houston's patch (but implemented somewhat differently).
3639
3640 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3641 allow it to read and process whole directories in Win32. This code was
3642 contributed by Lionel Fourquaux; it has not been tested by me.
3643
3644 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3645 used to name a group. Names consist of alphanumerics and underscores, and must
3646 be unique. Back references use the syntax (?P=name) and recursive calls use
3647 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3648 numbers. The function pcre_fullinfo() can be used after compilation to extract
3649 a name/number map. There are three relevant calls:
3650
3651 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3652 PCRE_INFO_NAMECOUNT yields the number of entries
3653 PCRE_INFO_NAMETABLE yields a pointer to the map.
3654
3655 The map is a vector of fixed-size entries. The size of each entry depends on
3656 the length of the longest name used. The first two bytes of each entry are the
3657 group number, most significant byte first. There follows the corresponding
3658 name, zero terminated. The names are in alphabetical order.
3659
3660 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3661 case instead of 255. Making it the same both with and without UTF-8 support
3662 means that the same test output works with both.
3663
3664 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3665 calling malloc() with a zero argument.
3666
3667 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3668 optimization. I've improved things by keeping a bitmap of backreferences with
3669 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3670 fact referenced, the optimization can be applied. It is unlikely that a
3671 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3672 the match to follow \n) will appear inside brackets with a number greater than
3673 31, but if it does, any back reference > 31 suppresses the optimization.
3674
3675 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3676 of disabling numbered capturing parentheses. Any opening parenthesis that is
3677 not followed by ? behaves as if it were followed by ?: but named parentheses
3678 can still be used for capturing (and they will acquire numbers in the usual
3679 way).
3680
3681 41. Redesigned the return codes from the match() function into yes/no/error so
3682 that errors can be passed back from deep inside the nested calls. A malloc
3683 failure while inside a recursive subpattern call now causes the
3684 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3685
3686 42. It is now possible to set a limit on the number of times the match()
3687 function is called in a call to pcre_exec(). This facility makes it possible to
3688 limit the amount of recursion and backtracking, though not in a directly
3689 obvious way, because the match() function is used in a number of different
3690 circumstances. The count starts from zero for each position in the subject
3691 string (for non-anchored patterns). The default limit is, for compatibility, a
3692 large number, namely 10 000 000. You can change this in two ways:
3693
3694 (a) When configuring PCRE before making, you can use --with-match-limit=n
3695 to set a default value for the compiled library.
3696
3697 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3698 a different value is set. See 45 below.
3699
3700 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3701
3702 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3703 of things that can be changed at compile time. The first argument specifies
3704 what is wanted and the second points to where the information is to be placed.
3705 The current list of available information is:
3706
3707 PCRE_CONFIG_UTF8
3708
3709 The output is an integer that is set to one if UTF-8 support is available;
3710 otherwise it is set to zero.
3711
3712 PCRE_CONFIG_NEWLINE
3713
3714 The output is an integer that it set to the value of the code that is used for
3715 newline. It is either LF (10) or CR (13).
3716
3717 PCRE_CONFIG_LINK_SIZE
3718
3719 The output is an integer that contains the number of bytes used for internal
3720 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3721
3722 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3723
3724 The output is an integer that contains the threshold above which the POSIX
3725 interface uses malloc() for output vectors. See item 31 above.
3726
3727 PCRE_CONFIG_MATCH_LIMIT
3728
3729 The output is an unsigned integer that contains the default limit of the number
3730 of match() calls in a pcre_exec() execution. See 42 above.
3731
3732 44. pcretest has been upgraded by the addition of the -C option. This causes it
3733 to extract all the available output from the new pcre_config() function, and to
3734 output it. The program then exits immediately.
3735
3736 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3737 order to support additional features. One way would have been to define
3738 pcre_exec2() (for example) with extra arguments, but this would not have been
3739 extensible, and would also have required all calls to the original function to
3740 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3741 is used for passing in "extra" data from pcre_study().
3742
3743 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3744 contains the following fields:
3745
3746 flags a bitmap indicating which of the following fields are set
3747 study_data opaque data from pcre_study()
3748 match_limit a way of specifying a limit on match() calls for a specific
3749 call to pcre_exec()
3750 callout_data data for callouts (see 49 below)
3751
3752 The flag bits are also defined in pcre.h, and are
3753
3754 PCRE_EXTRA_STUDY_DATA
3755 PCRE_EXTRA_MATCH_LIMIT
3756 PCRE_EXTRA_CALLOUT_DATA
3757
3758 The pcre_study() function now returns one of these new pcre_extra blocks, with
3759 the actual study data pointed to by the study_data field, and the
3760 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3761 before. That is, this change is entirely upwards-compatible and requires no
3762 change to existing code.
3763
3764 If you want to pass in additional data to pcre_exec(), you can either place it
3765 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3766 block.
3767
3768 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3769 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3770 times with different match limits, until it finds the minimum value needed for
3771 pcre_exec() to complete. The value is then output. This can be instructive; for
3772 most simple matches the number is quite small, but for pathological cases it
3773 gets very large very quickly.
3774
3775 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3776 returns the size of the data block pointed to by the study_data field in a
3777 pcre_extra block, that is, the value that was passed as the argument to
3778 pcre_malloc() when PCRE was getting memory in which to place the information
3779 created by pcre_study(). The fourth argument should point to a size_t variable.
3780 pcretest has been extended so that this information is shown after a successful
3781 pcre_study() call when information about the compiled regex is being displayed.
3782
3783 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3784 because what follows is always an absolute path. (Later: it turns out that this
3785 is more than cosmetic for MinGW, because it doesn't like empty path
3786 components.)
3787
3788 49. Some changes have been made to the callout feature (see 28 above):
3789
3790 (i) A callout function now has three choices for what it returns:
3791
3792 0 => success, carry on matching
3793 > 0 => failure at this point, but backtrack if possible
3794 < 0 => serious error, return this value from pcre_exec()
3795
3796 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3797 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3798 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3799 use by callout functions. It will never be used by PCRE itself.
3800
3801 (ii) The pcre_extra structure (see 45 above) has a void * field called
3802 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3803 pcre_callout_block structure has a field of the same name. The contents of
3804 the field passed in the pcre_extra structure are passed to the callout
3805 function in the corresponding field in the callout block. This makes it
3806 easier to use the same callout-containing regex from multiple threads. For
3807 testing, the pcretest program has a new data escape
3808
3809 \C*n pass the number n (may be negative) as callout_data
3810
3811 If the callout function in pcretest receives a non-zero value as
3812 callout_data, it returns that value.
3813
3814 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3815 there were some redundant $(CFLAGS) in commands that are now specified as
3816 $(LINK), which already includes $(CFLAGS).
3817
3818 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3819 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3820 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3821 one-byte characters throughout. Note that case-insensitive matching applies
3822 only to characters whose values are less than 256. PCRE doesn't support the
3823 notion of cases for higher-valued characters.
3824
3825 (i) A character class whose characters are all within 0-255 is handled as
3826 a bit map, and the map is inverted for negative classes. Previously, a
3827 character > 255 always failed to match such a class; however it should
3828 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3829
3830 (ii) A negated character class with a single character < 255 is coded as
3831 "not this character" (OP_NOT). This wasn't working properly when the test
3832 character was multibyte, either singly or repeated.
3833
3834 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3835 mode, for example: \x{100}{2,3}.
3836
3837 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3838 singly or repeated) now correctly test multibyte characters. However,
3839 PCRE doesn't recognize any characters with values greater than 255 as
3840 digits, spaces, or word characters. Such characters always match \D, \S,
3841 and \W, and never match \d, \s, or \w.
3842
3843 (v) Classes may now contain characters and character ranges with values
3844 greater than 255. For example: [ab\x{100}-\x{400}].
3845
3846 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3847 PCRE in UTF-8 mode.
3848
3849 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3850 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3851 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3852 value.)
3853
3854 53. The single man page has become too large. I have therefore split it up into
3855 a number of separate man pages. These also give rise to individual HTML pages;
3856 these are now put in a separate directory, and there is an index.html page that
3857 lists them all. Some hyperlinking between the pages has been installed.
3858
3859 54. Added convenience functions for handling named capturing parentheses.
3860
3861 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3862 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3863 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3864 are faulted.
3865
3866 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3867 calling configure. These values are used when compiling the dftables.c program
3868 which is run to generate the source of the default character tables. They
3869 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3870 you will need to set these values.
3871
3872 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3873
3874
3875 Version 3.9 02-Jan-02
3876 ---------------------
3877
3878 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3879
3880 2. If --disable-static was given, the building process failed when trying to
3881 build pcretest and pcregrep. (For some reason it was using libtool to compile
3882 them, which is not right, as they aren't part of the library.)
3883
3884
3885 Version 3.8 18-Dec-01
3886 ---------------------
3887
3888 1. The experimental UTF-8 code was completely screwed up. It was packing the
3889 bytes in the wrong order. How dumb can you get?
3890
3891
3892 Version 3.7 29-Oct-01
3893 ---------------------
3894
3895 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3896 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3897 this didn't happen under Solaris 8, where I normally test things.
3898
3899 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3900 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3901 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3902 this fix an hour or so after the initial 3.7 release.)
3903
3904
3905 Version 3.6 23-Oct-01
3906 ---------------------
3907
3908 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3909 offsets passed as NULL with zero offset count.
3910
3911 2. The config.guess and config.sub files had not been updated when I moved to
3912 the latest autoconf.
3913
3914
3915 Version 3.5 15-Aug-01
3916 ---------------------
3917
3918 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3919 had been forgotten.
3920
3921 2. By using declared but undefined structures, we can avoid using "void"
3922 definitions in pcre.h while keeping the internal definitions of the structures
3923 private.
3924
3925 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3926 user point of view, this means that both static and shared libraries are built
3927 by default, but this can be individually controlled. More of the work of
3928 handling this static/shared cases is now inside libtool instead of PCRE's make
3929 file.
3930
3931 4. The pcretest utility is now installed along with pcregrep because it is
3932 useful for users (to test regexs) and by doing this, it automatically gets
3933 relinked by libtool. The documentation has been turned into a man page, so
3934 there are now .1, .txt, and .html versions in /doc.
3935
3936 5. Upgrades to pcregrep:
3937 (i) Added long-form option names like gnu grep.
3938 (ii) Added --help to list all options with an explanatory phrase.
3939 (iii) Added -r, --recursive to recurse into sub-directories.
3940 (iv) Added -f, --file to read patterns from a file.
3941
3942 6. pcre_exec() was referring to its "code" argument before testing that
3943 argument for NULL (and giving an error if it was NULL).
3944
3945 7. Upgraded Makefile.in to allow for compiling in a different directory from
3946 the source directory.
3947
3948 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3949 options bits, the pointer it was passed was to an int instead of to an unsigned
3950 long int. This mattered only on 64-bit systems.
3951
3952 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3953 generated) instead of pcre.in, which it its source. Also made the same change
3954 in several of the .c files.
3955
3956 10. A new release of gcc defines printf() as a macro, which broke pcretest
3957 because it had an ifdef in the middle of a string argument for printf(). Fixed
3958 by using separate calls to printf().
3959
3960 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3961 script, to force use of CR or LF instead of \n in the source. On non-Unix
3962 systems, the value can be set in config.h.
3963
3964 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3965 absolute limit. Changed the text of the error message to make this clear, and
3966 likewise updated the man page.
3967
3968 13. The limit of 99 on the number of capturing subpatterns has been removed.
3969 The new limit is 65535, which I hope will not be a "real" limit.
3970
3971
3972 Version 3.4 22-Aug-00
3973 ---------------------
3974
3975 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3976
3977 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3978
3979
3980 Version 3.3 01-Aug-00
3981 ---------------------
3982
3983 1. If an octal character was given, but the value was greater than \377, it
3984 was not getting masked to the least significant bits, as documented. This could
3985 lead to crashes in some systems.
3986
3987 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3988 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3989
3990 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3991 These just pass their arguments on to (pcre_free)(), but they are provided
3992 because some uses of PCRE bind it to non-C systems that can call its functions,
3993 but cannot call free() or pcre_free() directly.
3994
3995 4. Add "make test" as a synonym for "make check". Corrected some comments in
3996 the Makefile.
3997
3998 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3999 Makefile.
4000
4001 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4002 command called pgrep for grepping around the active processes.
4003
4004 7. Added the beginnings of support for UTF-8 character strings.
4005
4006 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4007 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4008 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4009 out for the ar command.)
4010
4011
4012 Version 3.2 12-May-00
4013 ---------------------
4014
4015 This is purely a bug fixing release.
4016
4017 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4018 of ZA. This was just one example of several cases that could provoke this bug,
4019 which was introduced by change 9 of version 2.00. The code for breaking
4020 infinite loops after an iteration that matches an empty string was't working
4021 correctly.
4022
4023 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4024 when matched against abbab (for example). After matching an empty string, it
4025 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4026 caused it to match further down the string than it should.
4027
4028 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4029 was there because it doesn't seem to be needed, and it causes trouble on some
4030 systems, as it is not a Standard C header. It has been removed.
4031
4032 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4033 were reported on the Macintosh. The changes were from
4034
4035 while ((c = *(++ptr)) != 0 && c != '\n');
4036 to
4037 while ((c = *(++ptr)) != 0 && c != '\n') ;
4038
4039 Totally extraordinary, but if that's what it takes...
4040
4041 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4042 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4043 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4044 assumes the way PCRE uses memmove() (always moving upwards).
4045
4046 6. PCRE is being used in one environment where strchr() is not available. There
4047 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4048 faster code anyway.
4049
4050
4051 Version 3.1 09-Feb-00
4052 ---------------------
4053
4054 The only change in this release is the fixing of some bugs in Makefile.in for
4055 the "install" target:
4056
4057 (1) It was failing to install pcreposix.h.
4058
4059 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4060
4061
4062 Version 3.0 01-Feb-00
4063 ---------------------
4064
4065 1. Add support for the /+ modifier to perltest (to output $` like it does in
4066 pcretest).
4067
4068 2. Add support for the /g modifier to perltest.
4069
4070 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4071 matches null strings.
4072
4073 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4074 pattern. Perl treats empty patterns specially - it reuses the most recent
4075 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4076 effect.
4077
4078 5. The POSIX interface was broken in that it was just handing over the POSIX
4079 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4080 required a bigger vector, with some working space on the end. This means that
4081 the POSIX wrapper now has to get and free some memory, and copy the results.
4082
4083 6. Added some simple autoconf support, placing the test data and the
4084 documentation in separate directories, re-organizing some of the
4085 information files, and making it build pcre-config (a GNU standard). Also added
4086 libtool support for building PCRE as a shared library, which is now the
4087 default.
4088
4089 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4090 09 are not valid octal constants. Single digits will be used for minor values
4091 less than 10.
4092
4093 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4094 existing programs that set these in the POSIX interface can use PCRE without
4095 modification.
4096
4097 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4098 return all that pcre_info() returns, plus additional data. The pcre_info()
4099 function is retained for compatibility, but is considered to be obsolete.
4100
4101 10. Added experimental recursion feature (?R) to handle one common case that
4102 Perl 5.6 will be able to do with (?p{...}).
4103
4104 11. Added support for POSIX character classes like [:alpha:], which Perl is
4105 adopting.
4106
4107
4108 Version 2.08 31-Aug-99
4109 ----------------------
4110
4111 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4112 trying to match at the startoffset position, but instead was moving forward to
4113 the next newline as if a previous match had failed.
4114
4115 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4116 and could get into a loop if a null string was matched other than at the start
4117 of the subject.
4118
4119 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4120 be distinguished at compile time, and for completeness also added PCRE_DATE.
4121
4122 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4123 in GnuWin32 environments.
4124
4125
4126 Version 2.07 29-Jul-99
4127 ----------------------
4128
4129 1. The documentation is now supplied in plain text form and HTML as well as in
4130 the form of man page sources.
4131
4132 2. C++ compilers don't like assigning (void *) values to other pointer types.
4133 In particular this affects malloc(). Although there is no problem in Standard
4134 C, I've put in casts to keep C++ compilers happy.
4135
4136 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4137 should be (const char *).
4138
4139 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4140 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4141 However, I haven't made this a standard facility. The documentation doesn't
4142 mention it, and the Makefile doesn't support it.
4143
4144 5. The Makefile now contains an "install" target, with editable destinations at
4145 the top of the file. The pcretest program is not installed.
4146
4147 6. pgrep -V now gives the PCRE version number and date.
4148
4149 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4150 causing the entire string to be ignored, instead of just the last character.
4151
4152 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4153 non-matching string, it can take a very, very long time, even for strings of
4154 quite modest length, because of the nested recursion. PCRE now does better in
4155 some of these cases. It does this by remembering the last required literal
4156 character in the pattern, and pre-searching the subject to ensure it is present
4157 before running the real match. In other words, it applies a heuristic to detect
4158 some types of certain failure quickly, and in the above example, if presented
4159 with a string that has no trailing " it gives "no match" very quickly.
4160
4161 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4162 other alternatives are tried instead.
4163
4164
4165 Version 2.06 09-Jun-99
4166 ----------------------
4167
4168 1. Change pcretest's output for amount of store used to show just the code
4169 space, because the remainder (the data block) varies in size between 32-bit and
4170 64-bit systems.
4171
4172 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4173 start matching at. This allows lookbehinds to work when searching for multiple
4174 occurrences in a string.
4175
4176 3. Added additional options to pcretest for testing multiple occurrences:
4177
4178 /+ outputs the rest of the string that follows a match
4179 /g loops for multiple occurrences, using the new startoffset argument
4180 /G loops for multiple occurrences by passing an incremented pointer
4181
4182 4. PCRE wasn't doing the "first character" optimization for patterns starting
4183 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4184 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4185 the letter 'x'. On long subject strings, this gives a significant speed-up.
4186
4187
4188 Version 2.05 21-Apr-99
4189 ----------------------
4190
4191 1. Changed the type of magic_number from int to long int so that it works
4192 properly on 16-bit systems.
4193
4194 2. Fixed a bug which caused patterns starting with .* not to work correctly
4195 when the subject string contained newline characters. PCRE was assuming
4196 anchoring for such patterns in all cases, which is not correct because .* will
4197 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4198 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4199 must be retried after every newline in the subject.
4200
4201
4202 Version 2.04 18-Feb-99
4203 ----------------------
4204
4205 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4206 computation of the store needed to hold the pattern was incorrect (too large).
4207 If such patterns were nested a few deep, this could multiply and become a real
4208 problem.
4209
4210 2. Added /M option to pcretest to show the memory requirement of a specific
4211 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4212
4213 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4214 compiled in such a way that the backtracking after subsequent failure was
4215 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4216 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4217
4218
4219 Version 2.03 02-Feb-99
4220 ----------------------
4221
4222 1. Fixed typo and small mistake in man page.
4223
4224 2. Added 4th condition (GPL supersedes if conflict) and created separate
4225 LICENCE file containing the conditions.
4226
4227 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4228 Perl, that is the internal \ allows the delimiter to be included in the
4229 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4230 the final delimiter, add \ to the end of the pattern (to test the error).
4231
4232 4. Added the convenience functions for extracting substrings after a successful
4233 match. Updated pcretest to make it able to test these functions.
4234
4235
4236 Version 2.02 14-Jan-99
4237 ----------------------
4238
4239 1. Initialized the working variables associated with each extraction so that
4240 their saving and restoring doesn't refer to uninitialized store.
4241
4242 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4243 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4244 fix the problem.
4245
4246 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4247 calls, and wasn't printing the correct value for compiling calls. Increased the
4248 default value of LOOPREPEAT, and the number of significant figures in the
4249 times.
4250
4251 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4252
4253 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4254 a building problem on Windows NT with a FAT file system.
4255
4256
4257 Version 2.01 21-Oct-98
4258 ----------------------
4259
4260 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4261 to character tables built by pcre_maketables() in the current locale. If NULL
4262 is passed, the default tables are used.
4263
4264
4265 Version 2.00 24-Sep-98
4266 ----------------------
4267
4268 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4269 it any more.
4270
4271 2. Allow quantification of (?>) groups, and make it work correctly.
4272
4273 3. The first character computation wasn't working for (?>) groups.
4274
4275 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4276 end of the subject) and add 5.005's \z, which really does match only at the
4277 very end of the subject.
4278
4279 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4280
4281 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4282 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4283 localized options. All options to pcre_study() were also removed.
4284
4285 7. Add other new features from 5.005:
4286
4287 $(?<= positive lookbehind
4288 $(?<! negative lookbehind
4289 (?imsx-imsx) added the unsetting capability
4290 such a setting is global if at outer level; local otherwise
4291 (?imsx-imsx:) non-capturing groups with option setting
4292 (?(cond)re|re) conditional pattern matching
4293
4294 A backreference to itself in a repeated group matches the previous
4295 captured string.
4296
4297 8. General tidying up of studying (both automatic and via "study")
4298 consequential on the addition of new assertions.
4299
4300 9. As in 5.005, unlimited repeated groups that could match an empty substring
4301 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4302 runtime if any iteration does actually match an empty substring.
4303
4304 10. Include the RunTest script in the distribution.
4305
4306 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4307 discrepancies, some of which were old and were also with respect to 5.004. They
4308 have now been fixed.
4309
4310
4311 Version 1.09 28-Apr-98
4312 ----------------------
4313
4314 1. A negated single character class followed by a quantifier with a minimum
4315 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4316 program crashes, or just wrong answers. This did not apply to negated classes
4317 containing more than one character, or to minima other than one.
4318
4319
4320 Version 1.08 27-Mar-98
4321 ----------------------
4322
4323 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4324
4325 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4326 latter must appear before anything that relies on it in the pattern.
4327
4328
4329 Version 1.07 16-Feb-98
4330 ----------------------
4331
4332 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4333 repeat of a potentially empty string).
4334
4335
4336 Version 1.06 23-Jan-98
4337 ----------------------
4338
4339 1. Added Markus Oberhumer's little patches for C++.
4340
4341 2. Literal strings longer than 255 characters were broken.
4342
4343
4344 Version 1.05 23-Dec-97
4345 ----------------------
4346
4347 1. Negated character classes containing more than one character were failing if
4348 PCRE_CASELESS was set at run time.
4349
4350
4351 Version 1.04 19-Dec-97
4352 ----------------------
4353
4354 1. Corrected the man page, where some "const" qualifiers had been omitted.
4355
4356 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4357 input syntax.
4358
4359 3. Fixed memory leak which occurred when a regex with back references was
4360 matched with an offsets vector that wasn't big enough. The temporary memory
4361 that is used in this case wasn't being freed if the match failed.
4362
4363 4. Tidied pcretest to ensure it frees memory that it gets.
4364
4365 5. Temporary memory was being obtained in the case where the passed offsets
4366 vector was exactly big enough.
4367
4368 6. Corrected definition of offsetof() from change 5 below.
4369
4370 7. I had screwed up change 6 below and broken the rules for the use of
4371 setjmp(). Now fixed.
4372
4373
4374 Version 1.03 18-Dec-97
4375 ----------------------
4376
4377 1. A erroneous regex with a missing opening parenthesis was correctly
4378 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4379 on some systems.
4380
4381 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4382 it was reported that one broken compiler failed on the former because "code" is
4383 also an independent variable.
4384
4385 3. The erroneous regex a[]b caused an array overrun reference.
4386
4387 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4388 fail on data ending with that character. (It was going on too far, and checking
4389 the next character, typically a binary zero.) This was specific to the
4390 optimized code for single-character negative classes.
4391
4392 5. Added a contributed patch from the TIN world which does the following:
4393
4394 + Add an undef for memmove, in case the the system defines a macro for it.
4395
4396 + Add a definition of offsetof(), in case there isn't one. (I don't know
4397 the reason behind this - offsetof() is part of the ANSI standard - but
4398 it does no harm).
4399
4400 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4401 most of the places where whitespace preceded '#'. I have given up and
4402 allowed the remaining 2 cases to be at the margin.
4403
4404 + Rename some variables in pcre to eliminate shadowing. This seems very
4405 pedantic, but does no harm, of course.
4406
4407 6. Moved the call to setjmp() into its own function, to get rid of warnings
4408 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4409
4410 7. Constructs such as \d{8,} were compiling into the equivalent of
4411 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4412 outcome, but in this particular case used more store than had been allocated,
4413 which caused the bug to be discovered because it threw up an internal error.
4414
4415 8. The debugging code in both pcre and pcretest for outputting the compiled
4416 form of a regex was going wrong in the case of back references followed by
4417 curly-bracketed repeats.
4418
4419
4420 Version 1.02 12-Dec-97
4421 ----------------------
4422
4423 1. Typos in pcre.3 and comments in the source fixed.
4424
4425 2. Applied a contributed patch to get rid of places where it used to remove
4426 'const' from variables, and fixed some signed/unsigned and uninitialized
4427 variable warnings.
4428
4429 3. Added the "runtest" target to Makefile.
4430
4431 4. Set default compiler flag to -O2 rather than just -O.
4432
4433
4434 Version 1.01 19-Nov-97
4435 ----------------------
4436
4437 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4438 like /([ab]*)*/, that is, for classes with more than one character in them.
4439
4440 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4441 as /((?>a*))*/ (a PCRE_EXTRA facility).
4442
4443
4444 Version 1.00 18-Nov-97
4445 ----------------------
4446
4447 1. Added compile-time macros to support systems such as SunOS4 which don't have
4448 memmove() or strerror() but have other things that can be used instead.
4449
4450 2. Arranged that "make clean" removes the executables.
4451
4452
4453 Version 0.99 27-Oct-97
4454 ----------------------
4455
4456 1. Fixed bug in code for optimizing classes with only one character. It was
4457 initializing a 32-byte map regardless, which could cause it to run off the end
4458 of the memory it had got.
4459
4460 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4461
4462
4463 Version 0.98 22-Oct-97
4464 ----------------------
4465
4466 1. Fixed bug in code for handling temporary memory usage when there are more
4467 back references than supplied space in the ovector. This could cause segfaults.
4468
4469
4470 Version 0.97 21-Oct-97
4471 ----------------------
4472
4473 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4474
4475 2. Optimized negated single characters not to use a bit map.
4476
4477 3. Brought error texts together as macro definitions; clarified some of them;
4478 fixed one that was wrong - it said "range out of order" when it meant "invalid
4479 escape sequence".
4480
4481 4. Changed some char * arguments to const char *.
4482
4483 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4484
4485 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4486 pcretest.
4487
4488
4489 Version 0.96 16-Oct-97
4490 ----------------------
4491
4492 1. Added a simple "pgrep" utility to the distribution.
4493
4494 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4495 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4496 where "ddd" means "one or more decimal digits".
4497
4498 3. Fixed serious bug. If a pattern had a back reference, but the call to
4499 pcre_exec() didn't supply a large enough ovector to record the related
4500 identifying subpattern, the match always failed. PCRE now remembers the number
4501 of the largest back reference, and gets some temporary memory in which to save
4502 the offsets during matching if necessary, in order to ensure that
4503 backreferences always work.
4504
4505 4. Increased the compatibility with Perl in a number of ways:
4506
4507 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4508 to request this handling. The option can be set at compile or exec time.
4509
4510 (b) $ matches before a terminating newline by default; an option
4511 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4512 mode). The option can be set at compile or exec time.
4513
4514 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4515 the same as Perl's. If the decimal number it represents is less than 10
4516 or there aren't that many previous left capturing parentheses, an octal
4517 escape is read. Inside a character class, it's always an octal escape,
4518 even if it is a single digit.
4519
4520 (d) An escaped but undefined alphabetic character is taken as a literal,
4521 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4522 escapes.
4523
4524 (e) {0} is now permitted. (The previous item is removed from the compiled
4525 pattern).
4526
4527 5. Changed all the names of code files so that the basic parts are no longer
4528 than 10 characters, and abolished the teeny "globals.c" file.
4529
4530 6. Changed the handling of character classes; they are now done with a 32-byte
4531 bit map always.
4532
4533 7. Added the -d and /D options to pcretest to make it possible to look at the
4534 internals of compilation without having to recompile pcre.
4535
4536
4537 Version 0.95 23-Sep-97
4538 ----------------------
4539
4540 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4541 \x20 at the start of a run of normal characters. These were being treated as
4542 real characters, instead of the source characters being re-checked.
4543
4544
4545 Version 0.94 18-Sep-97
4546 ----------------------
4547
4548 1. The functions are now thread-safe, with the caveat that the global variables
4549 containing pointers to malloc() and free() or alternative functions are the
4550 same for all threads.
4551
4552 2. Get pcre_study() to generate a bitmap of initial characters for non-
4553 anchored patterns when this is possible, and use it if passed to pcre_exec().
4554
4555
4556 Version 0.93 15-Sep-97
4557 ----------------------
4558
4559 1. /(b)|(:+)/ was computing an incorrect first character.
4560
4561 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4562 but not actually doing anything yet.
4563
4564 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4565 as Perl does (e.g. [-az] or [az-]).
4566
4567 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4568 all possible positions.
4569
4570 5. Split up into different modules to avoid including unneeded functions in a
4571 compiled binary. However, compile and exec are still in one module. The "study"
4572 function is split off.
4573
4574 6. The character tables are now in a separate module whose source is generated
4575 by an auxiliary program - but can then be edited by hand if required. There are
4576 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4577 toupper() in the code.
4578
4579 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4580 make them global. Abolish the function for setting them, as the caller can now
4581 set them directly.
4582
4583
4584 Version 0.92 11-Sep-97
4585 ----------------------
4586
4587 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4588 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4589
4590 2. Caseless matching was not working in character classes if the characters in
4591 the pattern were in upper case.
4592
4593 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4594
4595 4. Make PCRE_ANCHORED public and accept as a compile option.
4596
4597 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4598 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4599 pass them.
4600
4601 6. Give an error if bad option bits passed at compile or run time.
4602
4603 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4604 pcretest to cause it to pass that flag.
4605
4606 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4607 options, and the first character, if set.
4608
4609 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4610
4611
4612 Version 0.91 10-Sep-97
4613 ----------------------
4614
4615 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4616 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4617
4618 2. PCRE was looping on encountering an indefinitely repeated back reference to
4619 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4620 Perl does - treats the match as successful.
4621
4622 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5