ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1653 - (show annotations)
Wed May 25 08:23:09 2016 UTC (4 days, 1 hour ago) by ph10
File size: 289434 byte(s)
Refactor to fix a typo.
1 ChangeLog for PCRE
2 ------------------
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
7 Version 8.39 21-May-2016
8 ------------------------
10 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
11 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
12 misbehaved. This bug was found by the LLVM fuzzer.
14 2. Similar to the above, if an isolated \E was present between an item and its
15 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
16 bug was found by the LLVM fuzzer.
18 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
19 working correctly in UCP mode.
21 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
22 was set when the pmatch argument was NULL. It now returns REG_INVARG.
24 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
26 6. An empty \Q\E sequence between an item and its qualifier caused
27 pcre_compile() to misbehave when auto callouts were enabled. This bug was
28 found by the LLVM fuzzer.
30 7. If a pattern that was compiled with PCRE_EXTENDED started with white
31 space or a #-type comment that was followed by (?-x), which turns off
32 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
33 pcre_compile() assumed that (?-x) applied to the whole pattern and
34 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
36 8. An call of pcre_copy_named_substring() for a named substring whose number
37 was greater than the space in the ovector could cause a crash.
39 9. Yet another buffer overflow bug involved duplicate named groups with a
40 group that reset capture numbers (compare 8.38/7 below). Once again, I have
41 just allowed for more memory, even if not needed. (A proper fix is
42 implemented in PCRE2, but it involves a lot of refactoring.)
44 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
45 start of the match to be earlier than the end.
47 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
49 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
50 assertion, caused pcretest to generate incorrect output, and also to read
51 uninitialized memory (detected by ASAN or valgrind).
53 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
54 nested set of parentheses of sufficient size caused an overflow of the
55 compiling workspace (which was diagnosed, but of course is not desirable).
57 14. And yet another buffer overflow bug involving duplicate named groups, this
58 time nested, with a nested back reference. Yet again, I have just allowed
59 for more memory, because anything more needs all the refactoring that has
60 been done for PCRE2. An example pattern that provoked this bug is:
61 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
62 registered as CVE-2016-1283.
64 15. pcretest went into a loop if global matching was requested with an ovector
65 size less than 2. It now gives an error message. This bug was found by
66 afl-fuzz.
68 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
69 ("assertion expected") when (?(?C) was not followed by an opening
70 parenthesis.
72 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
73 actually affect anything, by sheer luck.
75 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
76 static compilation.
78 19. Modified the RunTest script to incorporate a valgrind suppressions file so
79 that certain errors, provoked by the SSE2 instruction set when JIT is used,
80 are ignored.
82 20. A racing condition is fixed in JIT reported by Mozilla.
84 21. Minor code refactor to avoid "array subscript is below array bounds"
85 compiler warning.
87 22. Minor code refactor to avoid "left shift of negative number" warning.
89 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
90 UCP support.
92 24. Refactor to avoid compiler warnings in pcrecpp.cc.
94 25. Refactor to fix a typo in pcre_jit_test.c
97 Version 8.38 23-November-2015
98 -----------------------------
100 1. If a group that contained a recursive back reference also contained a
101 forward reference subroutine call followed by a non-forward-reference
102 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
103 compile correct code, leading to undefined behaviour or an internally
104 detected error. This bug was discovered by the LLVM fuzzer.
106 2. Quantification of certain items (e.g. atomic back references) could cause
107 incorrect code to be compiled when recursive forward references were
108 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
109 This bug was discovered by the LLVM fuzzer.
111 3. A repeated conditional group whose condition was a reference by name caused
112 a buffer overflow if there was more than one group with the given name.
113 This bug was discovered by the LLVM fuzzer.
115 4. A recursive back reference by name within a group that had the same name as
116 another group caused a buffer overflow. For example:
117 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
119 5. A forward reference by name to a group whose number is the same as the
120 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
121 a buffer overflow at compile time. This bug was discovered by the LLVM
122 fuzzer.
124 6. A lookbehind assertion within a set of mutually recursive subpatterns could
125 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
127 7. Another buffer overflow bug involved duplicate named groups with a
128 reference between their definition, with a group that reset capture
129 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
130 fixed by always allowing for more memory, even if not needed. (A proper fix
131 is implemented in PCRE2, but it involves more refactoring.)
133 8. There was no check for integer overflow in subroutine calls such as (?123).
135 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
136 being treated as a literal 'l' instead of causing an error.
138 10. There was a buffer overflow if pcre_exec() was called with an ovector of
139 size 1. This bug was found by american fuzzy lop.
141 11. If a non-capturing group containing a conditional group that could match
142 an empty string was repeated, it was not identified as matching an empty
143 string itself. For example: /^(?:(?(1)x|)+)+$()/.
145 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
146 \a and \e in test subject lines.
148 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
149 instead of the EBCDIC value.
151 14. The handling of \c in an EBCDIC environment has been revised so that it is
152 now compatible with the specification in Perl's perlebcdic page.
154 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
155 ASCII/Unicode. This has now been added to the list of characters that are
156 recognized as white space in EBCDIC.
158 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
159 error (correctly) when used outside a class, but did not give an error
160 within a class.
162 17. \h within a class was incorrectly compiled in EBCDIC environments.
164 18. A pattern with an unmatched closing parenthesis that contained a backward
165 assertion which itself contained a forward reference caused buffer
166 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
168 19. JIT should return with error when the compiled pattern requires more stack
169 space than the maximum.
171 20. A possessively repeated conditional group that could match an empty string,
172 for example, /(?(R))*+/, was incorrectly compiled.
174 21. Fix infinite recursion in the JIT compiler when certain patterns such as
175 /(?:|a|){100}x/ are analysed.
177 22. Some patterns with character classes involving [: and \\ were incorrectly
178 compiled and could cause reading from uninitialized memory or an incorrect
179 error diagnosis.
181 23. Pathological patterns containing many nested occurrences of [: caused
182 pcre_compile() to run for a very long time.
184 24. A conditional group with only one branch has an implicit empty alternative
185 branch and must therefore be treated as potentially matching an empty
186 string.
188 25. If (?R was followed by - or + incorrect behaviour happened instead of a
189 diagnostic.
191 26. Arrange to give up on finding the minimum matching length for overly
192 complex patterns.
194 27. Similar to (4) above: in a pattern with duplicated named groups and an
195 occurrence of (?| it is possible for an apparently non-recursive back
196 reference to become recursive if a later named group with the relevant
197 number is encountered. This could lead to a buffer overflow. Wen Guanxing
198 from Venustech ADLAB discovered this bug.
200 28. If pcregrep was given the -q option with -c or -l, or when handling a
201 binary file, it incorrectly wrote output to stdout.
203 29. The JIT compiler did not restore the control verb head in case of *THEN
204 control verbs. This issue was found by Karl Skomski with a custom LLVM
205 fuzzer.
207 30. Error messages for syntax errors following \g and \k were giving inaccurate
208 offsets in the pattern.
210 31. Added a check for integer overflow in conditions (?(<digits>) and
211 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
212 fuzzer.
214 32. Handling recursive references such as (?2) when the reference is to a group
215 later in the pattern uses code that is very hacked about and error-prone.
216 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
217 give an internal error if it is obvious that compiling has gone wrong.
219 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
220 This issue was found by Karl Skomski with a custom LLVM fuzzer.
222 34. The JIT compiler should restore the control chain for empty possessive
223 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
225 35. Match limit check added to JIT recursion. This issue was found by Karl
226 Skomski with a custom LLVM fuzzer.
228 36. Yet another case similar to 27 above has been circumvented by an
229 unconditional allocation of extra memory. This issue is fixed "properly" in
230 PCRE2 by refactoring the way references are handled. Wen Guanxing
231 from Venustech ADLAB discovered this bug.
233 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
234 with a custom LLVM fuzzer.
236 38. Fixed a corner case of range optimization in JIT.
238 39. An incorrect error "overran compiling workspace" was given if there were
239 exactly enough group forward references such that the last one extended
240 into the workspace safety margin. The next one would have expanded the
241 workspace. The test for overflow was not including the safety margin.
243 40. A match limit issue is fixed in JIT which was found by Karl Skomski
244 with a custom LLVM fuzzer.
246 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
247 work under Windows. (Why has it taken so long for anyone to notice?)
249 42. In a character class such as [\W\p{Any}] where both a negative-type escape
250 ("not a word character") and a property escape were present, the property
251 escape was being ignored.
253 43. Fix crash caused by very long (*MARK) or (*THEN) names.
255 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
256 by a single ASCII character in a class item, was incorrectly compiled in
257 UCP mode. The POSIX class got lost, but only if the single character
258 followed it.
260 45. [:punct:] in UCP mode was matching some characters in the range 128-255
261 that should not have been matched.
263 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
264 class, all characters with code points greater than 255 are in the class.
265 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
266 such as \w are turned into Unicode properties), wide characters were not
267 correctly handled, and could fail to match.
270 Version 8.37 28-April-2015
271 --------------------------
273 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
274 for those parentheses to be closed with whatever has been captured so far.
275 However, it was failing to mark any other groups between the hightest
276 capture so far and the currrent group as "unset". Thus, the ovector for
277 those groups contained whatever was previously there. An example is the
278 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
280 2. If an assertion condition was quantified with a minimum of zero (an odd
281 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
283 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
284 unrecognized modifier, a crash could occur.
286 4. An attempt to do global matching in pcretest with a zero-length ovector
287 caused a crash.
289 5. Fixed a memory leak during matching that could occur for a subpattern
290 subroutine call (recursive or otherwise) if the number of captured groups
291 that had to be saved was greater than ten.
293 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
294 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
295 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
297 7. A UTF pattern containing a "not" match of a non-ASCII character and a
298 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
300 8. When a pattern is compiled, it remembers the highest back reference so that
301 when matching, if the ovector is too small, extra memory can be obtained to
302 use instead. A conditional subpattern whose condition is a check on a
303 capture having happened, such as, for example in the pattern
304 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
305 setting the highest backreference number. This mattered only if pcre_exec()
306 was called with an ovector that was too small to hold the capture, and there
307 was no other kind of back reference (a situation which is probably quite
308 rare). The effect of the bug was that the condition was always treated as
309 FALSE when the capture could not be consulted, leading to a incorrect
310 behaviour by pcre_exec(). This bug has been fixed.
312 9. A reference to a duplicated named group (either a back reference or a test
313 for being set in a conditional) that occurred in a part of the pattern where
314 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
315 to be incorrectly calculated, leading to overwriting.
317 10. A mutually recursive set of back references such as (\2)(\1) caused a
318 segfault at study time (while trying to find the minimum matching length).
319 The infinite loop is now broken (with the minimum length unset, that is,
320 zero).
322 11. If an assertion that was used as a condition was quantified with a minimum
323 of zero, matching went wrong. In particular, if the whole group had
324 unlimited repetition and could match an empty string, a segfault was
325 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
326 assertions to be quantified, but not if they are being used as conditions,
327 so the above pattern is faulted by Perl. PCRE has now been changed so that
328 it also rejects such patterns.
330 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
331 failed to allow the zero-repeat case if pcre2_exec() was called with an
332 ovector too small to capture the group.
334 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
335 Red Hat Product Security:
337 (a) A crash if /K and /F were both set with the option to save the compiled
338 pattern.
340 (b) Another crash if the option to print captured substrings in a callout
341 was combined with setting a null ovector, for example \O\C+ as a subject
342 string.
344 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
345 forward reference repeated a large (but limited) number of times within a
346 repeated outer group that has a zero minimum quantifier, caused incorrect
347 code to be compiled, leading to the error "internal error:
348 previously-checked referenced subpattern not found" when an incorrect
349 memory address was read. This bug was reported as "heap overflow",
350 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
351 CVE-2015-2325.
353 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
354 call within a group that also contained a recursive back reference caused
355 incorrect code to be compiled. This bug was reported as "heap overflow",
356 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
357 number CVE-2015-2326.
359 24. Computing the size of the JIT read-only data in advance has been a source
360 of various issues, and new ones are still appear unfortunately. To fix
361 existing and future issues, size computation is eliminated from the code,
362 and replaced by on-demand memory allocation.
364 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
365 adjacent to the end of the range, and the range contained characters with
366 more than one other case, caused incorrect behaviour when compiled in UTF
367 mode. In that example, the range a-j was left out of the class.
369 26. Fix JIT compilation of conditional blocks, which assertion
370 is converted to (*FAIL). E.g: /(?(?!))/.
372 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
373 discovered by the LLVM fuzzer.
375 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
376 when this assertion was used as a condition, for example (?(?!)a|b). In
377 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
378 error about an unsupported item.
380 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
381 possessification code could take exponential time to complete. A recursion
382 depth limit of 1000 has been imposed to limit the resources used by this
383 optimization.
385 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
386 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
387 because \S ensures they are all in the class. The code for doing this was
388 interacting badly with the code for computing the amount of space needed to
389 compile the pattern, leading to a buffer overflow. This bug was discovered
390 by the LLVM fuzzer.
392 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
393 other kinds of group caused stack overflow at compile time. This bug was
394 discovered by the LLVM fuzzer.
396 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
397 between a subroutine call and its quantifier was incorrectly compiled,
398 leading to buffer overflow or other errors. This bug was discovered by the
399 LLVM fuzzer.
401 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
402 assertion after (?(. The code was failing to check the character after
403 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
404 was discovered by the LLVM fuzzer.
406 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
407 a fixed maximum following a group that contains a subroutine reference was
408 incorrectly compiled and could trigger buffer overflow. This bug was
409 discovered by the LLVM fuzzer.
411 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
412 caused a stack overflow instead of the diagnosis of a non-fixed length
413 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
415 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
416 (e.g. /(?<=\Ka)/) could make pcregrep loop.
418 37. There was a similar problem to 36 in pcretest for global matches.
420 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
421 and a subsequent item in the pattern caused a non-match, backtracking over
422 the repeated \X did not stop, but carried on past the start of the subject,
423 causing reference to random memory and/or a segfault. There were also some
424 other cases where backtracking after \C could crash. This set of bugs was
425 discovered by the LLVM fuzzer.
427 39. The function for finding the minimum length of a matching string could take
428 a very long time if mutual recursion was present many times in a pattern,
429 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
430 method has been implemented. This infelicity was discovered by the LLVM
431 fuzzer.
433 40. Static linking against the PCRE library using the pkg-config module was
434 failing on missing pthread symbols.
437 Version 8.36 26-September-2014
438 ------------------------------
440 1. Got rid of some compiler warnings in the C++ modules that were shown up by
441 -Wmissing-field-initializers and -Wunused-parameter.
443 2. The tests for quantifiers being too big (greater than 65535) were being
444 applied after reading the number, and stupidly assuming that integer
445 overflow would give a negative number. The tests are now applied as the
446 numbers are read.
448 3. Tidy code in pcre_exec.c where two branches that used to be different are
449 now the same.
451 4. The JIT compiler did not generate match limit checks for certain
452 bracketed expressions with quantifiers. This may lead to exponential
453 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
454 issue should be resolved now.
456 5. Fixed an issue, which occures when nested alternatives are optimized
457 with table jumps.
459 6. Inserted two casts and changed some ints to size_t in the light of some
460 reported 64-bit compiler warnings (Bugzilla 1477).
462 7. Fixed a bug concerned with zero-minimum possessive groups that could match
463 an empty string, which sometimes were behaving incorrectly in the
464 interpreter (though correctly in the JIT matcher). This pcretest input is
465 an example:
467 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
470 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
471 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
472 for an empty string was breaking the inner loop and carrying on at a lower
473 level, when possessive repeated groups should always return to a higher
474 level as they have no backtrack points in them. The empty string test now
475 occurs at the outer level.
477 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
478 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
480 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
481 doesn't).
483 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
484 to hold a pattern to be incorrectly computed (too small) when there were
485 named back references to duplicated names. This could cause "internal
486 error: code overflow" or "double free or corruption" or other memory
487 handling errors.
489 11. When named subpatterns had the same prefixes, back references could be
490 confused. For example, in this pattern:
492 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
494 the reference to 'Name' was incorrectly treated as a reference to a
495 duplicate name.
497 12. A pattern such as /^s?c/mi8 where the optional character has more than
498 one "other case" was incorrectly compiled such that it would only try to
499 match starting at "c".
501 13. When a pattern starting with \s was studied, VT was not included in the
502 list of possible starting characters; this should have been part of the
503 8.34/18 patch.
505 14. If a character class started [\Qx]... where x is any character, the class
506 was incorrectly terminated at the ].
508 15. If a pattern that started with a caseless match for a character with more
509 than one "other case" was studied, PCRE did not set up the starting code
510 unit bit map for the list of possible characters. Now it does. This is an
511 optimization improvement, not a bug fix.
513 16. The Unicode data tables have been updated to Unicode 7.0.0.
515 17. Fixed a number of memory leaks in pcregrep.
517 18. Avoid a compiler warning (from some compilers) for a function call with
518 a cast that removes "const" from an lvalue by using an intermediate
519 variable (to which the compiler does not object).
521 19. Incorrect code was compiled if a group that contained an internal recursive
522 back reference was optional (had quantifier with a minimum of zero). This
523 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
524 caused segmentation faults because of stack overflows at compile time.
526 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
527 group that is quantified with an indefinite repeat, caused a compile-time
528 loop which used up all the system stack and provoked a segmentation fault.
529 This was not the same bug as 19 above.
531 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
532 Patch by Mike Frysinger.
535 Version 8.35 04-April-2014
536 --------------------------
538 1. A new flag is set, when property checks are present in an XCLASS.
539 When this flag is not set, PCRE can perform certain optimizations
540 such as studying these XCLASS-es.
542 2. The auto-possessification of character sets were improved: a normal
543 and an extended character set can be compared now. Furthermore
544 the JIT compiler optimizes more character set checks.
546 3. Got rid of some compiler warnings for potentially uninitialized variables
547 that show up only when compiled with -O2.
549 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
550 of a match later then the end of the match. The pcretest program was not
551 handling the case sensibly - it was outputting from the start to the next
552 binary zero. It now reports this situation in a message, and outputs the
553 text from the end to the start.
555 5. Fast forward search is improved in JIT. Instead of the first three
556 characters, any three characters with fixed position can be searched.
557 Search order: first, last, middle.
559 6. Improve character range checks in JIT. Characters are read by an inprecise
560 function now, which returns with an unknown value if the character code is
561 above a certain threshold (e.g: 256). The only limitation is that the value
562 must be bigger than the threshold as well. This function is useful when
563 the characters above the threshold are handled in the same way.
565 7. The macros whose names start with RAWUCHAR are placeholders for a future
566 mode in which only the bottom 21 bits of 32-bit data items are used. To
567 make this more memorable for those maintaining the code, the names have
568 been changed to start with UCHAR21, and an extensive comment has been added
569 to their definition.
571 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
572 to the export list in Makefile.am (they were accidentally omitted from the
573 8.34 tarball).
575 9. The informational output from pcretest used the phrase "starting byte set"
576 which is inappropriate for the 16-bit and 32-bit libraries. As the output
577 for "first char" and "need char" really means "non-UTF-char", I've changed
578 "byte" to "char", and slightly reworded the output. The documentation about
579 these values has also been (I hope) clarified.
581 10. Another JIT related optimization: use table jumps for selecting the correct
582 backtracking path, when more than four alternatives are present inside a
583 bracket.
585 11. Empty match is not possible, when the minimum length is greater than zero,
586 and there is no \K in the pattern. JIT should avoid empty match checks in
587 such cases.
589 12. In a caseless character class with UCP support, when a character with more
590 than one alternative case was not the first character of a range, not all
591 the alternative cases were added to the class. For example, s and \x{17f}
592 are both alternative cases for S: the class [RST] was handled correctly,
593 but [R-T] was not.
595 13. The configure.ac file always checked for pthread support when JIT was
596 enabled. This is not used in Windows, so I have put this test inside a
597 check for the presence of windows.h (which was already tested for).
599 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
600 The algorithm provides a way to skip certain starting offsets, and usually
601 faster than linear prefix searches.
603 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
604 as for 'fr_FR' and 'french'. For some reason, however, it then used the
605 Windows-specific input and output files, which have 'french' screwed in.
606 So this could never have worked. One of the problems with locales is that
607 they aren't always the same. I have now updated RunTest so that it checks
608 the output of the locale test (test 3) against three different output
609 files, and it allows the test to pass if any one of them matches. With luck
610 this should make the test pass on some versions of Solaris where it was
611 failing. Because of the uncertainty, the script did not used to stop if
612 test 3 failed; it now does. If further versions of a French locale ever
613 come to light, they can now easily be added.
615 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
616 there was a message during ./configure, but it did not stop. This now
617 provokes an error. The invalid example in README has been corrected.
618 If a value less than the minimum is given, the minimum value has always
619 been used, but now a warning is given.
621 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
622 was a bug in the test system, which is now fixed. Also, the list of various
623 configurations that are tested for each release did not have one with both
624 16/32 bits and --enable-bar-anycrlf. It now does.
626 18. pcretest was missing "-C bsr" for displaying the \R default setting.
628 19. Little endian PowerPC systems are supported now by the JIT compiler.
630 20. The fast forward newline mechanism could enter to an infinite loop on
631 certain invalid UTF-8 input. Although we don't support these cases
632 this issue can be fixed by a performance optimization.
634 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
635 not take account if existing stack usage. There is now a new global
636 variable called pcre_stack_guard that can be set to point to an external
637 function to check stack availability. It is called at the start of
638 processing every parenthesized group.
640 22. A typo in the code meant that in ungreedy mode the max/min qualifier
641 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
642 not match "ab".
644 23. When UTF was disabled, the JIT program reported some incorrect compile
645 errors. These messages are silenced now.
647 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
648 compiler.
650 25. Change all the temporary files used in RunGrepTest to be different to those
651 used by RunTest so that the tests can be run simultaneously, for example by
652 "make -j check".
655 Version 8.34 15-December-2013
656 -----------------------------
658 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
659 executable memory. Patch inspired by Carsten Klein.
661 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
662 this macro is never tested and has no effect, because the work to support
663 coverage involves only compiling and linking options and special targets in
664 the Makefile. The comment in config.h implied that defining the macro would
665 enable coverage support, which is totally false. There was also support for
666 setting this macro in the CMake files (my fault, I just copied it from
667 configure). SUPPORT_GCOV has now been removed.
669 3. Make a small performance improvement in strlen16() and strlen32() in
670 pcretest.
672 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
673 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
674 these cases). There was also one in pcretest.c.
676 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
678 6. In UTF mode, the code for checking whether a group could match an empty
679 string (which is used for indefinitely repeated groups to allow for
680 breaking an infinite loop) was broken when the group contained a repeated
681 negated single-character class with a character that occupied more than one
682 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
683 UTF-8 mode). The effect was undefined: the group might or might not be
684 deemed as matching an empty string, or the program might have crashed.
686 7. The code for checking whether a group could match an empty string was not
687 recognizing that \h, \H, \v, \V, and \R must match a character.
689 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
690 an empty string. If it can, pcretest shows this in its information output.
692 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
693 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
694 matched by pcre_exec() without using JIT:
696 (a) If the rest of the pattern did not match after a maximal run of
697 grapheme clusters, the code for backing up to try with fewer of them
698 did not always back up over a full grapheme when characters that do not
699 have the modifier quality were involved, e.g. Hangul syllables.
701 (b) If the match point in a subject started with modifier character, and
702 there was no match, the code could incorrectly back up beyond the match
703 point, and potentially beyond the first character in the subject,
704 leading to a segfault or an incorrect match result.
706 10. A conditional group with an assertion condition could lead to PCRE
707 recording an incorrect first data item for a match if no other first data
708 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
709 first data item, and therefore matched "ca" after "c" instead of at the
710 start.
712 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
713 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
715 12. The source of pcregrep now includes z/OS-specific code so that it can be
716 compiled for z/OS as part of the special z/OS distribution.
718 13. Added the -T and -TM options to pcretest.
720 14. The code in pcre_compile.c for creating the table of named capturing groups
721 has been refactored. Instead of creating the table dynamically during the
722 actual compiling pass, the information is remembered during the pre-compile
723 pass (on the stack unless there are more than 20 named groups, in which
724 case malloc() is used) and the whole table is created before the actual
725 compile happens. This has simplified the code (it is now nearly 150 lines
726 shorter) and prepared the way for better handling of references to groups
727 with duplicate names.
729 15. A back reference to a named subpattern when there is more than one of the
730 same name now checks them in the order in which they appear in the pattern.
731 The first one that is set is used for the reference. Previously only the
732 first one was inspected. This change makes PCRE more compatible with Perl.
734 16. Unicode character properties were updated from Unicode 6.3.0.
736 17. The compile-time code for auto-possessification has been refactored, based
737 on a patch by Zoltan Herczeg. It now happens after instead of during
738 compilation. The code is cleaner, and more cases are handled. The option
739 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
740 options in pcretest are provided to set it. It can also be set by
741 (*NO_AUTO_POSSESS) at the start of a pattern.
743 18. The character VT has been added to the default ("C" locale) set of
744 characters that match \s and are generally treated as white space,
745 following this same change in Perl 5.18. There is now no difference between
746 "Perl space" and "POSIX space". Whether VT is treated as white space in
747 other locales depends on the locale.
749 19. The code for checking named groups as conditions, either for being set or
750 for being recursed, has been refactored (this is related to 14 and 15
751 above). Processing unduplicated named groups should now be as fast at
752 numerical groups, and processing duplicated groups should be faster than
753 before.
755 20. Two patches to the CMake build system, by Alexander Barkov:
757 (1) Replace the "source" command by "." in CMakeLists.txt because
758 "source" is a bash-ism.
760 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
761 without these the CMake build does not work on Solaris.
763 21. Perl has changed its handling of \8 and \9. If there is no previously
764 encountered capturing group of those numbers, they are treated as the
765 literal characters 8 and 9 instead of a binary zero followed by the
766 literals. PCRE now does the same.
768 22. Following Perl, added \o{} to specify codepoints in octal, making it
769 possible to specify values greater than 0777 and also making them
770 unambiguous.
772 23. Perl now gives an error for missing closing braces after \x{... instead of
773 treating the string as literal. PCRE now does the same.
775 24. RunTest used to grumble if an inappropriate test was selected explicitly,
776 but just skip it when running all tests. This make it awkward to run ranges
777 of tests when one of them was inappropriate. Now it just skips any
778 inappropriate tests, as it always did when running all tests.
780 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
781 character types such as \d or \w, too many callouts were inserted, and the
782 data that they returned was rubbish.
784 26. In UCP mode, \s was not matching two of the characters that Perl matches,
785 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
786 were matched by \h. The code has now been refactored so that the lists of
787 the horizontal and vertical whitespace characters used for \h and \v (which
788 are defined only in one place) are now also used for \s.
790 27. Add JIT support for the 64 bit TileGX architecture.
791 Patch by Jiong Wang (Tilera Corporation).
793 28. Possessive quantifiers for classes (both explicit and automatically
794 generated) now use special opcodes instead of wrapping in ONCE brackets.
796 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
797 (because it's meaningless), this was not happening when PCRE_CASELESS was
798 set. Not wrong, but inefficient.
800 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
801 properties for \w, \d, etc) is present in a test regex. Otherwise if the
802 test contains no characters greater than 255, Perl doesn't realise it
803 should be using Unicode semantics.
805 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
806 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
807 does in Unicode mode.
809 32. Added the "forbid" facility to pcretest so that putting tests into the
810 wrong test files can sometimes be quickly detected.
812 33. There is now a limit (default 250) on the depth of nesting of parentheses.
813 This limit is imposed to control the amount of system stack used at compile
814 time. It can be changed at build time by --with-parens-nest-limit=xxx or
815 the equivalent in CMake.
817 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
818 errors. Perl warns for these when in warning mode, but PCRE has no facility
819 for giving warnings.
821 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
822 However, this was not working for (?!) because it is optimized to (*FAIL),
823 for which PCRE does not allow quantifiers. The optimization is now disabled
824 when a quantifier follows (?!). I can't see any use for this, but it makes
825 things uniform.
827 36. Perl no longer allows group names to start with digits, so I have made this
828 change also in PCRE. It simplifies the code a bit.
830 37. In extended mode, Perl ignores spaces before a + that indicates a
831 possessive quantifier. PCRE allowed a space before the quantifier, but not
832 before the possessive +. It now does.
834 38. The use of \K (reset reported match start) within a repeated possessive
835 group such as (a\Kb)*+ was not working.
837 40. Document that the same character tables must be used at compile time and
838 run time, and that the facility to pass tables to pcre_exec() and
839 pcre_dfa_exec() is for use only with saved/restored patterns.
841 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
842 features for Builds with MSVC:
844 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
845 builds). The .rc files can be used to set FileDescription and many other
846 attributes.
848 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
849 This allows higher-level build scripts which want .pdb files to avoid
850 hard-coding the exact files needed."
852 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
853 mean "start of word" and "end of word", respectively, as a transition aid.
855 43. A minimizing repeat of a class containing codepoints greater than 255 in
856 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
857 compiled to use the heap for recursion.
859 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
860 is configured.
863 Version 8.33 28-May-2013
864 ------------------------
866 1. Added 'U' to some constants that are compared to unsigned integers, to
867 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
868 variables that are added to signed variables, to ensure the result is
869 signed and can be negated.
871 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
872 CMake config files.
874 3. Revise the creation of config.h.generic so that all boolean macros are
875 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
876 overriding via -D on the command line possible.
878 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
879 to unsigned int is reported to make a quite noticeable speed difference in
880 a specific Windows environment. Testing on Linux did also appear to show
881 some benefit (and it is clearly not harmful). Also fixed the definition of
882 Xop which should be unsigned.
884 5. Related to (4), changing the definition of the intermediate variable cc
885 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
886 improvements.
888 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
889 unnecessary spaces.
891 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
892 and later.
894 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
896 9. Optimizing fast_forward_start_bits in JIT.
898 10. Adding support for callouts in JIT, and fixing some issues revealed
899 during this work. Namely:
901 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
903 (b) Minimum length was not checked before the matching is started.
905 11. The value of capture_last that is passed to callouts was incorrect in some
906 cases when there was a capture on one path that was subsequently abandoned
907 after a backtrack. Also, the capture_last value is now reset after a
908 recursion, since all captures are also reset in this case.
910 12. The interpreter no longer returns the "too many substrings" error in the
911 case when an overflowing capture is in a branch that is subsequently
912 abandoned after a backtrack.
914 13. In the pathological case when an offset vector of size 2 is used, pcretest
915 now prints out the matched string after a yield of 0 or 1.
917 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
918 Only supported by the JIT compiler at the moment.
920 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
922 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
923 offset of the starting point of the matching process, provided the offsets
924 vector is large enough.
926 17. The \A escape now records a lookbehind value of 1, though its execution
927 does not actually inspect the previous character. This is to ensure that,
928 in partial multi-segment matching, at least one character from the old
929 segment is retained when a new segment is processed. Otherwise, if there
930 are no lookbehinds in the pattern, \A might match incorrectly at the start
931 of a new segment.
933 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
935 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
936 gives some modest performance improvement in 8-bit mode.
938 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
939 be expressed in certain programming languages using Universal Character
940 Names.
942 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
943 which points out that "non characters" are not "characters that may not
944 appear in Unicode strings" but rather "characters that are reserved for
945 internal use and have only local meaning".
947 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
948 there was a conditional group that depended on an assertion, if the
949 assertion was false, the callout that immediately followed the alternation
950 in the condition was skipped when pcre_exec() was used for matching.
952 23. Allow an explicit callout to be inserted before an assertion that is the
953 condition for a conditional group, for compatibility with automatic
954 callouts, which always insert a callout at this point.
956 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
957 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
959 25. (*PRUNE) is now supported by the JIT compiler.
961 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
963 27. Fix the case where there are two or more SKIPs with arguments that may be
964 ignored.
966 28. (*SKIP) is now supported by the JIT compiler.
968 29. (*THEN) is now supported by the JIT compiler.
970 30. Update RunTest with additional test selector options.
972 31. The way PCRE handles backtracking verbs has been changed in two ways.
974 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
975 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
976 backtracking. In some cases this makes it more Perl-compatible, but Perl's
977 rather obscure rules do not always do the same thing.
979 (2) Previously, backtracking verbs were confined within assertions. This is
980 no longer the case for positive assertions, except for (*ACCEPT). Again,
981 this sometimes improves Perl compatibility, and sometimes does not.
983 32. A number of tests that were in test 2 because Perl did things differently
984 have been moved to test 1, because either Perl or PCRE has changed, and
985 these tests are now compatible.
987 32. Backtracking control verbs are now handled in the same way in JIT and
988 interpreter.
990 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
991 contained a forward subroutine reference caused a compile error.
993 34. Auto-detect and optimize limited repetitions in JIT.
995 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
996 blocking (*UTF) etc.
998 36. In the interpreter, maximizing pattern repetitions for characters and
999 character types now use tail recursion, which reduces stack usage.
1001 37. The value of the max lookbehind was not correctly preserved if a compiled
1002 and saved regex was reloaded on a host of different endianness.
1004 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1005 of the compiled pattern block, expand the flags field from 16 to 32 bits
1006 because it was almost full.
1008 39. Try madvise first before posix_madvise.
1010 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1011 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1012 original change locked out all matches of empty strings. This has been
1013 changed so that one match of an empty string per line is recognized.
1014 Subsequent searches on the same line (for colouring or for --only-matching,
1015 for example) do not recognize empty strings.
1017 41. Applied a user patch to fix a number of spelling mistakes in comments.
1019 42. Data lines longer than 65536 caused pcretest to crash.
1021 43. Clarified the data type for length and startoffset arguments for pcre_exec
1022 and pcre_dfa_exec in the function-specific man pages, where they were
1023 explicitly stated to be in bytes, never having been updated. I also added
1024 some clarification to the pcreapi man page.
1026 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1027 a segmentation fault.
1030 Version 8.32 30-November-2012
1031 -----------------------------
1033 1. Improved JIT compiler optimizations for first character search and single
1034 character iterators.
1036 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1037 Patch by Daniel Richard G.
1039 3. Single character iterator optimizations in the JIT compiler.
1041 4. Improved JIT compiler optimizations for character ranges.
1043 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1044 Reported by Giuseppe D'Angelo.
1046 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1047 of a line, was being set incorrectly in cases where .* appeared inside
1048 atomic brackets at the start of a pattern, or where there was a subsequent
1049 *PRUNE or *SKIP.
1051 7. Improved instruction cache flush for POWER/PowerPC.
1052 Patch by Daniel Richard G.
1054 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1055 grep:
1057 (a) There is now no limit to the number of patterns to be matched.
1059 (b) An error is given if a pattern is too long.
1061 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1062 are now supported.
1064 (d) --exclude-from and --include-from (multiple use) have been added.
1066 (e) Exclusions and inclusions now apply to all files and directories, not
1067 just to those obtained from scanning a directory recursively.
1069 (f) Multiple uses of -f and --file-list are now supported.
1071 (g) In a Windows environment, the default for -d has been changed from
1072 "read" (the GNU grep default) to "skip", because otherwise the presence
1073 of a directory in the file list provokes an error.
1075 (h) The documentation has been revised and clarified in places.
1077 9. Improve the matching speed of capturing brackets.
1079 10. Changed the meaning of \X so that it now matches a Unicode extended
1080 grapheme cluster.
1082 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1083 out POSIX threads when JIT support is configured.
1085 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1087 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1088 structure could go wrong in environments where size_t is not the same size
1089 as int.
1091 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1092 set.
1094 15. The EBCDIC support had decayed; later updates to the code had included
1095 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1096 general tidy up of EBCDIC-related issues, and the documentation was also
1097 not quite right. There is now a test that can be run on ASCII systems to
1098 check some of the EBCDIC-related things (but is it not a full test).
1100 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1101 in a small tidy to the code.
1103 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1105 18. If the --only-matching (-o) option in pcregrep is specified multiple
1106 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1107 substrings matched by the 1st and 2nd capturing parentheses. A separating
1108 string can be specified by --om-separator (default empty).
1110 19. Improving the first n character searches.
1112 20. Turn case lists for horizontal and vertical white space into macros so that
1113 they are defined only once.
1115 21. This set of changes together give more compatible Unicode case-folding
1116 behaviour for characters that have more than one other case when UCP
1117 support is available.
1119 (a) The Unicode property table now has offsets into a new table of sets of
1120 three or more characters that are case-equivalent. The MultiStage2.py
1121 script that generates these tables (the pcre_ucd.c file) now scans
1122 CaseFolding.txt instead of UnicodeData.txt for character case
1123 information.
1125 (b) The code for adding characters or ranges of characters to a character
1126 class has been abstracted into a generalized function that also handles
1127 case-independence. In UTF-mode with UCP support, this uses the new data
1128 to handle characters with more than one other case.
1130 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1131 whose other case is greater than 256 are now correctly matched
1132 caselessly. Previously, the high codepoint matched the low one, but not
1133 vice versa.
1135 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1136 of the new class addition function, using character lists defined as
1137 macros alongside the case definitions of 20 above.
1139 (e) Caseless back references now work with characters that have more than
1140 one other case.
1142 (f) General caseless matching of characters with more than one other case
1143 is supported.
1145 22. Unicode character properties were updated from Unicode 6.2.0
1147 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1149 24. Add support for 32-bit character strings, and UTF-32
1151 25. Major JIT compiler update (code refactoring and bugfixing).
1152 Experimental Sparc 32 support is added.
1154 26. Applied a modified version of Daniel Richard G's patch to create
1155 pcre.h.generic and config.h.generic by "make" instead of in the
1156 PrepareRelease script.
1158 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1159 pcre_compile.c when checking for a zero character.
1161 28. Introducing a native interface for JIT. Through this interface, the compiled
1162 machine code can be directly executed. The purpose of this interface is to
1163 provide fast pattern matching, so several sanity checks are not performed.
1164 However, feature tests are still performed. The new interface provides
1165 1.4x speedup compared to the old one.
1167 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1168 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1169 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1171 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1172 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1173 the "old" RFC 2279). Instead, it ended up passing a negative length to
1174 pcre_exec().
1176 31. Add support for GCC's visibility feature to hide internal functions.
1178 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1179 "unknown -C option" after outputting 0 or 1.
1181 33. There is now support for generating a code coverage report for the test
1182 suite in environments where gcc is the compiler and lcov is installed. This
1183 is mainly for the benefit of the developers.
1185 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1186 unaddressable using valgrind annotations, allowing valgrind to detect
1187 invalid memory accesses. This is mainly for the benefit of the developers.
1189 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1191 26. Give configure error if --enable-cpp but no C++ compiler found.
1194 Version 8.31 06-July-2012
1195 -------------------------
1197 1. Fixing a wrong JIT test case and some compiler warnings.
1199 2. Removed a bashism from the RunTest script.
1201 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1202 to unsigned type, result still unsigned" that was given by an MS compiler
1203 on encountering the code "-sizeof(xxx)".
1205 4. Partial matching support is added to the JIT compiler.
1207 5. Fixed several bugs concerned with partial matching of items that consist
1208 of more than one character:
1210 (a) /^(..)\1/ did not partially match "aba" because checking references was
1211 done on an "all or nothing" basis. This also applied to repeated
1212 references.
1214 (b) \R did not give a hard partial match if \r was found at the end of the
1215 subject.
1217 (c) \X did not give a hard partial match after matching one or more
1218 characters at the end of the subject.
1220 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1221 a partial match for the string "\r".
1223 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1224 a partial match for a CR character at the end of the subject string.
1226 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1227 running pcretest, the text "(JIT)" added to the output whenever JIT is
1228 actually used to run the match.
1230 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1231 or /S+[+] with a digit between 1 and 7.
1233 8. OP_NOT now supports any UTF character not just single-byte ones.
1235 9. (*MARK) control verb is now supported by the JIT compiler.
1237 10. The command "./RunTest list" lists the available tests without actually
1238 running any of them. (Because I keep forgetting what they all are.)
1242 12. Applied a (slightly modified) user-supplied patch that improves performance
1243 when the heap is used for recursion (compiled with --disable-stack-for-
1244 recursion). Instead of malloc and free for each heap frame each time a
1245 logical recursion happens, frames are retained on a chain and re-used where
1246 possible. This sometimes gives as much as 30% improvement.
1248 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1249 call.
1251 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1253 15. It is now possible to link pcretest with libedit as an alternative to
1254 libreadline.
1256 16. (*COMMIT) control verb is now supported by the JIT compiler.
1258 17. The Unicode data tables have been updated to Unicode 6.1.0.
1260 18. Added --file-list option to pcregrep.
1262 19. Added binary file support to pcregrep, including the -a, --binary-files,
1263 -I, and --text options.
1265 20. The madvise function is renamed for posix_madvise for QNX compatibility
1266 reasons. Fixed by Giuseppe D'Angelo.
1268 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1270 22. Changed the option for creating symbolic links for 16-bit man pages from
1271 -s to -sf so that re-installing does not cause issues.
1273 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1275 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1276 matches in certain environments (the workspace was not being correctly
1277 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1278 some of the workspace data at the beginning of a restart.
1280 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1281 was not doing so when it should - probably a typo introduced by SVN 528
1282 (change 8.10/14).
1284 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1285 \w+ when the character tables indicated that \x{c4} was a word character.
1286 There were several related cases, all because the tests for doing a table
1287 lookup were testing for characters less than 127 instead of 255.
1289 27. If a pattern contains capturing parentheses that are not used in a match,
1290 their slots in the ovector are set to -1. For those that are higher than
1291 any matched groups, this happens at the end of processing. In the case when
1292 there were back references that the ovector was too small to contain
1293 (causing temporary malloc'd memory to be used during matching), and the
1294 highest capturing number was not used, memory off the end of the ovector
1295 was incorrectly being set to -1. (It was using the size of the temporary
1296 memory instead of the true size.)
1298 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1299 ovector size, it uses memory at the end of the block that it has got.
1301 29. Check for an overlong MARK name and give an error at compile time. The
1302 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1304 30. JIT compiler update.
1306 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1307 Rill for the patch.
1309 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1311 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1313 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1314 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1315 was enabled.
1317 35. Improve JIT code generation for greedy plus quantifier.
1319 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1320 1 to "aa" instead of to an empty string. The bug affected repeated groups
1321 that could potentially match an empty string.
1323 37. Optimizing single character iterators in JIT.
1325 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1326 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1327 codepoints that are too big for the mode are faulted, and in a UTF mode,
1328 disallowed codepoints are also faulted.
1330 39. If PCRE was compiled with UTF support, in three places in the DFA
1331 matcher there was code that should only have been obeyed in UTF mode, but
1332 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1333 processing when bytes with values greater than 127 were present. In 16-bit
1334 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1335 both cases the values are those that cannot be the first data item in a UTF
1336 character. The three items that might have provoked this were recursions,
1337 possessively repeated groups, and atomic groups.
1339 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1340 and pcregrep, because some OS require shared objects to be explicitly
1341 passed to ld, causing the link step to fail if they are not.
1343 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1344 mode, patterns that started with \h* or \R* might be incorrectly matched.
1347 Version 8.30 04-February-2012
1348 -----------------------------
1350 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1351 name is defined in ctype.h.
1353 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1354 only in quite long subpatterns.
1356 3. Removed the function pcre_info(), which has been obsolete and deprecated
1357 since it was replaced by pcre_fullinfo() in February 2000.
1359 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1360 match a (*MARK), and the match failed at the start of the subject, a
1361 reference to memory before the start of the subject could occur. This bug
1362 was introduced by fix 17 of release 8.21.
1364 5. A reference to an unset group with zero minimum repetition was giving
1365 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1366 /(another)?(\1?)test/ matched against "hello world test". This bug was
1367 introduced in release 8.13.
1369 6. Add support for 16-bit character strings (a large amount of work involving
1370 many changes and refactorings).
1372 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1373 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1374 from a file.
1376 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1377 rounding is not applied in this particular case).
1379 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1380 if they appear, or are escaped, in patterns.
1382 10. Get rid of a number of -Wunused-but-set-variable warnings.
1384 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1385 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1386 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1387 also returns the mark "x". This bug applied to capturing parentheses,
1388 non-capturing parentheses, and atomic parentheses. It also applied to some
1389 assertions.
1391 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1392 information out of configure.ac instead of relying on pcre.h.generic, which
1393 is not stored in the repository.
1395 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1396 -lreadline.
1398 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1400 15. Applied Graycode's patch to put the top-level frame on the stack rather
1401 than the heap when not using the stack for recursion. This gives a
1402 performance improvement in many cases when recursion is not deep.
1404 16. Experimental code added to "pcretest -C" to output the stack frame size.
1407 Version 8.21 12-Dec-2011
1408 ------------------------
1410 1. Updating the JIT compiler.
1412 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1413 are added as well.
1415 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1416 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1417 calling _pcre_jit_exec. Some extra comments are added.
1419 4. (*MARK) settings inside atomic groups that do not contain any capturing
1420 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1421 was introduced by change 18 for 8.20.
1423 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1424 ECMA-262 standard.
1426 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1427 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1428 This bug was probably introduced by change 9 of 8.13.
1430 7. While fixing 6 above, I noticed that a number of other items were being
1431 incorrectly rejected as "not fixed length". This arose partly because newer
1432 opcodes had not been added to the fixed-length checking code. I have (a)
1433 corrected the bug and added tests for these items, and (b) arranged for an
1434 error to occur if an unknown opcode is encountered while checking for fixed
1435 length instead of just assuming "not fixed length". The items that were
1436 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1437 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1438 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1440 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1441 being incorrectly compiled and would have given unpredicatble results.
1443 9. A possessively repeated subpattern with minimum repeat count greater than
1444 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1445 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1446 the first (A) could occur when it should not.
1448 10. Add a cast and remove a redundant test from the code.
1450 11. JIT should use pcre_malloc/pcre_free for allocation.
1452 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1453 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1454 is anything other than /usr, -L is still shown).
1456 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1458 14. Perl does not support \N without a following name in a [] class; PCRE now
1459 also gives an error.
1461 15. If a forward reference was repeated with an upper limit of around 2000,
1462 it caused the error "internal error: overran compiling workspace". The
1463 maximum number of forward references (including repeats) was limited by the
1464 internal workspace, and dependent on the LINK_SIZE. The code has been
1465 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1466 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1467 of around 200,000 forward references. While doing this, I also speeded up
1468 the filling in of repeated forward references.
1470 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1471 incorrectly expecting the subject to contain another "a" after the start.
1473 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1474 in the match, the SKIP should be ignored. This was not happening; instead
1475 the SKIP was being treated as NOMATCH. For patterns such as
1476 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1477 tested.
1479 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1480 now much more compatible with Perl, in particular in cases where the result
1481 is a non-match for a non-anchored pattern. For example, if
1482 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1483 "m", where previously it did not return a name. A side effect of this
1484 change is that for partial matches, the last encountered mark name is
1485 returned, as for non matches. A number of tests that were previously not
1486 Perl-compatible have been moved into the Perl-compatible test files. The
1487 refactoring has had the pleasing side effect of removing one argument from
1488 the match() function, thus reducing its stack requirements.
1490 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1491 subsequent uses of /S (without +) incorrectly behaved like /S+.
1493 21. Retrieve executable code size support for the JIT compiler and fixing
1494 some warnings.
1496 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1497 not work when the shorter character appeared right at the end of the
1498 subject string.
1500 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1501 systems.
1503 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1504 output it when the /M option is used in pcretest.
1506 25. The CheckMan script was not being included in the distribution. Also, added
1507 an explicit "perl" to run Perl scripts from the PrepareRelease script
1508 because this is reportedly needed in Windows.
1510 26. If study data was being save in a file and studying had not found a set of
1511 "starts with" bytes for the pattern, the data written to the file (though
1512 never used) was taken from uninitialized memory and so caused valgrind to
1513 complain.
1515 27. Updated RunTest.bat as provided by Sheri Pierce.
1517 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1519 29. Computation of memory usage for the table of capturing group names was
1520 giving an unnecessarily large value.
1523 Version 8.20 21-Oct-2011
1524 ------------------------
1526 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1527 a POSIX class. After further experiments with Perl, which convinced me that
1528 Perl has bugs and confusions, a closing square bracket is no longer allowed
1529 in a POSIX name. This bug also affected patterns with classes that started
1530 with full stops.
1532 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1533 captured substring, but while checking the failing first alternative,
1534 substring 1 is temporarily captured. If the output vector supplied to
1535 pcre_exec() was not big enough for this capture, the yield of the function
1536 was still zero ("insufficient space for captured substrings"). This cannot
1537 be totally fixed without adding another stack variable, which seems a lot
1538 of expense for a edge case. However, I have improved the situation in cases
1539 such as /(a)(b)x|abc/ matched against "abc", where the return code
1540 indicates that fewer than the maximum number of slots in the ovector have
1541 been set.
1543 3. Related to (2) above: when there are more back references in a pattern than
1544 slots in the output vector, pcre_exec() uses temporary memory during
1545 matching, and copies in the captures as far as possible afterwards. It was
1546 using the entire output vector, but this conflicts with the specification
1547 that only 2/3 is used for passing back captured substrings. Now it uses
1548 only the first 2/3, for compatibility. This is, of course, another edge
1549 case.
1551 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1552 main code base, and can be used by building with --enable-jit. When this is
1553 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1554 runtime --no-jit option is given.
1556 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1557 ovector, the return from the function was zero, implying that there were
1558 other matches that did not fit. The correct "exactly full" value is now
1559 returned.
1561 6. If a subpattern that was called recursively or as a subroutine contained
1562 (*PRUNE) or any other control that caused it to give a non-standard return,
1563 invalid errors such as "Error -26 (nested recursion at the same subject
1564 position)" or even infinite loops could occur.
1566 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1567 computing the minimum length on reaching *ACCEPT, and so ended up with the
1568 wrong value of 1 rather than 0. Further investigation indicates that
1569 computing a minimum subject length in the presence of *ACCEPT is difficult
1570 (think back references, subroutine calls), and so I have changed the code
1571 so that no minimum is registered for a pattern that contains *ACCEPT.
1573 8. If (*THEN) was present in the first (true) branch of a conditional group,
1574 it was not handled as intended. [But see 16 below.]
1576 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1577 Sheri Pierce.
1579 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1580 the first byte in a match must be "a".
1582 11. Change 17 for 8.13 increased the recursion depth for patterns like
1583 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1584 pattern contains any instances of (*THEN). If it does not, the old
1585 optimizations are restored. It would be nice to do this on a per-group
1586 basis, but at the moment that is not feasible.
1588 12. In some environments, the output of pcretest -C is CRLF terminated. This
1589 broke RunTest's code that checks for the link size. A single white space
1590 character after the value is now allowed for.
1592 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1593 For "fr", it uses the Windows-specific input and output files.
1595 14. If (*THEN) appeared in a group that was called recursively or as a
1596 subroutine, it did not work as intended. [But see next item.]
1598 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1599 pattern fragments (but not containing any | characters). If A and B are
1600 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1601 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1602 D. In other words, Perl considers parentheses that do not contain any |
1603 characters to be part of a surrounding alternative, whereas PCRE was
1604 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1605 differently. PCRE now behaves in the same way as Perl, except in the case
1606 of subroutine/recursion calls such as (?1) which have in any case always
1607 been different (but PCRE had them first :-).
1609 16. Related to 15 above: Perl does not treat the | in a conditional group as
1610 creating alternatives. Such a group is treated in the same way as an
1611 ordinary group without any | characters when processing (*THEN). PCRE has
1612 been changed to match Perl's behaviour.
1614 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1615 RunGrepTest script failed.
1617 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1618 inevitable for groups that contain captures, but it can lead to a lot of
1619 stack use in large patterns. The old behaviour has been restored for atomic
1620 groups that do not contain any capturing parentheses.
1622 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1623 suppress the check for a minimum subject length at run time. (If it was
1624 given to pcre_exec() or pcre_dfa_exec() it did work.)
1626 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1627 fail to work when decoding hex characters in data strings in EBCDIC
1628 environments.
1630 21. It appears that in at least one Mac OS environment, the isxdigit() function
1631 is implemented as a macro that evaluates to its argument more than once,
1632 contravening the C 90 Standard (I haven't checked a later standard). There
1633 was an instance in pcretest which caused it to go wrong when processing
1634 \x{...} escapes in subject strings. The has been rewritten to avoid using
1635 things like p++ in the argument of isxdigit().
1638 Version 8.13 16-Aug-2011
1639 ------------------------
1641 1. The Unicode data tables have been updated to Unicode 6.0.0.
1643 2. Two minor typos in pcre_internal.h have been fixed.
1645 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1646 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1647 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1649 4. There were a number of related bugs in the code for matching backrefences
1650 caselessly in UTF-8 mode when codes for the characters concerned were
1651 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1652 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1653 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1654 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1655 2-byte code at the end of the subject (it thought there wasn't enough data
1656 left).
1658 5. Comprehensive information about what went wrong is now returned by
1659 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1660 as the output vector has at least 2 elements. The offset of the start of
1661 the failing character and a reason code are placed in the vector.
1663 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1664 now returned is for the first byte of the failing character, instead of the
1665 last byte inspected. This is an incompatible change, but I hope it is small
1666 enough not to be a problem. It makes the returned offset consistent with
1667 pcre_exec() and pcre_dfa_exec().
1669 7. pcretest now gives a text phrase as well as the error number when
1670 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1671 failure, the offset and reason code are output.
1673 8. When \R was used with a maximizing quantifier it failed to skip backwards
1674 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1675 back over a single character (\n). This seems wrong (because it treated the
1676 two characters as a single entity when going forwards), conflicts with the
1677 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1678 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1679 has been changed.
1681 9. Some internal refactoring has changed the processing so that the handling
1682 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1683 time (the PCRE_DOTALL option was changed this way some time ago: version
1684 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1685 which was always a bit of a fudge. It also means that there is one less
1686 argument for the match() function, which reduces its stack requirements
1687 slightly. This change also fixes an incompatibility with Perl: the pattern
1688 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1690 10. More internal refactoring has drastically reduced the number of recursive
1691 calls to match() for possessively repeated groups such as (abc)++ when
1692 using pcre_exec().
1694 11. While implementing 10, a number of bugs in the handling of groups were
1695 discovered and fixed:
1697 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1698 (a|)*(?1) gave a compile-time internal error.
1699 ((a|)+)+ did not notice that the outer group could match an empty string.
1700 (^a|^)+ was not marked as anchored.
1701 (.*a|.*)+ was not marked as matching at start or after a newline.
1703 12. Yet more internal refactoring has removed another argument from the match()
1704 function. Special calls to this function are now indicated by setting a
1705 value in a variable in the "match data" data block.
1707 13. Be more explicit in pcre_study() instead of relying on "default" for
1708 opcodes that mean there is no starting character; this means that when new
1709 ones are added and accidentally left out of pcre_study(), testing should
1710 pick them up.
1712 14. The -s option of pcretest has been documented for ages as being an old
1713 synonym of -m (show memory usage). I have changed it to mean "force study
1714 for every regex", that is, assume /S for every regex. This is similar to -i
1715 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1716 using it. It makes it easier to run collections of tests with and without
1717 study enabled, and thereby test pcre_study() more easily. All the standard
1718 tests are now run with and without -s (but some patterns can be marked as
1719 "never study" - see 20 below).
1721 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1722 restoration of the capturing data to the outer values was not happening
1723 correctly.
1725 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1726 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1727 pattern had matched an empty string, and so incorrectly returned a no
1728 match.
1730 17. There was optimizing code for the last branch of non-capturing parentheses,
1731 and also for the obeyed branch of a conditional subexpression, which used
1732 tail recursion to cut down on stack usage. Unfortunately, now that there is
1733 the possibility of (*THEN) occurring in these branches, tail recursion is
1734 no longer possible because the return has to be checked for (*THEN). These
1735 two optimizations have therefore been removed. [But see 8.20/11 above.]
1737 18. If a pattern containing \R was studied, it was assumed that \R always
1738 matched two bytes, thus causing the minimum subject length to be
1739 incorrectly computed because \R can also match just one byte.
1741 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1742 was incorrectly computed.
1744 20. If /S is present twice on a test pattern in pcretest input, it now
1745 *disables* studying, thereby overriding the use of -s on the command line
1746 (see 14 above). This is necessary for one or two tests to keep the output
1747 identical in both cases.
1749 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1750 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1752 22. When an atomic group that contained a capturing parenthesis was
1753 successfully matched, but the branch in which it appeared failed, the
1754 capturing was not being forgotten if a higher numbered group was later
1755 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1756 group 1 to "a", when in fact it should be unset. This applied to multi-
1757 branched capturing and non-capturing groups, repeated or not, and also to
1758 positive assertions (capturing in negative assertions does not happen
1759 in PCRE) and also to nested atomic groups.
1761 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1762 subject after a captured substring, to make it easier to tell which of a
1763 number of identical substrings has been captured.
1765 24. The way atomic groups are processed by pcre_exec() has been changed so that
1766 if they are repeated, backtracking one repetition now resets captured
1767 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1768 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1769 "aaa". Previously, it would have been "a". As part of this code
1770 refactoring, the way recursive calls are handled has also been changed.
1772 25. If an assertion condition captured any substrings, they were not passed
1773 back unless some other capturing happened later. For example, if
1774 (?(?=(a))a) was matched against "a", no capturing was returned.
1776 26. When studying a pattern that contained subroutine calls or assertions,
1777 the code for finding the minimum length of a possible match was handling
1778 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1779 group 1 called group 2 while simultaneously a separate group 2 called group
1780 1). A stack overflow occurred in this case. I have fixed this by limiting
1781 the recursion depth to 10.
1783 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1784 Fortmann. This supports explicit test numbers on the command line, and has
1785 argument validation and error reporting.
1787 28. An instance of \X with an unlimited repeat could fail if at any point the
1788 first character it looked at was a mark character.
1790 29. Some minor code refactoring concerning Unicode properties and scripts
1791 should reduce the stack requirement of match() slightly.
1793 30. Added the '=' option to pcretest to check the setting of unused capturing
1794 slots at the end of the pattern, which are documented as being -1, but are
1795 not included in the return count.
1797 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1798 compiled something random. Now it gives a compile-time error (as does
1799 Perl).
1801 32. A *MARK encountered during the processing of a positive assertion is now
1802 recorded and passed back (compatible with Perl).
1804 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1805 had alternative anchored branches, the search for a second match in a line
1806 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1807 matched the line "0102" twice. The same bug affected patterns that started
1808 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1809 twice.
1811 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1812 does, and because of capturing effects, quantifying parenthesized
1813 assertions may at times be useful. Quantifiers are now allowed for
1814 parenthesized assertions.
1816 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1818 36. \g was being checked for fancy things in a character class, when it should
1819 just be a literal "g".
1821 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1822 appearance of a nested POSIX class supersedes an apparent external class.
1823 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1824 unescaped square brackets may also appear as part of class names. For
1825 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1826 more like Perl. (But see 8.20/1 above.)
1828 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1829 was because it thought it was \N{name}, which is not supported).
1831 39. Add minix to OS list not supporting the -S option in pcretest.
1833 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1834 cannot analyze patterns in sufficient detail to catch mutual recursions
1835 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1836 subgroup is called recursively as a subpattern for a second time at the
1837 same position in the subject string. In previous releases this might have
1838 been caught by the recursion limit, or it might have run out of stack.
1840 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1841 happen only once. PCRE was, however incorrectly giving a compile time error
1842 "recursive call could loop indefinitely" because it cannot analyze the
1843 pattern in sufficient detail. The compile time test no longer happens when
1844 PCRE is compiling a conditional subpattern, but actual runaway loops are
1845 now caught at runtime (see 40 above).
1847 42. It seems that Perl allows any characters other than a closing parenthesis
1848 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1849 has been changed to be the same.
1851 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1852 as not to get warnings when autogen.sh is called. Also changed
1853 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1855 44. To help people who use pcregrep to scan files containing exceedingly long
1856 lines, the following changes have been made:
1858 (a) The default value of the buffer size parameter has been increased from
1859 8K to 20K. (The actual buffer used is three times this size.)
1861 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1862 PCRE is built.
1864 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1865 to be set at run time.
1867 (d) Numerical values in pcregrep options can be followed by K or M, for
1868 example --buffer-size=50K.
1870 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1871 given and the return code is set to 2.
1873 45. Add a pointer to the latest mark to the callout data block.
1875 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1876 partial match of an empty string instead of no match. This was specific to
1877 the use of ".".
1879 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1880 complete match instead of a partial match. This bug was dependent on both
1881 the PCRE_UTF8 and PCRE_DOTALL options being set.
1883 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1884 starting byte set, because \b was not being ignored.
1887 Version 8.12 15-Jan-2011
1888 ------------------------
1890 1. Fixed some typos in the markup of the man pages, and wrote a script that
1891 checks for such things as part of the documentation building process.
1893 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1894 --match-limit and --recursion-limit options (added for 8.11). In
1895 particular, this made one of the standard tests fail. (The integer value
1896 went into the wrong half of a long int.)
1898 3. If the --colour option was given to pcregrep with -v (invert match), it
1899 did strange things, either producing crazy output, or crashing. It should,
1900 of course, ignore a request for colour when reporting lines that do not
1901 match.
1903 4. Another pcregrep bug caused similar problems if --colour was specified with
1904 -M (multiline) and the pattern match finished with a line ending.
1906 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1907 matched in multiline mode, the following line was shown as part of the
1908 match. This seems wrong, so I have changed it.
1910 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1911 the check for further matches in the same line (so they could be coloured)
1912 to overrun the end of the current line. If another match was found, it was
1913 incorrectly shown (and then shown again when found in the next line).
1915 7. If pcregrep was compiled under Windows, there was a reference to the
1916 function pcregrep_exit() before it was defined. I am assuming this was
1917 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1918 reported by a user. I've moved the definition above the reference.
1921 Version 8.11 10-Dec-2010
1922 ------------------------
1924 1. (*THEN) was not working properly if there were untried alternatives prior
1925 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1926 backtracked to try for "b" instead of moving to the next alternative branch
1927 at the same level (in this case, to look for "c"). The Perl documentation
1928 is clear that when (*THEN) is backtracked onto, it goes to the "next
1929 alternative in the innermost enclosing group".
1931 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1932 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1933 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1934 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1935 (*THEN).
1937 3. If \s appeared in a character class, it removed the VT character from
1938 the class, even if it had been included by some previous item, for example
1939 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1940 of \s, but is part of the POSIX "space" class.)
1942 4. A partial match never returns an empty string (because you can always
1943 match an empty string at the end of the subject); however the checking for
1944 an empty string was starting at the "start of match" point. This has been
1945 changed to the "earliest inspected character" point, because the returned
1946 data for a partial match starts at this character. This means that, for
1947 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1948 (previously it gave "no match").
1950 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1951 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1952 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1953 has an implication that the given string is incomplete (because a partial
1954 match is preferred over a full match). For this reason, these items now
1955 give a partial match in this situation. [Aside: previously, the one case
1956 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1957 match rather than a full match, which was wrong by the old rules, but is
1958 now correct.]
1960 6. There was a bug in the handling of #-introduced comments, recognized when
1961 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1962 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1963 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1964 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1965 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1966 places in pcre_compile().
1968 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1969 comments when looking ahead for named forward references to subpatterns,
1970 the only newline sequence it recognized was NL. It now handles newlines
1971 according to the set newline convention.
1973 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1974 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1975 cater for a lack of strerror(). These oversights have been fixed.
1977 9. Added --match-limit and --recursion-limit to pcregrep.
1979 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1981 11. When the -o option was used, pcregrep was setting a return code of 1, even
1982 when matches were found, and --line-buffered was not being honoured.
1984 12. Added an optional parentheses number to the -o and --only-matching options
1985 of pcregrep.
1987 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1988 can match an empty string. The code to do it in pcretest and pcredemo
1989 needed fixing:
1991 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1992 only one byte after an empty string match just before CRLF (this case
1993 just got forgotten; "any" and "anycrlf" were OK).
1995 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1996 mode when an empty string match preceded an ASCII character followed by
1997 a non-ASCII character. (The code for advancing by one character rather
1998 than one byte was nonsense.)
2000 (c) The pcredemo.c sample program did not have any code at all to handle
2001 the cases when CRLF is a valid newline sequence.
2003 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2004 as a starting offset was within the subject string. There is now a new
2005 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2006 negative or greater than the length of the string. In order to test this,
2007 pcretest is extended to allow the setting of negative starting offsets.
2009 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2010 starting offset points to the beginning of a UTF-8 character was
2011 unnecessarily clumsy. I tidied it up.
2013 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2014 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2016 17. Nobody had reported that the --include_dir option, which was added in
2017 release 7.7 should have been called --include-dir (hyphen, not underscore)
2018 for compatibility with GNU grep. I have changed it to --include-dir, but
2019 left --include_dir as an undocumented synonym, and the same for
2020 --exclude-dir, though that is not available in GNU grep, at least as of
2021 release 2.5.4.
2023 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2024 characters from a string of bytes) have been redefined so as not to use
2025 loops, in order to improve performance in some environments. At the same
2026 time, I abstracted some of the common code into auxiliary macros to save
2027 repetition (this should not affect the compiled code).
2029 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2030 compile-time error is now given if \c is not followed by an ASCII
2031 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2032 different, and any byte value is allowed.)
2034 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2035 START_OPTIMIZE option, which is now allowed at compile time - but just
2036 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2037 to pcregrep and other applications that have no direct access to PCRE
2038 options. The new /Y option in pcretest sets this option when calling
2039 pcre_compile().
2041 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2042 back references. Groups containing recursive back references were forced to
2043 be atomic by that change, but in the case of named groups, the amount of
2044 memory required was incorrectly computed, leading to "Failed: internal
2045 error: code overflow". This has been fixed.
2047 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2048 pcretest.c, to avoid build problems in some Borland environments.
2051 Version 8.10 25-Jun-2010
2052 ------------------------
2054 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2055 THEN.
2057 2. (*ACCEPT) was not working when inside an atomic group.
2059 3. Inside a character class, \B is treated as a literal by default, but
2060 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2061 causes the error). The code is unchanged, but I tidied the documentation.
2063 4. Inside a character class, PCRE always treated \R and \X as literals,
2064 whereas Perl faults them if its -w option is set. I have changed PCRE so
2065 that it faults them when PCRE_EXTRA is set.
2067 5. Added support for \N, which always matches any character other than
2068 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2070 6. When compiling pcregrep with newer versions of gcc which may have
2071 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2072 declared with attribute warn_unused_result" were given. Just casting the
2073 result to (void) does not stop the warnings; a more elaborate fudge is
2074 needed. I've used a macro to implement this.
2076 7. Minor change to pcretest.c to avoid a compiler warning.
2078 8. Added four artifical Unicode properties to help with an option to make
2079 \s etc use properties (see next item). The new properties are: Xan
2080 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2082 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2083 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2084 this option. Modified pcretest to add /W to test this facility. Added
2085 REG_UCP to make it available via the POSIX interface.
2087 10. Added --line-buffered to pcregrep.
2089 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2090 studied, and the match started with a letter with a code point greater than
2091 127 whose first byte was different to the first byte of the other case of
2092 the letter, the other case of this starting letter was not recognized
2093 (#976).
2095 12. If a pattern that was studied started with a repeated Unicode property
2096 test, for example, \p{Nd}+, there was the theoretical possibility of
2097 setting up an incorrect bitmap of starting bytes, but fortunately it could
2098 not have actually happened in practice until change 8 above was made (it
2099 added property types that matched character-matching opcodes).
2101 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2102 possible starting bytes for non-anchored patterns.
2104 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2105 \R, and also a number of cases that involve Unicode properties, both
2106 explicit and implicit when PCRE_UCP is set.
2108 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2109 input, it could crash or give wrong results if characters with values
2110 greater than 0xc0 were present in the subject string. (Detail: it assumed
2111 UTF-8 input when processing these items.)
2113 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2114 size_t is 64-bit (#991).
2116 17. Added a check for running out of memory when PCRE is compiled with
2117 --disable-stack-for-recursion (#990).
2119 18. If the last data line in a file for pcretest does not have a newline on
2120 the end, a newline was missing in the output.
2122 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2123 less than 128) in its various bitmaps. However, there is a facility for
2124 generating tables according to the current locale when PCRE is compiled. It
2125 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2126 characters, are recognized by isspace() and therefore were getting set in
2127 these tables, and indeed these tables seem to approximate to ISO 8859. This
2128 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2129 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2130 which of course cannot start UTF-8 characters. I have changed the code so
2131 that only real ASCII characters (less than 128) and the correct starting
2132 bytes for UTF-8 encodings are set for characters greater than 127 when in
2133 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2134 altogether.)
2136 20. Added the /T option to pcretest so as to be able to run tests with non-
2137 standard character tables, thus making it possible to include the tests
2138 used for 19 above in the standard set of tests.
2140 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2141 reference to a subpattern the other side of a comment that contains an
2142 opening parenthesis caused either an internal compiling error, or a
2143 reference to the wrong subpattern.
2146 Version 8.02 19-Mar-2010
2147 ------------------------
2149 1. The Unicode data tables have been updated to Unicode 5.2.0.
2151 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2152 configured.
2154 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2155 original author of that file, following a query about its status.
2157 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2158 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2160 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2161 quantifier applied to a forward-referencing subroutine call, could compile
2162 incorrect code or give the error "internal error: previously-checked
2163 referenced subpattern not found".
2165 6. Both MS Visual Studio and Symbian OS have problems with initializing
2166 variables to point to external functions. For these systems, therefore,
2167 pcre_malloc etc. are now initialized to local functions that call the
2168 relevant global functions.
2170 7. There were two entries missing in the vectors called coptable and poptable
2171 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2172 I've fixed the data, and added a kludgy way of testing at compile time that
2173 the lengths are correct (equal to the number of opcodes).
2175 8. Following on from 7, I added a similar kludge to check the length of the
2176 eint vector in pcreposix.c.
2178 9. Error texts for pcre_compile() are held as one long string to avoid too
2179 much relocation at load time. To find a text, the string is searched,
2180 counting zeros. There was no check for running off the end of the string,
2181 which could happen if a new error number was added without updating the
2182 string.
2184 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2186 11. \K was not working if it appeared in an atomic group or in a group that
2187 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2188 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2189 the assertion is positive, but not if it is negative.
2191 12. Change 11 fortuitously reduced the size of the stack frame used in the
2192 "match()" function of pcre_exec.c by one pointer. Forthcoming
2193 implementation of support for (*MARK) will need an extra pointer on the
2194 stack; I have reserved it now, so that the stack frame size does not
2195 decrease.
2197 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2198 item in branch that calls a recursion is a subroutine call - as in the
2199 second branch in the above example - was incorrectly given the compile-
2200 time error "recursive call could loop indefinitely" because pcre_compile()
2201 was not correctly checking the subroutine for matching a non-empty string.
2203 14. The checks for overrunning compiling workspace could trigger after an
2204 overrun had occurred. This is a "should never occur" error, but it can be
2205 triggered by pathological patterns such as hundreds of nested parentheses.
2206 The checks now trigger 100 bytes before the end of the workspace.
2208 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2211 Version 8.01 19-Jan-2010
2212 ------------------------
2214 1. If a pattern contained a conditional subpattern with only one branch (in
2215 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2216 computed the wrong minimum data length (which is of course zero for such
2217 subpatterns). This could cause incorrect "no match" results.
2219 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2220 the pattern is reset in the first branch, pcre_compile() failed with
2221 "internal error: code overflow at offset...". This happened only when
2222 the reset was to the original external option setting. (An optimization
2223 abstracts leading options settings into an external setting, which was the
2224 cause of this.)
2226 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2227 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2228 assertion pattern did not match (meaning that the assertion was true), it
2229 was incorrectly treated as false if the SKIP had been reached during the
2230 matching. This also applied to assertions used as conditions.
2232 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2233 assertion subpattern, including such a pattern used as a condition,
2234 unpredictable results occurred, instead of the error return
2237 5. The C++ GlobalReplace function was not working like Perl for the special
2238 situation when an empty string is matched. It now does the fancy magic
2239 stuff that is necessary.
2241 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2242 removed. (These were left over from very, very early versions of PCRE.)
2244 7. Some cosmetic changes to the code to make life easier when compiling it
2245 as part of something else:
2247 (a) Change DEBUG to PCRE_DEBUG.
2249 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2250 called "current" as "current_branch", to prevent a collision with the
2251 Linux macro when compiled as a kernel module.
2253 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2254 prevent a collision with the Linux macro when compiled as a kernel
2255 module.
2257 8. In pcre_compile() there are some checks for integer overflows that used to
2258 cast potentially large values to (double). This has been changed to that
2259 when building, a check for int64_t is made, and if it is found, it is used
2260 instead, thus avoiding the use of floating point arithmetic. (There is no
2261 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2262 double.
2264 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2265 2005 (difference between two addresses compared to an unsigned value).
2267 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2268 custom one, because of the following reported problem in Windows:
2270 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2271 under Win32.
2272 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2273 therefore missing the function definition.
2274 - The compiler thus generates a "C" signature for the test function.
2275 - The linker fails to find the "C" function.
2276 - PCRE fails to configure if asked to do so against libbz2.
2278 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2279 messages were output:
2281 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2282 rerunning libtoolize, to keep the correct libtool macros in-tree.
2283 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2285 I have done both of these things.
2287 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2288 most of the time, it *can* run out if it is given a pattern that contains a
2289 runaway infinite recursion. I updated the discussion in the pcrestack man
2290 page.
2292 13. Now that we have gone to the x.xx style of version numbers, the minor
2293 version may start with zero. Using 08 or 09 is a bad idea because users
2294 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2295 interpreted as invalid octal numbers. I've updated the previous comment in
2296 configure.ac, and also added a check that gives an error if 08 or 09 are
2297 used.
2299 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2300 causing partial matching to fail when the end of the subject matched \W
2301 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2303 15. There were some discrepancies between the declarations in pcre_internal.h
2304 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2305 their definitions. The declarations used "const uschar *" and the
2306 definitions used USPTR. Even though USPTR is normally defined as "const
2307 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2308 reported that: "This difference in casting confuses some C++ compilers, for
2309 example, SunCC recognizes above declarations as different functions and
2310 generates broken code for hbpcre." I have changed the declarations to use
2311 USPTR.
2313 16. GNU libtool is named differently on some systems. The autogen.sh script now
2314 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2315 (FreeBSD).
2317 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2318 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2319 comment: "Figure out how to create a longlong from a string: strtoll and
2320 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2321 instance, but it only takes 2 args instead of 3!"
2323 18. A subtle bug concerned with back references has been fixed by a change of
2324 specification, with a corresponding code fix. A pattern such as
2325 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2326 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2327 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2328 same bug. Such groups have to be quantified to be useful, or contained
2329 inside another quantified group. (If there's no repetition, the reference
2330 can never match.) The problem arises because, having left the group and
2331 moved on to the rest of the pattern, a later failure that backtracks into
2332 the group uses the captured value from the final iteration of the group
2333 rather than the correct earlier one. I have fixed this in PCRE by forcing
2334 any group that contains a reference to itself to be an atomic group; that
2335 is, there cannot be any backtracking into it once it has completed. This is
2336 similar to recursive and subroutine calls.
2339 Version 8.00 19-Oct-09
2340 ----------------------
2342 1. The table for translating pcre_compile() error codes into POSIX error codes
2343 was out-of-date, and there was no check on the pcre_compile() error code
2344 being within the table. This could lead to an OK return being given in
2345 error.
2347 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2348 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2349 in a Windows environment.
2351 3. The pcregrep --count option prints the count for each file even when it is
2352 zero, as does GNU grep. However, pcregrep was also printing all files when
2353 --files-with-matches was added. Now, when both options are given, it prints
2354 counts only for those files that have at least one match. (GNU grep just
2355 prints the file name in this circumstance, but including the count seems
2356 more useful - otherwise, why use --count?) Also ensured that the
2357 combination -clh just lists non-zero counts, with no names.
2359 4. The long form of the pcregrep -F option was incorrectly implemented as
2360 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2361 but it seems right to fix it, and I didn't think it was worth preserving
2362 the old behaviour.
2364 5. The command line items --regex=pattern and --regexp=pattern were not
2365 recognized by pcregrep, which required --regex pattern or --regexp pattern
2366 (with a space rather than an '='). The man page documented the '=' forms,
2367 which are compatible with GNU grep; these now work.
2369 6. No libpcreposix.pc file was created for pkg-config; there was just
2370 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2372 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2373 when UCP support is not needed, by modifying the Python script that
2374 generates it from Unicode data files. This should not matter if the module
2375 is correctly used as a library, but I received one complaint about 50K of
2376 unwanted data. My guess is that the person linked everything into his
2377 program rather than using a library. Anyway, it does no harm.
2379 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2380 was a minimum greater than 1 for a wide character in a possessive
2381 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2382 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2383 character. Chaos in the form of incorrect output or a compiling loop could
2384 result.
2386 9. The restrictions on what a pattern can contain when partial matching is
2387 requested for pcre_exec() have been removed. All patterns can now be
2388 partially matched by this function. In addition, if there are at least two
2389 slots in the offset vector, the offset of the earliest inspected character
2390 for the match and the offset of the end of the subject are set in them when
2391 PCRE_ERROR_PARTIAL is returned.
2393 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2394 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2395 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2396 and may be more useful for multi-segment matching.
2398 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2399 used to be given if ever the end of the subject was reached; now it is
2400 given only if matching could not proceed because another character was
2401 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2402 string "Z", which now yields "no match" instead of "partial match". In the
2403 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2404 final character ended with (*FAIL).
2406 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2407 if the pattern had a "must contain" character that was already found in the
2408 earlier partial match, unless partial matching was again requested. For
2409 example, with the pattern /dog.(body)?/, the "must contain" character is
2410 "g". If the first part-match was for the string "dog", restarting with
2411 "sbody" failed. This bug has been fixed.
2413 13. The string returned by pcre_dfa_exec() after a partial match has been
2414 changed so that it starts at the first inspected character rather than the
2415 first character of the match. This makes a difference only if the pattern
2416 starts with a lookbehind assertion or \b or \B (\K is not supported by
2417 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2418 matching functions compatible, and I think it's the right thing to do.
2420 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2421 so that the demonstration program is easily available in environments where
2422 PCRE has not been installed from source.
2424 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2425 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2426 library.
2428 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2429 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2430 is not the first non-POSIX option to be added. Clearly some people find
2431 these options useful.
2433 17. If a caller to the POSIX matching function regexec() passes a non-zero
2434 value for nmatch with a NULL value for pmatch, the value of
2435 nmatch is forced to zero.
2437 18. RunGrepTest did not have a test for the availability of the -u option of
2438 the diff command, as RunTest does. It now checks in the same way as
2439 RunTest, and also checks for the -b option.
2441 19. If an odd number of negated classes containing just a single character
2442 interposed, within parentheses, between a forward reference to a named
2443 subpattern and the definition of the subpattern, compilation crashed with
2444 an internal error, complaining that it could not find the referenced
2445 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2446 [The bug was that it was starting one character too far in when skipping
2447 over the character class, thus treating the ] as data rather than
2448 terminating the class. This meant it could skip too much.]
2450 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2451 /g option in pcretest when the pattern contains \K, which makes it possible
2452 to have an empty string match not at the start, even when the pattern is
2453 anchored. Updated pcretest and pcredemo to use this option.
2455 21. If the maximum number of capturing subpatterns in a recursion was greater
2456 than the maximum at the outer level, the higher number was returned, but
2457 with unset values at the outer level. The correct (outer level) value is
2458 now given.
2460 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2461 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2462 make it do so. The string so far is captured, making this feature
2463 compatible with Perl.
2465 23. The tests have been re-organized, adding tests 11 and 12, to make it
2466 possible to check the Perl 5.10 features against Perl 5.10.
2468 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2469 pattern matches a fixed length string. PCRE did not allow this; now it
2470 does. Neither allows recursion.
2472 25. I finally figured out how to implement a request to provide the minimum
2473 length of subject string that was needed in order to match a given pattern.
2474 (It was back references and recursion that I had previously got hung up
2475 on.) This code has now been added to pcre_study(); it finds a lower bound
2476 to the length of subject needed. It is not necessarily the greatest lower
2477 bound, but using it to avoid searching strings that are too short does give
2478 some useful speed-ups. The value is available to calling programs via
2479 pcre_fullinfo().
2481 26. While implementing 25, I discovered to my embarrassment that pcretest had
2482 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2483 study optimizations had never been tested with that matching function.
2484 Oops. What is worse, even when it was passed study data, there was a bug in
2485 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2486 were also very few tests of studied patterns with pcre_dfa_exec().
2488 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2489 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2490 on the other side of the coin, they are no longer allowed to have different
2491 names, because these cannot be distinguished in PCRE, and this has caused
2492 confusion. (This is a difference from Perl.)
2494 28. When duplicate subpattern names are present (necessarily with different
2495 numbers, as required by 27 above), and a test is made by name in a
2496 conditional pattern, either for a subpattern having been matched, or for
2497 recursion in such a pattern, all the associated numbered subpatterns are
2498 tested, and the overall condition is true if the condition is true for any
2499 one of them. This is the way Perl works, and is also more like the way
2500 testing by number works.
2503 Version 7.9 11-Apr-09
2504 ---------------------
2506 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2507 (pcretest), all targets were linked against these libraries. This included
2508 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2509 libraries. This caused unwanted dependencies to be created. This problem
2510 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2511 pcretest is linked with readline.
2513 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2514 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2515 moved outside it again, because FALSE and TRUE are already defined in AIX,
2516 but BOOL is not.
2518 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2519 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2521 4. The pcregrep documentation said spaces were inserted as well as colons (or
2522 hyphens) following file names and line numbers when outputting matching
2523 lines. This is not true; no spaces are inserted. I have also clarified the
2524 wording for the --colour (or --color) option.
2526 5. In pcregrep, when --colour was used with -o, the list of matching strings
2527 was not coloured; this is different to GNU grep, so I have changed it to be
2528 the same.
2530 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2531 each matching line was coloured. Now it goes on to look for further matches
2532 of any of the test patterns, which is the same behaviour as GNU grep.
2534 7. A pattern that could match an empty string could cause pcregrep to loop; it
2535 doesn't make sense to accept an empty string match in pcregrep, so I have
2536 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2537 seems to be how GNU grep behaves. [But see later change 40 for release
2538 8.33.]
2540 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2541 start or after a newline", because the conditional assertion was not being
2542 correctly handled. The rule now is that both the assertion and what follows
2543 in the first alternative must satisfy the test.
2545 9. If auto-callout was enabled in a pattern with a conditional group whose
2546 condition was an assertion, PCRE could crash during matching, both with
2547 pcre_exec() and pcre_dfa_exec().
2549 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2550 used for matching.
2552 11. Unicode property support in character classes was not working for
2553 characters (bytes) greater than 127 when not in UTF-8 mode.
2555 12. Added the -M command line option to pcretest.
2557 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2559 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2561 16. Added comments and documentation about mis-use of no_arg in the C++
2562 wrapper.
2564 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2565 from Martin Jerabek that uses macro names for all relevant character and
2566 string constants.
2568 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2569 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2570 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2571 these, but not everybody uses configure.
2573 19. A conditional group that had only one branch was not being correctly
2574 recognized as an item that could match an empty string. This meant that an
2575 enclosing group might also not be so recognized, causing infinite looping
2576 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2577 with the subject "ab", where knowledge that the repeated group can match
2578 nothing is needed in order to break the loop.
2580 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2581 exec(), but without supplying a callout function, matching went wrong.
2583 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2584 leak if the size of the offset vector was greater than 30. When the vector
2585 is smaller, the saved offsets during recursion go onto a local stack
2586 vector, but for larger vectors malloc() is used. It was failing to free
2587 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2588 error, in fact).
2590 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2591 heapframe that is used only when UTF-8 support is enabled. This caused no
2592 problem, but was untidy.
2594 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2595 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2596 included within another project.
2598 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2599 slightly modified by me:
2601 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2602 not building pcregrep.
2604 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2605 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2607 25. Forward references, both numeric and by name, in patterns that made use of
2608 duplicate group numbers, could behave incorrectly or give incorrect errors,
2609 because when scanning forward to find the reference group, PCRE was not
2610 taking into account the duplicate group numbers. A pattern such as
2611 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2613 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2614 the feature of a custom pointer more persuasive (as requested by a user).
2616 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2617 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2618 pre-defined.
2620 28. Added support for (*UTF8) at the start of a pattern.
2622 29. Arrange for flags added by the "release type" setting in CMake to be shown
2623 in the configuration summary.
2626 Version 7.8 05-Sep-08
2627 ---------------------
2629 1. Replaced UCP searching code with optimized version as implemented for Ad
2630 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2631 stage table and inline lookup instead of a function, giving speed ups of 2
2632 to 5 times on some simple patterns that I tested. Permission was given to
2633 distribute the MultiStage2.py script that generates the tables (it's not in
2634 the tarball, but is in the Subversion repository).
2636 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2637 scripts.
2639 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2640 a group with a zero qualifier. The result of the study could be incorrect,
2641 or the function might crash, depending on the pattern.
2643 4. Caseless matching was not working for non-ASCII characters in back
2644 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2645 It now works when Unicode Property Support is available.
2647 5. In pcretest, an escape such as \x{de} in the data was always generating
2648 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2649 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2650 truncation.
2652 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2654 7. Added two (int) casts to pcregrep when printing the difference of two
2655 pointers, in case they are 64-bit values.
2657 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2658 test 2 if it fails.
2660 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2661 and a #define of that name to empty if it is not externally set. This is to
2662 allow users of MSVC to set it if necessary.
2664 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2665 the convenience functions in the pcre_get.c source file.
2667 11. An option change at the start of a pattern that had top-level alternatives
2668 could cause overwriting and/or a crash. This command provoked a crash in
2669 some environments:
2671 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2673 This potential security problem was recorded as CVE-2008-2371.
2675 12. For a pattern where the match had to start at the beginning or immediately
2676 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2677 pcre_dfa_exec() could read past the end of the passed subject if there was
2678 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2679 pcretest so that it places the subject at the end of its malloc-ed buffer.
2681 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2682 exec() might read past the end of the data buffer in UTF-8 mode.
2684 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2685 the data contained the byte 0x85 as part of a UTF-8 character within its
2686 first line. This applied both to normal and DFA matching.
2688 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2689 /^[^d]*?$/8 failed to match "abc".
2691 16. Added a missing copyright notice to pcrecpp_internal.h.
2693 17. Make it more clear in the documentation that values returned from
2694 pcre_exec() in ovector are byte offsets, not character counts.
2696 18. Tidied a few places to stop certain compilers from issuing warnings.
2698 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2699 supplied by Stefan Weber. I made a further small update for 7.8 because
2700 there is a change of source arrangements: the pcre_searchfuncs.c module is
2701 replaced by pcre_ucd.c.
2704 Version 7.7 07-May-08
2705 ---------------------
2707 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2708 a string to a long long, pretend we don't even have a long long." This is
2709 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2711 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2712 pre-7.6 versions, which defined a global no_arg variable instead of putting
2713 it in the RE class. (See also #8 below.)
2715 3. Remove a line of dead code, identified by coverity and reported by Nuno
2716 Lopes.
2718 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2720 (1) The include/exclude patterns were being applied to the whole pathnames
2721 of files, instead of just to the final components.
2723 (2) If there was more than one level of directory, the subdirectories were
2724 skipped unless they satisfied the include/exclude conditions. This is
2725 inconsistent with GNU grep (and could even be seen as contrary to the
2726 pcregrep specification - which I improved to make it absolutely clear).
2727 The action now is always to scan all levels of directory, and just
2728 apply the include/exclude patterns to regular files.
2730 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2731 --exclude_dir in the tests to avoid scanning .svn directories.
2733 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2734 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2735 doesn't support NULs in patterns.
2737 7. Added some missing "const"s to declarations of static tables in
2738 pcre_compile.c and pcre_dfa_exec.c.
2740 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2741 caused by fix #2 above. (Subsequently also a second patch to fix the
2742 first patch. And a third patch - this was a messy problem.)
2744 9. Applied Craig's patch to remove the use of push_back().
2746 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2747 matching function regexec().
2749 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2750 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2751 references. PCRE supports relative numbers with this syntax (I don't think
2752 Oniguruma does).
2754 12. Previously, a group with a zero repeat such as (...){0} was completely
2755 omitted from the compiled regex. However, this means that if the group
2756 was called as a subroutine from elsewhere in the pattern, things went wrong
2757 (an internal error was given). Such groups are now left in the compiled
2758 pattern, with a new opcode that causes them to be skipped at execution
2759 time.
2761 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2762 to the way PCRE behaves:
2764 (a) A lone ] character is dis-allowed (Perl treats it as data).
2766 (b) A back reference to an unmatched subpattern matches an empty string
2767 (Perl fails the current match path).
2769 (c) A data ] in a character class must be notated as \] because if the
2770 first data character in a class is ], it defines an empty class. (In
2771 Perl it is not possible to have an empty class.) The empty class []
2772 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2773 The negative empty class [^] matches any one character, independently
2774 of the DOTALL setting.
2776 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2777 non-existent subpattern following a character class starting with ']' and
2778 containing () gave an internal compiling error instead of "reference to
2779 non-existent subpattern". Fortunately, when the pattern did exist, the
2780 compiled code was correct. (When scanning forwards to check for the
2781 existence of the subpattern, it was treating the data ']' as terminating
2782 the class, so got the count wrong. When actually compiling, the reference
2783 was subsequently set up correctly.)
2785 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2786 it was being rejected as not supported by pcre_dfa_exec(), even though
2787 other assertions are supported. I have made pcre_dfa_exec() support
2788 (*FAIL).
2790 16. The implementation of 13c above involved the invention of a new opcode,
2791 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2792 cannot be changed at match time, I realized I could make a small
2793 improvement to matching performance by compiling OP_ALLANY instead of
2794 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2795 on the OP_ANY path.
2797 17. Compiling pcretest on Windows with readline support failed without the
2798 following two fixes: (1) Make the unistd.h include conditional on
2799 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2801 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2802 ncurses library to be included for pcretest when ReadLine support is
2803 requested, but also to allow for it to be overridden. This patch came from
2804 Daniel Bergström.
2806 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2807 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2808 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2809 spotting this.
2812 Version 7.6 28-Jan-08
2813 ---------------------
2815 1. A character class containing a very large number of characters with
2816 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2817 overflow.
2819 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2820 HAVE_LONG_LONG is not defined.
2822 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2823 bring them up to date and include new features. This patch includes:
2825 - Fixed PH's badly added libz and libbz2 support.
2826 - Fixed a problem with static linking.
2827 - Added pcredemo. [But later removed - see 7 below.]
2828 - Fixed dftables problem and added an option.
2829 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2831 - Added readline support for pcretest.
2832 - Added an listing of the option settings after cmake has run.
2834 4. A user submitted a patch to Makefile that makes it easy to create
2835 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2836 Makefile.am that cause it to include this special target, without
2837 affecting anything else. Note that the same mingw target plus all
2838 the other distribution libraries and programs are now supported
2839 when configuring with CMake (see 6 below) instead of with
2840 Configure/Make.
2842 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2843 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2844 exported in the Windows port". It has not yet been confirmed that the patch
2845 solves the problem, but it does no harm.
2847 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2848 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2849 with CMake, and also correct the comment about stack recursion.
2851 7. Remove the automatic building of pcredemo from the ./configure system and
2852 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2853 of a program that users should build themselves after PCRE is installed, so
2854 building it automatically is not really right. What is more, it gave
2855 trouble in some build environments.
2857 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2860 Version 7.5 10-Jan-08
2861 ---------------------
2863 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2864 values in parens when parsing an RE using the C++ wrapper."
2866 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2867 Characters greater than 255 were excluded from the class instead of being
2868 included.
2870 3. The same bug as (2) above applied to negated POSIX classes such as
2871 [:^space:].
2873 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2874 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2875 I have changed it.
2877 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2878 first named subpattern) and a construct such as (?&a) would reference the
2879 first named subpattern whose name started with "a" (in other words, the
2880 length check was missing). Both these problems are fixed. "Subpattern name
2881 expected" is now given for (?&) (a zero-length name), and this patch also
2882 makes it give the same error for \k'' (previously it complained that that
2883 was a reference to a non-existent subpattern).
2885 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2886 this is right because (?- can be followed by option settings as well as by
2887 digits. I have, however, made the messages clearer.
2889 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2890 than the number used in the conditional) now cause a compile-time error.
2891 This is actually not compatible with Perl, which accepts such patterns, but
2892 treats the conditional as always being FALSE (as PCRE used to), but it
2893 seems to me that giving a diagnostic is better.
2895 8. Change "alphameric" to the more common word "alphanumeric" in comments
2896 and messages.
2898 9. Fix two occurrences of "backslash" in comments that should have been
2899 "backspace".
2901 10. Remove two redundant lines of code that can never be obeyed (their function
2902 was moved elsewhere).
2904 11. The program that makes PCRE's Unicode character property table had a bug
2905 which caused it to generate incorrect table entries for sequences of
2906 characters that have the same character type, but are in different scripts.
2907 It amalgamated them into a single range, with the script of the first of
2908 them. In other words, some characters were in the wrong script. There were
2909 thirteen such cases, affecting characters in the following ranges:
2911 U+002b0 - U+002c1
2912 U+0060c - U+0060d
2913 U+0061e - U+00612
2914 U+0064b - U+0065e
2915 U+0074d - U+0076d
2916 U+01800 - U+01805
2917 U+01d00 - U+01d77
2918 U+01d9b - U+01dbf
2919 U+0200b - U+0200f
2920 U+030fc - U+030fe
2921 U+03260 - U+0327f
2922 U+0fb46 - U+0fbb1
2923 U+10450 - U+1049d
2925 12. The -o option (show only the matching part of a line) for pcregrep was not
2926 compatible with GNU grep in that, if there was more than one match in a
2927 line, it showed only the first of them. It now behaves in the same way as
2928 GNU grep.
2930 13. If the -o and -v options were combined for pcregrep, it printed a blank
2931 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2932 does the same. The return code can be used to tell if there were any
2933 non-matching lines.
2935 14. Added --file-offsets and --line-offsets to pcregrep.
2937 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2938 infinitely looping recursion. The bug was that positive lookaheads were not
2939 being skipped when checking for a possible empty match (negative lookaheads
2940 and both kinds of lookbehind were skipped).
2942 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2943 inclusion of <windows.h> to before rather than after the definition of
2944 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2946 17. Specifying a possessive quantifier with a specific limit for a Unicode
2947 character property caused pcre_compile() to compile bad code, which led at
2948 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2949 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2950 caused the error; without that there was no problem.
2952 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2954 19. Added --enable-pcretest-libreadline.
2956 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2957 RE::GlobalReplace(). As a result, the number of replacements returned was
2958 double what it should be. I removed one of the increments, but Craig sent a
2959 later patch that removed the other one (the right fix) and added unit tests
2960 that check the return values (which was not done before).
2962 21. Several CMake things:
2964 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2965 the names libpcre and libpcreposix, not just pcre and pcreposix.
2967 (2) The above change means that pcretest and pcregrep are now correctly
2968 linked with the newly-built libraries, not previously installed ones.
2972 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2973 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2974 UTF-8 newline character). The key issue is that the pattern starts .*;
2975 this means that the match must be either at the beginning, or after a
2976 newline. The bug was in the code for advancing after a failed match and
2977 checking that the new position followed a newline. It was not taking
2978 account of UTF-8 characters correctly.
2980 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2981 character classes. PCRE was not treating the sequence [:...:] as a
2982 character class unless the ... were all letters. Perl, however, seems to
2983 allow any characters between [: and :], though of course it rejects as
2984 unknown any "names" that contain non-letters, because all the known class
2985 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2986 for example, whereas PCRE did not - it did not recognize a POSIX character
2987 class. This seemed a bit dangerous, so the code has been changed to be
2988 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2989 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2990 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2991 Perl does, and where it didn't before.
2993 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2994 Windows environments %n is disabled by default.
2997 Version 7.4 21-Sep-07
2998 ---------------------
3000 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3001 means that a class such as [\s] counted as "explicit reference to CR or
3002 LF". That isn't really right - the whole point of the change was to try to
3003 help when there was an actual mention of one of the two characters. So now
3004 the change happens only if \r or \n (or a literal CR or LF) character is
3005 encountered.
3007 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3008 of both had grown to the point where there were only 3 bits left.
3009 Fortunately, there was spare space in the data structure, and so I have
3010 moved the internal flags into a new 16-bit field to free up more option
3011 bits.
3013 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3014 but did not set the internal JCHANGED flag - either of these is enough to
3015 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3016 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3017 start sets both bits.
3019 4. Added options (at build time, compile time, exec time) to change \R from
3020 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3022 5. doc/pcresyntax.html was missing from the distribution.
3024 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3025 compatibility, even though it is no longer used.
3027 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3028 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3029 windows.h file is present (where different names are used). [This was
3030 reversed later after testing - see 16 below.]
3032 8. Changed all #include <config.h> to #include "config.h". There were also
3033 some further <pcre.h> cases that I changed to "pcre.h".
3035 9. When pcregrep was used with the --colour option, it missed the line ending
3036 sequence off the lines that it output.
3038 10. It was pointed out to me that arrays of string pointers cause lots of
3039 relocations when a shared library is dynamically loaded. A technique of
3040 using a single long string with a table of offsets can drastically reduce
3041 these. I have refactored PCRE in four places to do this. The result is
3042 dramatic:
3044 Originally: 290
3045 After changing UCP table: 187
3046 After changing error message table: 43
3047 After changing table of "verbs" 36
3048 After changing table of Posix names 22
3050 Thanks to the folks working on Gregex for glib for this insight.
3052 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3053 unicode-properties was also set.
3055 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3057 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3058 checked only for CRLF.
3060 14. Added casts to pcretest.c to avoid compiler warnings.
3062 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3064 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3065 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3066 entirely. This removes changes made in 7 above.
3068 17. The CMake files have been updated, and there is now more information about
3069 building with CMake in the NON-UNIX-USE document.
3072 Version 7.3 28-Aug-07
3073 ---------------------
3075 1. In the rejigging of the build system that eventually resulted in 7.1, the
3076 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3077 brackets there is not right, since it causes compilers to look for an
3078 installed pcre.h, not the version that is in the source that is being
3079 compiled (which of course may be different). I have changed it back to:
3081 #include "pcre.h"
3083 I have a vague recollection that the change was concerned with compiling in
3084 different directories, but in the new build system, that is taken care of
3085 by the VPATH setting the Makefile.
3087 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3088 when the subject happened to end in the byte 0x85 (e.g. if the last
3089 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3090 characters but of course it shouldn't be taken as a newline when it is part
3091 of another character. The bug was that, for an unlimited repeat of . in
3092 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3093 characters when looking for a newline.
3095 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3097 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3098 in debug output.
3100 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3101 long printing in the pcrecpp unittest when running under MinGW.
3103 6. ESC_K was left out of the EBCDIC table.
3105 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3106 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3107 limit also applies to "virtual nesting" when a pattern is recursive, and in
3108 this case 1000 isn't so big. I have been able to remove this limit at the
3109 expense of backing off one optimization in certain circumstances. Normally,
3110 when pcre_exec() would call its internal match() function recursively and
3111 immediately return the result unconditionally, it uses a "tail recursion"
3112 feature to save stack. However, when a subpattern that can match an empty
3113 string has an unlimited repetition quantifier, it no longer makes this
3114 optimization. That gives it a stack frame in which to save the data for
3115 checking that an empty string has been matched. Previously this was taken
3116 from the 1000-entry workspace that had been reserved. So now there is no
3117 explicit limit, but more stack is used.
3119 8. Applied Daniel's patches to solve problems with the import/export magic
3120 syntax that is required for Windows, and which was going wrong for the
3121 pcreposix and pcrecpp parts of the library. These were overlooked when this
3122 problem was solved for the main library.
3124 9. There were some crude static tests to avoid integer overflow when computing
3125 the size of patterns that contain repeated groups with explicit upper
3126 limits. As the maximum quantifier is 65535, the maximum group length was
3127 set at 30,000 so that the product of these two numbers did not overflow a
3128 32-bit integer. However, it turns out that people want to use groups that
3129 are longer than 30,000 bytes (though not repeat them that many times).
3130 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3131 made it possible to implement the integer overflow checks in a much more
3132 dynamic way, which I have now done. The artificial limitation on group
3133 length has been removed - we now have only the limit on the total length of
3134 the compiled pattern, which depends on the LINK_SIZE setting.
3136 10. Fixed a bug in the documentation for get/copy named substring when
3137 duplicate names are permitted. If none of the named substrings are set, the
3138 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3139 empty string.
3141 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3142 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3143 because the ] is interpreted as the first data character and the
3144 terminating ] is not found. PCRE has been made compatible with Perl in this
3145 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3146 cause memory overwriting.
3148 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3149 string has been matched (to stop an infinite loop). It was not recognizing
3150 a conditional subpattern that could match an empty string if that
3151 subpattern was within another subpattern. For example, it looped when
3152 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3153 condition was not nested. This bug has been fixed.
3155 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3156 past the start of the subject in the presence of bytes with the top bit
3157 set, for example "\x8aBCD".
3159 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3160 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3162 14. Optimized (?!) to (*FAIL).
3164 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3165 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3166 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3167 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3168 does: it's just the validity check that is more restrictive.
3170 16. Inserted checks for integer overflows during escape sequence (backslash)
3171 processing, and also fixed erroneous offset values for syntax errors during
3172 backslash processing.
3174 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3175 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3177 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3178 caused an overrun.
3180 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3181 something other than just ASCII characters) inside a group that had an
3182 unlimited repeat caused a loop at compile time (while checking to see
3183 whether the group could match an empty string).
3185 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3186 [\P{Any}] did so. (Error in the code for printing property names.)
3188 21. An orphan \E inside a character class could cause a crash.
3190 22. A repeated capturing bracket such as (A)? could cause a wild memory
3191 reference during compilation.
3193 23. There are several functions in pcre_compile() that scan along a compiled
3194 expression for various reasons (e.g. to see if it's fixed length for look
3195 behind). There were bugs in these functions when a repeated \p or \P was
3196 present in the pattern. These operators have additional parameters compared
3197 with \d, etc, and these were not being taken into account when moving along
3198 the compiled data. Specifically:
3200 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3201 length.
3203 (b) An item such as \pL+ within a repeated group could cause crashes or
3204 loops.
3206 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3207 "reference to non-existent subpattern" error.
3209 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3211 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3212 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3214 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3216 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3217 character were causing crashes (broken optimization).
3219 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3220 \p or \P) caused a compile-time loop.
3222 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3223 break. For example, the unstudied pattern [\r\n]A does not match the string
3224 "\r\nA" because change 7.0/46 below moves the current point on by two
3225 characters after failing to match at the start. However, the pattern \nA
3226 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3227 the same is true. There doesn't seem any very clean way out of this, but
3228 what I have chosen to do makes the common cases work: PCRE now takes note
3229 of whether there can be an explicit match for \r or \n anywhere in the
3230 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3231 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3232 pattern has explicit CR or LF references.
3234 29. Added (*CR) etc for changing newline setting at start of pattern.
3237 Version 7.2 19-Jun-07
3238 ---------------------
3240 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3241 which is apparently normally available under Windows.
3243 2. Re-jig the pcregrep tests with different newline settings in an attempt
3244 to make them independent of the local environment's newline setting.
3246 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3248 4. Some of the "internals" tests were previously cut out when the link size
3249 was not 2, because the output contained actual offsets. The recent new
3250 "Z" feature of pcretest means that these can be cut out, making the tests
3251 usable with all link sizes.
3253 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3254 stack recursion. This gives a massive performance boost under BSD, but just
3255 a small improvement under Linux. However, it saves one field in the frame
3256 in all cases.
3258 6. Added more features from the forthcoming Perl 5.10:
3260 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3261 recursion call. It refers to the nth most recently opened parentheses.
3263 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3264 to be opened parentheses.
3266 (c) Conditions that refer to capturing parentheses can be specified
3267 relatively, for example, (?(-2)... or (?(+3)...
3269 (d) \K resets the start of the current match so that everything before
3270 is not part of it.
3272 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3274 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3275 reference syntax.
3277 (g) (?| introduces a group in which the numbering of parentheses in each
3278 alternative starts with the same number.
3280 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3282 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3285 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3286 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3287 for detecting groups that can match an empty string.
3289 9. A pattern with a very large number of alternatives (more than several
3290 hundred) was running out of internal workspace during the pre-compile
3291 phase, where pcre_compile() figures out how much memory will be needed. A
3292 bit of new cunning has reduced the workspace needed for groups with
3293 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3294 workspace instead of running out of the 4096 that are available.
3296 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3298 11. Applied patch from Google to remove an optimization that didn't quite work.
3299 The report of the bug said:
3301 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3302 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3303 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3305 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3306 it matched the wrong number of bytes.
3309 Version 7.1 24-Apr-07
3310 ---------------------
3312 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3313 that is more "standard", making use of automake and other Autotools. There
3314 is some re-arrangement of the files and adjustment of comments consequent
3315 on this.
3317 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3318 for recursive directory scanning broke on some systems because the files
3319 are not scanned in any specific order and on different systems the order
3320 was different. A call to "sort" has been inserted into RunGrepTest for the
3321 approprate test as a short-term fix. In the longer term there may be an
3322 alternative.
3324 3. I had an email from Eric Raymond about problems translating some of PCRE's
3325 man pages to HTML (despite the fact that I distribute HTML pages, some
3326 people do their own conversions for various reasons). The problems
3327 concerned the use of low-level troff macros .br and .in. I have therefore
3328 removed all such uses from the man pages (some were redundant, some could
3329 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3330 HTML has been updated to handle .nf/.fi and to complain if it encounters
3331 .br or .in.
3333 4. Updated comments in configure.ac that get placed in config.h.in and also
3334 arranged for config.h to be included in the distribution, with the name
3335 config.h.generic, for the benefit of those who have to compile without
3336 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3338 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3339 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3340 makevp.bat for the current PCRE, using the additional files
3341 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3343 6. A Windows user reported a minor discrepancy with test 2, which turned out
3344 to be caused by a trailing space on an input line that had got lost in his
3345 copy. The trailing space was an accident, so I've just removed it.
3347 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3348 that is needed.
3350 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3351 as "const" (a) because they are and (b) because it helps the PHP
3352 maintainers who have recently made a script to detect big data structures
3353 in the php code that should be moved to the .rodata section. I remembered
3354 to update Builducptable as well, so it won't revert if ucptable.h is ever
3355 re-created.
3357 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3358 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3359 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3360 support is not required. This saves 1.5-2K of code, which is important in
3361 some applications.
3363 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3364 so as not to refer to the tables, even though these functions will never be
3365 called when UTF-8 support is disabled. Otherwise there are problems with a
3366 shared library.
3368 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3370 (a) It was defining its arguments as char * instead of void *.
3372 (b) It was assuming that all moves were upwards in memory; this was true
3373 a long time ago when I wrote it, but is no longer the case.
3375 The emulated memove() is provided for those environments that have neither
3376 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3377 is clearly not the case, as these two bugs were recently reported.
3379 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3380 and Detrail to create the HTML documentation, the .txt form of the man
3381 pages, and it removes trailing spaces from listed files. It also creates
3382 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3383 case, it wraps all the #defines with #ifndefs. This script should be run
3384 before "make dist".
3386 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3387 with Unicode property support.
3389 (a) For a maximizing quantifier, if the two different cases of the
3390 character were of different lengths in their UTF-8 codings (there are
3391 some cases like this - I found 11), and the matching function had to
3392 back up over a mixture of the two cases, it incorrectly assumed they
3393 were both the same length.
3395 (b) When PCRE was configured to use the heap rather than the stack for
3396 recursion during matching, it was not correctly preserving the data for
3397 the other case of a UTF-8 character when checking ahead for a match
3398 while processing a minimizing repeat. If the check also involved
3399 matching a wide character, but failed, corruption could cause an
3400 erroneous result when trying to check for a repeat of the original
3401 character.
3403 13. Some tidying changes to the testing mechanism:
3405 (a) The RunTest script now detects the internal link size and whether there
3406 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3407 values substituted by "configure". (The RunGrepTest script already did
3408 this for UTF-8.) The configure.ac script no longer substitutes the
3409 relevant variables.
3411 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3412 with length and offset values. This means that the output is different
3413 for different internal link sizes. Test 2 is skipped for link sizes
3414 other than 2 because of this, bypassing the problem. Unfortunately,
3415 there was also a test in test 3 (the locale tests) that used /B and
3416 failed for link sizes other than 2. Rather than cut the whole test out,
3417 I have added a new /Z option to pcretest that replaces the length and
3418 offset values with spaces. This is now used to make test 3 independent
3419 of link size. (Test 2 will be tidied up later.)
3421 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3422 segmentation fault instead of returning the appropriate error message.
3424 15. In multiline mode when the newline sequence was set to "any", the pattern
3425 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3426 This doesn't seem right; it now treats the CRLF combination as the line
3427 ending, and so does not match in that case. It's only a pattern such as ^$
3428 that would hit this one: something like ^ABC$ would have failed after \r
3429 and then tried again after \r\n.
3431 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3432 in an attempt to make files that differ only in their line terminators
3433 compare equal. This works on Linux.
3435 17. Under certain error circumstances pcregrep might try to free random memory
3436 as it exited. This is now fixed, thanks to valgrind.
3438 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3439 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3440 was because its rules for how to advance for /g after matching an empty
3441 string at the end of a line did not allow for this case. They now check for
3442 it specially.
3444 20. pcretest is supposed to handle patterns and data of any length, by
3445 extending its buffers when necessary. It was getting this wrong when the
3446 buffer for a data line had to be extended.
3448 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3449 CRLF as a newline sequence.
3451 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3452 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3453 I have nevertheless tidied it up.
3455 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3457 24. Added a man page for pcre-config.
3460 Version 7.0 19-Dec-06
3461 ---------------------
3463 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3464 moving to gcc 4.1.1.
3466 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3467 sys/time.h, which is documented as needed for this function. It doesn't
3468 seem to matter on Linux, but it showed up on some releases of OS X.
3470 3. It seems that there are systems where bytes whose values are greater than
3471 127 match isprint() in the "C" locale. The "C" locale should be the
3472 default when a C program starts up. In most systems, only ASCII printing
3473 characters match isprint(). This difference caused the output from pcretest
3474 to vary, making some of the tests fail. I have changed pcretest so that:
3476 (a) When it is outputting text in the compiled version of a pattern, bytes
3477 other than 32-126 are always shown as hex escapes.
3479 (b) When it is outputting text that is a matched part of a subject string,
3480 it does the same, unless a different locale has been set for the match
3481 (using the /L modifier). In this case, it uses isprint() to decide.
3483 4. Fixed a major bug that caused incorrect computation of the amount of memory
3484 required for a compiled pattern when options that changed within the
3485 pattern affected the logic of the preliminary scan that determines the
3486 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3487 that the computed length was too small. The symptoms of this bug were
3488 either the PCRE error "internal error: code overflow" from pcre_compile(),
3489 or a glibc crash with a message such as "pcretest: free(): invalid next
3490 size (fast)". Examples of patterns that provoked this bug (shown in
3491 pcretest format) are:
3493 /(?-x: )/x
3494 /(?x)(?-x: \s*#\s*)/
3495 /((?i)[\x{c0}])/8
3496 /(?i:[\x{c0}])/8
3498 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3499 is now done differently.
3501 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3502 wrapper classes; (b) implement a new function in the C++ scanner that is
3503 more efficient than the old way of doing things because it avoids levels of
3504 recursion in the regex matching; (c) add a paragraph to the documentation
3505 for the FullMatch() function.
3507 6. The escape sequence \n was being treated as whatever was defined as
3508 "newline". Not only was this contrary to the documentation, which states
3509 that \n is character 10 (hex 0A), but it also went horribly wrong when
3510 "newline" was defined as CRLF. This has been fixed.
3512 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3513 was being set to -1 for the "end of line" case (supposedly a value that no
3514 character can have). Though this value is never used (the check for end of
3515 line is "zero bytes in current character"), it caused compiler complaints.
3516 I've changed it to 0xffffffff.
3518 8. In pcre_version.c, the version string was being built by a sequence of
3519 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3520 string (as it is for production releases) called a macro with an empty
3521 argument. The C standard says the result of this is undefined. The gcc
3522 compiler treats it as an empty string (which was what was wanted) but it is
3523 reported that Visual C gives an error. The source has been hacked around to
3524 avoid this problem.
3526 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3527 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3528 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3529 of them did).
3531 10. Originally, pcretest opened its input and output without "b"; then I was
3532 told that "b" was needed in some environments, so it was added for release
3533 5.0 to both the input and output. (It makes no difference on Unix-like
3534 systems.) Later I was told that it is wrong for the input on Windows. I've
3535 now abstracted the modes into two macros, to make it easier to fiddle with
3536 them, and removed "b" from the input mode under Windows.
3538 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3540 12. Added -help and --help to pcretest as an official way of being reminded
3541 of the options.
3543 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3544 and pcrecpp.cc because they annoy compilers at high warning levels.
3546 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3548 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3549 scripts are not C programs :-) and which was not noticed because it works
3550 on Linux.
3552 16. pcretest is supposed to handle any length of pattern and data line (as one
3553 line or as a continued sequence of lines) by extending its input buffer if
3554 necessary. This feature was broken for very long pattern lines, leading to
3555 a string of junk being passed to pcre_compile() if the pattern was longer
3556 than about 50K.
3558 17. I have done a major re-factoring of the way pcre_compile() computes the
3559 amount of memory needed for a compiled pattern. Previously, there was code
3560 that made a preliminary scan of the pattern in order to do this. That was
3561 OK when PCRE was new, but as the facilities have expanded, it has become
3562 harder and harder to keep it in step with the real compile phase, and there
3563 have been a number of bugs (see for example, 4 above). I have now found a
3564 cunning way of running the real compile function in a "fake" mode that
3565 enables it to compute how much memory it would need, while actually only
3566 ever using a few hundred bytes of working memory and without too many
3567 tests of the mode. This should make future maintenance and development
3568 easier. A side effect of this work is that the limit of 200 on the nesting
3569 depth of parentheses has been removed (though this was never a serious
3570 limitation, I suspect). However, there is a downside: pcre_compile() now
3571 runs more slowly than before (30% or more, depending on the pattern). I
3572 hope this isn't a big issue. There is no effect on runtime performance.
3574 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3575 newline (only possible for the last line of a file) and it was a
3576 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3578 19. Added additional timing features to pcretest. (1) The -tm option now times
3579 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3580 separate command line item, by a number that specifies the number of
3581 repeats to use when timing. The default is 50000; this gives better
3582 precision, but takes uncomfortably long for very large patterns.
3584 20. Extended pcre_study() to be more clever in cases where a branch of a
3585 subpattern has no definite first character. For example, (a*|b*)[cd] would
3586 previously give no result from pcre_study(). Now it recognizes that the
3587 first character must be a, b, c, or d.
3589 21. There was an incorrect error "recursive call could loop indefinitely" if
3590 a subpattern (or the entire pattern) that was being tested for matching an
3591 empty string contained only one non-empty item after a nested subpattern.
3592 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3593 incorrectly, because the \d was being skipped in the check.
3595 22. The pcretest program now has a new pattern option /B and a command line
3596 option -b, which is equivalent to adding /B to every pattern. This causes
3597 it to show the compiled bytecode, without the additional information that
3598 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3599 is the same as /B/I).
3601 23. A new optimization is now able automatically to treat some sequences such
3602 as a*b as a*+b. More specifically, if something simple (such as a character
3603 or a simple class like \d) has an unlimited quantifier, and is followed by
3604 something that cannot possibly match the quantified thing, the quantifier
3605 is automatically "possessified".
3607 24. A recursive reference to a subpattern whose number was greater than 39
3608 went wrong under certain circumstances in UTF-8 mode. This bug could also
3609 have affected the operation of pcre_study().
3611 25. Realized that a little bit of performance could be had by replacing
3612 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3614 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3616 27. Possessive quantifiers such as a++ were previously implemented by turning
3617 them into atomic groups such as ($>a+). Now they have their own opcodes,
3618 which improves performance. This includes the automatically created ones
3619 from 23 above.
3621 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3622 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3623 the first matched character to be a colon. This applied both to named and
3624 numbered groups.
3626 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3628 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3629 building PCRE on Windows easier, so I have included it in the distribution.
3631 31. There is now a check in pcretest against a ridiculously large number being
3632 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3633 loop, the loop is abandoned.
3635 32. Forward references to subpatterns in conditions such as (?(2)...) where
3636 subpattern 2 is defined later cause pcre_compile() to search forwards in
3637 the pattern for the relevant set of parentheses. This search went wrong
3638 when there were unescaped parentheses in a character class, parentheses
3639 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3641 33. "Subroutine" calls and backreferences were previously restricted to
3642 referencing subpatterns earlier in the regex. This restriction has now
3643 been removed.
3645 34. Added a number of extra features that are going to be in Perl 5.10. On the
3646 whole, these are just syntactic alternatives for features that PCRE had
3647 previously implemented using the Python syntax or my own invention. The
3648 other formats are all retained for compatibility.
3650 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3651 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3652 also .NET compatible.
3654 (b) A recursion or subroutine call to a named group can now be defined as
3655 (?&name) as well as (?P>name).
3657 (c) A backreference to a named group can now be defined as \k<name> or
3658 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3659 5.10, are also .NET compatible.
3661 (d) A conditional reference to a named group can now use the syntax
3662 (?(<name>) or (?('name') as well as (?(name).
3664 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3665 groups (named and numbered) that are never evaluated inline, but can be
3666 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3667 is always false. There may be only one alternative in such a group.
3669 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3670 as the simple (?(R). The condition is true only if the most recent
3671 recursion is that of the given number or name. It does not search out
3672 through the entire recursion stack.
3674 (g) The escape \gN or \g{N} has been added, where N is a positive or
3675 negative number, specifying an absolute or relative reference.
3677 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3678 some "unreachable code" warnings.
3680 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3681 things, this adds five new scripts.
3683 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3684 There were also incompatibilities regarding the handling of \Q..\E inside
3685 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3686 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3688 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3689 matches an empty string, and forcibly breaks the loop. There were bugs in
3690 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3691 against aaaa the result was just "a" rather than "aaaa", for example. Two
3692 separate and independent bugs (that affected different cases) have been
3693 fixed.
3695 39. Refactored the code to abolish the use of different opcodes for small
3696 capturing bracket numbers. This is a tidy that I avoided doing when I
3697 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3698 The new approach is not only tidier, it makes it possible to reduce the
3699 memory needed to fix the previous bug (38).
3701 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3702 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3703 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3704 mode.
3706 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3707 report.
3709 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3710 copy construction and assignment in the C++ wrapper.
3712 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3713 couple of bugs that could have given wrong results in the "--newline=crlf"
3714 case.
3716 44. Added a number of casts and did some reorganization of signed/unsigned int
3717 variables following suggestions from Dair Grant. Also renamed the variable
3718 "this" as "item" because it is a C++ keyword.
3720 45. Arranged for dftables to add
3722 #include "pcre_internal.h"
3724 to pcre_chartables.c because without it, gcc 4.x may remove the array
3725 definition from the final binary if PCRE is built into a static library and
3726 dead code stripping is activated.
3728 46. For an unanchored pattern, if a match attempt fails at the start of a
3729 newline sequence, and the newline setting is CRLF or ANY, and the next two
3730 characters are CRLF, advance by two characters instead of one.
3733 Version 6.7 04-Jul-06
3734 ---------------------
3736 1. In order to handle tests when input lines are enormously long, pcretest has
3737 been re-factored so that it automatically extends its buffers when
3738 necessary. The code is crude, but this _is_ just a test program. The
3739 default size has been increased from 32K to 50K.
3741 2. The code in pcre_study() was using the value of the re argument before
3742 testing it for NULL. (Of course, in any sensible call of the function, it
3743 won't be NULL.)
3745 3. The memmove() emulation function in pcre_internal.h, which is used on
3746 systems that lack both memmove() and bcopy() - that is, hardly ever -
3747 was missing a "static" storage class specifier.
3749 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3750 containing an extended class (one that cannot be represented by a bitmap
3751 because it contains high-valued characters or Unicode property items, e.g.
3752 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3753 pattern, but PCRE should not loop if you do not (it no longer does).
3754 [Detail: two cases were found: (a) a repeated subpattern containing an
3755 extended class; (b) a recursive reference to a subpattern that followed a
3756 previous extended class. It wasn't skipping over the extended class
3757 correctly when UTF-8 mode was not set.]
3759 5. A negated single-character class was not being recognized as fixed-length
3760 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3761 compile error "lookbehind assertion is not fixed length".
3763 6. The RunPerlTest auxiliary script was showing an unexpected difference
3764 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3765 write a Perl script that can interpret lines of an input file either as
3766 byte characters or as UTF-8, which is what "perltest" was being required to
3767 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3768 can't do is switch easily at run time between having the "use utf8;" pragma
3769 or not. In the end, I fudged it by using the RunPerlTest script to insert
3770 "use utf8;" explicitly for the UTF-8 tests.
3772 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3773 the end of the subject string, contrary to the documentation and to what
3774 Perl does. This was true of both matching functions. Now it matches only at
3775 the start of the subject and immediately after *internal* newlines.
3777 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3778 a pointer to an int instead of a pointer to an unsigned long int. This
3779 caused problems on 64-bit systems.
3781 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3782 instance of the 'standard' template library not being so standard".
3784 10. There was no check on the number of named subpatterns nor the maximum
3785 length of a subpattern name. The product of these values is used to compute
3786 the size of the memory block for a compiled pattern. By supplying a very
3787 long subpattern name and a large number of named subpatterns, the size
3788 computation could be caused to overflow. This is now prevented by limiting
3789 the length of names to 32 characters, and the number of named subpatterns
3790 to 10,000.
3792 11. Subpatterns that are repeated with specific counts have to be replicated in
3793 the compiled pattern. The size of memory for this was computed from the
3794 length of the subpattern and the repeat count. The latter is limited to
3795 65535, but there was no limit on the former, meaning that integer overflow
3796 could in principle occur. The compiled length of a repeated subpattern is
3797 now limited to 30,000 bytes in order to prevent this.
3799 12. Added the optional facility to have named substrings with the same name.
3801 13. Added the ability to use a named substring as a condition, using the
3802 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3803 are numbers (not recommended). Forward references are permitted.
3805 14. Added forward references in named backreferences (if you see what I mean).
3807 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3808 pattern could run off the end of the subject. For example, the pattern
3809 "(?s)(.{1,5})"8 did this with the subject "ab".
3811 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3812 PCRE_CASELESS was set when matching characters that were quantified with ?
3813 or *.
3815 17. A character class other than a single negated character that had a minimum
3816 but no maximum quantifier - for example [ab]{6,} - was not handled
3817 correctly by pce_dfa_exec(). It would match only one character.
3819 18. A valid (though odd) pattern that looked like a POSIX character
3820 class but used an invalid character after [ (for example [[,abc,]]) caused
3821 pcre_compile() to give the error "Failed: internal error: code overflow" or
3822 in some cases to crash with a glibc free() error. This could even happen if
3823 the pattern terminated after [[ but there just happened to be a sequence of
3824 letters, a binary zero, and a closing ] in the memory that followed.
3826 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3827 over the years. Originally (before any Unicode support), just the bottom 8
3828 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3829 output from "man perlunicode" includes this:
3831 The regular expression compiler produces polymorphic opcodes. That
3832 is, the pattern adapts to the data and automatically switches to
3833 the Unicode character scheme when presented with Unicode data--or
3834 instead uses a traditional byte scheme when presented with byte
3835 data.
3837 Sadly, a wide octal escape does not cause a switch, and in a string with
3838 no other multibyte characters, these octal escapes are treated as before.
3839 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3840 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3841 Unicode string.
3843 I have not perpetrated such confusion in PCRE. Up till now, it took just
3844 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3845 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3846 translate to the appropriate multibyte character.
3848 29. Applied some refactoring to reduce the number of warnings from Microsoft
3849 and Borland compilers. This has included removing the fudge introduced
3850 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3851 a warning about an unused variable.
3853 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3854 characters since release 4.0, because Perl (from release 5.004) does not.
3855 [Or at least, is documented not to: some releases seem to be in conflict
3856 with the documentation.] However, when a pattern was studied with
3857 pcre_study() and all its branches started with \s, PCRE still included VT
3858 as a possible starting character. Of course, this did no harm; it just
3859 caused an unnecessary match attempt.
3861 22. Removed a now-redundant internal flag bit that recorded the fact that case
3862 dependency changed within the pattern. This was once needed for "required
3863 byte" processing, but is no longer used. This recovers a now-scarce options
3864 bit. Also moved the least significant internal flag bit to the most-
3865 significant bit of the word, which was not previously used (hangover from
3866 the days when it was an int rather than a uint) to free up another bit for
3867 the future.
3869 23. Added support for CRLF line endings as well as CR and LF. As well as the
3870 default being selectable at build time, it can now be changed at runtime
3871 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3872 specify that it is scanning data with non-default line endings.
3874 24. Changed the definition of CXXLINK to make it agree with the definition of
3875 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3877 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3878 recursions. This makes a big different to stack usage for some patterns.
3880 26. If a subpattern containing a named recursion or subroutine reference such
3881 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3882 the space required for the compiled pattern went wrong and gave too small a
3883 value. Depending on the environment, this could lead to "Failed: internal
3884 error: code overflow at offset 49" or "glibc detected double free or
3885 corruption" errors.
3887 27. Applied patches from Google (a) to support the new newline modes and (b) to
3888 advance over multibyte UTF-8 characters in GlobalReplace.
3890 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3891 difference for some implementation of PCRE in some Windows version.
3893 29. Added some extra testing facilities to pcretest:
3895 \q<number> in a data line sets the "match limit" value
3896 \Q<number> in a data line sets the "match recursion limt" value
3897 -S <number> sets the stack size, where <number> is in megabytes
3899 The -S option isn't available for Windows.
3902 Version 6.6 06-Feb-06
3903 ---------------------
3905 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3906 in pcreposix.h. I have copied the definition from pcre.h.
3908 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3909 because pcre.h is no longer a built file.
3911 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3912 not normally included in the compiled code.
3915 Version 6.5 01-Feb-06
3916 ---------------------
3918 1. When using the partial match feature with pcre_dfa_exec(), it was not
3919 anchoring the second and subsequent partial matches at the new starting
3920 point. This could lead to incorrect results. For example, with the pattern
3921 /1234/, partially matching against "123" and then "a4" gave a match.
3923 2. Changes to pcregrep:
3925 (a) All non-match returns from pcre_exec() were being treated as failures
3926 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3927 error message is output. Some extra information is given for the
3929 probably the only errors that are likely to be caused by users (by
3930 specifying a regex that has nested indefinite repeats, for instance).
3931 If there are more than 20 of these errors, pcregrep is abandoned.
3933 (b) A binary zero was treated as data while matching, but terminated the
3934 output line if it was written out. This has been fixed: binary zeroes
3935 are now no different to any other data bytes.
3937 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3938 used to set a locale for matching. The --locale=xxxx long option has
3939 been added (no short equivalent) to specify a locale explicitly on the
3940 pcregrep command, overriding the environment variables.
3942 (d) When -B was used with -n, some line numbers in the output were one less
3943 than they should have been.
3945 (e) Added the -o (--only-matching) option.
3947 (f) If -A or -C was used with -c (count only), some lines of context were
3948 accidentally printed for the final match.
3950 (g) Added the -H (--with-filename) option.
3952 (h) The combination of options -rh failed to suppress file names for files
3953 that were found from directory arguments.
3955 (i) Added the -D (--devices) and -d (--directories) options.
3957 (j) Added the -F (--fixed-strings) option.
3959 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3961 (l) Added the --colo(u)r option.
3963 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3964 is not present by default.
3966 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3967 items such as (?R) or (?1), when the recursion could match a number of
3968 alternatives. If it matched one of the alternatives, but subsequently,
3969 outside the recursion, there was a failure, the code tried to back up into
3970 the recursion. However, because of the way PCRE is implemented, this is not
3971 possible, and the result was an incorrect result from the match.
3973 In order to prevent this happening, the specification of recursion has
3974 been changed so that all such subpatterns are automatically treated as
3975 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3977 4. I had overlooked the fact that, in some locales, there are characters for
3978 which isalpha() is true but neither isupper() nor islower() are true. In
3979 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3980 and ordfeminine) are like this. This affected the treatment of \w and \W
3981 when they appeared in character classes, but not when they appeared outside
3982 a character class. The bit map for "word" characters is now created
3983 separately from the results of isalnum() instead of just taking it from the
3984 upper, lower, and digit maps. (Plus the underscore character, of course.)
3986 5. The above bug also affected the handling of POSIX character classes such as
3987 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3988 permanent tables. Instead, the bit maps for such a class were previously
3989 created as the appropriate unions of the upper, lower, and digit bitmaps.
3990 Now they are created by subtraction from the [[:word:]] class, which has
3991 its own bitmap.
3993 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3994 It is created by subtracting the vertical space characters (\x09, \x0a,
3995 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3996 subtraction was done in the overall bitmap for a character class, meaning
3997 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3998 be recognized. This bug has been fixed.
4000 7. Patches from the folks at Google:
4002 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4003 real life, but is still worth protecting against".
4005 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4006 regular expressions".
4008 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4009 have it.
4011 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4012 "configure" and the latter not, in order to fix a problem somebody had
4013 with compiling the Arg class on HP-UX.
4015 (e) Improve the error-handling of the C++ wrapper a little bit.
4017 (f) New tests for checking recursion limiting.
4019 8. The pcre_memmove() function, which is used only if the environment does not
4020 have a standard memmove() function (and is therefore rarely compiled),
4021 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4022 returning a result (though PCRE never actually uses the result).
4024 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4025 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4026 returned instead of calling malloc() with an overflowing number that would
4027 most likely cause subsequent chaos.
4029 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4031 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4032 with this option is matched, the nmatch and pmatch options of regexec() are
4033 ignored.
4035 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4036 provided in case anyone wants to the the POSIX interface with UTF-8
4037 strings.
4039 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4040 C++ linking (needed for some HP-UX environments).
4042 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4043 (unused parameter) and in the pcre_printint() function (omitted "default"
4044 switch label when the default is to do nothing).
4046 15. Added some code to make it possible, when PCRE is compiled as a C++
4047 library, to replace subject pointers for pcre_exec() with a smart pointer
4048 class, thus making it possible to process discontinuous strings.
4050 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4051 much the same function. They were added by different people who were trying
4052 to make PCRE easy to compile on non-Unix systems. It has been suggested
4053 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4054 for compiling on Windows systems. I have therefore replaced it with
4055 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4056 defaults to "extern" for C or "extern C" for C++, which works fine on
4057 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4058 SCOPE with something explicit in config.h. In addition:
4060 (a) pcreposix.h still had just "extern" instead of either of these macros;
4061 I have replaced it with PCRE_DATA_SCOPE.
4063 (b) Functions such as _pcre_xclass(), which are internal to the library,
4064 but external in the C sense, all had PCRE_EXPORT in their definitions.
4065 This is apparently wrong for the Windows case, so I have removed it.
4066 (It makes no difference on Unix-like systems.)
4068 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4069 of recursive calls to match(). This is different to MATCH_LIMIT because
4070 that limits the total number of calls to match(), not all of which increase
4071 the depth of recursion. Limiting the recursion depth limits the amount of
4072 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4073 when PCRE is compiled, and changed at run time. A patch from Google adds
4074 this functionality to the C++ interface.
4076 18. Changes to the handling of Unicode character properties:
4078 (a) Updated the table to Unicode 4.1.0.
4080 (b) Recognize characters that are not in the table as "Cn" (undefined).
4082 (c) I revised the way the table is implemented to a much improved format
4083 which includes recognition of ranges. It now supports the ranges that
4084 are defined in UnicodeData.txt, and it also amalgamates other
4085 characters into ranges. This has reduced the number of entries in the
4086 table from around 16,000 to around 3,000, thus reducing its size
4087 considerably. I realized I did not need to use a tree structure after
4088 all - a binary chop search is just as efficient. Having reduced the
4089 number of entries, I extended their size from 6 bytes to 8 bytes to
4090 allow for more data.
4092 (d) Added support for Unicode script names via properties such as \p{Han}.
4094 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4095 matching that character.
4097 20. When matching a repeated Unicode property with a minimum greater than zero,
4098 (for example \pL{2,}), PCRE could look past the end of the subject if it
4099 reached it while seeking the minimum number of characters. This could
4100 happen only if some of the characters were more than one byte long, because
4101 there is a check for at least the minimum number of bytes.
4103 21. Refactored the implementation of \p and \P so as to be more general, to
4104 allow for more different types of property in future. This has changed the
4105 compiled form incompatibly. Anybody with saved compiled patterns that use
4106 \p or \P will have to recompile them.
4108 22. Added "Any" and "L&" to the supported property types.
4110 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4111 but give a compile time error if the value is greater than 0xff.
4113 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4114 accidentally not being installed or uninstalled.
4116 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4117 made were to insert the current release number. This seemed silly, because
4118 it made things harder for people building PCRE on systems that don't run
4119 "configure". I have turned pcre.h into a distributed file, no longer built
4120 by "configure", with the version identification directly included. There is
4121 no longer a pcre.h.in file.
4123 However, this change necessitated a change to the pcre-config script as
4124 well. It is built from pcre-config.in, and one of the substitutions was the
4125 release number. I have updated configure.ac so that ./configure now finds
4126 the release number by grepping pcre.h.
4128 26. Added the ability to run the tests under valgrind.
4131 Version 6.4 05-Sep-05
4132 ---------------------
4134 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4135 "--" to be printed when multiple files were scanned, even when none of the
4136 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4137 consider it to be a bug, and have restored the previous behaviour.
4139 2. A couple of code tidies to get rid of compiler warnings.
4141 3. The pcretest program used to cheat by referring to symbols in the library
4142 whose names begin with _pcre_. These are internal symbols that are not
4143 really supposed to be visible externally, and in some environments it is
4144 possible to suppress them. The cheating is now confined to including
4145 certain files from the library's source, which is a bit cleaner.
4147 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4148 file's purpose clearer.
4150 5. Reorganized pcre_ucp_findchar().
4153 Version 6.3 15-Aug-05
4154 ---------------------
4156 1. The file libpcre.pc.in did not have general read permission in the tarball.
4158 2. There were some problems when building without C++ support:
4160 (a) If C++ support was not built, "make install" and "make test" still
4161 tried to test it.
4163 (b) There were problems when the value of CXX was explicitly set. Some
4164 changes have been made to try to fix these, and ...
4166 (c) --disable-cpp can now be used to explicitly disable C++ support.
4168 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4169 backslash in a target when C++ was disabled. This confuses some
4170 versions of "make", apparently. Using an intermediate variable solves
4171 this. (Same for CPP_LOBJ.)
4173 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4174 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4175 necessary on certain architectures.
4177 4. Added a setting of -export-symbols-regex to the link command to remove
4178 those symbols that are exported in the C sense, but actually are local
4179 within the library, and not documented. Their names all begin with
4180 "_pcre_". This is not a perfect job, because (a) we have to except some
4181 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4182 available (and never for static libraries). I have made a note to try to
4183 find a way round (a) in the future.
4186 Version 6.2 01-Aug-05
4187 ---------------------
4189 1. There was no test for integer overflow of quantifier values. A construction
4190 such as {1111111111111111} would give undefined results. What is worse, if
4191 a minimum quantifier for a parenthesized subpattern overflowed and became
4192 negative, the calculation of the memory size went wrong. This could have
4193 led to memory overwriting.
4195 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4197 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4198 operating environments where this matters.
4200 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4201 PCRE options from within the C++ wrapper.
4203 5. Named capturing subpatterns were not being correctly counted when a pattern
4204 was compiled. This caused two problems: (a) If there were more than 100
4205 such subpatterns, the calculation of the memory needed for the whole
4206 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4207 back references of the form \12, where the number was greater than 9, were
4208 not recognized as back references, even though there were sufficient
4209 previous subpatterns.
4211 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4212 versions of gcc, e.g. 2.95.4.
4215 Version 6.1 21-Jun-05
4216 ---------------------
4218 1. There was one reference to the variable "posix" in pcretest.c that was not
4219 surrounded by "#if !defined NOPOSIX".
4221 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4222 the cross-check on the old pcre_info() function, for the benefit of the
4223 cut-down version of PCRE that is currently imported into Exim.
4225 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4226 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4227 patterns that start (?i) but I don't think that matters. The use of (?i) is
4228 just an example; this all applies to the other options as well.
4230 4. Since libtool seems to echo the compile commands it is issuing, the output
4231 from "make" can be reduced a bit by putting "@" in front of each libtool
4232 compile command.
4234 5. Patch from the folks at Google for configure.in to be a bit more thorough
4235 in checking for a suitable C++ installation before trying to compile the
4236 C++ stuff. This should fix a reported problem when a compiler was present,
4237 but no suitable headers.
4239 6. The man pages all had just "PCRE" as their title. I have changed them to
4240 be the relevant file name. I have also arranged that these names are
4241 retained in the file doc/pcre.txt, which is a concatenation in text format
4242 of all the man pages except the little individual ones for each function.
4244 7. The NON-UNIX-USE file had not been updated for the different set of source
4245 files that come with release 6. I also added a few comments about the C++
4246 wrapper.
4249 Version 6.0 07-Jun-05
4250 ---------------------
4252 1. Some minor internal re-organization to help with my DFA experiments.
4254 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4255 didn't matter for the library itself when fully configured, but did matter
4256 when compiling without UCP support, or within Exim, where the ucp files are
4257 not imported.
4259 3. Refactoring of the library code to split up the various functions into
4260 different source modules. The addition of the new DFA matching code (see
4261 below) to a single monolithic source would have made it really too
4262 unwieldy, quite apart from causing all the code to be include in a
4263 statically linked application, when only some functions are used. This is
4264 relevant even without the DFA addition now that patterns can be compiled in
4265 one application and matched in another.
4267 The downside of splitting up is that there have to be some external
4268 functions and data tables that are used internally in different modules of
4269 the library but which are not part of the API. These have all had their
4270 names changed to start with "_pcre_" so that they are unlikely to clash
4271 with other external names.
4273 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4274 a different (DFA) algorithm. Although it is slower than the original
4275 function, it does have some advantages for certain types of matching
4276 problem.
4278 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4279 including restarting after a partial match.
4281 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4282 defined when compiling for Windows was sent to me. I have put it into the
4283 code, though I have no means of testing or verifying it.
4285 7. Added the pcre_refcount() auxiliary function.
4287 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4288 match before or at the first newline in the subject string. In pcretest,
4289 the /f option on a pattern can be used to set this.
4291 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4292 would behave wrongly. This has been present in PCRE since release 4.0.
4294 10. A number of changes to the pcregrep command:
4296 (a) Refactored how -x works; insert ^(...)$ instead of setting
4297 PCRE_ANCHORED and checking the length, in preparation for adding
4298 something similar for -w.
4300 (b) Added the -w (match as a word) option.
4302 (c) Refactored the way lines are read and buffered so as to have more
4303 than one at a time available.
4305 (d) Implemented a pcregrep test script.
4307 (e) Added the -M (multiline match) option. This allows patterns to match
4308 over several lines of the subject. The buffering ensures that at least
4309 8K, or the rest of the document (whichever is the shorter) is available
4310 for matching (and similarly the previous 8K for lookbehind assertions).
4312 (f) Changed the --help output so that it now says
4314 -w, --word-regex(p)
4316 instead of two lines, one with "regex" and the other with "regexp"
4317 because that confused at least one person since the short forms are the
4318 same. (This required a bit of code, as the output is generated
4319 automatically from a table. It wasn't just a text change.)
4321 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4322 option but starts with a hyphen. Could be a pattern or a path name
4323 starting with a hyphen, for instance.
4325 (h) "-" can be given as a file name to represent stdin.
4327 (i) When file names are being printed, "(standard input)" is used for
4328 the standard input, for compatibility with GNU grep. Previously
4329 "<stdin>" was used.
4331 (j) The option --label=xxx can be used to supply a name to be used for
4332 stdin when file names are being printed. There is no short form.
4334 (k) Re-factored the options decoding logic because we are going to add
4335 two more options that take data. Such options can now be given in four
4336 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4338 (l) Added the -A, -B, and -C options for requesting that lines of context
4339 around matches be printed.
4341 (m) Added the -L option to print the names of files that do not contain
4342 any matching lines, that is, the complement of -l.
4344 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4345 continue to scan other files.
4347 (o) The -s option was incorrectly implemented. For compatibility with other
4348 greps, it now suppresses the error message for a non-existent or non-
4349 accessible file (but not the return code). There is a new option called
4350 -q that suppresses the output of matching lines, which was what -s was
4351 previously doing.
4353 (p) Added --include and --exclude options to specify files for inclusion
4354 and exclusion when recursing.
4356 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4357 Hopefully, it now does.
4359 12. Missing cast in pcre_study().
4361 13. Added an "uninstall" target to the makefile.
4363 14. Replaced "extern" in the function prototypes in Makefile.in with
4364 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4365 world, but is set differently for Windows.
4367 15. Added a second compiling function called pcre_compile2(). The only
4368 difference is that it has an extra argument, which is a pointer to an
4369 integer error code. When there is a compile-time failure, this is set
4370 non-zero, in addition to the error test pointer being set to point to an
4371 error message. The new argument may be NULL if no error number is required
4372 (but then you may as well call pcre_compile(), which is now just a
4373 wrapper). This facility is provided because some applications need a
4374 numeric error indication, but it has also enabled me to tidy up the way
4375 compile-time errors are handled in the POSIX wrapper.
4377 16. Added VPATH=.libs to the makefile; this should help when building with one
4378 prefix path and installing with another. (Or so I'm told by someone who
4379 knows more about this stuff than I do.)
4381 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4382 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4383 match everything, including newlines. This is not POSIX-compatible, but
4384 somebody wanted the feature. From pcretest it can be activated by using
4385 both the P and the s flags.
4387 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4389 19. libpcre.pc was being incorrectly installed as executable.
4391 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4392 it now also looks for '\r' so that it will work unmodified on Windows.
4394 21. Added Google's contributed C++ wrapper to the distribution.
4396 22. Added some untidy missing memory free() calls in pcretest, to keep
4397 Electric Fence happy when testing.
4401 Version 5.0 13-Sep-04
4402 ---------------------
4404 1. Internal change: literal characters are no longer packed up into items
4405 containing multiple characters in a single byte-string. Each character
4406 is now matched using a separate opcode. However, there may be more than one
4407 byte in the character in UTF-8 mode.
4409 2. The pcre_callout_block structure has two new fields: pattern_position and
4410 next_item_length. These contain the offset in the pattern to the next match
4411 item, and its length, respectively.
4413 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4414 insertion of callouts before each pattern item. Added the /C option to
4415 pcretest to make use of this.
4417 4. On the advice of a Windows user, the lines
4419 #if defined(_WIN32) || defined(WIN32)
4420 _setmode( _fileno( stdout ), 0x8000 );
4421 #endif /* defined(_WIN32) || defined(WIN32) */
4423 have been added to the source of pcretest. This apparently does useful
4424 magic in relation to line terminators.
4426 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4427 for the benefit of those environments where the "b" makes a difference.
4429 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4430 to know about it. I have put a hack into configure.in that adds in code
4431 to set GCC=yes if CC=icc. This seems to end up at a point in the
4432 generated configure script that is early enough to affect the setting of
4433 compiler options, which is what is needed, but I have no means of testing
4434 whether it really works. (The user who reported this had patched the
4435 generated configure script, which of course I cannot do.)
4437 LATER: After change 22 below (new libtool files), the configure script
4438 seems to know about icc (and also ecc). Therefore, I have commented out
4439 this hack in configure.in.
4441 7. Added support for pkg-config (2 patches were sent in).
4443 8. Negated POSIX character classes that used a combination of internal tables
4444 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4445 [[:^ascii]]. Typically, they would match almost any characters. The other
4446 POSIX classes were not broken in this way.
4448 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4449 to find the match, as PCRE was deluded into thinking that the match had to
4450 start at the start point or following a newline. The same bug applied to
4451 patterns with negative forward assertions or any backward assertions
4452 preceding ".*" at the start, unless the pattern required a fixed first
4453 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4455 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4456 starting at the last subject character, bytes beyond the end of the subject
4457 string were read.
4459 11. Renamed the variable "class" as "classbits" to make life easier for C++
4460 users. (Previously there was a macro definition, but it apparently wasn't
4461 enough.)
4463 12. Added the new field "tables" to the extra data so that tables can be passed
4464 in at exec time, or the internal tables can be re-selected. This allows
4465 a compiled regex to be saved and re-used at a later time by a different
4466 program that might have everything at different addresses.
4468 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4469 -R library as well as a -L library.
4471 14. The debugging options of pcretest (-d on the command line or D on a
4472 pattern) showed incorrect output for anything following an extended class
4473 that contained multibyte characters and which was followed by a quantifier.
4475 15. Added optional support for general category Unicode character properties
4476 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4477 support. It adds about 90K to the size of the library. The meanings of the
4478 inbuilt class escapes such as \d and \s have NOT been changed.
4480 16. Updated pcredemo.c to include calls to free() to release the memory for the
4481 compiled pattern.
4483 17. The generated file chartables.c was being created in the source directory
4484 instead of in the building directory. This caused the build to fail if the
4485 source directory was different from the building directory, and was
4486 read-only.
4488 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4489 file. No doubt somebody will tell me if they don't make sense... Also added
4490 Dan Mooney's comments about building on OpenVMS.
4492 19. Added support for partial matching via the PCRE_PARTIAL option for
4493 pcre_exec() and the \P data escape in pcretest.
4495 20. Extended pcretest with 3 new pattern features:
4497 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4498 write the compiled pattern to the file whose name is "rest-of-line".
4499 This is a straight binary dump of the data, with the saved pointer to
4500 the character tables forced to be NULL. The study data, if any, is
4501 written too. After writing, pcretest reads a new pattern.
4503 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4504 compiled pattern from the given file. There must not be any
4505 occurrences of "<" in the file name (pretty unlikely); if there are,
4506 pcretest will instead treat the initial "<" as a pattern delimiter.
4507 After reading in the pattern, pcretest goes on to read data lines as
4508 usual.
4510 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4511 and 16-bit fields in a compiled pattern, to simulate a pattern that
4512 was compiled on a host of opposite endianness.
4514 21. The pcre-exec() function can now cope with patterns that were compiled on
4515 hosts of opposite endianness, with this restriction:
4517 As for any compiled expression that is saved and used later, the tables
4518 pointer field cannot be preserved; the extra_data field in the arguments
4519 to pcre_exec() should be used to pass in a tables address if a value
4520 other than the default internal tables were used at compile time.
4522 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4523 now diagnosed as an error. Previously, most of the time, a negative number
4524 would have been treated as zero, but if in addition "ovector" was passed as
4525 NULL, a crash could occur.
4527 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4528 new versions from the libtool 1.5 distribution (the last one is a copy of
4529 a file called libtool.m4). This seems to have fixed the need to patch
4530 "configure" to support Darwin 1.3 (which I used to do). However, I still
4531 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4532 workstation).
4534 24. Changed the PCRE licence to be the more standard "BSD" licence.
4537 Version 4.5 01-Dec-03
4538 ---------------------
4540 1. There has been some re-arrangement of the code for the match() function so
4541 that it can be compiled in a version that does not call itself recursively.
4542 Instead, it keeps those local variables that need separate instances for
4543 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4544 needs to "recurse". Keeping track of where control must go is done by means
4545 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4546 hide most of the details from the main code, and operates only if
4547 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4548 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4549 operating.
4551 To make it easier for callers to provide specially tailored get/free
4552 functions for this usage, two new functions, pcre_stack_malloc, and
4553 pcre_stack_free, are used. They are always called in strict stacking order,
4554 and the size of block requested is always the same.
4556 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4557 PCRE has been compiled to use the stack or the heap for recursion. The
4558 -C option of pcretest uses this to show which version is compiled.
4560 A new data escape \S, is added to pcretest; it causes the amounts of store
4561 obtained and freed by both kinds of malloc/free at match time to be added
4562 to the output.
4564 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4565 what's available on my current Linux desktop machine.
4567 3. When matching a UTF-8 string, the test for a valid string at the start has
4568 been extended. If start_offset is not zero, PCRE now checks that it points
4569 to a byte that is the start of a UTF-8 character. If not, it returns
4570 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4571 this is necessary because there may be backward assertions in the pattern.
4572 When matching the same subject several times, it may save resources to use
4573 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4575 4. The code for checking the validity of UTF-8 strings has been tightened so
4576 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4577 containing "overlong sequences".
4579 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4580 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4581 should have been "&", but it just so happened that all the cases this let
4582 through by mistake were picked up later in the function.
4584 6. I had used a variable called "isblank" - this is a C99 function, causing
4585 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4587 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4588 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4589 so the version is shown; (c) stop "make test" echoing "./RunTest".
4591 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4593 9. The prototype for memmove() for systems that don't have it was using
4594 size_t, but the inclusion of the header that defines size_t was later. I've
4595 moved the #includes for the C headers earlier to avoid this.
4597 10. Added some adjustments to the code to make it easier to compiler on certain
4598 special systems:
4600 (a) Some "const" qualifiers were missing.
4601 (b) Added the macro EXPORT before all exported functions; by default this
4602 is defined to be empty.
4603 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4604 that it reads its output file name as an argument instead of writing
4605 to the standard output and assuming this can be redirected.
4607 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4608 class containing characters with values greater than 255, PCRE compilation
4609 went into a loop.
4611 12. A recursive reference to a subpattern that was within another subpattern
4612 that had a minimum quantifier of zero caused PCRE to crash. For example,
4613 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4614 recursion. If the recursively-called subpattern itself had a zero repeat,
4615 that was OK.
4617 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4618 buffer into which it was copied (for escape processing) was still set at
4619 1024, so long lines caused crashes.
4621 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4622 "internal error: code overflow...". This applied to any character class
4623 that was followed by a possessive quantifier.
4625 15. Modified the Makefile to add libpcre.la as a prerequisite for
4626 libpcreposix.la because I was told this is needed for a parallel build to
4627 work.
4629 16. If a pattern that contained .* following optional items at the start was
4630 studied, the wrong optimizing data was generated, leading to matching
4631 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4632 matching string must start with a or b or c. The correct conclusion for
4633 this pattern is that a match can start with any character.
4636 Version 4.4 13-Aug-03
4637 ---------------------
4639 1. In UTF-8 mode, a character class containing characters with values between
4640 127 and 255 was not handled correctly if the compiled pattern was studied.
4641 In fixing this, I have also improved the studying algorithm for such
4642 classes (slightly).
4644 2. Three internal functions had redundant arguments passed to them. Removal
4645 might give a very teeny performance improvement.
4647 3. Documentation bug: the value of the capture_top field in a callout is *one
4648 more than* the number of the hightest numbered captured substring.
4650 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4651 in incorrectly linking with a previously installed version. They now link
4652 explicitly with libpcre.la.
4654 5. configure.in no longer needs to recognize Cygwin specially.
4656 6. A problem in pcre.in for Windows platforms is fixed.
4658 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4659 pcretest, it used to include the size of the study block as part of its
4660 output. Unfortunately, the structure contains a field that has a different
4661 size on different hardware architectures. This meant that the tests that
4662 showed this size failed. As the block is currently always of a fixed size,
4663 this information isn't actually particularly useful in pcretest output, so
4664 I have just removed it.
4666 8. Three pre-processor statements accidentally did not start in column 1.
4667 Sadly, there are *still* compilers around that complain, even though
4668 standard C has not required this for well over a decade. Sigh.
4670 9. In pcretest, the code for checking callouts passed small integers in the
4671 callout_data field, which is a void * field. However, some picky compilers
4672 complained about the casts involved for this on 64-bit systems. Now
4673 pcretest passes the address of the small integer instead, which should get
4674 rid of the warnings.
4676 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4677 both compile and run time, and gives an error if an invalid UTF-8 sequence
4678 is found. There is a option for disabling this check in cases where the
4679 string is known to be correct and/or the maximum performance is wanted.
4681 11. In response to a bug report, I changed one line in Makefile.in from
4683 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4684 to
4685 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4687 to look similar to other lines, but I have no way of telling whether this
4688 is the right thing to do, as I do not use Windows. No doubt I'll get told
4689 if it's wrong...
4692 Version 4.3 21-May-03
4693 ---------------------
4695 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4696 Makefile.
4698 2. Some refactoring to improve the quality of the code:
4700 (i) The utf8_table... variables are now declared "const".
4702 (ii) The code for \cx, which used the "case flipping" table to upper case
4703 lower case letters, now just substracts 32. This is ASCII-specific,
4704 but the whole concept of \cx is ASCII-specific, so it seems
4705 reasonable.
4707 (iii) PCRE was using its character types table to recognize decimal and
4708 hexadecimal digits in the pattern. This is silly, because it handles
4709 only 0-9, a-f, and A-F, but the character types table is locale-
4710 specific, which means strange things might happen. A private
4711 table is now used for this - though it costs 256 bytes, a table is
4712 much faster than multiple explicit tests. Of course, the standard
4713 character types table is still used for matching digits in subject
4714 strings against \d.
4716 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4717 ending in _t are). So I've renamed it as ESC_tee.
4719 3. The first argument for regexec() in the POSIX wrapper should have been
4720 defined as "const".
4722 4. Changed pcretest to use malloc() for its buffers so that they can be
4723 Electric Fenced for debugging.
4725 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4726 to read one or more bytes before the start of the subject string. Often this
4727 had no effect on PCRE's behaviour, but in some circumstances it could
4728 provoke a segmentation fault.
4730 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4731 to try to read one or more bytes before the start of the subject string.
4733 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4734 UTF-8 support could misbehave in various ways if the subject string
4735 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4736 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4737 back over UTF-8 characters.)
4740 Version 4.2 14-Apr-03
4741 ---------------------
4743 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4745 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4746 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4747 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4748 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4749 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4751 Note: automatic setting of the BUILD variables is not yet working
4752 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4753 compile-time but not at link-time
4754 [LINK]: use for linking executables only
4755 make different versions for Windows and non-Windows
4756 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4757 libraries
4758 [LINK_FOR_BUILD]: new variable
4759 [OBJEXT]: use throughout
4760 [EXEEXT]: use throughout
4761 <winshared>: new target
4762 <wininstall>: new target
4763 <dftables.o>: use native compiler
4764 <dftables>: use native linker
4765 <install>: handle Windows platform correctly
4766 <clean>: ditto
4767 <check>: ditto
4768 copy DLL to top builddir before testing
4770 As part of these changes, -no-undefined was removed again. This was reported
4771 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4772 in any case.
4774 3. Some tidies to get rid of compiler warnings:
4776 . In the match_data structure, match_limit was an unsigned long int, whereas
4777 match_call_count was an int. I've made them both unsigned long ints.
4779 . In pcretest the fact that a const uschar * doesn't automatically cast to
4780 a void * provoked a warning.
4782 . Turning on some more compiler warnings threw up some "shadow" variables
4783 and a few more missing casts.
4785 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4786 option, a class that contained a single character with a value between 128
4787 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4789 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4790 option, a class that contained several characters, but with at least one
4791 whose value was between 128 and 255 caused PCRE to crash.
4794 Version 4.1 12-Mar-03
4795 ---------------------
4797 1. Compiling with gcc -pedantic found a couple of places where casts were
4798 needed, and a string in dftables.c that was longer than standard compilers are
4799 required to support.
4801 2. Compiling with Sun's compiler found a few more places where the code could
4802 be tidied up in order to avoid warnings.
4804 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4805 first of these names is deprecated in the latest Autoconf in favour of the name
4806 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4807 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4808 analogy I have changed it to CFLAGS_FOR_BUILD.
4810 4. Added -no-undefined to the linking command in the Makefile, because this is
4811 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4812 linking step for the pcreposix library.
4814 5. PCRE was failing to diagnose the case of two named groups with the same
4815 name.
4817 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4818 literal character that is needed in the subject for a match, and scans along to
4819 ensure that it is present before embarking on the full matching process. This
4820 saves time in cases of nested unlimited repeats that are never going to match.
4821 Problem: the scan can take a lot of time if the subject is very long (e.g.
4822 megabytes), thus penalizing straightforward matches. It is now done only if the
4823 amount of subject to be scanned is less than 1000 bytes.
4825 7. A lesser problem with the same optimization is that it was recording the
4826 first character of an anchored pattern as "needed", thus provoking a search
4827 right along the subject, even when the first match of the pattern was going to
4828 fail. The "needed" character is now not set for anchored patterns, unless it
4829 follows something in the pattern that is of non-fixed length. Thus, it still
4830 fulfils its original purpose of finding quick non-matches in cases of nested
4831 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4834 Version 4.0 17-Feb-03
4835 ---------------------
4837 1. If a comment in an extended regex that started immediately after a meta-item
4838 extended to the end of string, PCRE compiled incorrect data. This could lead to
4839 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4841 2. Moved to autoconf 2.53 and libtool 1.4.2.
4843 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4844 the special perltest8 script is no longer needed - all the tests can be run
4845 from a single perltest script.
4847 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4848 by \s. It has now been removed in PCRE. This means it isn't recognized as
4849 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4850 class [:space:] *does* include VT, thereby creating a mess.
4852 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4853 space and tab.
4855 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4856 its new features into the main test script, reducing the number of scripts.
4858 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4859 were backward compatible, and made the (?i) apply to the whole pattern, as if
4860 /i were given. Now it behaves more logically, and applies the option setting
4861 only to what follows. PCRE has been changed to follow suit. However, if it
4862 finds options settings right at the start of the pattern, it extracts them into
4863 the global options, as before. Thus, they show up in the info data.
4865 8. Added support for the \Q...\E escape sequence. Characters in between are
4866 treated as literals. This is slightly different from Perl in that $ and @ are
4867 also handled as literals inside the quotes. In Perl, they will cause variable
4868 interpolation. Note the following examples:
4870 Pattern PCRE matches Perl matches
4872 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4873 \Qabc\$xyz\E abc\$xyz abc\$xyz
4874 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4876 For compatibility with Perl, \Q...\E sequences are recognized inside character
4877 classes as well as outside them.
4879 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4880 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4881 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4882 signed/unsigned warnings.
4884 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4885 option for pcretest, so I've replaced it by a simple function that does just
4886 that job.
4888 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4889 "pcregrep -".
4891 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4892 Java package. This provides some syntactic sugar for simple cases of what my
4893 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4894 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4895 item, you can use this simplified notation. Note that only makes sense with
4896 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4897 greediness, whatever the setting of the PCRE_UNGREEDY option.
4899 13. A change of greediness default within a pattern was not taking effect at
4900 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4901 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4902 was abstracted outside.
4904 14. PCRE now supports the \G assertion. It is true when the current matching
4905 position is at the start point of the match. This differs from \A when the
4906 starting offset is non-zero. Used with the /g option of pcretest (or similar
4907 code), it works in the same way as it does for Perl's /g option. If all
4908 alternatives of a regex begin with \G, the expression is anchored to the start
4909 match position, and the "anchored" flag is set in the compiled expression.
4911 15. Some bugs concerning the handling of certain option changes within patterns
4912 have been fixed. These applied to options other than (?ims). For example,
4913 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4914 been the other way round. Some of this was related to change 7 above.
4916 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4917 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4918 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4919 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4921 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4922 mode. Unlike ".", it always matches newline, whatever the setting of
4923 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4924 assertions. Perl allows it, but it doesn't (in general) work because it can't
4925 calculate the length of the lookbehind. At least, that's the case for Perl
4926 5.8.0 - I've been told they are going to document that it doesn't work in
4927 future.
4929 18. Added an error diagnosis for escapes that PCRE does not support: these are
4930 \L, \l, \N, \P, \p, \U, \u, and \X.
4932 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4933 reading past the end of the pattern in cases such as /[abcd/.
4935 20. PCRE was getting more memory than necessary for patterns with classes that
4936 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4938 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4939 compiling PCRE for use with Virtual Pascal.
4941 22. Small fix to the Makefile to make it work properly if the build is done
4942 outside the source tree.
4944 23. Added a new extension: a condition to go with recursion. If a conditional
4945 subpattern starts with (?(R) the "true" branch is used if recursion has
4946 happened, whereas the "false" branch is used only at the top level.
4948 24. When there was a very long string of literal characters (over 255 bytes
4949 without UTF support, over 250 bytes with UTF support), the computation of how
4950 much memory was required could be incorrect, leading to segfaults or other
4951 strange effects.
4953 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4954 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4955 there was a subsequent back reference to those brackets. This meant that, for
4956 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4957 possible to check for precisely this case. All we can do is abandon the
4958 optimization if .* occurs inside capturing brackets when there are any back
4959 references whatsoever. (See below for a better fix that came later.)
4961 26. The handling of the optimization for finding the first character of a
4962 non-anchored pattern, and for finding a character that is required later in the
4963 match were failing in some cases. This didn't break the matching; it just
4964 failed to optimize when it could. The way this is done has been re-implemented.
4966 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4968 28. Added a new feature that provides some of the functionality that Perl
4969 provides with (?{...}). The facility is termed a "callout". The way it is done
4970 in PCRE is for the caller to provide an optional function, by setting
4971 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4972 global variable. By default it is unset, which disables all calling out. To get
4973 the function called, the regex must include (?C) at appropriate points. This
4974 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4975 This provides a means of identifying different callout points. When PCRE
4976 reaches such a point in the regex, if pcre_callout has been set, the external
4977 function is called. It is provided with data in a structure called
4978 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4979 matching continues; if it returns a non-zero value, the match at the current
4980 point fails. However, backtracking will occur if possible. [This was changed
4981 later and other features added - see item 49 below.]
4983 29. pcretest is upgraded to test the callout functionality. It provides a
4984 callout function that displays information. By default, it shows the start of
4985 the match and the current position in the text. There are some new data escapes
4986 to vary what happens:
4988 \C+ in addition, show current contents of captured substrings
4989 \C- do not supply a callout function
4990 \C!n return 1 when callout number n is reached
4991 \C!n!m return 1 when callout number n is reached for the mth time
4993 30. If pcregrep was called with the -l option and just a single file name, it
4994 output "<stdin>" if a match was found, instead of the file name.
4996 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4997 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4998 pcre_exec(). This saves a malloc/free per call. The default value of
4999 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5000 when configuring.
5002 32. The default maximum size of a compiled pattern is 64K. There have been a
5003 few cases of people hitting this limit. The code now uses macros to handle the
5004 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5005 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5006 configuring. Tests 2 and 5 work only with 2-byte links because they output
5007 debugging information about compiled patterns.
5009 33. Internal code re-arrangements:
5011 (a) Moved the debugging function for printing out a compiled regex into
5012 its own source file (printint.c) and used #include to pull it into
5013 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5014 separate copies.
5016 (b) Defined the list of op-code names for debugging as a macro in
5017 internal.h so that it is next to the definition of the opcodes.
5019 (c) Defined a table of op-code lengths for simpler skipping along compiled
5020 code. This is again a macro in internal.h so that it is next to the
5021 definition of the opcodes.
5023 34. Added support for recursive calls to individual subpatterns, along the
5024 lines of Robin Houston's patch (but implemented somewhat differently).
5026 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5027 allow it to read and process whole directories in Win32. This code was
5028 contributed by Lionel Fourquaux; it has not been tested by me.
5030 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5031 used to name a group. Names consist of alphanumerics and underscores, and must
5032 be unique. Back references use the syntax (?P=name) and recursive calls use
5033 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5034 numbers. The function pcre_fullinfo() can be used after compilation to extract
5035 a name/number map. There are three relevant calls:
5037 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5038 PCRE_INFO_NAMECOUNT yields the number of entries
5039 PCRE_INFO_NAMETABLE yields a pointer to the map.
5041 The map is a vector of fixed-size entries. The size of each entry depends on
5042 the length of the longest name used. The first two bytes of each entry are the
5043 group number, most significant byte first. There follows the corresponding
5044 name, zero terminated. The names are in alphabetical order.
5046 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5047 case instead of 255. Making it the same both with and without UTF-8 support
5048 means that the same test output works with both.
5050 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5051 calling malloc() with a zero argument.
5053 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5054 optimization. I've improved things by keeping a bitmap of backreferences with
5055 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5056 fact referenced, the optimization can be applied. It is unlikely that a
5057 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5058 the match to follow \n) will appear inside brackets with a number greater than
5059 31, but if it does, any back reference > 31 suppresses the optimization.
5061 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5062 of disabling numbered capturing parentheses. Any opening parenthesis that is
5063 not followed by ? behaves as if it were followed by ?: but named parentheses
5064 can still be used for capturing (and they will acquire numbers in the usual
5065 way).
5067 41. Redesigned the return codes from the match() function into yes/no/error so
5068 that errors can be passed back from deep inside the nested calls. A malloc
5069 failure while inside a recursive subpattern call now causes the
5070 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5072 42. It is now possible to set a limit on the number of times the match()
5073 function is called in a call to pcre_exec(). This facility makes it possible to
5074 limit the amount of recursion and backtracking, though not in a directly
5075 obvious way, because the match() function is used in a number of different
5076 circumstances. The count starts from zero for each position in the subject
5077 string (for non-anchored patterns). The default limit is, for compatibility, a
5078 large number, namely 10 000 000. You can change this in two ways:
5080 (a) When configuring PCRE before making, you can use --with-match-limit=n
5081 to set a default value for the compiled library.
5083 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5084 a different value is set. See 45 below.
5086 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5088 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5089 of things that can be changed at compile time. The first argument specifies
5090 what is wanted and the second points to where the information is to be placed.
5091 The current list of available information is:
5095 The output is an integer that is set to one if UTF-8 support is available;
5096 otherwise it is set to zero.
5100 The output is an integer that it set to the value of the code that is used for
5101 newline. It is either LF (10) or CR (13).
5105 The output is an integer that contains the number of bytes used for internal
5106 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5110 The output is an integer that contains the threshold above which the POSIX
5111 interface uses malloc() for output vectors. See item 31 above.
5115 The output is an unsigned integer that contains the default limit of the number
5116 of match() calls in a pcre_exec() execution. See 42 above.
5118 44. pcretest has been upgraded by the addition of the -C option. This causes it
5119 to extract all the available output from the new pcre_config() function, and to
5120 output it. The program then exits immediately.
5122 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5123 order to support additional features. One way would have been to define
5124 pcre_exec2() (for example) with extra arguments, but this would not have been
5125 extensible, and would also have required all calls to the original function to
5126 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5127 is used for passing in "extra" data from pcre_study().
5129 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5130 contains the following fields:
5132 flags a bitmap indicating which of the following fields are set
5133 study_data opaque data from pcre_study()
5134 match_limit a way of specifying a limit on match() calls for a specific
5135 call to pcre_exec()
5136 callout_data data for callouts (see 49 below)
5138 The flag bits are also defined in pcre.h, and are
5144 The pcre_study() function now returns one of these new pcre_extra blocks, with
5145 the actual study data pointed to by the study_data field, and the
5146 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5147 before. That is, this change is entirely upwards-compatible and requires no
5148 change to existing code.
5150 If you want to pass in additional data to pcre_exec(), you can either place it
5151 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5152 block.
5154 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5155 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5156 times with different match limits, until it finds the minimum value needed for
5157 pcre_exec() to complete. The value is then output. This can be instructive; for
5158 most simple matches the number is quite small, but for pathological cases it
5159 gets very large very quickly.
5161 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5162 returns the size of the data block pointed to by the study_data field in a
5163 pcre_extra block, that is, the value that was passed as the argument to
5164 pcre_malloc() when PCRE was getting memory in which to place the information
5165 created by pcre_study(). The fourth argument should point to a size_t variable.
5166 pcretest has been extended so that this information is shown after a successful
5167 pcre_study() call when information about the compiled regex is being displayed.
5169 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5170 because what follows is always an absolute path. (Later: it turns out that this
5171 is more than cosmetic for MinGW, because it doesn't like empty path
5172 components.)
5174 49. Some changes have been made to the callout feature (see 28 above):
5176 (i) A callout function now has three choices for what it returns:
5178 0 => success, carry on matching
5179 > 0 => failure at this point, but backtrack if possible
5180 < 0 => serious error, return this value from pcre_exec()
5182 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5183 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5184 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5185 use by callout functions. It will never be used by PCRE itself.
5187 (ii) The pcre_extra structure (see 45 above) has a void * field called
5188 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5189 pcre_callout_block structure has a field of the same name. The contents of
5190 the field passed in the pcre_extra structure are passed to the callout
5191 function in the corresponding field in the callout block. This makes it
5192 easier to use the same callout-containing regex from multiple threads. For
5193 testing, the pcretest program has a new data escape
5195 \C*n pass the number n (may be negative) as callout_data
5197 If the callout function in pcretest receives a non-zero value as
5198 callout_data, it returns that value.
5200 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5201 there were some redundant $(CFLAGS) in commands that are now specified as
5202 $(LINK), which already includes $(CFLAGS).
5204 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5205 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5206 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5207 one-byte characters throughout. Note that case-insensitive matching applies
5208 only to characters whose values are less than 256. PCRE doesn't support the
5209 notion of cases for higher-valued characters.
5211 (i) A character class whose characters are all within 0-255 is handled as
5212 a bit map, and the map is inverted for negative classes. Previously, a
5213 character > 255 always failed to match such a class; however it should
5214 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5216 (ii) A negated character class with a single character < 255 is coded as
5217 "not this character" (OP_NOT). This wasn't working properly when the test
5218 character was multibyte, either singly or repeated.
5220 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5221 mode, for example: \x{100}{2,3}.
5223 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5224 singly or repeated) now correctly test multibyte characters. However,
5225 PCRE doesn't recognize any characters with values greater than 255 as
5226 digits, spaces, or word characters. Such characters always match \D, \S,
5227 and \W, and never match \d, \s, or \w.
5229 (v) Classes may now contain characters and character ranges with values
5230 greater than 255. For example: [ab\x{100}-\x{400}].
5232 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5233 PCRE in UTF-8 mode.
5235 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5236 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5237 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5238 value.)
5240 53. The single man page has become too large. I have therefore split it up into
5241 a number of separate man pages. These also give rise to individual HTML pages;
5242 these are now put in a separate directory, and there is an index.html page that
5243 lists them all. Some hyperlinking between the pages has been installed.
5245 54. Added convenience functions for handling named capturing parentheses.
5247 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5248 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5249 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5250 are faulted.
5252 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5253 calling configure. These values are used when compiling the dftables.c program
5254 which is run to generate the source of the default character tables. They
5255 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5256 you will need to set these values.
5258 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5261 Version 3.9 02-Jan-02
5262 ---------------------
5264 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5266 2. If --disable-static was given, the building process failed when trying to
5267 build pcretest and pcregrep. (For some reason it was using libtool to compile
5268 them, which is not right, as they aren't part of the library.)
5271 Version 3.8 18-Dec-01
5272 ---------------------
5274 1. The experimental UTF-8 code was completely screwed up. It was packing the
5275 bytes in the wrong order. How dumb can you get?
5278 Version 3.7 29-Oct-01
5279 ---------------------
5281 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5282 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5283 this didn't happen under Solaris 8, where I normally test things.
5285 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5286 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5287 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5288 this fix an hour or so after the initial 3.7 release.)
5291 Version 3.6 23-Oct-01
5292 ---------------------
5294 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5295 offsets passed as NULL with zero offset count.
5297 2. The config.guess and config.sub files had not been updated when I moved to
5298 the latest autoconf.
5301 Version 3.5 15-Aug-01
5302 ---------------------
5304 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5305 had been forgotten.
5307 2. By using declared but undefined structures, we can avoid using "void"
5308 definitions in pcre.h while keeping the internal definitions of the structures
5309 private.
5311 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5312 user point of view, this means that both static and shared libraries are built
5313 by default, but this can be individually controlled. More of the work of
5314 handling this static/shared cases is now inside libtool instead of PCRE's make
5315 file.
5317 4. The pcretest utility is now installed along with pcregrep because it is
5318 useful for users (to test regexs) and by doing this, it automatically gets
5319 relinked by libtool. The documentation has been turned into a man page, so
5320 there are now .1, .txt, and .html versions in /doc.
5322 5. Upgrades to pcregrep:
5323 (i) Added long-form option names like gnu grep.
5324 (ii) Added --help to list all options with an explanatory phrase.
5325 (iii) Added -r, --recursive to recurse into sub-directories.
5326 (iv) Added -f, --file to read patterns from a file.
5328 6. pcre_exec() was referring to its "code" argument before testing that
5329 argument for NULL (and giving an error if it was NULL).
5331 7. Upgraded Makefile.in to allow for compiling in a different directory from
5332 the source directory.
5334 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5335 options bits, the pointer it was passed was to an int instead of to an unsigned
5336 long int. This mattered only on 64-bit systems.
5338 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5339 generated) instead of pcre.in, which it its source. Also made the same change
5340 in several of the .c files.
5342 10. A new release of gcc defines printf() as a macro, which broke pcretest
5343 because it had an ifdef in the middle of a string argument for printf(). Fixed
5344 by using separate calls to printf().
5346 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5347 script, to force use of CR or LF instead of \n in the source. On non-Unix
5348 systems, the value can be set in config.h.
5350 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5351 absolute limit. Changed the text of the error message to make this clear, and
5352 likewise updated the man page.
5354 13. The limit of 99 on the number of capturing subpatterns has been removed.
5355 The new limit is 65535, which I hope will not be a "real" limit.
5358 Version 3.4 22-Aug-00
5359 ---------------------
5361 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5363 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5366 Version 3.3 01-Aug-00
5367 ---------------------
5369 1. If an octal character was given, but the value was greater than \377, it
5370 was not getting masked to the least significant bits, as documented. This could
5371 lead to crashes in some systems.
5373 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5374 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5376 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5377 These just pass their arguments on to (pcre_free)(), but they are provided
5378 because some uses of PCRE bind it to non-C systems that can call its functions,
5379 but cannot call free() or pcre_free() directly.
5381 4. Add "make test" as a synonym for "make check". Corrected some comments in
5382 the Makefile.
5384 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5385 Makefile.
5387 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5388 command called pgrep for grepping around the active processes.
5390 7. Added the beginnings of support for UTF-8 character strings.
5392 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5393 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5394 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5395 out for the ar command.)
5398 Version 3.2 12-May-00
5399 ---------------------
5401 This is purely a bug fixing release.
5403 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5404 of ZA. This was just one example of several cases that could provoke this bug,
5405 which was introduced by change 9 of version 2.00. The code for breaking
5406 infinite loops after an iteration that matches an empty string was't working
5407 correctly.
5409 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5410 when matched against abbab (for example). After matching an empty string, it
5411 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5412 caused it to match further down the string than it should.
5414 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5415 was there because it doesn't seem to be needed, and it causes trouble on some
5416 systems, as it is not a Standard C header. It has been removed.
5418 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5419 were reported on the Macintosh. The changes were from
5421 while ((c = *(++ptr)) != 0 && c != '\n');
5422 to
5423 while ((c = *(++ptr)) != 0 && c != '\n') ;
5425 Totally extraordinary, but if that's what it takes...
5427 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5428 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5429 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5430 assumes the way PCRE uses memmove() (always moving upwards).
5432 6. PCRE is being used in one environment where strchr() is not available. There
5433 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5434 faster code anyway.
5437 Version 3.1 09-Feb-00
5438 ---------------------
5440 The only change in this release is the fixing of some bugs in Makefile.in for
5441 the "install" target:
5443 (1) It was failing to install pcreposix.h.
5445 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5448 Version 3.0 01-Feb-00
5449 ---------------------
5451 1. Add support for the /+ modifier to perltest (to output $` like it does in
5452 pcretest).
5454 2. Add support for the /g modifier to perltest.
5456 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5457 matches null strings.
5459 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5460 pattern. Perl treats empty patterns specially - it reuses the most recent
5461 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5462 effect.
5464 5. The POSIX interface was broken in that it was just handing over the POSIX
5465 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5466 required a bigger vector, with some working space on the end. This means that
5467 the POSIX wrapper now has to get and free some memory, and copy the results.
5469 6. Added some simple autoconf support, placing the test data and the
5470 documentation in separate directories, re-organizing some of the
5471 information files, and making it build pcre-config (a GNU standard). Also added
5472 libtool support for building PCRE as a shared library, which is now the
5473 default.
5475 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5476 09 are not valid octal constants. Single digits will be used for minor values
5477 less than 10.
5479 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5480 existing programs that set these in the POSIX interface can use PCRE without
5481 modification.
5483 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5484 return all that pcre_info() returns, plus additional data. The pcre_info()
5485 function is retained for compatibility, but is considered to be obsolete.
5487 10. Added experimental recursion feature (?R) to handle one common case that
5488 Perl 5.6 will be able to do with (?p{...}).
5490 11. Added support for POSIX character classes like [:alpha:], which Perl is
5491 adopting.
5494 Version 2.08 31-Aug-99
5495 ----------------------
5497 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5498 trying to match at the startoffset position, but instead was moving forward to
5499 the next newline as if a previous match had failed.
5501 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5502 and could get into a loop if a null string was matched other than at the start
5503 of the subject.
5505 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5506 be distinguished at compile time, and for completeness also added PCRE_DATE.
5508 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5509 in GnuWin32 environments.
5512 Version 2.07 29-Jul-99
5513 ----------------------
5515 1. The documentation is now supplied in plain text form and HTML as well as in
5516 the form of man page sources.
5518 2. C++ compilers don't like assigning (void *) values to other pointer types.
5519 In particular this affects malloc(). Although there is no problem in Standard
5520 C, I've put in casts to keep C++ compilers happy.
5522 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5523 should be (const char *).
5525 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5526 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5527 However, I haven't made this a standard facility. The documentation doesn't
5528 mention it, and the Makefile doesn't support it.
5530 5. The Makefile now contains an "install" target, with editable destinations at
5531 the top of the file. The pcretest program is not installed.
5533 6. pgrep -V now gives the PCRE version number and date.
5535 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5536 causing the entire string to be ignored, instead of just the last character.
5538 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5539 non-matching string, it can take a very, very long time, even for strings of
5540 quite modest length, because of the nested recursion. PCRE now does better in
5541 some of these cases. It does this by remembering the last required literal
5542 character in the pattern, and pre-searching the subject to ensure it is present
5543 before running the real match. In other words, it applies a heuristic to detect
5544 some types of certain failure quickly, and in the above example, if presented
5545 with a string that has no trailing " it gives "no match" very quickly.
5547 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5548 other alternatives are tried instead.
5551 Version 2.06 09-Jun-99
5552 ----------------------
5554 1. Change pcretest's output for amount of store used to show just the code
5555 space, because the remainder (the data block) varies in size between 32-bit and
5556 64-bit systems.
5558 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5559 start matching at. This allows lookbehinds to work when searching for multiple
5560 occurrences in a string.
5562 3. Added additional options to pcretest for testing multiple occurrences:
5564 /+ outputs the rest of the string that follows a match
5565 /g loops for multiple occurrences, using the new startoffset argument
5566 /G loops for multiple occurrences by passing an incremented pointer
5568 4. PCRE wasn't doing the "first character" optimization for patterns starting
5569 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5570 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5571 the letter 'x'. On long subject strings, this gives a significant speed-up.
5574 Version 2.05 21-Apr-99
5575 ----------------------
5577 1. Changed the type of magic_number from int to long int so that it works
5578 properly on 16-bit systems.
5580 2. Fixed a bug which caused patterns starting with .* not to work correctly
5581 when the subject string contained newline characters. PCRE was assuming
5582 anchoring for such patterns in all cases, which is not correct because .* will
5583 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5584 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5585 must be retried after every newline in the subject.
5588 Version 2.04 18-Feb-99
5589 ----------------------
5591 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5592 computation of the store needed to hold the pattern was incorrect (too large).
5593 If such patterns were nested a few deep, this could multiply and become a real
5594 problem.
5596 2. Added /M option to pcretest to show the memory requirement of a specific
5597 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5599 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5600 compiled in such a way that the backtracking after subsequent failure was
5601 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5602 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5605 Version 2.03 02-Feb-99
5606 ----------------------
5608 1. Fixed typo and small mistake in man page.
5610 2. Added 4th condition (GPL supersedes if conflict) and created separate
5611 LICENCE file containing the conditions.
5613 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5614 Perl, that is the internal \ allows the delimiter to be included in the
5615 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5616 the final delimiter, add \ to the end of the pattern (to test the error).
5618 4. Added the convenience functions for extracting substrings after a successful
5619 match. Updated pcretest to make it able to test these functions.
5622 Version 2.02 14-Jan-99
5623 ----------------------
5625 1. Initialized the working variables associated with each extraction so that
5626 their saving and restoring doesn't refer to uninitialized store.
5628 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5629 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5630 fix the problem.
5632 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5633 calls, and wasn't printing the correct value for compiling calls. Increased the
5634 default value of LOOPREPEAT, and the number of significant figures in the
5635 times.
5637 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5639 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5640 a building problem on Windows NT with a FAT file system.
5643 Version 2.01 21-Oct-98
5644 ----------------------
5646 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5647 to character tables built by pcre_maketables() in the current locale. If NULL
5648 is passed, the default tables are used.
5651 Version 2.00 24-Sep-98
5652 ----------------------
5654 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5655 it any more.
5657 2. Allow quantification of (?>) groups, and make it work correctly.
5659 3. The first character computation wasn't working for (?>) groups.
5661 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5662 end of the subject) and add 5.005's \z, which really does match only at the
5663 very end of the subject.
5665 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5667 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5668 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5669 localized options. All options to pcre_study() were also removed.
5671 7. Add other new features from 5.005:
5673 $(?<= positive lookbehind
5674 $(?<! negative lookbehind
5675 (?imsx-imsx) added the unsetting capability
5676 such a setting is global if at outer level; local otherwise
5677 (?imsx-imsx:) non-capturing groups with option setting
5678 (?(cond)re|re) conditional pattern matching
5680 A backreference to itself in a repeated group matches the previous
5681 captured string.
5683 8. General tidying up of studying (both automatic and via "study")
5684 consequential on the addition of new assertions.
5686 9. As in 5.005, unlimited repeated groups that could match an empty substring
5687 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5688 runtime if any iteration does actually match an empty substring.
5690 10. Include the RunTest script in the distribution.
5692 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5693 discrepancies, some of which were old and were also with respect to 5.004. They
5694 have now been fixed.
5697 Version 1.09 28-Apr-98
5698 ----------------------
5700 1. A negated single character class followed by a quantifier with a minimum
5701 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5702 program crashes, or just wrong answers. This did not apply to negated classes
5703 containing more than one character, or to minima other than one.
5706 Version 1.08 27-Mar-98
5707 ----------------------
5709 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5711 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5712 latter must appear before anything that relies on it in the pattern.
5715 Version 1.07 16-Feb-98
5716 ----------------------
5718 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5719 repeat of a potentially empty string).
5722 Version 1.06 23-Jan-98
5723 ----------------------
5725 1. Added Markus Oberhumer's little patches for C++.
5727 2. Literal strings longer than 255 characters were broken.
5730 Version 1.05 23-Dec-97
5731 ----------------------
5733 1. Negated character classes containing more than one character were failing if
5734 PCRE_CASELESS was set at run time.
5737 Version 1.04 19-Dec-97
5738 ----------------------
5740 1. Corrected the man page, where some "const" qualifiers had been omitted.
5742 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5743 input syntax.
5745 3. Fixed memory leak which occurred when a regex with back references was
5746 matched with an offsets vector that wasn't big enough. The temporary memory
5747 that is used in this case wasn't being freed if the match failed.
5749 4. Tidied pcretest to ensure it frees memory that it gets.
5751 5. Temporary memory was being obtained in the case where the passed offsets
5752 vector was exactly big enough.
5754 6. Corrected definition of offsetof() from change 5 below.
5756 7. I had screwed up change 6 below and broken the rules for the use of
5757 setjmp(). Now fixed.
5760 Version 1.03 18-Dec-97
5761 ----------------------
5763 1. A erroneous regex with a missing opening parenthesis was correctly
5764 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5765 on some systems.
5767 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5768 it was reported that one broken compiler failed on the former because "code" is
5769 also an independent variable.
5771 3. The erroneous regex a[]b caused an array overrun reference.
5773 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5774 fail on data ending with that character. (It was going on too far, and checking
5775 the next character, typically a binary zero.) This was specific to the
5776 optimized code for single-character negative classes.
5778 5. Added a contributed patch from the TIN world which does the following:
5780 + Add an undef for memmove, in case the the system defines a macro for it.
5782 + Add a definition of offsetof(), in case there isn't one. (I don't know
5783 the reason behind this - offsetof() is part of the ANSI standard - but
5784 it does no harm).
5786 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5787 most of the places where whitespace preceded '#'. I have given up and
5788 allowed the remaining 2 cases to be at the margin.
5790 + Rename some variables in pcre to eliminate shadowing. This seems very
5791 pedantic, but does no harm, of course.
5793 6. Moved the call to setjmp() into its own function, to get rid of warnings
5794 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5796 7. Constructs such as \d{8,} were compiling into the equivalent of
5797 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5798 outcome, but in this particular case used more store than had been allocated,
5799 which caused the bug to be discovered because it threw up an internal error.
5801 8. The debugging code in both pcre and pcretest for outputting the compiled
5802 form of a regex was going wrong in the case of back references followed by
5803 curly-bracketed repeats.
5806 Version 1.02 12-Dec-97
5807 ----------------------
5809 1. Typos in pcre.3 and comments in the source fixed.
5811 2. Applied a contributed patch to get rid of places where it used to remove
5812 'const' from variables, and fixed some signed/unsigned and uninitialized
5813 variable warnings.
5815 3. Added the "runtest" target to Makefile.
5817 4. Set default compiler flag to -O2 rather than just -O.
5820 Version 1.01 19-Nov-97
5821 ----------------------
5823 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5824 like /([ab]*)*/, that is, for classes with more than one character in them.
5826 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5827 as /((?>a*))*/ (a PCRE_EXTRA facility).
5830 Version 1.00 18-Nov-97
5831 ----------------------
5833 1. Added compile-time macros to support systems such as SunOS4 which don't have
5834 memmove() or strerror() but have other things that can be used instead.
5836 2. Arranged that "make clean" removes the executables.
5839 Version 0.99 27-Oct-97
5840 ----------------------
5842 1. Fixed bug in code for optimizing classes with only one character. It was
5843 initializing a 32-byte map regardless, which could cause it to run off the end
5844 of the memory it had got.
5846 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5849 Version 0.98 22-Oct-97
5850 ----------------------
5852 1. Fixed bug in code for handling temporary memory usage when there are more
5853 back references than supplied space in the ovector. This could cause segfaults.
5856 Version 0.97 21-Oct-97
5857 ----------------------
5859 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5861 2. Optimized negated single characters not to use a bit map.
5863 3. Brought error texts together as macro definitions; clarified some of them;
5864 fixed one that was wrong - it said "range out of order" when it meant "invalid
5865 escape sequence".
5867 4. Changed some char * arguments to const char *.
5869 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5871 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5872 pcretest.
5875 Version 0.96 16-Oct-97
5876 ----------------------
5878 1. Added a simple "pgrep" utility to the distribution.
5880 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5881 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5882 where "ddd" means "one or more decimal digits".
5884 3. Fixed serious bug. If a pattern had a back reference, but the call to
5885 pcre_exec() didn't supply a large enough ovector to record the related
5886 identifying subpattern, the match always failed. PCRE now remembers the number
5887 of the largest back reference, and gets some temporary memory in which to save
5888 the offsets during matching if necessary, in order to ensure that
5889 backreferences always work.
5891 4. Increased the compatibility with Perl in a number of ways:
5893 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5894 to request this handling. The option can be set at compile or exec time.
5896 (b) $ matches before a terminating newline by default; an option
5897 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5898 mode). The option can be set at compile or exec time.
5900 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5901 the same as Perl's. If the decimal number it represents is less than 10
5902 or there aren't that many previous left capturing parentheses, an octal
5903 escape is read. Inside a character class, it's always an octal escape,
5904 even if it is a single digit.
5906 (d) An escaped but undefined alphabetic character is taken as a literal,
5907 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5908 escapes.
5910 (e) {0} is now permitted. (The previous item is removed from the compiled
5911 pattern).
5913 5. Changed all the names of code files so that the basic parts are no longer
5914 than 10 characters, and abolished the teeny "globals.c" file.
5916 6. Changed the handling of character classes; they are now done with a 32-byte
5917 bit map always.
5919 7. Added the -d and /D options to pcretest to make it possible to look at the
5920 internals of compilation without having to recompile pcre.
5923 Version 0.95 23-Sep-97
5924 ----------------------
5926 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5927 \x20 at the start of a run of normal characters. These were being treated as
5928 real characters, instead of the source characters being re-checked.
5931 Version 0.94 18-Sep-97
5932 ----------------------
5934 1. The functions are now thread-safe, with the caveat that the global variables
5935 containing pointers to malloc() and free() or alternative functions are the
5936 same for all threads.
5938 2. Get pcre_study() to generate a bitmap of initial characters for non-
5939 anchored patterns when this is possible, and use it if passed to pcre_exec().
5942 Version 0.93 15-Sep-97
5943 ----------------------
5945 1. /(b)|(:+)/ was computing an incorrect first character.
5947 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5948 but not actually doing anything yet.
5950 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5951 as Perl does (e.g. [-az] or [az-]).
5953 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5954 all possible positions.
5956 5. Split up into different modules to avoid including unneeded functions in a
5957 compiled binary. However, compile and exec are still in one module. The "study"
5958 function is split off.
5960 6. The character tables are now in a separate module whose source is generated
5961 by an auxiliary program - but can then be edited by hand if required. There are
5962 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5963 toupper() in the code.
5965 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5966 make them global. Abolish the function for setting them, as the caller can now
5967 set them directly.
5970 Version 0.92 11-Sep-97
5971 ----------------------
5973 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5974 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5976 2. Caseless matching was not working in character classes if the characters in
5977 the pattern were in upper case.
5979 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5981 4. Make PCRE_ANCHORED public and accept as a compile option.
5983 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5984 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5985 pass them.
5987 6. Give an error if bad option bits passed at compile or run time.
5989 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5990 pcretest to cause it to pass that flag.
5992 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5993 options, and the first character, if set.
5995 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5998 Version 0.91 10-Sep-97
5999 ----------------------
6001 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
6002 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
6004 2. PCRE was looping on encountering an indefinitely repeated back reference to
6005 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
6006 Perl does - treats the match as successful.
6008 ****


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

  ViewVC Help
Powered by ViewVC 1.1.5