ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1669 - (show annotations)
Tue Oct 18 15:10:09 2016 UTC (2 days, 20 hours ago) by ph10
File size: 291797 byte(s)
Error occurred while calculating annotation data.
Fix optimization bugs for patterns starting with lookaheads.
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 8.40 xx-xxxx-2016
8 -------------------------
10 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
11 the match extended over a line boundary.
13 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
14 MSVC static compilation, putting the first patch under a new option.
16 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
18 4. Ignore "show all captures" (/=) for DFA matching.
20 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
22 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
23 without PCRE_UCP set, a negative character type such as \D in a positive
24 class should cause all characters greater than 255 to match, whatever else
25 is in the class. There was a bug that caused this not to happen if a
26 Unicode property item was added to such a class, for example [\D\P{Nd}] or
27 [\W\pL].
29 7. When pcretest was outputing information from a callout, the caret indicator
30 for the current position in the subject line was incorrect if it was after
31 an escape sequence for a character whose code point was greater than
32 \x{ff}.
34 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
35 the conditional was interpreted as a reference to capturing group 1 instead
36 of a test for recursion. Any group whose name began with R was
37 misinterpreted in this way. (The reference interpretation should only
38 happen if the group's name is precisely "R".)
40 9. A number of bugs have been mended relating to match start-up optimizations
41 when the first thing in a pattern is a positive lookahead. These all
42 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
44 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
45 both an initial 'X' and a following 'X'.
46 (b) Some patterns starting with an assertion that started with .* were
47 incorrectly optimized as having to match at the start of the subject or
48 after a newline. There are cases where this is not true, for example,
49 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
50 start with spaces. Starting .* in an assertion is no longer taken as an
51 indication of matching at the start (or after a newline).
54 Version 8.39 14-June-2016
55 -------------------------
57 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
58 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
59 misbehaved. This bug was found by the LLVM fuzzer.
61 2. Similar to the above, if an isolated \E was present between an item and its
62 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
63 bug was found by the LLVM fuzzer.
65 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
66 working correctly in UCP mode.
68 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
69 was set when the pmatch argument was NULL. It now returns REG_INVARG.
71 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
73 6. An empty \Q\E sequence between an item and its qualifier caused
74 pcre_compile() to misbehave when auto callouts were enabled. This bug was
75 found by the LLVM fuzzer.
77 7. If a pattern that was compiled with PCRE_EXTENDED started with white
78 space or a #-type comment that was followed by (?-x), which turns off
79 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
80 pcre_compile() assumed that (?-x) applied to the whole pattern and
81 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
83 8. A call of pcre_copy_named_substring() for a named substring whose number
84 was greater than the space in the ovector could cause a crash.
86 9. Yet another buffer overflow bug involved duplicate named groups with a
87 group that reset capture numbers (compare 8.38/7 below). Once again, I have
88 just allowed for more memory, even if not needed. (A proper fix is
89 implemented in PCRE2, but it involves a lot of refactoring.)
91 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
92 start of the match to be earlier than the end.
94 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
96 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
97 assertion, caused pcretest to generate incorrect output, and also to read
98 uninitialized memory (detected by ASAN or valgrind).
100 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
101 nested set of parentheses of sufficient size caused an overflow of the
102 compiling workspace (which was diagnosed, but of course is not desirable).
104 14. And yet another buffer overflow bug involving duplicate named groups, this
105 time nested, with a nested back reference. Yet again, I have just allowed
106 for more memory, because anything more needs all the refactoring that has
107 been done for PCRE2. An example pattern that provoked this bug is:
108 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
109 registered as CVE-2016-1283.
111 15. pcretest went into a loop if global matching was requested with an ovector
112 size less than 2. It now gives an error message. This bug was found by
113 afl-fuzz.
115 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
116 ("assertion expected") when (?(?C) was not followed by an opening
117 parenthesis.
119 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
120 actually affect anything, by sheer luck.
122 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
123 static compilation.
125 19. Modified the RunTest script to incorporate a valgrind suppressions file so
126 that certain errors, provoked by the SSE2 instruction set when JIT is used,
127 are ignored.
129 20. A racing condition is fixed in JIT reported by Mozilla.
131 21. Minor code refactor to avoid "array subscript is below array bounds"
132 compiler warning.
134 22. Minor code refactor to avoid "left shift of negative number" warning.
136 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
137 UCP support.
139 24. Refactor to avoid compiler warnings in pcrecpp.cc.
141 25. Refactor to fix a typo in pcre_jit_test.c
143 26. Patch to support compiling pcrecpp.cc with Intel compiler.
146 Version 8.38 23-November-2015
147 -----------------------------
149 1. If a group that contained a recursive back reference also contained a
150 forward reference subroutine call followed by a non-forward-reference
151 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
152 compile correct code, leading to undefined behaviour or an internally
153 detected error. This bug was discovered by the LLVM fuzzer.
155 2. Quantification of certain items (e.g. atomic back references) could cause
156 incorrect code to be compiled when recursive forward references were
157 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
158 This bug was discovered by the LLVM fuzzer.
160 3. A repeated conditional group whose condition was a reference by name caused
161 a buffer overflow if there was more than one group with the given name.
162 This bug was discovered by the LLVM fuzzer.
164 4. A recursive back reference by name within a group that had the same name as
165 another group caused a buffer overflow. For example:
166 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
168 5. A forward reference by name to a group whose number is the same as the
169 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
170 a buffer overflow at compile time. This bug was discovered by the LLVM
171 fuzzer.
173 6. A lookbehind assertion within a set of mutually recursive subpatterns could
174 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
176 7. Another buffer overflow bug involved duplicate named groups with a
177 reference between their definition, with a group that reset capture
178 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
179 fixed by always allowing for more memory, even if not needed. (A proper fix
180 is implemented in PCRE2, but it involves more refactoring.)
182 8. There was no check for integer overflow in subroutine calls such as (?123).
184 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
185 being treated as a literal 'l' instead of causing an error.
187 10. There was a buffer overflow if pcre_exec() was called with an ovector of
188 size 1. This bug was found by american fuzzy lop.
190 11. If a non-capturing group containing a conditional group that could match
191 an empty string was repeated, it was not identified as matching an empty
192 string itself. For example: /^(?:(?(1)x|)+)+$()/.
194 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
195 \a and \e in test subject lines.
197 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
198 instead of the EBCDIC value.
200 14. The handling of \c in an EBCDIC environment has been revised so that it is
201 now compatible with the specification in Perl's perlebcdic page.
203 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
204 ASCII/Unicode. This has now been added to the list of characters that are
205 recognized as white space in EBCDIC.
207 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
208 error (correctly) when used outside a class, but did not give an error
209 within a class.
211 17. \h within a class was incorrectly compiled in EBCDIC environments.
213 18. A pattern with an unmatched closing parenthesis that contained a backward
214 assertion which itself contained a forward reference caused buffer
215 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
217 19. JIT should return with error when the compiled pattern requires more stack
218 space than the maximum.
220 20. A possessively repeated conditional group that could match an empty string,
221 for example, /(?(R))*+/, was incorrectly compiled.
223 21. Fix infinite recursion in the JIT compiler when certain patterns such as
224 /(?:|a|){100}x/ are analysed.
226 22. Some patterns with character classes involving [: and \\ were incorrectly
227 compiled and could cause reading from uninitialized memory or an incorrect
228 error diagnosis.
230 23. Pathological patterns containing many nested occurrences of [: caused
231 pcre_compile() to run for a very long time.
233 24. A conditional group with only one branch has an implicit empty alternative
234 branch and must therefore be treated as potentially matching an empty
235 string.
237 25. If (?R was followed by - or + incorrect behaviour happened instead of a
238 diagnostic.
240 26. Arrange to give up on finding the minimum matching length for overly
241 complex patterns.
243 27. Similar to (4) above: in a pattern with duplicated named groups and an
244 occurrence of (?| it is possible for an apparently non-recursive back
245 reference to become recursive if a later named group with the relevant
246 number is encountered. This could lead to a buffer overflow. Wen Guanxing
247 from Venustech ADLAB discovered this bug.
249 28. If pcregrep was given the -q option with -c or -l, or when handling a
250 binary file, it incorrectly wrote output to stdout.
252 29. The JIT compiler did not restore the control verb head in case of *THEN
253 control verbs. This issue was found by Karl Skomski with a custom LLVM
254 fuzzer.
256 30. Error messages for syntax errors following \g and \k were giving inaccurate
257 offsets in the pattern.
259 31. Added a check for integer overflow in conditions (?(<digits>) and
260 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
261 fuzzer.
263 32. Handling recursive references such as (?2) when the reference is to a group
264 later in the pattern uses code that is very hacked about and error-prone.
265 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
266 give an internal error if it is obvious that compiling has gone wrong.
268 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
269 This issue was found by Karl Skomski with a custom LLVM fuzzer.
271 34. The JIT compiler should restore the control chain for empty possessive
272 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
274 35. Match limit check added to JIT recursion. This issue was found by Karl
275 Skomski with a custom LLVM fuzzer.
277 36. Yet another case similar to 27 above has been circumvented by an
278 unconditional allocation of extra memory. This issue is fixed "properly" in
279 PCRE2 by refactoring the way references are handled. Wen Guanxing
280 from Venustech ADLAB discovered this bug.
282 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
283 with a custom LLVM fuzzer.
285 38. Fixed a corner case of range optimization in JIT.
287 39. An incorrect error "overran compiling workspace" was given if there were
288 exactly enough group forward references such that the last one extended
289 into the workspace safety margin. The next one would have expanded the
290 workspace. The test for overflow was not including the safety margin.
292 40. A match limit issue is fixed in JIT which was found by Karl Skomski
293 with a custom LLVM fuzzer.
295 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
296 work under Windows. (Why has it taken so long for anyone to notice?)
298 42. In a character class such as [\W\p{Any}] where both a negative-type escape
299 ("not a word character") and a property escape were present, the property
300 escape was being ignored.
302 43. Fix crash caused by very long (*MARK) or (*THEN) names.
304 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
305 by a single ASCII character in a class item, was incorrectly compiled in
306 UCP mode. The POSIX class got lost, but only if the single character
307 followed it.
309 45. [:punct:] in UCP mode was matching some characters in the range 128-255
310 that should not have been matched.
312 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
313 class, all characters with code points greater than 255 are in the class.
314 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
315 such as \w are turned into Unicode properties), wide characters were not
316 correctly handled, and could fail to match.
319 Version 8.37 28-April-2015
320 --------------------------
322 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
323 for those parentheses to be closed with whatever has been captured so far.
324 However, it was failing to mark any other groups between the hightest
325 capture so far and the currrent group as "unset". Thus, the ovector for
326 those groups contained whatever was previously there. An example is the
327 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
329 2. If an assertion condition was quantified with a minimum of zero (an odd
330 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
332 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
333 unrecognized modifier, a crash could occur.
335 4. An attempt to do global matching in pcretest with a zero-length ovector
336 caused a crash.
338 5. Fixed a memory leak during matching that could occur for a subpattern
339 subroutine call (recursive or otherwise) if the number of captured groups
340 that had to be saved was greater than ten.
342 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
343 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
344 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
346 7. A UTF pattern containing a "not" match of a non-ASCII character and a
347 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
349 8. When a pattern is compiled, it remembers the highest back reference so that
350 when matching, if the ovector is too small, extra memory can be obtained to
351 use instead. A conditional subpattern whose condition is a check on a
352 capture having happened, such as, for example in the pattern
353 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
354 setting the highest backreference number. This mattered only if pcre_exec()
355 was called with an ovector that was too small to hold the capture, and there
356 was no other kind of back reference (a situation which is probably quite
357 rare). The effect of the bug was that the condition was always treated as
358 FALSE when the capture could not be consulted, leading to a incorrect
359 behaviour by pcre_exec(). This bug has been fixed.
361 9. A reference to a duplicated named group (either a back reference or a test
362 for being set in a conditional) that occurred in a part of the pattern where
363 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
364 to be incorrectly calculated, leading to overwriting.
366 10. A mutually recursive set of back references such as (\2)(\1) caused a
367 segfault at study time (while trying to find the minimum matching length).
368 The infinite loop is now broken (with the minimum length unset, that is,
369 zero).
371 11. If an assertion that was used as a condition was quantified with a minimum
372 of zero, matching went wrong. In particular, if the whole group had
373 unlimited repetition and could match an empty string, a segfault was
374 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
375 assertions to be quantified, but not if they are being used as conditions,
376 so the above pattern is faulted by Perl. PCRE has now been changed so that
377 it also rejects such patterns.
379 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
380 failed to allow the zero-repeat case if pcre2_exec() was called with an
381 ovector too small to capture the group.
383 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
384 Red Hat Product Security:
386 (a) A crash if /K and /F were both set with the option to save the compiled
387 pattern.
389 (b) Another crash if the option to print captured substrings in a callout
390 was combined with setting a null ovector, for example \O\C+ as a subject
391 string.
393 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
394 forward reference repeated a large (but limited) number of times within a
395 repeated outer group that has a zero minimum quantifier, caused incorrect
396 code to be compiled, leading to the error "internal error:
397 previously-checked referenced subpattern not found" when an incorrect
398 memory address was read. This bug was reported as "heap overflow",
399 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
400 CVE-2015-2325.
402 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
403 call within a group that also contained a recursive back reference caused
404 incorrect code to be compiled. This bug was reported as "heap overflow",
405 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
406 number CVE-2015-2326.
408 24. Computing the size of the JIT read-only data in advance has been a source
409 of various issues, and new ones are still appear unfortunately. To fix
410 existing and future issues, size computation is eliminated from the code,
411 and replaced by on-demand memory allocation.
413 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
414 adjacent to the end of the range, and the range contained characters with
415 more than one other case, caused incorrect behaviour when compiled in UTF
416 mode. In that example, the range a-j was left out of the class.
418 26. Fix JIT compilation of conditional blocks, which assertion
419 is converted to (*FAIL). E.g: /(?(?!))/.
421 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
422 discovered by the LLVM fuzzer.
424 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
425 when this assertion was used as a condition, for example (?(?!)a|b). In
426 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
427 error about an unsupported item.
429 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
430 possessification code could take exponential time to complete. A recursion
431 depth limit of 1000 has been imposed to limit the resources used by this
432 optimization.
434 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
435 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
436 because \S ensures they are all in the class. The code for doing this was
437 interacting badly with the code for computing the amount of space needed to
438 compile the pattern, leading to a buffer overflow. This bug was discovered
439 by the LLVM fuzzer.
441 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
442 other kinds of group caused stack overflow at compile time. This bug was
443 discovered by the LLVM fuzzer.
445 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
446 between a subroutine call and its quantifier was incorrectly compiled,
447 leading to buffer overflow or other errors. This bug was discovered by the
448 LLVM fuzzer.
450 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
451 assertion after (?(. The code was failing to check the character after
452 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
453 was discovered by the LLVM fuzzer.
455 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
456 a fixed maximum following a group that contains a subroutine reference was
457 incorrectly compiled and could trigger buffer overflow. This bug was
458 discovered by the LLVM fuzzer.
460 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
461 caused a stack overflow instead of the diagnosis of a non-fixed length
462 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
464 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
465 (e.g. /(?<=\Ka)/) could make pcregrep loop.
467 37. There was a similar problem to 36 in pcretest for global matches.
469 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
470 and a subsequent item in the pattern caused a non-match, backtracking over
471 the repeated \X did not stop, but carried on past the start of the subject,
472 causing reference to random memory and/or a segfault. There were also some
473 other cases where backtracking after \C could crash. This set of bugs was
474 discovered by the LLVM fuzzer.
476 39. The function for finding the minimum length of a matching string could take
477 a very long time if mutual recursion was present many times in a pattern,
478 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
479 method has been implemented. This infelicity was discovered by the LLVM
480 fuzzer.
482 40. Static linking against the PCRE library using the pkg-config module was
483 failing on missing pthread symbols.
486 Version 8.36 26-September-2014
487 ------------------------------
489 1. Got rid of some compiler warnings in the C++ modules that were shown up by
490 -Wmissing-field-initializers and -Wunused-parameter.
492 2. The tests for quantifiers being too big (greater than 65535) were being
493 applied after reading the number, and stupidly assuming that integer
494 overflow would give a negative number. The tests are now applied as the
495 numbers are read.
497 3. Tidy code in pcre_exec.c where two branches that used to be different are
498 now the same.
500 4. The JIT compiler did not generate match limit checks for certain
501 bracketed expressions with quantifiers. This may lead to exponential
502 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
503 issue should be resolved now.
505 5. Fixed an issue, which occures when nested alternatives are optimized
506 with table jumps.
508 6. Inserted two casts and changed some ints to size_t in the light of some
509 reported 64-bit compiler warnings (Bugzilla 1477).
511 7. Fixed a bug concerned with zero-minimum possessive groups that could match
512 an empty string, which sometimes were behaving incorrectly in the
513 interpreter (though correctly in the JIT matcher). This pcretest input is
514 an example:
516 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
519 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
520 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
521 for an empty string was breaking the inner loop and carrying on at a lower
522 level, when possessive repeated groups should always return to a higher
523 level as they have no backtrack points in them. The empty string test now
524 occurs at the outer level.
526 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
527 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
529 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
530 doesn't).
532 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
533 to hold a pattern to be incorrectly computed (too small) when there were
534 named back references to duplicated names. This could cause "internal
535 error: code overflow" or "double free or corruption" or other memory
536 handling errors.
538 11. When named subpatterns had the same prefixes, back references could be
539 confused. For example, in this pattern:
541 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
543 the reference to 'Name' was incorrectly treated as a reference to a
544 duplicate name.
546 12. A pattern such as /^s?c/mi8 where the optional character has more than
547 one "other case" was incorrectly compiled such that it would only try to
548 match starting at "c".
550 13. When a pattern starting with \s was studied, VT was not included in the
551 list of possible starting characters; this should have been part of the
552 8.34/18 patch.
554 14. If a character class started [\Qx]... where x is any character, the class
555 was incorrectly terminated at the ].
557 15. If a pattern that started with a caseless match for a character with more
558 than one "other case" was studied, PCRE did not set up the starting code
559 unit bit map for the list of possible characters. Now it does. This is an
560 optimization improvement, not a bug fix.
562 16. The Unicode data tables have been updated to Unicode 7.0.0.
564 17. Fixed a number of memory leaks in pcregrep.
566 18. Avoid a compiler warning (from some compilers) for a function call with
567 a cast that removes "const" from an lvalue by using an intermediate
568 variable (to which the compiler does not object).
570 19. Incorrect code was compiled if a group that contained an internal recursive
571 back reference was optional (had quantifier with a minimum of zero). This
572 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
573 caused segmentation faults because of stack overflows at compile time.
575 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
576 group that is quantified with an indefinite repeat, caused a compile-time
577 loop which used up all the system stack and provoked a segmentation fault.
578 This was not the same bug as 19 above.
580 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
581 Patch by Mike Frysinger.
584 Version 8.35 04-April-2014
585 --------------------------
587 1. A new flag is set, when property checks are present in an XCLASS.
588 When this flag is not set, PCRE can perform certain optimizations
589 such as studying these XCLASS-es.
591 2. The auto-possessification of character sets were improved: a normal
592 and an extended character set can be compared now. Furthermore
593 the JIT compiler optimizes more character set checks.
595 3. Got rid of some compiler warnings for potentially uninitialized variables
596 that show up only when compiled with -O2.
598 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
599 of a match later then the end of the match. The pcretest program was not
600 handling the case sensibly - it was outputting from the start to the next
601 binary zero. It now reports this situation in a message, and outputs the
602 text from the end to the start.
604 5. Fast forward search is improved in JIT. Instead of the first three
605 characters, any three characters with fixed position can be searched.
606 Search order: first, last, middle.
608 6. Improve character range checks in JIT. Characters are read by an inprecise
609 function now, which returns with an unknown value if the character code is
610 above a certain threshold (e.g: 256). The only limitation is that the value
611 must be bigger than the threshold as well. This function is useful when
612 the characters above the threshold are handled in the same way.
614 7. The macros whose names start with RAWUCHAR are placeholders for a future
615 mode in which only the bottom 21 bits of 32-bit data items are used. To
616 make this more memorable for those maintaining the code, the names have
617 been changed to start with UCHAR21, and an extensive comment has been added
618 to their definition.
620 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
621 to the export list in Makefile.am (they were accidentally omitted from the
622 8.34 tarball).
624 9. The informational output from pcretest used the phrase "starting byte set"
625 which is inappropriate for the 16-bit and 32-bit libraries. As the output
626 for "first char" and "need char" really means "non-UTF-char", I've changed
627 "byte" to "char", and slightly reworded the output. The documentation about
628 these values has also been (I hope) clarified.
630 10. Another JIT related optimization: use table jumps for selecting the correct
631 backtracking path, when more than four alternatives are present inside a
632 bracket.
634 11. Empty match is not possible, when the minimum length is greater than zero,
635 and there is no \K in the pattern. JIT should avoid empty match checks in
636 such cases.
638 12. In a caseless character class with UCP support, when a character with more
639 than one alternative case was not the first character of a range, not all
640 the alternative cases were added to the class. For example, s and \x{17f}
641 are both alternative cases for S: the class [RST] was handled correctly,
642 but [R-T] was not.
644 13. The configure.ac file always checked for pthread support when JIT was
645 enabled. This is not used in Windows, so I have put this test inside a
646 check for the presence of windows.h (which was already tested for).
648 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
649 The algorithm provides a way to skip certain starting offsets, and usually
650 faster than linear prefix searches.
652 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
653 as for 'fr_FR' and 'french'. For some reason, however, it then used the
654 Windows-specific input and output files, which have 'french' screwed in.
655 So this could never have worked. One of the problems with locales is that
656 they aren't always the same. I have now updated RunTest so that it checks
657 the output of the locale test (test 3) against three different output
658 files, and it allows the test to pass if any one of them matches. With luck
659 this should make the test pass on some versions of Solaris where it was
660 failing. Because of the uncertainty, the script did not used to stop if
661 test 3 failed; it now does. If further versions of a French locale ever
662 come to light, they can now easily be added.
664 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
665 there was a message during ./configure, but it did not stop. This now
666 provokes an error. The invalid example in README has been corrected.
667 If a value less than the minimum is given, the minimum value has always
668 been used, but now a warning is given.
670 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
671 was a bug in the test system, which is now fixed. Also, the list of various
672 configurations that are tested for each release did not have one with both
673 16/32 bits and --enable-bar-anycrlf. It now does.
675 18. pcretest was missing "-C bsr" for displaying the \R default setting.
677 19. Little endian PowerPC systems are supported now by the JIT compiler.
679 20. The fast forward newline mechanism could enter to an infinite loop on
680 certain invalid UTF-8 input. Although we don't support these cases
681 this issue can be fixed by a performance optimization.
683 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
684 not take account if existing stack usage. There is now a new global
685 variable called pcre_stack_guard that can be set to point to an external
686 function to check stack availability. It is called at the start of
687 processing every parenthesized group.
689 22. A typo in the code meant that in ungreedy mode the max/min qualifier
690 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
691 not match "ab".
693 23. When UTF was disabled, the JIT program reported some incorrect compile
694 errors. These messages are silenced now.
696 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
697 compiler.
699 25. Change all the temporary files used in RunGrepTest to be different to those
700 used by RunTest so that the tests can be run simultaneously, for example by
701 "make -j check".
704 Version 8.34 15-December-2013
705 -----------------------------
707 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
708 executable memory. Patch inspired by Carsten Klein.
710 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
711 this macro is never tested and has no effect, because the work to support
712 coverage involves only compiling and linking options and special targets in
713 the Makefile. The comment in config.h implied that defining the macro would
714 enable coverage support, which is totally false. There was also support for
715 setting this macro in the CMake files (my fault, I just copied it from
716 configure). SUPPORT_GCOV has now been removed.
718 3. Make a small performance improvement in strlen16() and strlen32() in
719 pcretest.
721 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
722 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
723 these cases). There was also one in pcretest.c.
725 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
727 6. In UTF mode, the code for checking whether a group could match an empty
728 string (which is used for indefinitely repeated groups to allow for
729 breaking an infinite loop) was broken when the group contained a repeated
730 negated single-character class with a character that occupied more than one
731 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
732 UTF-8 mode). The effect was undefined: the group might or might not be
733 deemed as matching an empty string, or the program might have crashed.
735 7. The code for checking whether a group could match an empty string was not
736 recognizing that \h, \H, \v, \V, and \R must match a character.
738 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
739 an empty string. If it can, pcretest shows this in its information output.
741 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
742 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
743 matched by pcre_exec() without using JIT:
745 (a) If the rest of the pattern did not match after a maximal run of
746 grapheme clusters, the code for backing up to try with fewer of them
747 did not always back up over a full grapheme when characters that do not
748 have the modifier quality were involved, e.g. Hangul syllables.
750 (b) If the match point in a subject started with modifier character, and
751 there was no match, the code could incorrectly back up beyond the match
752 point, and potentially beyond the first character in the subject,
753 leading to a segfault or an incorrect match result.
755 10. A conditional group with an assertion condition could lead to PCRE
756 recording an incorrect first data item for a match if no other first data
757 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
758 first data item, and therefore matched "ca" after "c" instead of at the
759 start.
761 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
762 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
764 12. The source of pcregrep now includes z/OS-specific code so that it can be
765 compiled for z/OS as part of the special z/OS distribution.
767 13. Added the -T and -TM options to pcretest.
769 14. The code in pcre_compile.c for creating the table of named capturing groups
770 has been refactored. Instead of creating the table dynamically during the
771 actual compiling pass, the information is remembered during the pre-compile
772 pass (on the stack unless there are more than 20 named groups, in which
773 case malloc() is used) and the whole table is created before the actual
774 compile happens. This has simplified the code (it is now nearly 150 lines
775 shorter) and prepared the way for better handling of references to groups
776 with duplicate names.
778 15. A back reference to a named subpattern when there is more than one of the
779 same name now checks them in the order in which they appear in the pattern.
780 The first one that is set is used for the reference. Previously only the
781 first one was inspected. This change makes PCRE more compatible with Perl.
783 16. Unicode character properties were updated from Unicode 6.3.0.
785 17. The compile-time code for auto-possessification has been refactored, based
786 on a patch by Zoltan Herczeg. It now happens after instead of during
787 compilation. The code is cleaner, and more cases are handled. The option
788 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
789 options in pcretest are provided to set it. It can also be set by
790 (*NO_AUTO_POSSESS) at the start of a pattern.
792 18. The character VT has been added to the default ("C" locale) set of
793 characters that match \s and are generally treated as white space,
794 following this same change in Perl 5.18. There is now no difference between
795 "Perl space" and "POSIX space". Whether VT is treated as white space in
796 other locales depends on the locale.
798 19. The code for checking named groups as conditions, either for being set or
799 for being recursed, has been refactored (this is related to 14 and 15
800 above). Processing unduplicated named groups should now be as fast at
801 numerical groups, and processing duplicated groups should be faster than
802 before.
804 20. Two patches to the CMake build system, by Alexander Barkov:
806 (1) Replace the "source" command by "." in CMakeLists.txt because
807 "source" is a bash-ism.
809 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
810 without these the CMake build does not work on Solaris.
812 21. Perl has changed its handling of \8 and \9. If there is no previously
813 encountered capturing group of those numbers, they are treated as the
814 literal characters 8 and 9 instead of a binary zero followed by the
815 literals. PCRE now does the same.
817 22. Following Perl, added \o{} to specify codepoints in octal, making it
818 possible to specify values greater than 0777 and also making them
819 unambiguous.
821 23. Perl now gives an error for missing closing braces after \x{... instead of
822 treating the string as literal. PCRE now does the same.
824 24. RunTest used to grumble if an inappropriate test was selected explicitly,
825 but just skip it when running all tests. This make it awkward to run ranges
826 of tests when one of them was inappropriate. Now it just skips any
827 inappropriate tests, as it always did when running all tests.
829 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
830 character types such as \d or \w, too many callouts were inserted, and the
831 data that they returned was rubbish.
833 26. In UCP mode, \s was not matching two of the characters that Perl matches,
834 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
835 were matched by \h. The code has now been refactored so that the lists of
836 the horizontal and vertical whitespace characters used for \h and \v (which
837 are defined only in one place) are now also used for \s.
839 27. Add JIT support for the 64 bit TileGX architecture.
840 Patch by Jiong Wang (Tilera Corporation).
842 28. Possessive quantifiers for classes (both explicit and automatically
843 generated) now use special opcodes instead of wrapping in ONCE brackets.
845 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
846 (because it's meaningless), this was not happening when PCRE_CASELESS was
847 set. Not wrong, but inefficient.
849 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
850 properties for \w, \d, etc) is present in a test regex. Otherwise if the
851 test contains no characters greater than 255, Perl doesn't realise it
852 should be using Unicode semantics.
854 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
855 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
856 does in Unicode mode.
858 32. Added the "forbid" facility to pcretest so that putting tests into the
859 wrong test files can sometimes be quickly detected.
861 33. There is now a limit (default 250) on the depth of nesting of parentheses.
862 This limit is imposed to control the amount of system stack used at compile
863 time. It can be changed at build time by --with-parens-nest-limit=xxx or
864 the equivalent in CMake.
866 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
867 errors. Perl warns for these when in warning mode, but PCRE has no facility
868 for giving warnings.
870 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
871 However, this was not working for (?!) because it is optimized to (*FAIL),
872 for which PCRE does not allow quantifiers. The optimization is now disabled
873 when a quantifier follows (?!). I can't see any use for this, but it makes
874 things uniform.
876 36. Perl no longer allows group names to start with digits, so I have made this
877 change also in PCRE. It simplifies the code a bit.
879 37. In extended mode, Perl ignores spaces before a + that indicates a
880 possessive quantifier. PCRE allowed a space before the quantifier, but not
881 before the possessive +. It now does.
883 38. The use of \K (reset reported match start) within a repeated possessive
884 group such as (a\Kb)*+ was not working.
886 40. Document that the same character tables must be used at compile time and
887 run time, and that the facility to pass tables to pcre_exec() and
888 pcre_dfa_exec() is for use only with saved/restored patterns.
890 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
891 features for Builds with MSVC:
893 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
894 builds). The .rc files can be used to set FileDescription and many other
895 attributes.
897 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
898 This allows higher-level build scripts which want .pdb files to avoid
899 hard-coding the exact files needed."
901 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
902 mean "start of word" and "end of word", respectively, as a transition aid.
904 43. A minimizing repeat of a class containing codepoints greater than 255 in
905 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
906 compiled to use the heap for recursion.
908 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
909 is configured.
912 Version 8.33 28-May-2013
913 ------------------------
915 1. Added 'U' to some constants that are compared to unsigned integers, to
916 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
917 variables that are added to signed variables, to ensure the result is
918 signed and can be negated.
920 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
921 CMake config files.
923 3. Revise the creation of config.h.generic so that all boolean macros are
924 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
925 overriding via -D on the command line possible.
927 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
928 to unsigned int is reported to make a quite noticeable speed difference in
929 a specific Windows environment. Testing on Linux did also appear to show
930 some benefit (and it is clearly not harmful). Also fixed the definition of
931 Xop which should be unsigned.
933 5. Related to (4), changing the definition of the intermediate variable cc
934 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
935 improvements.
937 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
938 unnecessary spaces.
940 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
941 and later.
943 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
945 9. Optimizing fast_forward_start_bits in JIT.
947 10. Adding support for callouts in JIT, and fixing some issues revealed
948 during this work. Namely:
950 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
952 (b) Minimum length was not checked before the matching is started.
954 11. The value of capture_last that is passed to callouts was incorrect in some
955 cases when there was a capture on one path that was subsequently abandoned
956 after a backtrack. Also, the capture_last value is now reset after a
957 recursion, since all captures are also reset in this case.
959 12. The interpreter no longer returns the "too many substrings" error in the
960 case when an overflowing capture is in a branch that is subsequently
961 abandoned after a backtrack.
963 13. In the pathological case when an offset vector of size 2 is used, pcretest
964 now prints out the matched string after a yield of 0 or 1.
966 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
967 Only supported by the JIT compiler at the moment.
969 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
971 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
972 offset of the starting point of the matching process, provided the offsets
973 vector is large enough.
975 17. The \A escape now records a lookbehind value of 1, though its execution
976 does not actually inspect the previous character. This is to ensure that,
977 in partial multi-segment matching, at least one character from the old
978 segment is retained when a new segment is processed. Otherwise, if there
979 are no lookbehinds in the pattern, \A might match incorrectly at the start
980 of a new segment.
982 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
984 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
985 gives some modest performance improvement in 8-bit mode.
987 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
988 be expressed in certain programming languages using Universal Character
989 Names.
991 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
992 which points out that "non characters" are not "characters that may not
993 appear in Unicode strings" but rather "characters that are reserved for
994 internal use and have only local meaning".
996 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
997 there was a conditional group that depended on an assertion, if the
998 assertion was false, the callout that immediately followed the alternation
999 in the condition was skipped when pcre_exec() was used for matching.
1001 23. Allow an explicit callout to be inserted before an assertion that is the
1002 condition for a conditional group, for compatibility with automatic
1003 callouts, which always insert a callout at this point.
1005 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1006 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1008 25. (*PRUNE) is now supported by the JIT compiler.
1010 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1012 27. Fix the case where there are two or more SKIPs with arguments that may be
1013 ignored.
1015 28. (*SKIP) is now supported by the JIT compiler.
1017 29. (*THEN) is now supported by the JIT compiler.
1019 30. Update RunTest with additional test selector options.
1021 31. The way PCRE handles backtracking verbs has been changed in two ways.
1023 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1024 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1025 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1026 rather obscure rules do not always do the same thing.
1028 (2) Previously, backtracking verbs were confined within assertions. This is
1029 no longer the case for positive assertions, except for (*ACCEPT). Again,
1030 this sometimes improves Perl compatibility, and sometimes does not.
1032 32. A number of tests that were in test 2 because Perl did things differently
1033 have been moved to test 1, because either Perl or PCRE has changed, and
1034 these tests are now compatible.
1036 32. Backtracking control verbs are now handled in the same way in JIT and
1037 interpreter.
1039 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1040 contained a forward subroutine reference caused a compile error.
1042 34. Auto-detect and optimize limited repetitions in JIT.
1044 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1045 blocking (*UTF) etc.
1047 36. In the interpreter, maximizing pattern repetitions for characters and
1048 character types now use tail recursion, which reduces stack usage.
1050 37. The value of the max lookbehind was not correctly preserved if a compiled
1051 and saved regex was reloaded on a host of different endianness.
1053 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1054 of the compiled pattern block, expand the flags field from 16 to 32 bits
1055 because it was almost full.
1057 39. Try madvise first before posix_madvise.
1059 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1060 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1061 original change locked out all matches of empty strings. This has been
1062 changed so that one match of an empty string per line is recognized.
1063 Subsequent searches on the same line (for colouring or for --only-matching,
1064 for example) do not recognize empty strings.
1066 41. Applied a user patch to fix a number of spelling mistakes in comments.
1068 42. Data lines longer than 65536 caused pcretest to crash.
1070 43. Clarified the data type for length and startoffset arguments for pcre_exec
1071 and pcre_dfa_exec in the function-specific man pages, where they were
1072 explicitly stated to be in bytes, never having been updated. I also added
1073 some clarification to the pcreapi man page.
1075 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1076 a segmentation fault.
1079 Version 8.32 30-November-2012
1080 -----------------------------
1082 1. Improved JIT compiler optimizations for first character search and single
1083 character iterators.
1085 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1086 Patch by Daniel Richard G.
1088 3. Single character iterator optimizations in the JIT compiler.
1090 4. Improved JIT compiler optimizations for character ranges.
1092 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1093 Reported by Giuseppe D'Angelo.
1095 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1096 of a line, was being set incorrectly in cases where .* appeared inside
1097 atomic brackets at the start of a pattern, or where there was a subsequent
1098 *PRUNE or *SKIP.
1100 7. Improved instruction cache flush for POWER/PowerPC.
1101 Patch by Daniel Richard G.
1103 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1104 grep:
1106 (a) There is now no limit to the number of patterns to be matched.
1108 (b) An error is given if a pattern is too long.
1110 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1111 are now supported.
1113 (d) --exclude-from and --include-from (multiple use) have been added.
1115 (e) Exclusions and inclusions now apply to all files and directories, not
1116 just to those obtained from scanning a directory recursively.
1118 (f) Multiple uses of -f and --file-list are now supported.
1120 (g) In a Windows environment, the default for -d has been changed from
1121 "read" (the GNU grep default) to "skip", because otherwise the presence
1122 of a directory in the file list provokes an error.
1124 (h) The documentation has been revised and clarified in places.
1126 9. Improve the matching speed of capturing brackets.
1128 10. Changed the meaning of \X so that it now matches a Unicode extended
1129 grapheme cluster.
1131 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1132 out POSIX threads when JIT support is configured.
1134 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1136 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1137 structure could go wrong in environments where size_t is not the same size
1138 as int.
1140 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1141 set.
1143 15. The EBCDIC support had decayed; later updates to the code had included
1144 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1145 general tidy up of EBCDIC-related issues, and the documentation was also
1146 not quite right. There is now a test that can be run on ASCII systems to
1147 check some of the EBCDIC-related things (but is it not a full test).
1149 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1150 in a small tidy to the code.
1152 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1154 18. If the --only-matching (-o) option in pcregrep is specified multiple
1155 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1156 substrings matched by the 1st and 2nd capturing parentheses. A separating
1157 string can be specified by --om-separator (default empty).
1159 19. Improving the first n character searches.
1161 20. Turn case lists for horizontal and vertical white space into macros so that
1162 they are defined only once.
1164 21. This set of changes together give more compatible Unicode case-folding
1165 behaviour for characters that have more than one other case when UCP
1166 support is available.
1168 (a) The Unicode property table now has offsets into a new table of sets of
1169 three or more characters that are case-equivalent. The MultiStage2.py
1170 script that generates these tables (the pcre_ucd.c file) now scans
1171 CaseFolding.txt instead of UnicodeData.txt for character case
1172 information.
1174 (b) The code for adding characters or ranges of characters to a character
1175 class has been abstracted into a generalized function that also handles
1176 case-independence. In UTF-mode with UCP support, this uses the new data
1177 to handle characters with more than one other case.
1179 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1180 whose other case is greater than 256 are now correctly matched
1181 caselessly. Previously, the high codepoint matched the low one, but not
1182 vice versa.
1184 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1185 of the new class addition function, using character lists defined as
1186 macros alongside the case definitions of 20 above.
1188 (e) Caseless back references now work with characters that have more than
1189 one other case.
1191 (f) General caseless matching of characters with more than one other case
1192 is supported.
1194 22. Unicode character properties were updated from Unicode 6.2.0
1196 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1198 24. Add support for 32-bit character strings, and UTF-32
1200 25. Major JIT compiler update (code refactoring and bugfixing).
1201 Experimental Sparc 32 support is added.
1203 26. Applied a modified version of Daniel Richard G's patch to create
1204 pcre.h.generic and config.h.generic by "make" instead of in the
1205 PrepareRelease script.
1207 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1208 pcre_compile.c when checking for a zero character.
1210 28. Introducing a native interface for JIT. Through this interface, the compiled
1211 machine code can be directly executed. The purpose of this interface is to
1212 provide fast pattern matching, so several sanity checks are not performed.
1213 However, feature tests are still performed. The new interface provides
1214 1.4x speedup compared to the old one.
1216 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1217 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1218 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1220 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1221 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1222 the "old" RFC 2279). Instead, it ended up passing a negative length to
1223 pcre_exec().
1225 31. Add support for GCC's visibility feature to hide internal functions.
1227 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1228 "unknown -C option" after outputting 0 or 1.
1230 33. There is now support for generating a code coverage report for the test
1231 suite in environments where gcc is the compiler and lcov is installed. This
1232 is mainly for the benefit of the developers.
1234 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1235 unaddressable using valgrind annotations, allowing valgrind to detect
1236 invalid memory accesses. This is mainly for the benefit of the developers.
1238 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1240 26. Give configure error if --enable-cpp but no C++ compiler found.
1243 Version 8.31 06-July-2012
1244 -------------------------
1246 1. Fixing a wrong JIT test case and some compiler warnings.
1248 2. Removed a bashism from the RunTest script.
1250 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1251 to unsigned type, result still unsigned" that was given by an MS compiler
1252 on encountering the code "-sizeof(xxx)".
1254 4. Partial matching support is added to the JIT compiler.
1256 5. Fixed several bugs concerned with partial matching of items that consist
1257 of more than one character:
1259 (a) /^(..)\1/ did not partially match "aba" because checking references was
1260 done on an "all or nothing" basis. This also applied to repeated
1261 references.
1263 (b) \R did not give a hard partial match if \r was found at the end of the
1264 subject.
1266 (c) \X did not give a hard partial match after matching one or more
1267 characters at the end of the subject.
1269 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1270 a partial match for the string "\r".
1272 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1273 a partial match for a CR character at the end of the subject string.
1275 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1276 running pcretest, the text "(JIT)" added to the output whenever JIT is
1277 actually used to run the match.
1279 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1280 or /S+[+] with a digit between 1 and 7.
1282 8. OP_NOT now supports any UTF character not just single-byte ones.
1284 9. (*MARK) control verb is now supported by the JIT compiler.
1286 10. The command "./RunTest list" lists the available tests without actually
1287 running any of them. (Because I keep forgetting what they all are.)
1291 12. Applied a (slightly modified) user-supplied patch that improves performance
1292 when the heap is used for recursion (compiled with --disable-stack-for-
1293 recursion). Instead of malloc and free for each heap frame each time a
1294 logical recursion happens, frames are retained on a chain and re-used where
1295 possible. This sometimes gives as much as 30% improvement.
1297 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1298 call.
1300 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1302 15. It is now possible to link pcretest with libedit as an alternative to
1303 libreadline.
1305 16. (*COMMIT) control verb is now supported by the JIT compiler.
1307 17. The Unicode data tables have been updated to Unicode 6.1.0.
1309 18. Added --file-list option to pcregrep.
1311 19. Added binary file support to pcregrep, including the -a, --binary-files,
1312 -I, and --text options.
1314 20. The madvise function is renamed for posix_madvise for QNX compatibility
1315 reasons. Fixed by Giuseppe D'Angelo.
1317 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1319 22. Changed the option for creating symbolic links for 16-bit man pages from
1320 -s to -sf so that re-installing does not cause issues.
1322 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1324 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1325 matches in certain environments (the workspace was not being correctly
1326 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1327 some of the workspace data at the beginning of a restart.
1329 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1330 was not doing so when it should - probably a typo introduced by SVN 528
1331 (change 8.10/14).
1333 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1334 \w+ when the character tables indicated that \x{c4} was a word character.
1335 There were several related cases, all because the tests for doing a table
1336 lookup were testing for characters less than 127 instead of 255.
1338 27. If a pattern contains capturing parentheses that are not used in a match,
1339 their slots in the ovector are set to -1. For those that are higher than
1340 any matched groups, this happens at the end of processing. In the case when
1341 there were back references that the ovector was too small to contain
1342 (causing temporary malloc'd memory to be used during matching), and the
1343 highest capturing number was not used, memory off the end of the ovector
1344 was incorrectly being set to -1. (It was using the size of the temporary
1345 memory instead of the true size.)
1347 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1348 ovector size, it uses memory at the end of the block that it has got.
1350 29. Check for an overlong MARK name and give an error at compile time. The
1351 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1353 30. JIT compiler update.
1355 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1356 Rill for the patch.
1358 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1360 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1362 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1363 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1364 was enabled.
1366 35. Improve JIT code generation for greedy plus quantifier.
1368 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1369 1 to "aa" instead of to an empty string. The bug affected repeated groups
1370 that could potentially match an empty string.
1372 37. Optimizing single character iterators in JIT.
1374 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1375 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1376 codepoints that are too big for the mode are faulted, and in a UTF mode,
1377 disallowed codepoints are also faulted.
1379 39. If PCRE was compiled with UTF support, in three places in the DFA
1380 matcher there was code that should only have been obeyed in UTF mode, but
1381 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1382 processing when bytes with values greater than 127 were present. In 16-bit
1383 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1384 both cases the values are those that cannot be the first data item in a UTF
1385 character. The three items that might have provoked this were recursions,
1386 possessively repeated groups, and atomic groups.
1388 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1389 and pcregrep, because some OS require shared objects to be explicitly
1390 passed to ld, causing the link step to fail if they are not.
1392 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1393 mode, patterns that started with \h* or \R* might be incorrectly matched.
1396 Version 8.30 04-February-2012
1397 -----------------------------
1399 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1400 name is defined in ctype.h.
1402 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1403 only in quite long subpatterns.
1405 3. Removed the function pcre_info(), which has been obsolete and deprecated
1406 since it was replaced by pcre_fullinfo() in February 2000.
1408 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1409 match a (*MARK), and the match failed at the start of the subject, a
1410 reference to memory before the start of the subject could occur. This bug
1411 was introduced by fix 17 of release 8.21.
1413 5. A reference to an unset group with zero minimum repetition was giving
1414 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1415 /(another)?(\1?)test/ matched against "hello world test". This bug was
1416 introduced in release 8.13.
1418 6. Add support for 16-bit character strings (a large amount of work involving
1419 many changes and refactorings).
1421 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1422 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1423 from a file.
1425 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1426 rounding is not applied in this particular case).
1428 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1429 if they appear, or are escaped, in patterns.
1431 10. Get rid of a number of -Wunused-but-set-variable warnings.
1433 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1434 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1435 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1436 also returns the mark "x". This bug applied to capturing parentheses,
1437 non-capturing parentheses, and atomic parentheses. It also applied to some
1438 assertions.
1440 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1441 information out of configure.ac instead of relying on pcre.h.generic, which
1442 is not stored in the repository.
1444 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1445 -lreadline.
1447 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1449 15. Applied Graycode's patch to put the top-level frame on the stack rather
1450 than the heap when not using the stack for recursion. This gives a
1451 performance improvement in many cases when recursion is not deep.
1453 16. Experimental code added to "pcretest -C" to output the stack frame size.
1456 Version 8.21 12-Dec-2011
1457 ------------------------
1459 1. Updating the JIT compiler.
1461 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1462 are added as well.
1464 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1465 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1466 calling _pcre_jit_exec. Some extra comments are added.
1468 4. (*MARK) settings inside atomic groups that do not contain any capturing
1469 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1470 was introduced by change 18 for 8.20.
1472 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1473 ECMA-262 standard.
1475 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1476 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1477 This bug was probably introduced by change 9 of 8.13.
1479 7. While fixing 6 above, I noticed that a number of other items were being
1480 incorrectly rejected as "not fixed length". This arose partly because newer
1481 opcodes had not been added to the fixed-length checking code. I have (a)
1482 corrected the bug and added tests for these items, and (b) arranged for an
1483 error to occur if an unknown opcode is encountered while checking for fixed
1484 length instead of just assuming "not fixed length". The items that were
1485 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1486 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1487 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1489 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1490 being incorrectly compiled and would have given unpredicatble results.
1492 9. A possessively repeated subpattern with minimum repeat count greater than
1493 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1494 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1495 the first (A) could occur when it should not.
1497 10. Add a cast and remove a redundant test from the code.
1499 11. JIT should use pcre_malloc/pcre_free for allocation.
1501 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1502 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1503 is anything other than /usr, -L is still shown).
1505 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1507 14. Perl does not support \N without a following name in a [] class; PCRE now
1508 also gives an error.
1510 15. If a forward reference was repeated with an upper limit of around 2000,
1511 it caused the error "internal error: overran compiling workspace". The
1512 maximum number of forward references (including repeats) was limited by the
1513 internal workspace, and dependent on the LINK_SIZE. The code has been
1514 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1515 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1516 of around 200,000 forward references. While doing this, I also speeded up
1517 the filling in of repeated forward references.
1519 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1520 incorrectly expecting the subject to contain another "a" after the start.
1522 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1523 in the match, the SKIP should be ignored. This was not happening; instead
1524 the SKIP was being treated as NOMATCH. For patterns such as
1525 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1526 tested.
1528 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1529 now much more compatible with Perl, in particular in cases where the result
1530 is a non-match for a non-anchored pattern. For example, if
1531 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1532 "m", where previously it did not return a name. A side effect of this
1533 change is that for partial matches, the last encountered mark name is
1534 returned, as for non matches. A number of tests that were previously not
1535 Perl-compatible have been moved into the Perl-compatible test files. The
1536 refactoring has had the pleasing side effect of removing one argument from
1537 the match() function, thus reducing its stack requirements.
1539 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1540 subsequent uses of /S (without +) incorrectly behaved like /S+.
1542 21. Retrieve executable code size support for the JIT compiler and fixing
1543 some warnings.
1545 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1546 not work when the shorter character appeared right at the end of the
1547 subject string.
1549 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1550 systems.
1552 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1553 output it when the /M option is used in pcretest.
1555 25. The CheckMan script was not being included in the distribution. Also, added
1556 an explicit "perl" to run Perl scripts from the PrepareRelease script
1557 because this is reportedly needed in Windows.
1559 26. If study data was being save in a file and studying had not found a set of
1560 "starts with" bytes for the pattern, the data written to the file (though
1561 never used) was taken from uninitialized memory and so caused valgrind to
1562 complain.
1564 27. Updated RunTest.bat as provided by Sheri Pierce.
1566 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1568 29. Computation of memory usage for the table of capturing group names was
1569 giving an unnecessarily large value.
1572 Version 8.20 21-Oct-2011
1573 ------------------------
1575 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1576 a POSIX class. After further experiments with Perl, which convinced me that
1577 Perl has bugs and confusions, a closing square bracket is no longer allowed
1578 in a POSIX name. This bug also affected patterns with classes that started
1579 with full stops.
1581 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1582 captured substring, but while checking the failing first alternative,
1583 substring 1 is temporarily captured. If the output vector supplied to
1584 pcre_exec() was not big enough for this capture, the yield of the function
1585 was still zero ("insufficient space for captured substrings"). This cannot
1586 be totally fixed without adding another stack variable, which seems a lot
1587 of expense for a edge case. However, I have improved the situation in cases
1588 such as /(a)(b)x|abc/ matched against "abc", where the return code
1589 indicates that fewer than the maximum number of slots in the ovector have
1590 been set.
1592 3. Related to (2) above: when there are more back references in a pattern than
1593 slots in the output vector, pcre_exec() uses temporary memory during
1594 matching, and copies in the captures as far as possible afterwards. It was
1595 using the entire output vector, but this conflicts with the specification
1596 that only 2/3 is used for passing back captured substrings. Now it uses
1597 only the first 2/3, for compatibility. This is, of course, another edge
1598 case.
1600 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1601 main code base, and can be used by building with --enable-jit. When this is
1602 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1603 runtime --no-jit option is given.
1605 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1606 ovector, the return from the function was zero, implying that there were
1607 other matches that did not fit. The correct "exactly full" value is now
1608 returned.
1610 6. If a subpattern that was called recursively or as a subroutine contained
1611 (*PRUNE) or any other control that caused it to give a non-standard return,
1612 invalid errors such as "Error -26 (nested recursion at the same subject
1613 position)" or even infinite loops could occur.
1615 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1616 computing the minimum length on reaching *ACCEPT, and so ended up with the
1617 wrong value of 1 rather than 0. Further investigation indicates that
1618 computing a minimum subject length in the presence of *ACCEPT is difficult
1619 (think back references, subroutine calls), and so I have changed the code
1620 so that no minimum is registered for a pattern that contains *ACCEPT.
1622 8. If (*THEN) was present in the first (true) branch of a conditional group,
1623 it was not handled as intended. [But see 16 below.]
1625 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1626 Sheri Pierce.
1628 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1629 the first byte in a match must be "a".
1631 11. Change 17 for 8.13 increased the recursion depth for patterns like
1632 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1633 pattern contains any instances of (*THEN). If it does not, the old
1634 optimizations are restored. It would be nice to do this on a per-group
1635 basis, but at the moment that is not feasible.
1637 12. In some environments, the output of pcretest -C is CRLF terminated. This
1638 broke RunTest's code that checks for the link size. A single white space
1639 character after the value is now allowed for.
1641 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1642 For "fr", it uses the Windows-specific input and output files.
1644 14. If (*THEN) appeared in a group that was called recursively or as a
1645 subroutine, it did not work as intended. [But see next item.]
1647 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1648 pattern fragments (but not containing any | characters). If A and B are
1649 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1650 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1651 D. In other words, Perl considers parentheses that do not contain any |
1652 characters to be part of a surrounding alternative, whereas PCRE was
1653 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1654 differently. PCRE now behaves in the same way as Perl, except in the case
1655 of subroutine/recursion calls such as (?1) which have in any case always
1656 been different (but PCRE had them first :-).
1658 16. Related to 15 above: Perl does not treat the | in a conditional group as
1659 creating alternatives. Such a group is treated in the same way as an
1660 ordinary group without any | characters when processing (*THEN). PCRE has
1661 been changed to match Perl's behaviour.
1663 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1664 RunGrepTest script failed.
1666 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1667 inevitable for groups that contain captures, but it can lead to a lot of
1668 stack use in large patterns. The old behaviour has been restored for atomic
1669 groups that do not contain any capturing parentheses.
1671 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1672 suppress the check for a minimum subject length at run time. (If it was
1673 given to pcre_exec() or pcre_dfa_exec() it did work.)
1675 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1676 fail to work when decoding hex characters in data strings in EBCDIC
1677 environments.
1679 21. It appears that in at least one Mac OS environment, the isxdigit() function
1680 is implemented as a macro that evaluates to its argument more than once,
1681 contravening the C 90 Standard (I haven't checked a later standard). There
1682 was an instance in pcretest which caused it to go wrong when processing
1683 \x{...} escapes in subject strings. The has been rewritten to avoid using
1684 things like p++ in the argument of isxdigit().
1687 Version 8.13 16-Aug-2011
1688 ------------------------
1690 1. The Unicode data tables have been updated to Unicode 6.0.0.
1692 2. Two minor typos in pcre_internal.h have been fixed.
1694 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1695 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1696 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1698 4. There were a number of related bugs in the code for matching backrefences
1699 caselessly in UTF-8 mode when codes for the characters concerned were
1700 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1701 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1702 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1703 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1704 2-byte code at the end of the subject (it thought there wasn't enough data
1705 left).
1707 5. Comprehensive information about what went wrong is now returned by
1708 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1709 as the output vector has at least 2 elements. The offset of the start of
1710 the failing character and a reason code are placed in the vector.
1712 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1713 now returned is for the first byte of the failing character, instead of the
1714 last byte inspected. This is an incompatible change, but I hope it is small
1715 enough not to be a problem. It makes the returned offset consistent with
1716 pcre_exec() and pcre_dfa_exec().
1718 7. pcretest now gives a text phrase as well as the error number when
1719 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1720 failure, the offset and reason code are output.
1722 8. When \R was used with a maximizing quantifier it failed to skip backwards
1723 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1724 back over a single character (\n). This seems wrong (because it treated the
1725 two characters as a single entity when going forwards), conflicts with the
1726 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1727 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1728 has been changed.
1730 9. Some internal refactoring has changed the processing so that the handling
1731 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1732 time (the PCRE_DOTALL option was changed this way some time ago: version
1733 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1734 which was always a bit of a fudge. It also means that there is one less
1735 argument for the match() function, which reduces its stack requirements
1736 slightly. This change also fixes an incompatibility with Perl: the pattern
1737 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1739 10. More internal refactoring has drastically reduced the number of recursive
1740 calls to match() for possessively repeated groups such as (abc)++ when
1741 using pcre_exec().
1743 11. While implementing 10, a number of bugs in the handling of groups were
1744 discovered and fixed:
1746 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1747 (a|)*(?1) gave a compile-time internal error.
1748 ((a|)+)+ did not notice that the outer group could match an empty string.
1749 (^a|^)+ was not marked as anchored.
1750 (.*a|.*)+ was not marked as matching at start or after a newline.
1752 12. Yet more internal refactoring has removed another argument from the match()
1753 function. Special calls to this function are now indicated by setting a
1754 value in a variable in the "match data" data block.
1756 13. Be more explicit in pcre_study() instead of relying on "default" for
1757 opcodes that mean there is no starting character; this means that when new
1758 ones are added and accidentally left out of pcre_study(), testing should
1759 pick them up.
1761 14. The -s option of pcretest has been documented for ages as being an old
1762 synonym of -m (show memory usage). I have changed it to mean "force study
1763 for every regex", that is, assume /S for every regex. This is similar to -i
1764 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1765 using it. It makes it easier to run collections of tests with and without
1766 study enabled, and thereby test pcre_study() more easily. All the standard
1767 tests are now run with and without -s (but some patterns can be marked as
1768 "never study" - see 20 below).
1770 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1771 restoration of the capturing data to the outer values was not happening
1772 correctly.
1774 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1775 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1776 pattern had matched an empty string, and so incorrectly returned a no
1777 match.
1779 17. There was optimizing code for the last branch of non-capturing parentheses,
1780 and also for the obeyed branch of a conditional subexpression, which used
1781 tail recursion to cut down on stack usage. Unfortunately, now that there is
1782 the possibility of (*THEN) occurring in these branches, tail recursion is
1783 no longer possible because the return has to be checked for (*THEN). These
1784 two optimizations have therefore been removed. [But see 8.20/11 above.]
1786 18. If a pattern containing \R was studied, it was assumed that \R always
1787 matched two bytes, thus causing the minimum subject length to be
1788 incorrectly computed because \R can also match just one byte.
1790 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1791 was incorrectly computed.
1793 20. If /S is present twice on a test pattern in pcretest input, it now
1794 *disables* studying, thereby overriding the use of -s on the command line
1795 (see 14 above). This is necessary for one or two tests to keep the output
1796 identical in both cases.
1798 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1799 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1801 22. When an atomic group that contained a capturing parenthesis was
1802 successfully matched, but the branch in which it appeared failed, the
1803 capturing was not being forgotten if a higher numbered group was later
1804 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1805 group 1 to "a", when in fact it should be unset. This applied to multi-
1806 branched capturing and non-capturing groups, repeated or not, and also to
1807 positive assertions (capturing in negative assertions does not happen
1808 in PCRE) and also to nested atomic groups.
1810 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1811 subject after a captured substring, to make it easier to tell which of a
1812 number of identical substrings has been captured.
1814 24. The way atomic groups are processed by pcre_exec() has been changed so that
1815 if they are repeated, backtracking one repetition now resets captured
1816 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1817 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1818 "aaa". Previously, it would have been "a". As part of this code
1819 refactoring, the way recursive calls are handled has also been changed.
1821 25. If an assertion condition captured any substrings, they were not passed
1822 back unless some other capturing happened later. For example, if
1823 (?(?=(a))a) was matched against "a", no capturing was returned.
1825 26. When studying a pattern that contained subroutine calls or assertions,
1826 the code for finding the minimum length of a possible match was handling
1827 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1828 group 1 called group 2 while simultaneously a separate group 2 called group
1829 1). A stack overflow occurred in this case. I have fixed this by limiting
1830 the recursion depth to 10.
1832 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1833 Fortmann. This supports explicit test numbers on the command line, and has
1834 argument validation and error reporting.
1836 28. An instance of \X with an unlimited repeat could fail if at any point the
1837 first character it looked at was a mark character.
1839 29. Some minor code refactoring concerning Unicode properties and scripts
1840 should reduce the stack requirement of match() slightly.
1842 30. Added the '=' option to pcretest to check the setting of unused capturing
1843 slots at the end of the pattern, which are documented as being -1, but are
1844 not included in the return count.
1846 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1847 compiled something random. Now it gives a compile-time error (as does
1848 Perl).
1850 32. A *MARK encountered during the processing of a positive assertion is now
1851 recorded and passed back (compatible with Perl).
1853 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1854 had alternative anchored branches, the search for a second match in a line
1855 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1856 matched the line "0102" twice. The same bug affected patterns that started
1857 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1858 twice.
1860 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1861 does, and because of capturing effects, quantifying parenthesized
1862 assertions may at times be useful. Quantifiers are now allowed for
1863 parenthesized assertions.
1865 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1867 36. \g was being checked for fancy things in a character class, when it should
1868 just be a literal "g".
1870 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1871 appearance of a nested POSIX class supersedes an apparent external class.
1872 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1873 unescaped square brackets may also appear as part of class names. For
1874 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1875 more like Perl. (But see 8.20/1 above.)
1877 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1878 was because it thought it was \N{name}, which is not supported).
1880 39. Add minix to OS list not supporting the -S option in pcretest.
1882 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1883 cannot analyze patterns in sufficient detail to catch mutual recursions
1884 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1885 subgroup is called recursively as a subpattern for a second time at the
1886 same position in the subject string. In previous releases this might have
1887 been caught by the recursion limit, or it might have run out of stack.
1889 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1890 happen only once. PCRE was, however incorrectly giving a compile time error
1891 "recursive call could loop indefinitely" because it cannot analyze the
1892 pattern in sufficient detail. The compile time test no longer happens when
1893 PCRE is compiling a conditional subpattern, but actual runaway loops are
1894 now caught at runtime (see 40 above).
1896 42. It seems that Perl allows any characters other than a closing parenthesis
1897 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1898 has been changed to be the same.
1900 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1901 as not to get warnings when autogen.sh is called. Also changed
1902 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1904 44. To help people who use pcregrep to scan files containing exceedingly long
1905 lines, the following changes have been made:
1907 (a) The default value of the buffer size parameter has been increased from
1908 8K to 20K. (The actual buffer used is three times this size.)
1910 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1911 PCRE is built.
1913 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1914 to be set at run time.
1916 (d) Numerical values in pcregrep options can be followed by K or M, for
1917 example --buffer-size=50K.
1919 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1920 given and the return code is set to 2.
1922 45. Add a pointer to the latest mark to the callout data block.
1924 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1925 partial match of an empty string instead of no match. This was specific to
1926 the use of ".".
1928 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1929 complete match instead of a partial match. This bug was dependent on both
1930 the PCRE_UTF8 and PCRE_DOTALL options being set.
1932 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1933 starting byte set, because \b was not being ignored.
1936 Version 8.12 15-Jan-2011
1937 ------------------------
1939 1. Fixed some typos in the markup of the man pages, and wrote a script that
1940 checks for such things as part of the documentation building process.
1942 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1943 --match-limit and --recursion-limit options (added for 8.11). In
1944 particular, this made one of the standard tests fail. (The integer value
1945 went into the wrong half of a long int.)
1947 3. If the --colour option was given to pcregrep with -v (invert match), it
1948 did strange things, either producing crazy output, or crashing. It should,
1949 of course, ignore a request for colour when reporting lines that do not
1950 match.
1952 4. Another pcregrep bug caused similar problems if --colour was specified with
1953 -M (multiline) and the pattern match finished with a line ending.
1955 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1956 matched in multiline mode, the following line was shown as part of the
1957 match. This seems wrong, so I have changed it.
1959 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1960 the check for further matches in the same line (so they could be coloured)
1961 to overrun the end of the current line. If another match was found, it was
1962 incorrectly shown (and then shown again when found in the next line).
1964 7. If pcregrep was compiled under Windows, there was a reference to the
1965 function pcregrep_exit() before it was defined. I am assuming this was
1966 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1967 reported by a user. I've moved the definition above the reference.
1970 Version 8.11 10-Dec-2010
1971 ------------------------
1973 1. (*THEN) was not working properly if there were untried alternatives prior
1974 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1975 backtracked to try for "b" instead of moving to the next alternative branch
1976 at the same level (in this case, to look for "c"). The Perl documentation
1977 is clear that when (*THEN) is backtracked onto, it goes to the "next
1978 alternative in the innermost enclosing group".
1980 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1981 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1982 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1983 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1984 (*THEN).
1986 3. If \s appeared in a character class, it removed the VT character from
1987 the class, even if it had been included by some previous item, for example
1988 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1989 of \s, but is part of the POSIX "space" class.)
1991 4. A partial match never returns an empty string (because you can always
1992 match an empty string at the end of the subject); however the checking for
1993 an empty string was starting at the "start of match" point. This has been
1994 changed to the "earliest inspected character" point, because the returned
1995 data for a partial match starts at this character. This means that, for
1996 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1997 (previously it gave "no match").
1999 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2000 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2001 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2002 has an implication that the given string is incomplete (because a partial
2003 match is preferred over a full match). For this reason, these items now
2004 give a partial match in this situation. [Aside: previously, the one case
2005 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2006 match rather than a full match, which was wrong by the old rules, but is
2007 now correct.]
2009 6. There was a bug in the handling of #-introduced comments, recognized when
2010 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2011 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2012 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2013 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2014 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2015 places in pcre_compile().
2017 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2018 comments when looking ahead for named forward references to subpatterns,
2019 the only newline sequence it recognized was NL. It now handles newlines
2020 according to the set newline convention.
2022 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2023 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2024 cater for a lack of strerror(). These oversights have been fixed.
2026 9. Added --match-limit and --recursion-limit to pcregrep.
2028 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2030 11. When the -o option was used, pcregrep was setting a return code of 1, even
2031 when matches were found, and --line-buffered was not being honoured.
2033 12. Added an optional parentheses number to the -o and --only-matching options
2034 of pcregrep.
2036 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2037 can match an empty string. The code to do it in pcretest and pcredemo
2038 needed fixing:
2040 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2041 only one byte after an empty string match just before CRLF (this case
2042 just got forgotten; "any" and "anycrlf" were OK).
2044 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2045 mode when an empty string match preceded an ASCII character followed by
2046 a non-ASCII character. (The code for advancing by one character rather
2047 than one byte was nonsense.)
2049 (c) The pcredemo.c sample program did not have any code at all to handle
2050 the cases when CRLF is a valid newline sequence.
2052 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2053 as a starting offset was within the subject string. There is now a new
2054 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2055 negative or greater than the length of the string. In order to test this,
2056 pcretest is extended to allow the setting of negative starting offsets.
2058 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2059 starting offset points to the beginning of a UTF-8 character was
2060 unnecessarily clumsy. I tidied it up.
2062 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2063 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2065 17. Nobody had reported that the --include_dir option, which was added in
2066 release 7.7 should have been called --include-dir (hyphen, not underscore)
2067 for compatibility with GNU grep. I have changed it to --include-dir, but
2068 left --include_dir as an undocumented synonym, and the same for
2069 --exclude-dir, though that is not available in GNU grep, at least as of
2070 release 2.5.4.
2072 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2073 characters from a string of bytes) have been redefined so as not to use
2074 loops, in order to improve performance in some environments. At the same
2075 time, I abstracted some of the common code into auxiliary macros to save
2076 repetition (this should not affect the compiled code).
2078 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2079 compile-time error is now given if \c is not followed by an ASCII
2080 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2081 different, and any byte value is allowed.)
2083 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2084 START_OPTIMIZE option, which is now allowed at compile time - but just
2085 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2086 to pcregrep and other applications that have no direct access to PCRE
2087 options. The new /Y option in pcretest sets this option when calling
2088 pcre_compile().
2090 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2091 back references. Groups containing recursive back references were forced to
2092 be atomic by that change, but in the case of named groups, the amount of
2093 memory required was incorrectly computed, leading to "Failed: internal
2094 error: code overflow". This has been fixed.
2096 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2097 pcretest.c, to avoid build problems in some Borland environments.
2100 Version 8.10 25-Jun-2010
2101 ------------------------
2103 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2104 THEN.
2106 2. (*ACCEPT) was not working when inside an atomic group.
2108 3. Inside a character class, \B is treated as a literal by default, but
2109 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2110 causes the error). The code is unchanged, but I tidied the documentation.
2112 4. Inside a character class, PCRE always treated \R and \X as literals,
2113 whereas Perl faults them if its -w option is set. I have changed PCRE so
2114 that it faults them when PCRE_EXTRA is set.
2116 5. Added support for \N, which always matches any character other than
2117 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2119 6. When compiling pcregrep with newer versions of gcc which may have
2120 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2121 declared with attribute warn_unused_result" were given. Just casting the
2122 result to (void) does not stop the warnings; a more elaborate fudge is
2123 needed. I've used a macro to implement this.
2125 7. Minor change to pcretest.c to avoid a compiler warning.
2127 8. Added four artifical Unicode properties to help with an option to make
2128 \s etc use properties (see next item). The new properties are: Xan
2129 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2131 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2132 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2133 this option. Modified pcretest to add /W to test this facility. Added
2134 REG_UCP to make it available via the POSIX interface.
2136 10. Added --line-buffered to pcregrep.
2138 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2139 studied, and the match started with a letter with a code point greater than
2140 127 whose first byte was different to the first byte of the other case of
2141 the letter, the other case of this starting letter was not recognized
2142 (#976).
2144 12. If a pattern that was studied started with a repeated Unicode property
2145 test, for example, \p{Nd}+, there was the theoretical possibility of
2146 setting up an incorrect bitmap of starting bytes, but fortunately it could
2147 not have actually happened in practice until change 8 above was made (it
2148 added property types that matched character-matching opcodes).
2150 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2151 possible starting bytes for non-anchored patterns.
2153 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2154 \R, and also a number of cases that involve Unicode properties, both
2155 explicit and implicit when PCRE_UCP is set.
2157 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2158 input, it could crash or give wrong results if characters with values
2159 greater than 0xc0 were present in the subject string. (Detail: it assumed
2160 UTF-8 input when processing these items.)
2162 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2163 size_t is 64-bit (#991).
2165 17. Added a check for running out of memory when PCRE is compiled with
2166 --disable-stack-for-recursion (#990).
2168 18. If the last data line in a file for pcretest does not have a newline on
2169 the end, a newline was missing in the output.
2171 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2172 less than 128) in its various bitmaps. However, there is a facility for
2173 generating tables according to the current locale when PCRE is compiled. It
2174 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2175 characters, are recognized by isspace() and therefore were getting set in
2176 these tables, and indeed these tables seem to approximate to ISO 8859. This
2177 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2178 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2179 which of course cannot start UTF-8 characters. I have changed the code so
2180 that only real ASCII characters (less than 128) and the correct starting
2181 bytes for UTF-8 encodings are set for characters greater than 127 when in
2182 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2183 altogether.)
2185 20. Added the /T option to pcretest so as to be able to run tests with non-
2186 standard character tables, thus making it possible to include the tests
2187 used for 19 above in the standard set of tests.
2189 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2190 reference to a subpattern the other side of a comment that contains an
2191 opening parenthesis caused either an internal compiling error, or a
2192 reference to the wrong subpattern.
2195 Version 8.02 19-Mar-2010
2196 ------------------------
2198 1. The Unicode data tables have been updated to Unicode 5.2.0.
2200 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2201 configured.
2203 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2204 original author of that file, following a query about its status.
2206 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2207 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2209 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2210 quantifier applied to a forward-referencing subroutine call, could compile
2211 incorrect code or give the error "internal error: previously-checked
2212 referenced subpattern not found".
2214 6. Both MS Visual Studio and Symbian OS have problems with initializing
2215 variables to point to external functions. For these systems, therefore,
2216 pcre_malloc etc. are now initialized to local functions that call the
2217 relevant global functions.
2219 7. There were two entries missing in the vectors called coptable and poptable
2220 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2221 I've fixed the data, and added a kludgy way of testing at compile time that
2222 the lengths are correct (equal to the number of opcodes).
2224 8. Following on from 7, I added a similar kludge to check the length of the
2225 eint vector in pcreposix.c.
2227 9. Error texts for pcre_compile() are held as one long string to avoid too
2228 much relocation at load time. To find a text, the string is searched,
2229 counting zeros. There was no check for running off the end of the string,
2230 which could happen if a new error number was added without updating the
2231 string.
2233 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2235 11. \K was not working if it appeared in an atomic group or in a group that
2236 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2237 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2238 the assertion is positive, but not if it is negative.
2240 12. Change 11 fortuitously reduced the size of the stack frame used in the
2241 "match()" function of pcre_exec.c by one pointer. Forthcoming
2242 implementation of support for (*MARK) will need an extra pointer on the
2243 stack; I have reserved it now, so that the stack frame size does not
2244 decrease.
2246 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2247 item in branch that calls a recursion is a subroutine call - as in the
2248 second branch in the above example - was incorrectly given the compile-
2249 time error "recursive call could loop indefinitely" because pcre_compile()
2250 was not correctly checking the subroutine for matching a non-empty string.
2252 14. The checks for overrunning compiling workspace could trigger after an
2253 overrun had occurred. This is a "should never occur" error, but it can be
2254 triggered by pathological patterns such as hundreds of nested parentheses.
2255 The checks now trigger 100 bytes before the end of the workspace.
2257 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2260 Version 8.01 19-Jan-2010
2261 ------------------------
2263 1. If a pattern contained a conditional subpattern with only one branch (in
2264 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2265 computed the wrong minimum data length (which is of course zero for such
2266 subpatterns). This could cause incorrect "no match" results.
2268 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2269 the pattern is reset in the first branch, pcre_compile() failed with
2270 "internal error: code overflow at offset...". This happened only when
2271 the reset was to the original external option setting. (An optimization
2272 abstracts leading options settings into an external setting, which was the
2273 cause of this.)
2275 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2276 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2277 assertion pattern did not match (meaning that the assertion was true), it
2278 was incorrectly treated as false if the SKIP had been reached during the
2279 matching. This also applied to assertions used as conditions.
2281 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2282 assertion subpattern, including such a pattern used as a condition,
2283 unpredictable results occurred, instead of the error return
2286 5. The C++ GlobalReplace function was not working like Perl for the special
2287 situation when an empty string is matched. It now does the fancy magic
2288 stuff that is necessary.
2290 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2291 removed. (These were left over from very, very early versions of PCRE.)
2293 7. Some cosmetic changes to the code to make life easier when compiling it
2294 as part of something else:
2296 (a) Change DEBUG to PCRE_DEBUG.
2298 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2299 called "current" as "current_branch", to prevent a collision with the
2300 Linux macro when compiled as a kernel module.
2302 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2303 prevent a collision with the Linux macro when compiled as a kernel
2304 module.
2306 8. In pcre_compile() there are some checks for integer overflows that used to
2307 cast potentially large values to (double). This has been changed to that
2308 when building, a check for int64_t is made, and if it is found, it is used
2309 instead, thus avoiding the use of floating point arithmetic. (There is no
2310 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2311 double.
2313 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2314 2005 (difference between two addresses compared to an unsigned value).
2316 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2317 custom one, because of the following reported problem in Windows:
2319 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2320 under Win32.
2321 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2322 therefore missing the function definition.
2323 - The compiler thus generates a "C" signature for the test function.
2324 - The linker fails to find the "C" function.
2325 - PCRE fails to configure if asked to do so against libbz2.
2327 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2328 messages were output:
2330 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2331 rerunning libtoolize, to keep the correct libtool macros in-tree.
2332 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2334 I have done both of these things.
2336 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2337 most of the time, it *can* run out if it is given a pattern that contains a
2338 runaway infinite recursion. I updated the discussion in the pcrestack man
2339 page.
2341 13. Now that we have gone to the x.xx style of version numbers, the minor
2342 version may start with zero. Using 08 or 09 is a bad idea because users
2343 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2344 interpreted as invalid octal numbers. I've updated the previous comment in
2345 configure.ac, and also added a check that gives an error if 08 or 09 are
2346 used.
2348 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2349 causing partial matching to fail when the end of the subject matched \W
2350 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2352 15. There were some discrepancies between the declarations in pcre_internal.h
2353 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2354 their definitions. The declarations used "const uschar *" and the
2355 definitions used USPTR. Even though USPTR is normally defined as "const
2356 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2357 reported that: "This difference in casting confuses some C++ compilers, for
2358 example, SunCC recognizes above declarations as different functions and
2359 generates broken code for hbpcre." I have changed the declarations to use
2360 USPTR.
2362 16. GNU libtool is named differently on some systems. The autogen.sh script now
2363 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2364 (FreeBSD).
2366 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2367 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2368 comment: "Figure out how to create a longlong from a string: strtoll and
2369 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2370 instance, but it only takes 2 args instead of 3!"
2372 18. A subtle bug concerned with back references has been fixed by a change of
2373 specification, with a corresponding code fix. A pattern such as
2374 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2375 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2376 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2377 same bug. Such groups have to be quantified to be useful, or contained
2378 inside another quantified group. (If there's no repetition, the reference
2379 can never match.) The problem arises because, having left the group and
2380 moved on to the rest of the pattern, a later failure that backtracks into
2381 the group uses the captured value from the final iteration of the group
2382 rather than the correct earlier one. I have fixed this in PCRE by forcing
2383 any group that contains a reference to itself to be an atomic group; that
2384 is, there cannot be any backtracking into it once it has completed. This is
2385 similar to recursive and subroutine calls.
2388 Version 8.00 19-Oct-09
2389 ----------------------
2391 1. The table for translating pcre_compile() error codes into POSIX error codes
2392 was out-of-date, and there was no check on the pcre_compile() error code
2393 being within the table. This could lead to an OK return being given in
2394 error.
2396 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2397 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2398 in a Windows environment.
2400 3. The pcregrep --count option prints the count for each file even when it is
2401 zero, as does GNU grep. However, pcregrep was also printing all files when
2402 --files-with-matches was added. Now, when both options are given, it prints
2403 counts only for those files that have at least one match. (GNU grep just
2404 prints the file name in this circumstance, but including the count seems
2405 more useful - otherwise, why use --count?) Also ensured that the
2406 combination -clh just lists non-zero counts, with no names.
2408 4. The long form of the pcregrep -F option was incorrectly implemented as
2409 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2410 but it seems right to fix it, and I didn't think it was worth preserving
2411 the old behaviour.
2413 5. The command line items --regex=pattern and --regexp=pattern were not
2414 recognized by pcregrep, which required --regex pattern or --regexp pattern
2415 (with a space rather than an '='). The man page documented the '=' forms,
2416 which are compatible with GNU grep; these now work.
2418 6. No libpcreposix.pc file was created for pkg-config; there was just
2419 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2421 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2422 when UCP support is not needed, by modifying the Python script that
2423 generates it from Unicode data files. This should not matter if the module
2424 is correctly used as a library, but I received one complaint about 50K of
2425 unwanted data. My guess is that the person linked everything into his
2426 program rather than using a library. Anyway, it does no harm.
2428 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2429 was a minimum greater than 1 for a wide character in a possessive
2430 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2431 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2432 character. Chaos in the form of incorrect output or a compiling loop could
2433 result.
2435 9. The restrictions on what a pattern can contain when partial matching is
2436 requested for pcre_exec() have been removed. All patterns can now be
2437 partially matched by this function. In addition, if there are at least two
2438 slots in the offset vector, the offset of the earliest inspected character
2439 for the match and the offset of the end of the subject are set in them when
2440 PCRE_ERROR_PARTIAL is returned.
2442 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2443 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2444 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2445 and may be more useful for multi-segment matching.
2447 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2448 used to be given if ever the end of the subject was reached; now it is
2449 given only if matching could not proceed because another character was
2450 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2451 string "Z", which now yields "no match" instead of "partial match". In the
2452 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2453 final character ended with (*FAIL).
2455 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2456 if the pattern had a "must contain" character that was already found in the
2457 earlier partial match, unless partial matching was again requested. For
2458 example, with the pattern /dog.(body)?/, the "must contain" character is
2459 "g". If the first part-match was for the string "dog", restarting with
2460 "sbody" failed. This bug has been fixed.
2462 13. The string returned by pcre_dfa_exec() after a partial match has been
2463 changed so that it starts at the first inspected character rather than the
2464 first character of the match. This makes a difference only if the pattern
2465 starts with a lookbehind assertion or \b or \B (\K is not supported by
2466 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2467 matching functions compatible, and I think it's the right thing to do.
2469 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2470 so that the demonstration program is easily available in environments where
2471 PCRE has not been installed from source.
2473 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2474 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2475 library.
2477 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2478 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2479 is not the first non-POSIX option to be added. Clearly some people find
2480 these options useful.
2482 17. If a caller to the POSIX matching function regexec() passes a non-zero
2483 value for nmatch with a NULL value for pmatch, the value of
2484 nmatch is forced to zero.
2486 18. RunGrepTest did not have a test for the availability of the -u option of
2487 the diff command, as RunTest does. It now checks in the same way as
2488 RunTest, and also checks for the -b option.
2490 19. If an odd number of negated classes containing just a single character
2491 interposed, within parentheses, between a forward reference to a named
2492 subpattern and the definition of the subpattern, compilation crashed with
2493 an internal error, complaining that it could not find the referenced
2494 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2495 [The bug was that it was starting one character too far in when skipping
2496 over the character class, thus treating the ] as data rather than
2497 terminating the class. This meant it could skip too much.]
2499 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2500 /g option in pcretest when the pattern contains \K, which makes it possible
2501 to have an empty string match not at the start, even when the pattern is
2502 anchored. Updated pcretest and pcredemo to use this option.
2504 21. If the maximum number of capturing subpatterns in a recursion was greater
2505 than the maximum at the outer level, the higher number was returned, but
2506 with unset values at the outer level. The correct (outer level) value is
2507 now given.
2509 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2510 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2511 make it do so. The string so far is captured, making this feature
2512 compatible with Perl.
2514 23. The tests have been re-organized, adding tests 11 and 12, to make it
2515 possible to check the Perl 5.10 features against Perl 5.10.
2517 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2518 pattern matches a fixed length string. PCRE did not allow this; now it
2519 does. Neither allows recursion.
2521 25. I finally figured out how to implement a request to provide the minimum
2522 length of subject string that was needed in order to match a given pattern.
2523 (It was back references and recursion that I had previously got hung up
2524 on.) This code has now been added to pcre_study(); it finds a lower bound
2525 to the length of subject needed. It is not necessarily the greatest lower
2526 bound, but using it to avoid searching strings that are too short does give
2527 some useful speed-ups. The value is available to calling programs via
2528 pcre_fullinfo().
2530 26. While implementing 25, I discovered to my embarrassment that pcretest had
2531 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2532 study optimizations had never been tested with that matching function.
2533 Oops. What is worse, even when it was passed study data, there was a bug in
2534 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2535 were also very few tests of studied patterns with pcre_dfa_exec().
2537 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2538 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2539 on the other side of the coin, they are no longer allowed to have different
2540 names, because these cannot be distinguished in PCRE, and this has caused
2541 confusion. (This is a difference from Perl.)
2543 28. When duplicate subpattern names are present (necessarily with different
2544 numbers, as required by 27 above), and a test is made by name in a
2545 conditional pattern, either for a subpattern having been matched, or for
2546 recursion in such a pattern, all the associated numbered subpatterns are
2547 tested, and the overall condition is true if the condition is true for any
2548 one of them. This is the way Perl works, and is also more like the way
2549 testing by number works.
2552 Version 7.9 11-Apr-09
2553 ---------------------
2555 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2556 (pcretest), all targets were linked against these libraries. This included
2557 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2558 libraries. This caused unwanted dependencies to be created. This problem
2559 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2560 pcretest is linked with readline.
2562 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2563 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2564 moved outside it again, because FALSE and TRUE are already defined in AIX,
2565 but BOOL is not.
2567 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2568 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2570 4. The pcregrep documentation said spaces were inserted as well as colons (or
2571 hyphens) following file names and line numbers when outputting matching
2572 lines. This is not true; no spaces are inserted. I have also clarified the
2573 wording for the --colour (or --color) option.
2575 5. In pcregrep, when --colour was used with -o, the list of matching strings
2576 was not coloured; this is different to GNU grep, so I have changed it to be
2577 the same.
2579 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2580 each matching line was coloured. Now it goes on to look for further matches
2581 of any of the test patterns, which is the same behaviour as GNU grep.
2583 7. A pattern that could match an empty string could cause pcregrep to loop; it
2584 doesn't make sense to accept an empty string match in pcregrep, so I have
2585 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2586 seems to be how GNU grep behaves. [But see later change 40 for release
2587 8.33.]
2589 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2590 start or after a newline", because the conditional assertion was not being
2591 correctly handled. The rule now is that both the assertion and what follows
2592 in the first alternative must satisfy the test.
2594 9. If auto-callout was enabled in a pattern with a conditional group whose
2595 condition was an assertion, PCRE could crash during matching, both with
2596 pcre_exec() and pcre_dfa_exec().
2598 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2599 used for matching.
2601 11. Unicode property support in character classes was not working for
2602 characters (bytes) greater than 127 when not in UTF-8 mode.
2604 12. Added the -M command line option to pcretest.
2606 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2608 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2610 16. Added comments and documentation about mis-use of no_arg in the C++
2611 wrapper.
2613 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2614 from Martin Jerabek that uses macro names for all relevant character and
2615 string constants.
2617 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2618 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2619 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2620 these, but not everybody uses configure.
2622 19. A conditional group that had only one branch was not being correctly
2623 recognized as an item that could match an empty string. This meant that an
2624 enclosing group might also not be so recognized, causing infinite looping
2625 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2626 with the subject "ab", where knowledge that the repeated group can match
2627 nothing is needed in order to break the loop.
2629 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2630 exec(), but without supplying a callout function, matching went wrong.
2632 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2633 leak if the size of the offset vector was greater than 30. When the vector
2634 is smaller, the saved offsets during recursion go onto a local stack
2635 vector, but for larger vectors malloc() is used. It was failing to free
2636 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2637 error, in fact).
2639 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2640 heapframe that is used only when UTF-8 support is enabled. This caused no
2641 problem, but was untidy.
2643 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2644 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2645 included within another project.
2647 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2648 slightly modified by me:
2650 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2651 not building pcregrep.
2653 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2654 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2656 25. Forward references, both numeric and by name, in patterns that made use of
2657 duplicate group numbers, could behave incorrectly or give incorrect errors,
2658 because when scanning forward to find the reference group, PCRE was not
2659 taking into account the duplicate group numbers. A pattern such as
2660 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2662 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2663 the feature of a custom pointer more persuasive (as requested by a user).
2665 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2666 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2667 pre-defined.
2669 28. Added support for (*UTF8) at the start of a pattern.
2671 29. Arrange for flags added by the "release type" setting in CMake to be shown
2672 in the configuration summary.
2675 Version 7.8 05-Sep-08
2676 ---------------------
2678 1. Replaced UCP searching code with optimized version as implemented for Ad
2679 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2680 stage table and inline lookup instead of a function, giving speed ups of 2
2681 to 5 times on some simple patterns that I tested. Permission was given to
2682 distribute the MultiStage2.py script that generates the tables (it's not in
2683 the tarball, but is in the Subversion repository).
2685 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2686 scripts.
2688 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2689 a group with a zero qualifier. The result of the study could be incorrect,
2690 or the function might crash, depending on the pattern.
2692 4. Caseless matching was not working for non-ASCII characters in back
2693 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2694 It now works when Unicode Property Support is available.
2696 5. In pcretest, an escape such as \x{de} in the data was always generating
2697 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2698 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2699 truncation.
2701 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2703 7. Added two (int) casts to pcregrep when printing the difference of two
2704 pointers, in case they are 64-bit values.
2706 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2707 test 2 if it fails.
2709 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2710 and a #define of that name to empty if it is not externally set. This is to
2711 allow users of MSVC to set it if necessary.
2713 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2714 the convenience functions in the pcre_get.c source file.
2716 11. An option change at the start of a pattern that had top-level alternatives
2717 could cause overwriting and/or a crash. This command provoked a crash in
2718 some environments:
2720 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2722 This potential security problem was recorded as CVE-2008-2371.
2724 12. For a pattern where the match had to start at the beginning or immediately
2725 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2726 pcre_dfa_exec() could read past the end of the passed subject if there was
2727 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2728 pcretest so that it places the subject at the end of its malloc-ed buffer.
2730 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2731 exec() might read past the end of the data buffer in UTF-8 mode.
2733 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2734 the data contained the byte 0x85 as part of a UTF-8 character within its
2735 first line. This applied both to normal and DFA matching.
2737 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2738 /^[^d]*?$/8 failed to match "abc".
2740 16. Added a missing copyright notice to pcrecpp_internal.h.
2742 17. Make it more clear in the documentation that values returned from
2743 pcre_exec() in ovector are byte offsets, not character counts.
2745 18. Tidied a few places to stop certain compilers from issuing warnings.
2747 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2748 supplied by Stefan Weber. I made a further small update for 7.8 because
2749 there is a change of source arrangements: the pcre_searchfuncs.c module is
2750 replaced by pcre_ucd.c.
2753 Version 7.7 07-May-08
2754 ---------------------
2756 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2757 a string to a long long, pretend we don't even have a long long." This is
2758 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2760 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2761 pre-7.6 versions, which defined a global no_arg variable instead of putting
2762 it in the RE class. (See also #8 below.)
2764 3. Remove a line of dead code, identified by coverity and reported by Nuno
2765 Lopes.
2767 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2769 (1) The include/exclude patterns were being applied to the whole pathnames
2770 of files, instead of just to the final components.
2772 (2) If there was more than one level of directory, the subdirectories were
2773 skipped unless they satisfied the include/exclude conditions. This is
2774 inconsistent with GNU grep (and could even be seen as contrary to the
2775 pcregrep specification - which I improved to make it absolutely clear).
2776 The action now is always to scan all levels of directory, and just
2777 apply the include/exclude patterns to regular files.
2779 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2780 --exclude_dir in the tests to avoid scanning .svn directories.
2782 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2783 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2784 doesn't support NULs in patterns.
2786 7. Added some missing "const"s to declarations of static tables in
2787 pcre_compile.c and pcre_dfa_exec.c.
2789 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2790 caused by fix #2 above. (Subsequently also a second patch to fix the
2791 first patch. And a third patch - this was a messy problem.)
2793 9. Applied Craig's patch to remove the use of push_back().
2795 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2796 matching function regexec().
2798 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2799 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2800 references. PCRE supports relative numbers with this syntax (I don't think
2801 Oniguruma does).
2803 12. Previously, a group with a zero repeat such as (...){0} was completely
2804 omitted from the compiled regex. However, this means that if the group
2805 was called as a subroutine from elsewhere in the pattern, things went wrong
2806 (an internal error was given). Such groups are now left in the compiled
2807 pattern, with a new opcode that causes them to be skipped at execution
2808 time.
2810 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2811 to the way PCRE behaves:
2813 (a) A lone ] character is dis-allowed (Perl treats it as data).
2815 (b) A back reference to an unmatched subpattern matches an empty string
2816 (Perl fails the current match path).
2818 (c) A data ] in a character class must be notated as \] because if the
2819 first data character in a class is ], it defines an empty class. (In
2820 Perl it is not possible to have an empty class.) The empty class []
2821 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2822 The negative empty class [^] matches any one character, independently
2823 of the DOTALL setting.
2825 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2826 non-existent subpattern following a character class starting with ']' and
2827 containing () gave an internal compiling error instead of "reference to
2828 non-existent subpattern". Fortunately, when the pattern did exist, the
2829 compiled code was correct. (When scanning forwards to check for the
2830 existence of the subpattern, it was treating the data ']' as terminating
2831 the class, so got the count wrong. When actually compiling, the reference
2832 was subsequently set up correctly.)
2834 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2835 it was being rejected as not supported by pcre_dfa_exec(), even though
2836 other assertions are supported. I have made pcre_dfa_exec() support
2837 (*FAIL).
2839 16. The implementation of 13c above involved the invention of a new opcode,
2840 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2841 cannot be changed at match time, I realized I could make a small
2842 improvement to matching performance by compiling OP_ALLANY instead of
2843 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2844 on the OP_ANY path.
2846 17. Compiling pcretest on Windows with readline support failed without the
2847 following two fixes: (1) Make the unistd.h include conditional on
2848 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2850 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2851 ncurses library to be included for pcretest when ReadLine support is
2852 requested, but also to allow for it to be overridden. This patch came from
2853 Daniel Bergström.
2855 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2856 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2857 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2858 spotting this.
2861 Version 7.6 28-Jan-08
2862 ---------------------
2864 1. A character class containing a very large number of characters with
2865 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2866 overflow.
2868 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2869 HAVE_LONG_LONG is not defined.
2871 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2872 bring them up to date and include new features. This patch includes:
2874 - Fixed PH's badly added libz and libbz2 support.
2875 - Fixed a problem with static linking.
2876 - Added pcredemo. [But later removed - see 7 below.]
2877 - Fixed dftables problem and added an option.
2878 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2880 - Added readline support for pcretest.
2881 - Added an listing of the option settings after cmake has run.
2883 4. A user submitted a patch to Makefile that makes it easy to create
2884 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2885 Makefile.am that cause it to include this special target, without
2886 affecting anything else. Note that the same mingw target plus all
2887 the other distribution libraries and programs are now supported
2888 when configuring with CMake (see 6 below) instead of with
2889 Configure/Make.
2891 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2892 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2893 exported in the Windows port". It has not yet been confirmed that the patch
2894 solves the problem, but it does no harm.
2896 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2897 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2898 with CMake, and also correct the comment about stack recursion.
2900 7. Remove the automatic building of pcredemo from the ./configure system and
2901 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2902 of a program that users should build themselves after PCRE is installed, so
2903 building it automatically is not really right. What is more, it gave
2904 trouble in some build environments.
2906 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2909 Version 7.5 10-Jan-08
2910 ---------------------
2912 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2913 values in parens when parsing an RE using the C++ wrapper."
2915 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2916 Characters greater than 255 were excluded from the class instead of being
2917 included.
2919 3. The same bug as (2) above applied to negated POSIX classes such as
2920 [:^space:].
2922 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2923 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2924 I have changed it.
2926 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2927 first named subpattern) and a construct such as (?&a) would reference the
2928 first named subpattern whose name started with "a" (in other words, the
2929 length check was missing). Both these problems are fixed. "Subpattern name
2930 expected" is now given for (?&) (a zero-length name), and this patch also
2931 makes it give the same error for \k'' (previously it complained that that
2932 was a reference to a non-existent subpattern).
2934 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2935 this is right because (?- can be followed by option settings as well as by
2936 digits. I have, however, made the messages clearer.
2938 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2939 than the number used in the conditional) now cause a compile-time error.
2940 This is actually not compatible with Perl, which accepts such patterns, but
2941 treats the conditional as always being FALSE (as PCRE used to), but it
2942 seems to me that giving a diagnostic is better.
2944 8. Change "alphameric" to the more common word "alphanumeric" in comments
2945 and messages.
2947 9. Fix two occurrences of "backslash" in comments that should have been
2948 "backspace".
2950 10. Remove two redundant lines of code that can never be obeyed (their function
2951 was moved elsewhere).
2953 11. The program that makes PCRE's Unicode character property table had a bug
2954 which caused it to generate incorrect table entries for sequences of
2955 characters that have the same character type, but are in different scripts.
2956 It amalgamated them into a single range, with the script of the first of
2957 them. In other words, some characters were in the wrong script. There were
2958 thirteen such cases, affecting characters in the following ranges:
2960 U+002b0 - U+002c1
2961 U+0060c - U+0060d
2962 U+0061e - U+00612
2963 U+0064b - U+0065e
2964 U+0074d - U+0076d
2965 U+01800 - U+01805
2966 U+01d00 - U+01d77
2967 U+01d9b - U+01dbf
2968 U+0200b - U+0200f
2969 U+030fc - U+030fe
2970 U+03260 - U+0327f
2971 U+0fb46 - U+0fbb1
2972 U+10450 - U+1049d
2974 12. The -o option (show only the matching part of a line) for pcregrep was not
2975 compatible with GNU grep in that, if there was more than one match in a
2976 line, it showed only the first of them. It now behaves in the same way as
2977 GNU grep.
2979 13. If the -o and -v options were combined for pcregrep, it printed a blank
2980 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2981 does the same. The return code can be used to tell if there were any
2982 non-matching lines.
2984 14. Added --file-offsets and --line-offsets to pcregrep.
2986 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2987 infinitely looping recursion. The bug was that positive lookaheads were not
2988 being skipped when checking for a possible empty match (negative lookaheads
2989 and both kinds of lookbehind were skipped).
2991 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2992 inclusion of <windows.h> to before rather than after the definition of
2993 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2995 17. Specifying a possessive quantifier with a specific limit for a Unicode
2996 character property caused pcre_compile() to compile bad code, which led at
2997 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2998 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2999 caused the error; without that there was no problem.
3001 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3003 19. Added --enable-pcretest-libreadline.
3005 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3006 RE::GlobalReplace(). As a result, the number of replacements returned was
3007 double what it should be. I removed one of the increments, but Craig sent a
3008 later patch that removed the other one (the right fix) and added unit tests
3009 that check the return values (which was not done before).
3011 21. Several CMake things:
3013 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3014 the names libpcre and libpcreposix, not just pcre and pcreposix.
3016 (2) The above change means that pcretest and pcregrep are now correctly
3017 linked with the newly-built libraries, not previously installed ones.
3021 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3022 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3023 UTF-8 newline character). The key issue is that the pattern starts .*;
3024 this means that the match must be either at the beginning, or after a
3025 newline. The bug was in the code for advancing after a failed match and
3026 checking that the new position followed a newline. It was not taking
3027 account of UTF-8 characters correctly.
3029 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3030 character classes. PCRE was not treating the sequence [:...:] as a
3031 character class unless the ... were all letters. Perl, however, seems to
3032 allow any characters between [: and :], though of course it rejects as
3033 unknown any "names" that contain non-letters, because all the known class
3034 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3035 for example, whereas PCRE did not - it did not recognize a POSIX character
3036 class. This seemed a bit dangerous, so the code has been changed to be
3037 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3038 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3039 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3040 Perl does, and where it didn't before.
3042 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3043 Windows environments %n is disabled by default.
3046 Version 7.4 21-Sep-07
3047 ---------------------
3049 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3050 means that a class such as [\s] counted as "explicit reference to CR or
3051 LF". That isn't really right - the whole point of the change was to try to
3052 help when there was an actual mention of one of the two characters. So now
3053 the change happens only if \r or \n (or a literal CR or LF) character is
3054 encountered.
3056 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3057 of both had grown to the point where there were only 3 bits left.
3058 Fortunately, there was spare space in the data structure, and so I have
3059 moved the internal flags into a new 16-bit field to free up more option
3060 bits.
3062 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3063 but did not set the internal JCHANGED flag - either of these is enough to
3064 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3065 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3066 start sets both bits.
3068 4. Added options (at build time, compile time, exec time) to change \R from
3069 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3071 5. doc/pcresyntax.html was missing from the distribution.
3073 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3074 compatibility, even though it is no longer used.
3076 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3077 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3078 windows.h file is present (where different names are used). [This was
3079 reversed later after testing - see 16 below.]
3081 8. Changed all #include <config.h> to #include "config.h". There were also
3082 some further <pcre.h> cases that I changed to "pcre.h".
3084 9. When pcregrep was used with the --colour option, it missed the line ending
3085 sequence off the lines that it output.
3087 10. It was pointed out to me that arrays of string pointers cause lots of
3088 relocations when a shared library is dynamically loaded. A technique of
3089 using a single long string with a table of offsets can drastically reduce
3090 these. I have refactored PCRE in four places to do this. The result is
3091 dramatic:
3093 Originally: 290
3094 After changing UCP table: 187
3095 After changing error message table: 43
3096 After changing table of "verbs" 36
3097 After changing table of Posix names 22
3099 Thanks to the folks working on Gregex for glib for this insight.
3101 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3102 unicode-properties was also set.
3104 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3106 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3107 checked only for CRLF.
3109 14. Added casts to pcretest.c to avoid compiler warnings.
3111 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3113 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3114 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3115 entirely. This removes changes made in 7 above.
3117 17. The CMake files have been updated, and there is now more information about
3118 building with CMake in the NON-UNIX-USE document.
3121 Version 7.3 28-Aug-07
3122 ---------------------
3124 1. In the rejigging of the build system that eventually resulted in 7.1, the
3125 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3126 brackets there is not right, since it causes compilers to look for an
3127 installed pcre.h, not the version that is in the source that is being
3128 compiled (which of course may be different). I have changed it back to:
3130 #include "pcre.h"
3132 I have a vague recollection that the change was concerned with compiling in
3133 different directories, but in the new build system, that is taken care of
3134 by the VPATH setting the Makefile.
3136 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3137 when the subject happened to end in the byte 0x85 (e.g. if the last
3138 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3139 characters but of course it shouldn't be taken as a newline when it is part
3140 of another character. The bug was that, for an unlimited repeat of . in
3141 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3142 characters when looking for a newline.
3144 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3146 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3147 in debug output.
3149 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3150 long printing in the pcrecpp unittest when running under MinGW.
3152 6. ESC_K was left out of the EBCDIC table.
3154 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3155 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3156 limit also applies to "virtual nesting" when a pattern is recursive, and in
3157 this case 1000 isn't so big. I have been able to remove this limit at the
3158 expense of backing off one optimization in certain circumstances. Normally,
3159 when pcre_exec() would call its internal match() function recursively and
3160 immediately return the result unconditionally, it uses a "tail recursion"
3161 feature to save stack. However, when a subpattern that can match an empty
3162 string has an unlimited repetition quantifier, it no longer makes this
3163 optimization. That gives it a stack frame in which to save the data for
3164 checking that an empty string has been matched. Previously this was taken
3165 from the 1000-entry workspace that had been reserved. So now there is no
3166 explicit limit, but more stack is used.
3168 8. Applied Daniel's patches to solve problems with the import/export magic
3169 syntax that is required for Windows, and which was going wrong for the
3170 pcreposix and pcrecpp parts of the library. These were overlooked when this
3171 problem was solved for the main library.
3173 9. There were some crude static tests to avoid integer overflow when computing
3174 the size of patterns that contain repeated groups with explicit upper
3175 limits. As the maximum quantifier is 65535, the maximum group length was
3176 set at 30,000 so that the product of these two numbers did not overflow a
3177 32-bit integer. However, it turns out that people want to use groups that
3178 are longer than 30,000 bytes (though not repeat them that many times).
3179 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3180 made it possible to implement the integer overflow checks in a much more
3181 dynamic way, which I have now done. The artificial limitation on group
3182 length has been removed - we now have only the limit on the total length of
3183 the compiled pattern, which depends on the LINK_SIZE setting.
3185 10. Fixed a bug in the documentation for get/copy named substring when
3186 duplicate names are permitted. If none of the named substrings are set, the
3187 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3188 empty string.
3190 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3191 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3192 because the ] is interpreted as the first data character and the
3193 terminating ] is not found. PCRE has been made compatible with Perl in this
3194 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3195 cause memory overwriting.
3197 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3198 string has been matched (to stop an infinite loop). It was not recognizing
3199 a conditional subpattern that could match an empty string if that
3200 subpattern was within another subpattern. For example, it looped when
3201 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3202 condition was not nested. This bug has been fixed.
3204 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3205 past the start of the subject in the presence of bytes with the top bit
3206 set, for example "\x8aBCD".
3208 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3209 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3211 14. Optimized (?!) to (*FAIL).
3213 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3214 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3215 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3216 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3217 does: it's just the validity check that is more restrictive.
3219 16. Inserted checks for integer overflows during escape sequence (backslash)
3220 processing, and also fixed erroneous offset values for syntax errors during
3221 backslash processing.
3223 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3224 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3226 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3227 caused an overrun.
3229 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3230 something other than just ASCII characters) inside a group that had an
3231 unlimited repeat caused a loop at compile time (while checking to see
3232 whether the group could match an empty string).
3234 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3235 [\P{Any}] did so. (Error in the code for printing property names.)
3237 21. An orphan \E inside a character class could cause a crash.
3239 22. A repeated capturing bracket such as (A)? could cause a wild memory
3240 reference during compilation.
3242 23. There are several functions in pcre_compile() that scan along a compiled
3243 expression for various reasons (e.g. to see if it's fixed length for look
3244 behind). There were bugs in these functions when a repeated \p or \P was
3245 present in the pattern. These operators have additional parameters compared
3246 with \d, etc, and these were not being taken into account when moving along
3247 the compiled data. Specifically:
3249 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3250 length.
3252 (b) An item such as \pL+ within a repeated group could cause crashes or
3253 loops.
3255 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3256 "reference to non-existent subpattern" error.
3258 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3260 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3261 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3263 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3265 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3266 character were causing crashes (broken optimization).
3268 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3269 \p or \P) caused a compile-time loop.
3271 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3272 break. For example, the unstudied pattern [\r\n]A does not match the string
3273 "\r\nA" because change 7.0/46 below moves the current point on by two
3274 characters after failing to match at the start. However, the pattern \nA
3275 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3276 the same is true. There doesn't seem any very clean way out of this, but
3277 what I have chosen to do makes the common cases work: PCRE now takes note
3278 of whether there can be an explicit match for \r or \n anywhere in the
3279 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3280 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3281 pattern has explicit CR or LF references.
3283 29. Added (*CR) etc for changing newline setting at start of pattern.
3286 Version 7.2 19-Jun-07
3287 ---------------------
3289 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3290 which is apparently normally available under Windows.
3292 2. Re-jig the pcregrep tests with different newline settings in an attempt
3293 to make them independent of the local environment's newline setting.
3295 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3297 4. Some of the "internals" tests were previously cut out when the link size
3298 was not 2, because the output contained actual offsets. The recent new
3299 "Z" feature of pcretest means that these can be cut out, making the tests
3300 usable with all link sizes.
3302 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3303 stack recursion. This gives a massive performance boost under BSD, but just
3304 a small improvement under Linux. However, it saves one field in the frame
3305 in all cases.
3307 6. Added more features from the forthcoming Perl 5.10:
3309 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3310 recursion call. It refers to the nth most recently opened parentheses.
3312 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3313 to be opened parentheses.
3315 (c) Conditions that refer to capturing parentheses can be specified
3316 relatively, for example, (?(-2)... or (?(+3)...
3318 (d) \K resets the start of the current match so that everything before
3319 is not part of it.
3321 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3323 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3324 reference syntax.
3326 (g) (?| introduces a group in which the numbering of parentheses in each
3327 alternative starts with the same number.
3329 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3331 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3334 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3335 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3336 for detecting groups that can match an empty string.
3338 9. A pattern with a very large number of alternatives (more than several
3339 hundred) was running out of internal workspace during the pre-compile
3340 phase, where pcre_compile() figures out how much memory will be needed. A
3341 bit of new cunning has reduced the workspace needed for groups with
3342 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3343 workspace instead of running out of the 4096 that are available.
3345 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3347 11. Applied patch from Google to remove an optimization that didn't quite work.
3348 The report of the bug said:
3350 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3351 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3352 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3354 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3355 it matched the wrong number of bytes.
3358 Version 7.1 24-Apr-07
3359 ---------------------
3361 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3362 that is more "standard", making use of automake and other Autotools. There
3363 is some re-arrangement of the files and adjustment of comments consequent
3364 on this.
3366 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3367 for recursive directory scanning broke on some systems because the files
3368 are not scanned in any specific order and on different systems the order
3369 was different. A call to "sort" has been inserted into RunGrepTest for the
3370 approprate test as a short-term fix. In the longer term there may be an
3371 alternative.
3373 3. I had an email from Eric Raymond about problems translating some of PCRE's
3374 man pages to HTML (despite the fact that I distribute HTML pages, some
3375 people do their own conversions for various reasons). The problems
3376 concerned the use of low-level troff macros .br and .in. I have therefore
3377 removed all such uses from the man pages (some were redundant, some could
3378 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3379 HTML has been updated to handle .nf/.fi and to complain if it encounters
3380 .br or .in.
3382 4. Updated comments in configure.ac that get placed in config.h.in and also
3383 arranged for config.h to be included in the distribution, with the name
3384 config.h.generic, for the benefit of those who have to compile without
3385 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3387 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3388 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3389 makevp.bat for the current PCRE, using the additional files
3390 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3392 6. A Windows user reported a minor discrepancy with test 2, which turned out
3393 to be caused by a trailing space on an input line that had got lost in his
3394 copy. The trailing space was an accident, so I've just removed it.
3396 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3397 that is needed.
3399 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3400 as "const" (a) because they are and (b) because it helps the PHP
3401 maintainers who have recently made a script to detect big data structures
3402 in the php code that should be moved to the .rodata section. I remembered
3403 to update Builducptable as well, so it won't revert if ucptable.h is ever
3404 re-created.
3406 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3407 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3408 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3409 support is not required. This saves 1.5-2K of code, which is important in
3410 some applications.
3412 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3413 so as not to refer to the tables, even though these functions will never be
3414 called when UTF-8 support is disabled. Otherwise there are problems with a
3415 shared library.
3417 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3419 (a) It was defining its arguments as char * instead of void *.
3421 (b) It was assuming that all moves were upwards in memory; this was true
3422 a long time ago when I wrote it, but is no longer the case.
3424 The emulated memove() is provided for those environments that have neither
3425 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3426 is clearly not the case, as these two bugs were recently reported.
3428 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3429 and Detrail to create the HTML documentation, the .txt form of the man
3430 pages, and it removes trailing spaces from listed files. It also creates
3431 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3432 case, it wraps all the #defines with #ifndefs. This script should be run
3433 before "make dist".
3435 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3436 with Unicode property support.
3438 (a) For a maximizing quantifier, if the two different cases of the
3439 character were of different lengths in their UTF-8 codings (there are
3440 some cases like this - I found 11), and the matching function had to
3441 back up over a mixture of the two cases, it incorrectly assumed they
3442 were both the same length.
3444 (b) When PCRE was configured to use the heap rather than the stack for
3445 recursion during matching, it was not correctly preserving the data for
3446 the other case of a UTF-8 character when checking ahead for a match
3447 while processing a minimizing repeat. If the check also involved
3448 matching a wide character, but failed, corruption could cause an
3449 erroneous result when trying to check for a repeat of the original
3450 character.
3452 13. Some tidying changes to the testing mechanism:
3454 (a) The RunTest script now detects the internal link size and whether there
3455 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3456 values substituted by "configure". (The RunGrepTest script already did
3457 this for UTF-8.) The configure.ac script no longer substitutes the
3458 relevant variables.
3460 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3461 with length and offset values. This means that the output is different
3462 for different internal link sizes. Test 2 is skipped for link sizes
3463 other than 2 because of this, bypassing the problem. Unfortunately,
3464 there was also a test in test 3 (the locale tests) that used /B and
3465 failed for link sizes other than 2. Rather than cut the whole test out,
3466 I have added a new /Z option to pcretest that replaces the length and
3467 offset values with spaces. This is now used to make test 3 independent
3468 of link size. (Test 2 will be tidied up later.)
3470 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3471 segmentation fault instead of returning the appropriate error message.
3473 15. In multiline mode when the newline sequence was set to "any", the pattern
3474 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3475 This doesn't seem right; it now treats the CRLF combination as the line
3476 ending, and so does not match in that case. It's only a pattern such as ^$
3477 that would hit this one: something like ^ABC$ would have failed after \r
3478 and then tried again after \r\n.
3480 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3481 in an attempt to make files that differ only in their line terminators
3482 compare equal. This works on Linux.
3484 17. Under certain error circumstances pcregrep might try to free random memory
3485 as it exited. This is now fixed, thanks to valgrind.
3487 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3488 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3489 was because its rules for how to advance for /g after matching an empty
3490 string at the end of a line did not allow for this case. They now check for
3491 it specially.
3493 20. pcretest is supposed to handle patterns and data of any length, by
3494 extending its buffers when necessary. It was getting this wrong when the
3495 buffer for a data line had to be extended.
3497 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3498 CRLF as a newline sequence.
3500 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3501 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3502 I have nevertheless tidied it up.
3504 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3506 24. Added a man page for pcre-config.
3509 Version 7.0 19-Dec-06
3510 ---------------------
3512 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3513 moving to gcc 4.1.1.
3515 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3516 sys/time.h, which is documented as needed for this function. It doesn't
3517 seem to matter on Linux, but it showed up on some releases of OS X.
3519 3. It seems that there are systems where bytes whose values are greater than
3520 127 match isprint() in the "C" locale. The "C" locale should be the
3521 default when a C program starts up. In most systems, only ASCII printing
3522 characters match isprint(). This difference caused the output from pcretest
3523 to vary, making some of the tests fail. I have changed pcretest so that:
3525 (a) When it is outputting text in the compiled version of a pattern, bytes
3526 other than 32-126 are always shown as hex escapes.
3528 (b) When it is outputting text that is a matched part of a subject string,
3529 it does the same, unless a different locale has been set for the match
3530 (using the /L modifier). In this case, it uses isprint() to decide.
3532 4. Fixed a major bug that caused incorrect computation of the amount of memory
3533 required for a compiled pattern when options that changed within the
3534 pattern affected the logic of the preliminary scan that determines the
3535 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3536 that the computed length was too small. The symptoms of this bug were
3537 either the PCRE error "internal error: code overflow" from pcre_compile(),
3538 or a glibc crash with a message such as "pcretest: free(): invalid next
3539 size (fast)". Examples of patterns that provoked this bug (shown in
3540 pcretest format) are:
3542 /(?-x: )/x
3543 /(?x)(?-x: \s*#\s*)/
3544 /((?i)[\x{c0}])/8
3545 /(?i:[\x{c0}])/8
3547 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3548 is now done differently.
3550 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3551 wrapper classes; (b) implement a new function in the C++ scanner that is
3552 more efficient than the old way of doing things because it avoids levels of
3553 recursion in the regex matching; (c) add a paragraph to the documentation
3554 for the FullMatch() function.
3556 6. The escape sequence \n was being treated as whatever was defined as
3557 "newline". Not only was this contrary to the documentation, which states
3558 that \n is character 10 (hex 0A), but it also went horribly wrong when
3559 "newline" was defined as CRLF. This has been fixed.
3561 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3562 was being set to -1 for the "end of line" case (supposedly a value that no
3563 character can have). Though this value is never used (the check for end of
3564 line is "zero bytes in current character"), it caused compiler complaints.
3565 I've changed it to 0xffffffff.
3567 8. In pcre_version.c, the version string was being built by a sequence of
3568 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3569 string (as it is for production releases) called a macro with an empty
3570 argument. The C standard says the result of this is undefined. The gcc
3571 compiler treats it as an empty string (which was what was wanted) but it is
3572 reported that Visual C gives an error. The source has been hacked around to
3573 avoid this problem.
3575 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3576 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3577 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3578 of them did).
3580 10. Originally, pcretest opened its input and output without "b"; then I was
3581 told that "b" was needed in some environments, so it was added for release
3582 5.0 to both the input and output. (It makes no difference on Unix-like
3583 systems.) Later I was told that it is wrong for the input on Windows. I've
3584 now abstracted the modes into two macros, to make it easier to fiddle with
3585 them, and removed "b" from the input mode under Windows.
3587 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3589 12. Added -help and --help to pcretest as an official way of being reminded
3590 of the options.
3592 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3593 and pcrecpp.cc because they annoy compilers at high warning levels.
3595 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3597 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3598 scripts are not C programs :-) and which was not noticed because it works
3599 on Linux.
3601 16. pcretest is supposed to handle any length of pattern and data line (as one
3602 line or as a continued sequence of lines) by extending its input buffer if
3603 necessary. This feature was broken for very long pattern lines, leading to
3604 a string of junk being passed to pcre_compile() if the pattern was longer
3605 than about 50K.
3607 17. I have done a major re-factoring of the way pcre_compile() computes the
3608 amount of memory needed for a compiled pattern. Previously, there was code
3609 that made a preliminary scan of the pattern in order to do this. That was
3610 OK when PCRE was new, but as the facilities have expanded, it has become
3611 harder and harder to keep it in step with the real compile phase, and there
3612 have been a number of bugs (see for example, 4 above). I have now found a
3613 cunning way of running the real compile function in a "fake" mode that
3614 enables it to compute how much memory it would need, while actually only
3615 ever using a few hundred bytes of working memory and without too many
3616 tests of the mode. This should make future maintenance and development
3617 easier. A side effect of this work is that the limit of 200 on the nesting
3618 depth of parentheses has been removed (though this was never a serious
3619 limitation, I suspect). However, there is a downside: pcre_compile() now
3620 runs more slowly than before (30% or more, depending on the pattern). I
3621 hope this isn't a big issue. There is no effect on runtime performance.
3623 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3624 newline (only possible for the last line of a file) and it was a
3625 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3627 19. Added additional timing features to pcretest. (1) The -tm option now times
3628 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3629 separate command line item, by a number that specifies the number of
3630 repeats to use when timing. The default is 50000; this gives better
3631 precision, but takes uncomfortably long for very large patterns.
3633 20. Extended pcre_study() to be more clever in cases where a branch of a
3634 subpattern has no definite first character. For example, (a*|b*)[cd] would
3635 previously give no result from pcre_study(). Now it recognizes that the
3636 first character must be a, b, c, or d.
3638 21. There was an incorrect error "recursive call could loop indefinitely" if
3639 a subpattern (or the entire pattern) that was being tested for matching an
3640 empty string contained only one non-empty item after a nested subpattern.
3641 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3642 incorrectly, because the \d was being skipped in the check.
3644 22. The pcretest program now has a new pattern option /B and a command line
3645 option -b, which is equivalent to adding /B to every pattern. This causes
3646 it to show the compiled bytecode, without the additional information that
3647 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3648 is the same as /B/I).
3650 23. A new optimization is now able automatically to treat some sequences such
3651 as a*b as a*+b. More specifically, if something simple (such as a character
3652 or a simple class like \d) has an unlimited quantifier, and is followed by
3653 something that cannot possibly match the quantified thing, the quantifier
3654 is automatically "possessified".
3656 24. A recursive reference to a subpattern whose number was greater than 39
3657 went wrong under certain circumstances in UTF-8 mode. This bug could also
3658 have affected the operation of pcre_study().
3660 25. Realized that a little bit of performance could be had by replacing
3661 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3663 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3665 27. Possessive quantifiers such as a++ were previously implemented by turning
3666 them into atomic groups such as ($>a+). Now they have their own opcodes,
3667 which improves performance. This includes the automatically created ones
3668 from 23 above.
3670 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3671 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3672 the first matched character to be a colon. This applied both to named and
3673 numbered groups.
3675 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3677 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3678 building PCRE on Windows easier, so I have included it in the distribution.
3680 31. There is now a check in pcretest against a ridiculously large number being
3681 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3682 loop, the loop is abandoned.
3684 32. Forward references to subpatterns in conditions such as (?(2)...) where
3685 subpattern 2 is defined later cause pcre_compile() to search forwards in
3686 the pattern for the relevant set of parentheses. This search went wrong
3687 when there were unescaped parentheses in a character class, parentheses
3688 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3690 33. "Subroutine" calls and backreferences were previously restricted to
3691 referencing subpatterns earlier in the regex. This restriction has now
3692 been removed.
3694 34. Added a number of extra features that are going to be in Perl 5.10. On the
3695 whole, these are just syntactic alternatives for features that PCRE had
3696 previously implemented using the Python syntax or my own invention. The
3697 other formats are all retained for compatibility.
3699 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3700 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3701 also .NET compatible.
3703 (b) A recursion or subroutine call to a named group can now be defined as
3704 (?&name) as well as (?P>name).
3706 (c) A backreference to a named group can now be defined as \k<name> or
3707 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3708 5.10, are also .NET compatible.
3710 (d) A conditional reference to a named group can now use the syntax
3711 (?(<name>) or (?('name') as well as (?(name).
3713 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3714 groups (named and numbered) that are never evaluated inline, but can be
3715 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3716 is always false. There may be only one alternative in such a group.
3718 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3719 as the simple (?(R). The condition is true only if the most recent
3720 recursion is that of the given number or name. It does not search out
3721 through the entire recursion stack.
3723 (g) The escape \gN or \g{N} has been added, where N is a positive or
3724 negative number, specifying an absolute or relative reference.
3726 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3727 some "unreachable code" warnings.
3729 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3730 things, this adds five new scripts.
3732 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3733 There were also incompatibilities regarding the handling of \Q..\E inside
3734 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3735 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3737 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3738 matches an empty string, and forcibly breaks the loop. There were bugs in
3739 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3740 against aaaa the result was just "a" rather than "aaaa", for example. Two
3741 separate and independent bugs (that affected different cases) have been
3742 fixed.
3744 39. Refactored the code to abolish the use of different opcodes for small
3745 capturing bracket numbers. This is a tidy that I avoided doing when I
3746 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3747 The new approach is not only tidier, it makes it possible to reduce the
3748 memory needed to fix the previous bug (38).
3750 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3751 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3752 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3753 mode.
3755 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3756 report.
3758 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3759 copy construction and assignment in the C++ wrapper.
3761 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3762 couple of bugs that could have given wrong results in the "--newline=crlf"
3763 case.
3765 44. Added a number of casts and did some reorganization of signed/unsigned int
3766 variables following suggestions from Dair Grant. Also renamed the variable
3767 "this" as "item" because it is a C++ keyword.
3769 45. Arranged for dftables to add
3771 #include "pcre_internal.h"
3773 to pcre_chartables.c because without it, gcc 4.x may remove the array
3774 definition from the final binary if PCRE is built into a static library and
3775 dead code stripping is activated.
3777 46. For an unanchored pattern, if a match attempt fails at the start of a
3778 newline sequence, and the newline setting is CRLF or ANY, and the next two
3779 characters are CRLF, advance by two characters instead of one.
3782 Version 6.7 04-Jul-06
3783 ---------------------
3785 1. In order to handle tests when input lines are enormously long, pcretest has
3786 been re-factored so that it automatically extends its buffers when
3787 necessary. The code is crude, but this _is_ just a test program. The
3788 default size has been increased from 32K to 50K.
3790 2. The code in pcre_study() was using the value of the re argument before
3791 testing it for NULL. (Of course, in any sensible call of the function, it
3792 won't be NULL.)
3794 3. The memmove() emulation function in pcre_internal.h, which is used on
3795 systems that lack both memmove() and bcopy() - that is, hardly ever -
3796 was missing a "static" storage class specifier.
3798 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3799 containing an extended class (one that cannot be represented by a bitmap
3800 because it contains high-valued characters or Unicode property items, e.g.
3801 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3802 pattern, but PCRE should not loop if you do not (it no longer does).
3803 [Detail: two cases were found: (a) a repeated subpattern containing an
3804 extended class; (b) a recursive reference to a subpattern that followed a
3805 previous extended class. It wasn't skipping over the extended class
3806 correctly when UTF-8 mode was not set.]
3808 5. A negated single-character class was not being recognized as fixed-length
3809 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3810 compile error "lookbehind assertion is not fixed length".
3812 6. The RunPerlTest auxiliary script was showing an unexpected difference
3813 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3814 write a Perl script that can interpret lines of an input file either as
3815 byte characters or as UTF-8, which is what "perltest" was being required to
3816 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3817 can't do is switch easily at run time between having the "use utf8;" pragma
3818 or not. In the end, I fudged it by using the RunPerlTest script to insert
3819 "use utf8;" explicitly for the UTF-8 tests.
3821 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3822 the end of the subject string, contrary to the documentation and to what
3823 Perl does. This was true of both matching functions. Now it matches only at
3824 the start of the subject and immediately after *internal* newlines.
3826 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3827 a pointer to an int instead of a pointer to an unsigned long int. This
3828 caused problems on 64-bit systems.
3830 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3831 instance of the 'standard' template library not being so standard".
3833 10. There was no check on the number of named subpatterns nor the maximum
3834 length of a subpattern name. The product of these values is used to compute
3835 the size of the memory block for a compiled pattern. By supplying a very
3836 long subpattern name and a large number of named subpatterns, the size
3837 computation could be caused to overflow. This is now prevented by limiting
3838 the length of names to 32 characters, and the number of named subpatterns
3839 to 10,000.
3841 11. Subpatterns that are repeated with specific counts have to be replicated in
3842 the compiled pattern. The size of memory for this was computed from the
3843 length of the subpattern and the repeat count. The latter is limited to
3844 65535, but there was no limit on the former, meaning that integer overflow
3845 could in principle occur. The compiled length of a repeated subpattern is
3846 now limited to 30,000 bytes in order to prevent this.
3848 12. Added the optional facility to have named substrings with the same name.
3850 13. Added the ability to use a named substring as a condition, using the
3851 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3852 are numbers (not recommended). Forward references are permitted.
3854 14. Added forward references in named backreferences (if you see what I mean).
3856 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3857 pattern could run off the end of the subject. For example, the pattern
3858 "(?s)(.{1,5})"8 did this with the subject "ab".
3860 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3861 PCRE_CASELESS was set when matching characters that were quantified with ?
3862 or *.
3864 17. A character class other than a single negated character that had a minimum
3865 but no maximum quantifier - for example [ab]{6,} - was not handled
3866 correctly by pce_dfa_exec(). It would match only one character.
3868 18. A valid (though odd) pattern that looked like a POSIX character
3869 class but used an invalid character after [ (for example [[,abc,]]) caused
3870 pcre_compile() to give the error "Failed: internal error: code overflow" or
3871 in some cases to crash with a glibc free() error. This could even happen if
3872 the pattern terminated after [[ but there just happened to be a sequence of
3873 letters, a binary zero, and a closing ] in the memory that followed.
3875 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3876 over the years. Originally (before any Unicode support), just the bottom 8
3877 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3878 output from "man perlunicode" includes this:
3880 The regular expression compiler produces polymorphic opcodes. That
3881 is, the pattern adapts to the data and automatically switches to
3882 the Unicode character scheme when presented with Unicode data--or
3883 instead uses a traditional byte scheme when presented with byte
3884 data.
3886 Sadly, a wide octal escape does not cause a switch, and in a string with
3887 no other multibyte characters, these octal escapes are treated as before.
3888 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3889 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3890 Unicode string.
3892 I have not perpetrated such confusion in PCRE. Up till now, it took just
3893 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3894 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3895 translate to the appropriate multibyte character.
3897 29. Applied some refactoring to reduce the number of warnings from Microsoft
3898 and Borland compilers. This has included removing the fudge introduced
3899 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3900 a warning about an unused variable.
3902 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3903 characters since release 4.0, because Perl (from release 5.004) does not.
3904 [Or at least, is documented not to: some releases seem to be in conflict
3905 with the documentation.] However, when a pattern was studied with
3906 pcre_study() and all its branches started with \s, PCRE still included VT
3907 as a possible starting character. Of course, this did no harm; it just
3908 caused an unnecessary match attempt.
3910 22. Removed a now-redundant internal flag bit that recorded the fact that case
3911 dependency changed within the pattern. This was once needed for "required
3912 byte" processing, but is no longer used. This recovers a now-scarce options
3913 bit. Also moved the least significant internal flag bit to the most-
3914 significant bit of the word, which was not previously used (hangover from
3915 the days when it was an int rather than a uint) to free up another bit for
3916 the future.
3918 23. Added support for CRLF line endings as well as CR and LF. As well as the
3919 default being selectable at build time, it can now be changed at runtime
3920 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3921 specify that it is scanning data with non-default line endings.
3923 24. Changed the definition of CXXLINK to make it agree with the definition of
3924 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3926 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3927 recursions. This makes a big different to stack usage for some patterns.
3929 26. If a subpattern containing a named recursion or subroutine reference such
3930 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3931 the space required for the compiled pattern went wrong and gave too small a
3932 value. Depending on the environment, this could lead to "Failed: internal
3933 error: code overflow at offset 49" or "glibc detected double free or
3934 corruption" errors.
3936 27. Applied patches from Google (a) to support the new newline modes and (b) to
3937 advance over multibyte UTF-8 characters in GlobalReplace.
3939 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3940 difference for some implementation of PCRE in some Windows version.
3942 29. Added some extra testing facilities to pcretest:
3944 \q<number> in a data line sets the "match limit" value
3945 \Q<number> in a data line sets the "match recursion limt" value
3946 -S <number> sets the stack size, where <number> is in megabytes
3948 The -S option isn't available for Windows.
3951 Version 6.6 06-Feb-06
3952 ---------------------
3954 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3955 in pcreposix.h. I have copied the definition from pcre.h.
3957 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3958 because pcre.h is no longer a built file.
3960 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3961 not normally included in the compiled code.
3964 Version 6.5 01-Feb-06
3965 ---------------------
3967 1. When using the partial match feature with pcre_dfa_exec(), it was not
3968 anchoring the second and subsequent partial matches at the new starting
3969 point. This could lead to incorrect results. For example, with the pattern
3970 /1234/, partially matching against "123" and then "a4" gave a match.
3972 2. Changes to pcregrep:
3974 (a) All non-match returns from pcre_exec() were being treated as failures
3975 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3976 error message is output. Some extra information is given for the
3978 probably the only errors that are likely to be caused by users (by
3979 specifying a regex that has nested indefinite repeats, for instance).
3980 If there are more than 20 of these errors, pcregrep is abandoned.
3982 (b) A binary zero was treated as data while matching, but terminated the
3983 output line if it was written out. This has been fixed: binary zeroes
3984 are now no different to any other data bytes.
3986 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3987 used to set a locale for matching. The --locale=xxxx long option has
3988 been added (no short equivalent) to specify a locale explicitly on the
3989 pcregrep command, overriding the environment variables.
3991 (d) When -B was used with -n, some line numbers in the output were one less
3992 than they should have been.
3994 (e) Added the -o (--only-matching) option.
3996 (f) If -A or -C was used with -c (count only), some lines of context were
3997 accidentally printed for the final match.
3999 (g) Added the -H (--with-filename) option.
4001 (h) The combination of options -rh failed to suppress file names for files
4002 that were found from directory arguments.
4004 (i) Added the -D (--devices) and -d (--directories) options.
4006 (j) Added the -F (--fixed-strings) option.
4008 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4010 (l) Added the --colo(u)r option.
4012 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4013 is not present by default.
4015 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4016 items such as (?R) or (?1), when the recursion could match a number of
4017 alternatives. If it matched one of the alternatives, but subsequently,
4018 outside the recursion, there was a failure, the code tried to back up into
4019 the recursion. However, because of the way PCRE is implemented, this is not
4020 possible, and the result was an incorrect result from the match.
4022 In order to prevent this happening, the specification of recursion has
4023 been changed so that all such subpatterns are automatically treated as
4024 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4026 4. I had overlooked the fact that, in some locales, there are characters for
4027 which isalpha() is true but neither isupper() nor islower() are true. In
4028 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4029 and ordfeminine) are like this. This affected the treatment of \w and \W
4030 when they appeared in character classes, but not when they appeared outside
4031 a character class. The bit map for "word" characters is now created
4032 separately from the results of isalnum() instead of just taking it from the
4033 upper, lower, and digit maps. (Plus the underscore character, of course.)
4035 5. The above bug also affected the handling of POSIX character classes such as
4036 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4037 permanent tables. Instead, the bit maps for such a class were previously
4038 created as the appropriate unions of the upper, lower, and digit bitmaps.
4039 Now they are created by subtraction from the [[:word:]] class, which has
4040 its own bitmap.
4042 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4043 It is created by subtracting the vertical space characters (\x09, \x0a,
4044 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4045 subtraction was done in the overall bitmap for a character class, meaning
4046 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4047 be recognized. This bug has been fixed.
4049 7. Patches from the folks at Google:
4051 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4052 real life, but is still worth protecting against".
4054 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4055 regular expressions".
4057 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4058 have it.
4060 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4061 "configure" and the latter not, in order to fix a problem somebody had
4062 with compiling the Arg class on HP-UX.
4064 (e) Improve the error-handling of the C++ wrapper a little bit.
4066 (f) New tests for checking recursion limiting.
4068 8. The pcre_memmove() function, which is used only if the environment does not
4069 have a standard memmove() function (and is therefore rarely compiled),
4070 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4071 returning a result (though PCRE never actually uses the result).
4073 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4074 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4075 returned instead of calling malloc() with an overflowing number that would
4076 most likely cause subsequent chaos.
4078 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4080 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4081 with this option is matched, the nmatch and pmatch options of regexec() are
4082 ignored.
4084 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4085 provided in case anyone wants to the the POSIX interface with UTF-8
4086 strings.
4088 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4089 C++ linking (needed for some HP-UX environments).
4091 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4092 (unused parameter) and in the pcre_printint() function (omitted "default"
4093 switch label when the default is to do nothing).
4095 15. Added some code to make it possible, when PCRE is compiled as a C++
4096 library, to replace subject pointers for pcre_exec() with a smart pointer
4097 class, thus making it possible to process discontinuous strings.
4099 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4100 much the same function. They were added by different people who were trying
4101 to make PCRE easy to compile on non-Unix systems. It has been suggested
4102 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4103 for compiling on Windows systems. I have therefore replaced it with
4104 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4105 defaults to "extern" for C or "extern C" for C++, which works fine on
4106 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4107 SCOPE with something explicit in config.h. In addition:
4109 (a) pcreposix.h still had just "extern" instead of either of these macros;
4110 I have replaced it with PCRE_DATA_SCOPE.
4112 (b) Functions such as _pcre_xclass(), which are internal to the library,
4113 but external in the C sense, all had PCRE_EXPORT in their definitions.
4114 This is apparently wrong for the Windows case, so I have removed it.
4115 (It makes no difference on Unix-like systems.)
4117 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4118 of recursive calls to match(). This is different to MATCH_LIMIT because
4119 that limits the total number of calls to match(), not all of which increase
4120 the depth of recursion. Limiting the recursion depth limits the amount of
4121 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4122 when PCRE is compiled, and changed at run time. A patch from Google adds
4123 this functionality to the C++ interface.
4125 18. Changes to the handling of Unicode character properties:
4127 (a) Updated the table to Unicode 4.1.0.
4129 (b) Recognize characters that are not in the table as "Cn" (undefined).
4131 (c) I revised the way the table is implemented to a much improved format
4132 which includes recognition of ranges. It now supports the ranges that
4133 are defined in UnicodeData.txt, and it also amalgamates other
4134 characters into ranges. This has reduced the number of entries in the
4135 table from around 16,000 to around 3,000, thus reducing its size
4136 considerably. I realized I did not need to use a tree structure after
4137 all - a binary chop search is just as efficient. Having reduced the
4138 number of entries, I extended their size from 6 bytes to 8 bytes to
4139 allow for more data.
4141 (d) Added support for Unicode script names via properties such as \p{Han}.
4143 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4144 matching that character.
4146 20. When matching a repeated Unicode property with a minimum greater than zero,
4147 (for example \pL{2,}), PCRE could look past the end of the subject if it
4148 reached it while seeking the minimum number of characters. This could
4149 happen only if some of the characters were more than one byte long, because
4150 there is a check for at least the minimum number of bytes.
4152 21. Refactored the implementation of \p and \P so as to be more general, to
4153 allow for more different types of property in future. This has changed the
4154 compiled form incompatibly. Anybody with saved compiled patterns that use
4155 \p or \P will have to recompile them.
4157 22. Added "Any" and "L&" to the supported property types.
4159 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4160 but give a compile time error if the value is greater than 0xff.
4162 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4163 accidentally not being installed or uninstalled.
4165 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4166 made were to insert the current release number. This seemed silly, because
4167 it made things harder for people building PCRE on systems that don't run
4168 "configure". I have turned pcre.h into a distributed file, no longer built
4169 by "configure", with the version identification directly included. There is
4170 no longer a pcre.h.in file.
4172 However, this change necessitated a change to the pcre-config script as
4173 well. It is built from pcre-config.in, and one of the substitutions was the
4174 release number. I have updated configure.ac so that ./configure now finds
4175 the release number by grepping pcre.h.
4177 26. Added the ability to run the tests under valgrind.
4180 Version 6.4 05-Sep-05
4181 ---------------------
4183 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4184 "--" to be printed when multiple files were scanned, even when none of the
4185 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4186 consider it to be a bug, and have restored the previous behaviour.
4188 2. A couple of code tidies to get rid of compiler warnings.
4190 3. The pcretest program used to cheat by referring to symbols in the library
4191 whose names begin with _pcre_. These are internal symbols that are not
4192 really supposed to be visible externally, and in some environments it is
4193 possible to suppress them. The cheating is now confined to including
4194 certain files from the library's source, which is a bit cleaner.
4196 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4197 file's purpose clearer.
4199 5. Reorganized pcre_ucp_findchar().
4202 Version 6.3 15-Aug-05
4203 ---------------------
4205 1. The file libpcre.pc.in did not have general read permission in the tarball.
4207 2. There were some problems when building without C++ support:
4209 (a) If C++ support was not built, "make install" and "make test" still
4210 tried to test it.
4212 (b) There were problems when the value of CXX was explicitly set. Some
4213 changes have been made to try to fix these, and ...
4215 (c) --disable-cpp can now be used to explicitly disable C++ support.
4217 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4218 backslash in a target when C++ was disabled. This confuses some
4219 versions of "make", apparently. Using an intermediate variable solves
4220 this. (Same for CPP_LOBJ.)
4222 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4223 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4224 necessary on certain architectures.
4226 4. Added a setting of -export-symbols-regex to the link command to remove
4227 those symbols that are exported in the C sense, but actually are local
4228 within the library, and not documented. Their names all begin with
4229 "_pcre_". This is not a perfect job, because (a) we have to except some
4230 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4231 available (and never for static libraries). I have made a note to try to
4232 find a way round (a) in the future.
4235 Version 6.2 01-Aug-05
4236 ---------------------
4238 1. There was no test for integer overflow of quantifier values. A construction
4239 such as {1111111111111111} would give undefined results. What is worse, if
4240 a minimum quantifier for a parenthesized subpattern overflowed and became
4241 negative, the calculation of the memory size went wrong. This could have
4242 led to memory overwriting.
4244 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4246 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4247 operating environments where this matters.
4249 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4250 PCRE options from within the C++ wrapper.
4252 5. Named capturing subpatterns were not being correctly counted when a pattern
4253 was compiled. This caused two problems: (a) If there were more than 100
4254 such subpatterns, the calculation of the memory needed for the whole
4255 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4256 back references of the form \12, where the number was greater than 9, were
4257 not recognized as back references, even though there were sufficient
4258 previous subpatterns.
4260 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4261 versions of gcc, e.g. 2.95.4.
4264 Version 6.1 21-Jun-05
4265 ---------------------
4267 1. There was one reference to the variable "posix" in pcretest.c that was not
4268 surrounded by "#if !defined NOPOSIX".
4270 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4271 the cross-check on the old pcre_info() function, for the benefit of the
4272 cut-down version of PCRE that is currently imported into Exim.
4274 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4275 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4276 patterns that start (?i) but I don't think that matters. The use of (?i) is
4277 just an example; this all applies to the other options as well.
4279 4. Since libtool seems to echo the compile commands it is issuing, the output
4280 from "make" can be reduced a bit by putting "@" in front of each libtool
4281 compile command.
4283 5. Patch from the folks at Google for configure.in to be a bit more thorough
4284 in checking for a suitable C++ installation before trying to compile the
4285 C++ stuff. This should fix a reported problem when a compiler was present,
4286 but no suitable headers.
4288 6. The man pages all had just "PCRE" as their title. I have changed them to
4289 be the relevant file name. I have also arranged that these names are
4290 retained in the file doc/pcre.txt, which is a concatenation in text format
4291 of all the man pages except the little individual ones for each function.
4293 7. The NON-UNIX-USE file had not been updated for the different set of source
4294 files that come with release 6. I also added a few comments about the C++
4295 wrapper.
4298 Version 6.0 07-Jun-05
4299 ---------------------
4301 1. Some minor internal re-organization to help with my DFA experiments.
4303 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4304 didn't matter for the library itself when fully configured, but did matter
4305 when compiling without UCP support, or within Exim, where the ucp files are
4306 not imported.
4308 3. Refactoring of the library code to split up the various functions into
4309 different source modules. The addition of the new DFA matching code (see
4310 below) to a single monolithic source would have made it really too
4311 unwieldy, quite apart from causing all the code to be include in a
4312 statically linked application, when only some functions are used. This is
4313 relevant even without the DFA addition now that patterns can be compiled in
4314 one application and matched in another.
4316 The downside of splitting up is that there have to be some external
4317 functions and data tables that are used internally in different modules of
4318 the library but which are not part of the API. These have all had their
4319 names changed to start with "_pcre_" so that they are unlikely to clash
4320 with other external names.
4322 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4323 a different (DFA) algorithm. Although it is slower than the original
4324 function, it does have some advantages for certain types of matching
4325 problem.
4327 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4328 including restarting after a partial match.
4330 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4331 defined when compiling for Windows was sent to me. I have put it into the
4332 code, though I have no means of testing or verifying it.
4334 7. Added the pcre_refcount() auxiliary function.
4336 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4337 match before or at the first newline in the subject string. In pcretest,
4338 the /f option on a pattern can be used to set this.
4340 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4341 would behave wrongly. This has been present in PCRE since release 4.0.
4343 10. A number of changes to the pcregrep command:
4345 (a) Refactored how -x works; insert ^(...)$ instead of setting
4346 PCRE_ANCHORED and checking the length, in preparation for adding
4347 something similar for -w.
4349 (b) Added the -w (match as a word) option.
4351 (c) Refactored the way lines are read and buffered so as to have more
4352 than one at a time available.
4354 (d) Implemented a pcregrep test script.
4356 (e) Added the -M (multiline match) option. This allows patterns to match
4357 over several lines of the subject. The buffering ensures that at least
4358 8K, or the rest of the document (whichever is the shorter) is available
4359 for matching (and similarly the previous 8K for lookbehind assertions).
4361 (f) Changed the --help output so that it now says
4363 -w, --word-regex(p)
4365 instead of two lines, one with "regex" and the other with "regexp"
4366 because that confused at least one person since the short forms are the
4367 same. (This required a bit of code, as the output is generated
4368 automatically from a table. It wasn't just a text change.)
4370 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4371 option but starts with a hyphen. Could be a pattern or a path name
4372 starting with a hyphen, for instance.
4374 (h) "-" can be given as a file name to represent stdin.
4376 (i) When file names are being printed, "(standard input)" is used for
4377 the standard input, for compatibility with GNU grep. Previously
4378 "<stdin>" was used.
4380 (j) The option --label=xxx can be used to supply a name to be used for
4381 stdin when file names are being printed. There is no short form.
4383 (k) Re-factored the options decoding logic because we are going to add
4384 two more options that take data. Such options can now be given in four
4385 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4387 (l) Added the -A, -B, and -C options for requesting that lines of context
4388 around matches be printed.
4390 (m) Added the -L option to print the names of files that do not contain
4391 any matching lines, that is, the complement of -l.
4393 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4394 continue to scan other files.
4396 (o) The -s option was incorrectly implemented. For compatibility with other
4397 greps, it now suppresses the error message for a non-existent or non-
4398 accessible file (but not the return code). There is a new option called
4399 -q that suppresses the output of matching lines, which was what -s was
4400 previously doing.
4402 (p) Added --include and --exclude options to specify files for inclusion
4403 and exclusion when recursing.
4405 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4406 Hopefully, it now does.
4408 12. Missing cast in pcre_study().
4410 13. Added an "uninstall" target to the makefile.
4412 14. Replaced "extern" in the function prototypes in Makefile.in with
4413 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4414 world, but is set differently for Windows.
4416 15. Added a second compiling function called pcre_compile2(). The only
4417 difference is that it has an extra argument, which is a pointer to an
4418 integer error code. When there is a compile-time failure, this is set
4419 non-zero, in addition to the error test pointer being set to point to an
4420 error message. The new argument may be NULL if no error number is required
4421 (but then you may as well call pcre_compile(), which is now just a
4422 wrapper). This facility is provided because some applications need a
4423 numeric error indication, but it has also enabled me to tidy up the way
4424 compile-time errors are handled in the POSIX wrapper.
4426 16. Added VPATH=.libs to the makefile; this should help when building with one
4427 prefix path and installing with another. (Or so I'm told by someone who
4428 knows more about this stuff than I do.)
4430 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4431 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4432 match everything, including newlines. This is not POSIX-compatible, but
4433 somebody wanted the feature. From pcretest it can be activated by using
4434 both the P and the s flags.
4436 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4438 19. libpcre.pc was being incorrectly installed as executable.
4440 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4441 it now also looks for '\r' so that it will work unmodified on Windows.
4443 21. Added Google's contributed C++ wrapper to the distribution.
4445 22. Added some untidy missing memory free() calls in pcretest, to keep
4446 Electric Fence happy when testing.
4450 Version 5.0 13-Sep-04
4451 ---------------------
4453 1. Internal change: literal characters are no longer packed up into items
4454 containing multiple characters in a single byte-string. Each character
4455 is now matched using a separate opcode. However, there may be more than one
4456 byte in the character in UTF-8 mode.
4458 2. The pcre_callout_block structure has two new fields: pattern_position and
4459 next_item_length. These contain the offset in the pattern to the next match
4460 item, and its length, respectively.
4462 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4463 insertion of callouts before each pattern item. Added the /C option to
4464 pcretest to make use of this.
4466 4. On the advice of a Windows user, the lines
4468 #if defined(_WIN32) || defined(WIN32)
4469 _setmode( _fileno( stdout ), 0x8000 );
4470 #endif /* defined(_WIN32) || defined(WIN32) */
4472 have been added to the source of pcretest. This apparently does useful
4473 magic in relation to line terminators.
4475 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4476 for the benefit of those environments where the "b" makes a difference.
4478 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4479 to know about it. I have put a hack into configure.in that adds in code
4480 to set GCC=yes if CC=icc. This seems to end up at a point in the
4481 generated configure script that is early enough to affect the setting of
4482 compiler options, which is what is needed, but I have no means of testing
4483 whether it really works. (The user who reported this had patched the
4484 generated configure script, which of course I cannot do.)
4486 LATER: After change 22 below (new libtool files), the configure script
4487 seems to know about icc (and also ecc). Therefore, I have commented out
4488 this hack in configure.in.
4490 7. Added support for pkg-config (2 patches were sent in).
4492 8. Negated POSIX character classes that used a combination of internal tables
4493 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4494 [[:^ascii]]. Typically, they would match almost any characters. The other
4495 POSIX classes were not broken in this way.
4497 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4498 to find the match, as PCRE was deluded into thinking that the match had to
4499 start at the start point or following a newline. The same bug applied to
4500 patterns with negative forward assertions or any backward assertions
4501 preceding ".*" at the start, unless the pattern required a fixed first
4502 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4504 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4505 starting at the last subject character, bytes beyond the end of the subject
4506 string were read.
4508 11. Renamed the variable "class" as "classbits" to make life easier for C++
4509 users. (Previously there was a macro definition, but it apparently wasn't
4510 enough.)
4512 12. Added the new field "tables" to the extra data so that tables can be passed
4513 in at exec time, or the internal tables can be re-selected. This allows
4514 a compiled regex to be saved and re-used at a later time by a different
4515 program that might have everything at different addresses.
4517 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4518 -R library as well as a -L library.
4520 14. The debugging options of pcretest (-d on the command line or D on a
4521 pattern) showed incorrect output for anything following an extended class
4522 that contained multibyte characters and which was followed by a quantifier.
4524 15. Added optional support for general category Unicode character properties
4525 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4526 support. It adds about 90K to the size of the library. The meanings of the
4527 inbuilt class escapes such as \d and \s have NOT been changed.
4529 16. Updated pcredemo.c to include calls to free() to release the memory for the
4530 compiled pattern.
4532 17. The generated file chartables.c was being created in the source directory
4533 instead of in the building directory. This caused the build to fail if the
4534 source directory was different from the building directory, and was
4535 read-only.
4537 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4538 file. No doubt somebody will tell me if they don't make sense... Also added
4539 Dan Mooney's comments about building on OpenVMS.
4541 19. Added support for partial matching via the PCRE_PARTIAL option for
4542 pcre_exec() and the \P data escape in pcretest.
4544 20. Extended pcretest with 3 new pattern features:
4546 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4547 write the compiled pattern to the file whose name is "rest-of-line".
4548 This is a straight binary dump of the data, with the saved pointer to
4549 the character tables forced to be NULL. The study data, if any, is
4550 written too. After writing, pcretest reads a new pattern.
4552 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4553 compiled pattern from the given file. There must not be any
4554 occurrences of "<" in the file name (pretty unlikely); if there are,
4555 pcretest will instead treat the initial "<" as a pattern delimiter.
4556 After reading in the pattern, pcretest goes on to read data lines as
4557 usual.
4559 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4560 and 16-bit fields in a compiled pattern, to simulate a pattern that
4561 was compiled on a host of opposite endianness.
4563 21. The pcre-exec() function can now cope with patterns that were compiled on
4564 hosts of opposite endianness, with this restriction:
4566 As for any compiled expression that is saved and used later, the tables
4567 pointer field cannot be preserved; the extra_data field in the arguments
4568 to pcre_exec() should be used to pass in a tables address if a value
4569 other than the default internal tables were used at compile time.
4571 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4572 now diagnosed as an error. Previously, most of the time, a negative number
4573 would have been treated as zero, but if in addition "ovector" was passed as
4574 NULL, a crash could occur.
4576 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4577 new versions from the libtool 1.5 distribution (the last one is a copy of
4578 a file called libtool.m4). This seems to have fixed the need to patch
4579 "configure" to support Darwin 1.3 (which I used to do). However, I still
4580 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4581 workstation).
4583 24. Changed the PCRE licence to be the more standard "BSD" licence.
4586 Version 4.5 01-Dec-03
4587 ---------------------
4589 1. There has been some re-arrangement of the code for the match() function so
4590 that it can be compiled in a version that does not call itself recursively.
4591 Instead, it keeps those local variables that need separate instances for
4592 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4593 needs to "recurse". Keeping track of where control must go is done by means
4594 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4595 hide most of the details from the main code, and operates only if
4596 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4597 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4598 operating.
4600 To make it easier for callers to provide specially tailored get/free
4601 functions for this usage, two new functions, pcre_stack_malloc, and
4602 pcre_stack_free, are used. They are always called in strict stacking order,
4603 and the size of block requested is always the same.
4605 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4606 PCRE has been compiled to use the stack or the heap for recursion. The
4607 -C option of pcretest uses this to show which version is compiled.
4609 A new data escape \S, is added to pcretest; it causes the amounts of store
4610 obtained and freed by both kinds of malloc/free at match time to be added
4611 to the output.
4613 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4614 what's available on my current Linux desktop machine.
4616 3. When matching a UTF-8 string, the test for a valid string at the start has
4617 been extended. If start_offset is not zero, PCRE now checks that it points
4618 to a byte that is the start of a UTF-8 character. If not, it returns
4619 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4620 this is necessary because there may be backward assertions in the pattern.
4621 When matching the same subject several times, it may save resources to use
4622 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4624 4. The code for checking the validity of UTF-8 strings has been tightened so
4625 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4626 containing "overlong sequences".
4628 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4629 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4630 should have been "&", but it just so happened that all the cases this let
4631 through by mistake were picked up later in the function.
4633 6. I had used a variable called "isblank" - this is a C99 function, causing
4634 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4636 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4637 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4638 so the version is shown; (c) stop "make test" echoing "./RunTest".
4640 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4642 9. The prototype for memmove() for systems that don't have it was using
4643 size_t, but the inclusion of the header that defines size_t was later. I've
4644 moved the #includes for the C headers earlier to avoid this.
4646 10. Added some adjustments to the code to make it easier to compiler on certain
4647 special systems:
4649 (a) Some "const" qualifiers were missing.
4650 (b) Added the macro EXPORT before all exported functions; by default this
4651 is defined to be empty.
4652 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4653 that it reads its output file name as an argument instead of writing
4654 to the standard output and assuming this can be redirected.
4656 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4657 class containing characters with values greater than 255, PCRE compilation
4658 went into a loop.
4660 12. A recursive reference to a subpattern that was within another subpattern
4661 that had a minimum quantifier of zero caused PCRE to crash. For example,
4662 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4663 recursion. If the recursively-called subpattern itself had a zero repeat,
4664 that was OK.
4666 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4667 buffer into which it was copied (for escape processing) was still set at
4668 1024, so long lines caused crashes.
4670 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4671 "internal error: code overflow...". This applied to any character class
4672 that was followed by a possessive quantifier.
4674 15. Modified the Makefile to add libpcre.la as a prerequisite for
4675 libpcreposix.la because I was told this is needed for a parallel build to
4676 work.
4678 16. If a pattern that contained .* following optional items at the start was
4679 studied, the wrong optimizing data was generated, leading to matching
4680 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4681 matching string must start with a or b or c. The correct conclusion for
4682 this pattern is that a match can start with any character.
4685 Version 4.4 13-Aug-03
4686 ---------------------
4688 1. In UTF-8 mode, a character class containing characters with values between
4689 127 and 255 was not handled correctly if the compiled pattern was studied.
4690 In fixing this, I have also improved the studying algorithm for such
4691 classes (slightly).
4693 2. Three internal functions had redundant arguments passed to them. Removal
4694 might give a very teeny performance improvement.
4696 3. Documentation bug: the value of the capture_top field in a callout is *one
4697 more than* the number of the hightest numbered captured substring.
4699 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4700 in incorrectly linking with a previously installed version. They now link
4701 explicitly with libpcre.la.
4703 5. configure.in no longer needs to recognize Cygwin specially.
4705 6. A problem in pcre.in for Windows platforms is fixed.
4707 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4708 pcretest, it used to include the size of the study block as part of its
4709 output. Unfortunately, the structure contains a field that has a different
4710 size on different hardware architectures. This meant that the tests that
4711 showed this size failed. As the block is currently always of a fixed size,
4712 this information isn't actually particularly useful in pcretest output, so
4713 I have just removed it.
4715 8. Three pre-processor statements accidentally did not start in column 1.
4716 Sadly, there are *still* compilers around that complain, even though
4717 standard C has not required this for well over a decade. Sigh.
4719 9. In pcretest, the code for checking callouts passed small integers in the
4720 callout_data field, which is a void * field. However, some picky compilers
4721 complained about the casts involved for this on 64-bit systems. Now
4722 pcretest passes the address of the small integer instead, which should get
4723 rid of the warnings.
4725 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4726 both compile and run time, and gives an error if an invalid UTF-8 sequence
4727 is found. There is a option for disabling this check in cases where the
4728 string is known to be correct and/or the maximum performance is wanted.
4730 11. In response to a bug report, I changed one line in Makefile.in from
4732 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4733 to
4734 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4736 to look similar to other lines, but I have no way of telling whether this
4737 is the right thing to do, as I do not use Windows. No doubt I'll get told
4738 if it's wrong...
4741 Version 4.3 21-May-03
4742 ---------------------
4744 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4745 Makefile.
4747 2. Some refactoring to improve the quality of the code:
4749 (i) The utf8_table... variables are now declared "const".
4751 (ii) The code for \cx, which used the "case flipping" table to upper case
4752 lower case letters, now just substracts 32. This is ASCII-specific,
4753 but the whole concept of \cx is ASCII-specific, so it seems
4754 reasonable.
4756 (iii) PCRE was using its character types table to recognize decimal and
4757 hexadecimal digits in the pattern. This is silly, because it handles
4758 only 0-9, a-f, and A-F, but the character types table is locale-
4759 specific, which means strange things might happen. A private
4760 table is now used for this - though it costs 256 bytes, a table is
4761 much faster than multiple explicit tests. Of course, the standard
4762 character types table is still used for matching digits in subject
4763 strings against \d.
4765 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4766 ending in _t are). So I've renamed it as ESC_tee.
4768 3. The first argument for regexec() in the POSIX wrapper should have been
4769 defined as "const".
4771 4. Changed pcretest to use malloc() for its buffers so that they can be
4772 Electric Fenced for debugging.
4774 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4775 to read one or more bytes before the start of the subject string. Often this
4776 had no effect on PCRE's behaviour, but in some circumstances it could
4777 provoke a segmentation fault.
4779 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4780 to try to read one or more bytes before the start of the subject string.
4782 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4783 UTF-8 support could misbehave in various ways if the subject string
4784 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4785 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4786 back over UTF-8 characters.)
4789 Version 4.2 14-Apr-03
4790 ---------------------
4792 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4794 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4795 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4796 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4797 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4798 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4800 Note: automatic setting of the BUILD variables is not yet working
4801 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4802 compile-time but not at link-time
4803 [LINK]: use for linking executables only
4804 make different versions for Windows and non-Windows
4805 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4806 libraries
4807 [LINK_FOR_BUILD]: new variable
4808 [OBJEXT]: use throughout
4809 [EXEEXT]: use throughout
4810 <winshared>: new target
4811 <wininstall>: new target
4812 <dftables.o>: use native compiler
4813 <dftables>: use native linker
4814 <install>: handle Windows platform correctly
4815 <clean>: ditto
4816 <check>: ditto
4817 copy DLL to top builddir before testing
4819 As part of these changes, -no-undefined was removed again. This was reported
4820 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4821 in any case.
4823 3. Some tidies to get rid of compiler warnings:
4825 . In the match_data structure, match_limit was an unsigned long int, whereas
4826 match_call_count was an int. I've made them both unsigned long ints.
4828 . In pcretest the fact that a const uschar * doesn't automatically cast to
4829 a void * provoked a warning.
4831 . Turning on some more compiler warnings threw up some "shadow" variables
4832 and a few more missing casts.
4834 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4835 option, a class that contained a single character with a value between 128
4836 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4838 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4839 option, a class that contained several characters, but with at least one
4840 whose value was between 128 and 255 caused PCRE to crash.
4843 Version 4.1 12-Mar-03
4844 ---------------------
4846 1. Compiling with gcc -pedantic found a couple of places where casts were
4847 needed, and a string in dftables.c that was longer than standard compilers are
4848 required to support.
4850 2. Compiling with Sun's compiler found a few more places where the code could
4851 be tidied up in order to avoid warnings.
4853 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4854 first of these names is deprecated in the latest Autoconf in favour of the name
4855 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4856 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4857 analogy I have changed it to CFLAGS_FOR_BUILD.
4859 4. Added -no-undefined to the linking command in the Makefile, because this is
4860 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4861 linking step for the pcreposix library.
4863 5. PCRE was failing to diagnose the case of two named groups with the same
4864 name.
4866 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4867 literal character that is needed in the subject for a match, and scans along to
4868 ensure that it is present before embarking on the full matching process. This
4869 saves time in cases of nested unlimited repeats that are never going to match.
4870 Problem: the scan can take a lot of time if the subject is very long (e.g.
4871 megabytes), thus penalizing straightforward matches. It is now done only if the
4872 amount of subject to be scanned is less than 1000 bytes.
4874 7. A lesser problem with the same optimization is that it was recording the
4875 first character of an anchored pattern as "needed", thus provoking a search
4876 right along the subject, even when the first match of the pattern was going to
4877 fail. The "needed" character is now not set for anchored patterns, unless it
4878 follows something in the pattern that is of non-fixed length. Thus, it still
4879 fulfils its original purpose of finding quick non-matches in cases of nested
4880 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4883 Version 4.0 17-Feb-03
4884 ---------------------
4886 1. If a comment in an extended regex that started immediately after a meta-item
4887 extended to the end of string, PCRE compiled incorrect data. This could lead to
4888 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4890 2. Moved to autoconf 2.53 and libtool 1.4.2.
4892 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4893 the special perltest8 script is no longer needed - all the tests can be run
4894 from a single perltest script.
4896 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4897 by \s. It has now been removed in PCRE. This means it isn't recognized as
4898 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4899 class [:space:] *does* include VT, thereby creating a mess.
4901 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4902 space and tab.
4904 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4905 its new features into the main test script, reducing the number of scripts.
4907 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4908 were backward compatible, and made the (?i) apply to the whole pattern, as if
4909 /i were given. Now it behaves more logically, and applies the option setting
4910 only to what follows. PCRE has been changed to follow suit. However, if it
4911 finds options settings right at the start of the pattern, it extracts them into
4912 the global options, as before. Thus, they show up in the info data.
4914 8. Added support for the \Q...\E escape sequence. Characters in between are
4915 treated as literals. This is slightly different from Perl in that $ and @ are
4916 also handled as literals inside the quotes. In Perl, they will cause variable
4917 interpolation. Note the following examples:
4919 Pattern PCRE matches Perl matches
4921 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4922 \Qabc\$xyz\E abc\$xyz abc\$xyz
4923 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4925 For compatibility with Perl, \Q...\E sequences are recognized inside character
4926 classes as well as outside them.
4928 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4929 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4930 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4931 signed/unsigned warnings.
4933 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4934 option for pcretest, so I've replaced it by a simple function that does just
4935 that job.
4937 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4938 "pcregrep -".
4940 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4941 Java package. This provides some syntactic sugar for simple cases of what my
4942 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4943 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4944 item, you can use this simplified notation. Note that only makes sense with
4945 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4946 greediness, whatever the setting of the PCRE_UNGREEDY option.
4948 13. A change of greediness default within a pattern was not taking effect at
4949 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4950 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4951 was abstracted outside.
4953 14. PCRE now supports the \G assertion. It is true when the current matching
4954 position is at the start point of the match. This differs from \A when the
4955 starting offset is non-zero. Used with the /g option of pcretest (or similar
4956 code), it works in the same way as it does for Perl's /g option. If all
4957 alternatives of a regex begin with \G, the expression is anchored to the start
4958 match position, and the "anchored" flag is set in the compiled expression.
4960 15. Some bugs concerning the handling of certain option changes within patterns
4961 have been fixed. These applied to options other than (?ims). For example,
4962 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4963 been the other way round. Some of this was related to change 7 above.
4965 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4966 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4967 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4968 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4970 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4971 mode. Unlike ".", it always matches newline, whatever the setting of
4972 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4973 assertions. Perl allows it, but it doesn't (in general) work because it can't
4974 calculate the length of the lookbehind. At least, that's the case for Perl
4975 5.8.0 - I've been told they are going to document that it doesn't work in
4976 future.
4978 18. Added an error diagnosis for escapes that PCRE does not support: these are
4979 \L, \l, \N, \P, \p, \U, \u, and \X.
4981 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4982 reading past the end of the pattern in cases such as /[abcd/.
4984 20. PCRE was getting more memory than necessary for patterns with classes that
4985 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4987 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4988 compiling PCRE for use with Virtual Pascal.
4990 22. Small fix to the Makefile to make it work properly if the build is done
4991 outside the source tree.
4993 23. Added a new extension: a condition to go with recursion. If a conditional
4994 subpattern starts with (?(R) the "true" branch is used if recursion has
4995 happened, whereas the "false" branch is used only at the top level.
4997 24. When there was a very long string of literal characters (over 255 bytes
4998 without UTF support, over 250 bytes with UTF support), the computation of how
4999 much memory was required could be incorrect, leading to segfaults or other
5000 strange effects.
5002 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5003 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5004 there was a subsequent back reference to those brackets. This meant that, for
5005 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5006 possible to check for precisely this case. All we can do is abandon the
5007 optimization if .* occurs inside capturing brackets when there are any back
5008 references whatsoever. (See below for a better fix that came later.)
5010 26. The handling of the optimization for finding the first character of a
5011 non-anchored pattern, and for finding a character that is required later in the
5012 match were failing in some cases. This didn't break the matching; it just
5013 failed to optimize when it could. The way this is done has been re-implemented.
5015 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5017 28. Added a new feature that provides some of the functionality that Perl
5018 provides with (?{...}). The facility is termed a "callout". The way it is done
5019 in PCRE is for the caller to provide an optional function, by setting
5020 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5021 global variable. By default it is unset, which disables all calling out. To get
5022 the function called, the regex must include (?C) at appropriate points. This
5023 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5024 This provides a means of identifying different callout points. When PCRE
5025 reaches such a point in the regex, if pcre_callout has been set, the external
5026 function is called. It is provided with data in a structure called
5027 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5028 matching continues; if it returns a non-zero value, the match at the current
5029 point fails. However, backtracking will occur if possible. [This was changed
5030 later and other features added - see item 49 below.]
5032 29. pcretest is upgraded to test the callout functionality. It provides a
5033 callout function that displays information. By default, it shows the start of
5034 the match and the current position in the text. There are some new data escapes
5035 to vary what happens:
5037 \C+ in addition, show current contents of captured substrings
5038 \C- do not supply a callout function
5039 \C!n return 1 when callout number n is reached
5040 \C!n!m return 1 when callout number n is reached for the mth time
5042 30. If pcregrep was called with the -l option and just a single file name, it
5043 output "<stdin>" if a match was found, instead of the file name.
5045 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5046 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5047 pcre_exec(). This saves a malloc/free per call. The default value of
5048 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5049 when configuring.
5051 32. The default maximum size of a compiled pattern is 64K. There have been a
5052 few cases of people hitting this limit. The code now uses macros to handle the
5053 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5054 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5055 configuring. Tests 2 and 5 work only with 2-byte links because they output
5056 debugging information about compiled patterns.
5058 33. Internal code re-arrangements:
5060 (a) Moved the debugging function for printing out a compiled regex into
5061 its own source file (printint.c) and used #include to pull it into
5062 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5063 separate copies.
5065 (b) Defined the list of op-code names for debugging as a macro in
5066 internal.h so that it is next to the definition of the opcodes.
5068 (c) Defined a table of op-code lengths for simpler skipping along compiled
5069 code. This is again a macro in internal.h so that it is next to the
5070 definition of the opcodes.
5072 34. Added support for recursive calls to individual subpatterns, along the
5073 lines of Robin Houston's patch (but implemented somewhat differently).
5075 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5076 allow it to read and process whole directories in Win32. This code was
5077 contributed by Lionel Fourquaux; it has not been tested by me.
5079 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5080 used to name a group. Names consist of alphanumerics and underscores, and must
5081 be unique. Back references use the syntax (?P=name) and recursive calls use
5082 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5083 numbers. The function pcre_fullinfo() can be used after compilation to extract
5084 a name/number map. There are three relevant calls:
5086 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5087 PCRE_INFO_NAMECOUNT yields the number of entries
5088 PCRE_INFO_NAMETABLE yields a pointer to the map.
5090 The map is a vector of fixed-size entries. The size of each entry depends on
5091 the length of the longest name used. The first two bytes of each entry are the
5092 group number, most significant byte first. There follows the corresponding
5093 name, zero terminated. The names are in alphabetical order.
5095 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5096 case instead of 255. Making it the same both with and without UTF-8 support
5097 means that the same test output works with both.
5099 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5100 calling malloc() with a zero argument.
5102 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5103 optimization. I've improved things by keeping a bitmap of backreferences with
5104 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5105 fact referenced, the optimization can be applied. It is unlikely that a
5106 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5107 the match to follow \n) will appear inside brackets with a number greater than
5108 31, but if it does, any back reference > 31 suppresses the optimization.
5110 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5111 of disabling numbered capturing parentheses. Any opening parenthesis that is
5112 not followed by ? behaves as if it were followed by ?: but named parentheses
5113 can still be used for capturing (and they will acquire numbers in the usual
5114 way).
5116 41. Redesigned the return codes from the match() function into yes/no/error so
5117 that errors can be passed back from deep inside the nested calls. A malloc
5118 failure while inside a recursive subpattern call now causes the
5119 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5121 42. It is now possible to set a limit on the number of times the match()
5122 function is called in a call to pcre_exec(). This facility makes it possible to
5123 limit the amount of recursion and backtracking, though not in a directly
5124 obvious way, because the match() function is used in a number of different
5125 circumstances. The count starts from zero for each position in the subject
5126 string (for non-anchored patterns). The default limit is, for compatibility, a
5127 large number, namely 10 000 000. You can change this in two ways:
5129 (a) When configuring PCRE before making, you can use --with-match-limit=n
5130 to set a default value for the compiled library.
5132 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5133 a different value is set. See 45 below.
5135 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5137 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5138 of things that can be changed at compile time. The first argument specifies
5139 what is wanted and the second points to where the information is to be placed.
5140 The current list of available information is:
5144 The output is an integer that is set to one if UTF-8 support is available;
5145 otherwise it is set to zero.
5149 The output is an integer that it set to the value of the code that is used for
5150 newline. It is either LF (10) or CR (13).
5154 The output is an integer that contains the number of bytes used for internal
5155 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5159 The output is an integer that contains the threshold above which the POSIX
5160 interface uses malloc() for output vectors. See item 31 above.
5164 The output is an unsigned integer that contains the default limit of the number
5165 of match() calls in a pcre_exec() execution. See 42 above.
5167 44. pcretest has been upgraded by the addition of the -C option. This causes it
5168 to extract all the available output from the new pcre_config() function, and to
5169 output it. The program then exits immediately.
5171 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5172 order to support additional features. One way would have been to define
5173 pcre_exec2() (for example) with extra arguments, but this would not have been
5174 extensible, and would also have required all calls to the original function to
5175 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5176 is used for passing in "extra" data from pcre_study().
5178 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5179 contains the following fields:
5181 flags a bitmap indicating which of the following fields are set
5182 study_data opaque data from pcre_study()
5183 match_limit a way of specifying a limit on match() calls for a specific
5184 call to pcre_exec()
5185 callout_data data for callouts (see 49 below)
5187 The flag bits are also defined in pcre.h, and are
5193 The pcre_study() function now returns one of these new pcre_extra blocks, with
5194 the actual study data pointed to by the study_data field, and the
5195 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5196 before. That is, this change is entirely upwards-compatible and requires no
5197 change to existing code.
5199 If you want to pass in additional data to pcre_exec(), you can either place it
5200 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5201 block.
5203 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5204 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5205 times with different match limits, until it finds the minimum value needed for
5206 pcre_exec() to complete. The value is then output. This can be instructive; for
5207 most simple matches the number is quite small, but for pathological cases it
5208 gets very large very quickly.
5210 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5211 returns the size of the data block pointed to by the study_data field in a
5212 pcre_extra block, that is, the value that was passed as the argument to
5213 pcre_malloc() when PCRE was getting memory in which to place the information
5214 created by pcre_study(). The fourth argument should point to a size_t variable.
5215 pcretest has been extended so that this information is shown after a successful
5216 pcre_study() call when information about the compiled regex is being displayed.
5218 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5219 because what follows is always an absolute path. (Later: it turns out that this
5220 is more than cosmetic for MinGW, because it doesn't like empty path
5221 components.)
5223 49. Some changes have been made to the callout feature (see 28 above):
5225 (i) A callout function now has three choices for what it returns:
5227 0 => success, carry on matching
5228 > 0 => failure at this point, but backtrack if possible
5229 < 0 => serious error, return this value from pcre_exec()
5231 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5232 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5233 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5234 use by callout functions. It will never be used by PCRE itself.
5236 (ii) The pcre_extra structure (see 45 above) has a void * field called
5237 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5238 pcre_callout_block structure has a field of the same name. The contents of
5239 the field passed in the pcre_extra structure are passed to the callout
5240 function in the corresponding field in the callout block. This makes it
5241 easier to use the same callout-containing regex from multiple threads. For
5242 testing, the pcretest program has a new data escape
5244 \C*n pass the number n (may be negative) as callout_data
5246 If the callout function in pcretest receives a non-zero value as
5247 callout_data, it returns that value.
5249 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5250 there were some redundant $(CFLAGS) in commands that are now specified as
5251 $(LINK), which already includes $(CFLAGS).
5253 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5254 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5255 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5256 one-byte characters throughout. Note that case-insensitive matching applies
5257 only to characters whose values are less than 256. PCRE doesn't support the
5258 notion of cases for higher-valued characters.
5260 (i) A character class whose characters are all within 0-255 is handled as
5261 a bit map, and the map is inverted for negative classes. Previously, a
5262 character > 255 always failed to match such a class; however it should
5263 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5265 (ii) A negated character class with a single character < 255 is coded as
5266 "not this character" (OP_NOT). This wasn't working properly when the test
5267 character was multibyte, either singly or repeated.
5269 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5270 mode, for example: \x{100}{2,3}.
5272 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5273 singly or repeated) now correctly test multibyte characters. However,
5274 PCRE doesn't recognize any characters with values greater than 255 as
5275 digits, spaces, or word characters. Such characters always match \D, \S,
5276 and \W, and never match \d, \s, or \w.
5278 (v) Classes may now contain characters and character ranges with values
5279 greater than 255. For example: [ab\x{100}-\x{400}].
5281 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5282 PCRE in UTF-8 mode.
5284 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5285 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5286 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5287 value.)
5289 53. The single man page has become too large. I have therefore split it up into
5290 a number of separate man pages. These also give rise to individual HTML pages;
5291 these are now put in a separate directory, and there is an index.html page that
5292 lists them all. Some hyperlinking between the pages has been installed.
5294 54. Added convenience functions for handling named capturing parentheses.
5296 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5297 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5298 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5299 are faulted.
5301 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5302 calling configure. These values are used when compiling the dftables.c program
5303 which is run to generate the source of the default character tables. They
5304 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5305 you will need to set these values.
5307 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5310 Version 3.9 02-Jan-02
5311 ---------------------
5313 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5315 2. If --disable-static was given, the building process failed when trying to
5316 build pcretest and pcregrep. (For some reason it was using libtool to compile
5317 them, which is not right, as they aren't part of the library.)
5320 Version 3.8 18-Dec-01
5321 ---------------------
5323 1. The experimental UTF-8 code was completely screwed up. It was packing the
5324 bytes in the wrong order. How dumb can you get?
5327 Version 3.7 29-Oct-01
5328 ---------------------
5330 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5331 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5332 this didn't happen under Solaris 8, where I normally test things.
5334 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5335 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5336 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5337 this fix an hour or so after the initial 3.7 release.)
5340 Version 3.6 23-Oct-01
5341 ---------------------
5343 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5344 offsets passed as NULL with zero offset count.
5346 2. The config.guess and config.sub files had not been updated when I moved to
5347 the latest autoconf.
5350 Version 3.5 15-Aug-01
5351 ---------------------
5353 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5354 had been forgotten.
5356 2. By using declared but undefined structures, we can avoid using "void"
5357 definitions in pcre.h while keeping the internal definitions of the structures
5358 private.
5360 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5361 user point of view, this means that both static and shared libraries are built
5362 by default, but this can be individually controlled. More of the work of
5363 handling this static/shared cases is now inside libtool instead of PCRE's make
5364 file.
5366 4. The pcretest utility is now installed along with pcregrep because it is
5367 useful for users (to test regexs) and by doing this, it automatically gets
5368 relinked by libtool. The documentation has been turned into a man page, so
5369 there are now .1, .txt, and .html versions in /doc.
5371 5. Upgrades to pcregrep:
5372 (i) Added long-form option names like gnu grep.
5373 (ii) Added --help to list all options with an explanatory phrase.
5374 (iii) Added -r, --recursive to recurse into sub-directories.
5375 (iv) Added -f, --file to read patterns from a file.
5377 6. pcre_exec() was referring to its "code" argument before testing that
5378 argument for NULL (and giving an error if it was NULL).
5380 7. Upgraded Makefile.in to allow for compiling in a different directory from
5381 the source directory.
5383 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5384 options bits, the pointer it was passed was to an int instead of to an unsigned
5385 long int. This mattered only on 64-bit systems.
5387 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5388 generated) instead of pcre.in, which it its source. Also made the same change
5389 in several of the .c files.
5391 10. A new release of gcc defines printf() as a macro, which broke pcretest
5392 because it had an ifdef in the middle of a string argument for printf(). Fixed
5393 by using separate calls to printf().
5395 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5396 script, to force use of CR or LF instead of \n in the source. On non-Unix
5397 systems, the value can be set in config.h.
5399 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5400 absolute limit. Changed the text of the error message to make this clear, and
5401 likewise updated the man page.
5403 13. The limit of 99 on the number of capturing subpatterns has been removed.
5404 The new limit is 65535, which I hope will not be a "real" limit.
5407 Version 3.4 22-Aug-00
5408 ---------------------
5410 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5412 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5415 Version 3.3 01-Aug-00
5416 ---------------------
5418 1. If an octal character was given, but the value was greater than \377, it
5419 was not getting masked to the least significant bits, as documented. This could
5420 lead to crashes in some systems.
5422 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5423 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5425 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5426 These just pass their arguments on to (pcre_free)(), but they are provided
5427 because some uses of PCRE bind it to non-C systems that can call its functions,
5428 but cannot call free() or pcre_free() directly.
5430 4. Add "make test" as a synonym for "make check". Corrected some comments in
5431 the Makefile.
5433 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5434 Makefile.
5436 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5437 command called pgrep for grepping around the active processes.
5439 7. Added the beginnings of support for UTF-8 character strings.
5441 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5442 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5443 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5444 out for the ar command.)
5447 Version 3.2 12-May-00
5448 ---------------------
5450 This is purely a bug fixing release.
5452 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5453 of ZA. This was just one example of several cases that could provoke this bug,
5454 which was introduced by change 9 of version 2.00. The code for breaking
5455 infinite loops after an iteration that matches an empty string was't working
5456 correctly.
5458 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5459 when matched against abbab (for example). After matching an empty string, it
5460 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5461 caused it to match further down the string than it should.
5463 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5464 was there because it doesn't seem to be needed, and it causes trouble on some
5465 systems, as it is not a Standard C header. It has been removed.
5467 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5468 were reported on the Macintosh. The changes were from
5470 while ((c = *(++ptr)) != 0 && c != '\n');
5471 to
5472 while ((c = *(++ptr)) != 0 && c != '\n') ;
5474 Totally extraordinary, but if that's what it takes...
5476 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5477 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5478 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5479 assumes the way PCRE uses memmove() (always moving upwards).
5481 6. PCRE is being used in one environment where strchr() is not available. There
5482 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5483 faster code anyway.
5486 Version 3.1 09-Feb-00
5487 ---------------------
5489 The only change in this release is the fixing of some bugs in Makefile.in for
5490 the "install" target:
5492 (1) It was failing to install pcreposix.h.
5494 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5497 Version 3.0 01-Feb-00
5498 ---------------------
5500 1. Add support for the /+ modifier to perltest (to output $` like it does in
5501 pcretest).
5503 2. Add support for the /g modifier to perltest.
5505 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5506 matches null strings.
5508 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5509 pattern. Perl treats empty patterns specially - it reuses the most recent
5510 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5511 effect.
5513 5. The POSIX interface was broken in that it was just handing over the POSIX
5514 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5515 required a bigger vector, with some working space on the end. This means that
5516 the POSIX wrapper now has to get and free some memory, and copy the results.
5518 6. Added some simple autoconf support, placing the test data and the
5519 documentation in separate directories, re-organizing some of the
5520 information files, and making it build pcre-config (a GNU standard). Also added
5521 libtool support for building PCRE as a shared library, which is now the
5522 default.
5524 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5525 09 are not valid octal constants. Single digits will be used for minor values
5526 less than 10.
5528 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5529 existing programs that set these in the POSIX interface can use PCRE without
5530 modification.
5532 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5533 return all that pcre_info() returns, plus additional data. The pcre_info()
5534 function is retained for compatibility, but is considered to be obsolete.
5536 10. Added experimental recursion feature (?R) to handle one common case that
5537 Perl 5.6 will be able to do with (?p{...}).
5539 11. Added support for POSIX character classes like [:alpha:], which Perl is
5540 adopting.
5543 Version 2.08 31-Aug-99
5544 ----------------------
5546 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5547 trying to match at the startoffset position, but instead was moving forward to
5548 the next newline as if a previous match had failed.
5550 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5551 and could get into a loop if a null string was matched other than at the start
5552 of the subject.
5554 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5555 be distinguished at compile time, and for completeness also added PCRE_DATE.
5557 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5558 in GnuWin32 environments.
5561 Version 2.07 29-Jul-99
5562 ----------------------
5564 1. The documentation is now supplied in plain text form and HTML as well as in
5565 the form of man page sources.
5567 2. C++ compilers don't like assigning (void *) values to other pointer types.
5568 In particular this affects malloc(). Although there is no problem in Standard
5569 C, I've put in casts to keep C++ compilers happy.
5571 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5572 should be (const char *).
5574 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5575 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5576 However, I haven't made this a standard facility. The documentation doesn't
5577 mention it, and the Makefile doesn't support it.
5579 5. The Makefile now contains an "install" target, with editable destinations at
5580 the top of the file. The pcretest program is not installed.
5582 6. pgrep -V now gives the PCRE version number and date.
5584 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5585 causing the entire string to be ignored, instead of just the last character.
5587 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5588 non-matching string, it can take a very, very long time, even for strings of
5589 quite modest length, because of the nested recursion. PCRE now does better in
5590 some of these cases. It does this by remembering the last required literal
5591 character in the pattern, and pre-searching the subject to ensure it is present
5592 before running the real match. In other words, it applies a heuristic to detect
5593 some types of certain failure quickly, and in the above example, if presented
5594 with a string that has no trailing " it gives "no match" very quickly.
5596 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5597 other alternatives are tried instead.
5600 Version 2.06 09-Jun-99
5601 ----------------------
5603 1. Change pcretest's output for amount of store used to show just the code
5604 space, because the remainder (the data block) varies in size between 32-bit and
5605 64-bit systems.
5607 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5608 start matching at. This allows lookbehinds to work when searching for multiple
5609 occurrences in a string.
5611 3. Added additional options to pcretest for testing multiple occurrences:
5613 /+ outputs the rest of the string that follows a match
5614 /g loops for multiple occurrences, using the new startoffset argument
5615 /G loops for multiple occurrences by passing an incremented pointer
5617 4. PCRE wasn't doing the "first character" optimization for patterns starting
5618 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5619 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5620 the letter 'x'. On long subject strings, this gives a significant speed-up.
5623 Version 2.05 21-Apr-99
5624 ----------------------
5626 1. Changed the type of magic_number from int to long int so that it works
5627 properly on 16-bit systems.
5629 2. Fixed a bug which caused patterns starting with .* not to work correctly
5630 when the subject string contained newline characters. PCRE was assuming
5631 anchoring for such patterns in all cases, which is not correct because .* will
5632 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5633 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5634 must be retried after every newline in the subject.
5637 Version 2.04 18-Feb-99
5638 ----------------------
5640 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5641 computation of the store needed to hold the pattern was incorrect (too large).
5642 If such patterns were nested a few deep, this could multiply and become a real
5643 problem.
5645 2. Added /M option to pcretest to show the memory requirement of a specific
5646 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5648 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5649 compiled in such a way that the backtracking after subsequent failure was
5650 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5651 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5654 Version 2.03 02-Feb-99
5655 ----------------------
5657 1. Fixed typo and small mistake in man page.
5659 2. Added 4th condition (GPL supersedes if conflict) and created separate
5660 LICENCE file containing the conditions.
5662 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5663 Perl, that is the internal \ allows the delimiter to be included in the
5664 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5665 the final delimiter, add \ to the end of the pattern (to test the error).
5667 4. Added the convenience functions for extracting substrings after a successful
5668 match. Updated pcretest to make it able to test these functions.
5671 Version 2.02 14-Jan-99
5672 ----------------------
5674 1. Initialized the working variables associated with each extraction so that
5675 their saving and restoring doesn't refer to uninitialized store.
5677 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5678 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5679 fix the problem.
5681 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5682 calls, and wasn't printing the correct value for compiling calls. Increased the
5683 default value of LOOPREPEAT, and the number of significant figures in the
5684 times.
5686 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5688 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5689 a building problem on Windows NT with a FAT file system.
5692 Version 2.01 21-Oct-98
5693 ----------------------
5695 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5696 to character tables built by pcre_maketables() in the current locale. If NULL
5697 is passed, the default tables are used.
5700 Version 2.00 24-Sep-98
5701 ----------------------
5703 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5704 it any more.
5706 2. Allow quantification of (?>) groups, and make it work correctly.
5708 3. The first character computation wasn't working for (?>) groups.
5710 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5711 end of the subject) and add 5.005's \z, which really does match only at the
5712 very end of the subject.
5714 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5716 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5717 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5718 localized options. All options to pcre_study() were also removed.
5720 7. Add other new features from 5.005:
5722 $(?<= positive lookbehind
5723 $(?<! negative lookbehind
5724 (?imsx-imsx) added the unsetting capability
5725 such a setting is global if at outer level; local otherwise
5726 (?imsx-imsx:) non-capturing groups with option setting
5727 (?(cond)re|re) conditional pattern matching
5729 A backreference to itself in a repeated group matches the previous
5730 captured string.
5732 8. General tidying up of studying (both automatic and via "study")
5733 consequential on the addition of new assertions.
5735 9. As in 5.005, unlimited repeated groups that could match an empty substring
5736 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5737 runtime if any iteration does actually match an empty substring.
5739 10. Include the RunTest script in the distribution.
5741 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5742 discrepancies, some of which were old and were also with respect to 5.004. They
5743 have now been fixed.
5746 Version 1.09 28-Apr-98
5747 ----------------------
5749 1. A negated single character class followed by a quantifier with a minimum
5750 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5751 program crashes, or just wrong answers. This did not apply to negated classes
5752 containing more than one character, or to minima other than one.
5755 Version 1.08 27-Mar-98
5756 ----------------------
5758 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5760 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5761 latter must appear before anything that relies on it in the pattern.
5764 Version 1.07 16-Feb-98
5765 ----------------------
5767 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5768 repeat of a potentially empty string).
5771 Version 1.06 23-Jan-98
5772 ----------------------
5774 1. Added Markus Oberhumer's little patches for C++.
5776 2. Literal strings longer than 255 characters were broken.
5779 Version 1.05 23-Dec-97
5780 ----------------------
5782 1. Negated character classes containing more than one character were failing if
5783 PCRE_CASELESS was set at run time.
5786 Version 1.04 19-Dec-97
5787 ----------------------
5789 1. Corrected the man page, where some "const" qualifiers had been omitted.
5791 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5792 input syntax.
5794 3. Fixed memory leak which occurred when a regex with back references was
5795 matched with an offsets vector that wasn't big enough. The temporary memory
5796 that is used in this case wasn't being freed if the match failed.
5798 4. Tidied pcretest to ensure it frees memory that it gets.
5800 5. Temporary memory was being obtained in the case where the passed offsets
5801 vector was exactly big enough.
5803 6. Corrected definition of offsetof() from change 5 below.
5805 7. I had screwed up change 6 below and broken the rules for the use of
5806 setjmp(). Now fixed.
5809 Version 1.03 18-Dec-97
5810 ----------------------
5812 1. A erroneous regex with a missing opening parenthesis was correctly
5813 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5814 on some systems.
5816 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5817 it was reported that one broken compiler failed on the former because "code" is
5818 also an independent variable.
5820 3. The erroneous regex a[]b caused an array overrun reference.
5822 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5823 fail on data ending with that character. (It was going on too far, and checking
5824 the next character, typically a binary zero.) This was specific to the
5825 optimized code for single-character negative classes.
5827 5. Added a contributed patch from the TIN world which does the following:
5829 + Add an undef for memmove, in case the the system defines a macro for it.
5831 + Add a definition of offsetof(), in case there isn't one. (I don't know
5832 the reason behind this - offsetof() is part of the ANSI standard - but
5833 it does no harm).
5835 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5836 most of the places where whitespace preceded '#'. I have given up and
5837 allowed the remaining 2 cases to be at the margin.
5839 + Rename some variables in pcre to eliminate shadowing. This seems very
5840 pedantic, but does no harm, of course.
5842 6. Moved the call to setjmp() into its own function, to get rid of warnings
5843 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5845 7. Constructs such as \d{8,} were compiling into the equivalent of
5846 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5847 outcome, but in this particular case used more store than had been allocated,
5848 which caused the bug to be discovered because it threw up an internal error.
5850 8. The debugging code in both pcre and pcretest for outputting the compiled
5851 form of a regex was going wrong in the case of back references followed by
5852 curly-bracketed repeats.
5855 Version 1.02 12-Dec-97
5856 ----------------------
5858 1. Typos in pcre.3 and comments in the source fixed.
5860 2. Applied a contributed patch to get rid of places where it used to remove
5861 'const' from variables, and fixed some signed/unsigned and uninitialized
5862 variable warnings.
5864 3. Added the "runtest" target to Makefile.
5866 4. Set default compiler flag to -O2 rather than just -O.
5869 Version 1.01 19-Nov-97
5870 ----------------------
5872 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5873 like /([ab]*)*/, that is, for classes with more than one character in them.
5875 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5876 as /((?>a*))*/ (a PCRE_EXTRA facility).
5879 Version 1.00 18-Nov-97
5880 ----------------------
5882 1. Added compile-time macros to support systems such as SunOS4 which don't have
5883 memmove() or strerror() but have other things that can be used instead.
5885 2. Arranged that "make clean" removes the executables.
5888 Version 0.99 27-Oct-97
5889 ----------------------
5891 1. Fixed bug in code for optimizing classes with only one character. It was
5892 initializing a 32-byte map regardless, which could cause it to run off the end
5893 of the memory it had got.
5895 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5898 Version 0.98 22-Oct-97
5899 ----------------------
5901 1. Fixed bug in code for handling temporary memory usage when there are more
5902 back references than supplied space in the ovector. This could cause segfaults.
5905 Version 0.97 21-Oct-97
5906 ----------------------
5908 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5910 2. Optimized negated single characters not to use a bit map.
5912 3. Brought error texts together as macro definitions; clarified some of them;
5913 fixed one that was wrong - it said "range out of order" when it meant "invalid
5914 escape sequence".
5916 4. Changed some char * arguments to const char *.
5918 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5920 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5921 pcretest.
5924 Version 0.96 16-Oct-97
5925 ----------------------
5927 1. Added a simple "pgrep" utility to the distribution.
5929 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5930 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5931 where "ddd" means "one or more decimal digits".
5933 3. Fixed serious bug. If a pattern had a back reference, but the call to
5934 pcre_exec() didn't supply a large enough ovector to record the related
5935 identifying subpattern, the match always failed. PCRE now remembers the number
5936 of the largest back reference, and gets some temporary memory in which to save
5937 the offsets during matching if necessary, in order to ensure that
5938 backreferences always work.
5940 4. Increased the compatibility with Perl in a number of ways:
5942 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5943 to request this handling. The option can be set at compile or exec time.
5945 (b) $ matches before a terminating newline by default; an option
5946 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5947 mode). The option can be set at compile or exec time.
5949 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5950 the same as Perl's. If the decimal number it represents is less than 10
5951 or there aren't that many previous left capturing parentheses, an octal
5952 escape is read. Inside a character class, it's always an octal escape,
5953 even if it is a single digit.
5955 (d) An escaped but undefined alphabetic character is taken as a literal,
5956 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5957 escapes.
5959 (e) {0} is now permitted. (The previous item is removed from the compiled
5960 pattern).
5962 5. Changed all the names of code files so that the basic parts are no longer
5963 than 10 characters, and abolished the teeny "globals.c" file.
5965 6. Changed the handling of character classes; they are now done with a 32-byte
5966 bit map always.
5968 7. Added the -d and /D options to pcretest to make it possible to look at the
5969 internals of compilation without having to recompile pcre.
5972 Version 0.95 23-Sep-97
5973 ----------------------
5975 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5976 \x20 at the start of a run of normal characters. These were being treated as
5977 real characters, instead of the source characters being re-checked.
5980 Version 0.94 18-Sep-97
5981 ----------------------
5983 1. The functions are now thread-safe, with the caveat that the global variables
5984 containing pointers to malloc() and free() or alternative functions are the
5985 same for all threads.
5987 2. Get pcre_study() to generate a bitmap of initial characters for non-
5988 anchored patterns when this is possible, and use it if passed to pcre_exec().
5991 Version 0.93 15-Sep-97
5992 ----------------------
5994 1. /(b)|(:+)/ was computing an incorrect first character.
5996 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5997 but not actually doing anything yet.
5999 3. Treat "-" characters in classes that cannot be part of ranges as literals,
6000 as Perl does (e.g. [-az] or [az-]).
6002 4. Set the anchored flag if a branch starts with .* or .*? because that tests
6003 all possible positions.
6005 5. Split up into different modules to avoid including unneeded functions in a
6006 compiled binary. However, compile and exec are still in one module. The "study"
6007 function is split off.
6009 6. The character tables are now in a separate module whose source is generated
6010 by an auxiliary program - but can then be edited by hand if required. There are
6011 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
6012 toupper() in the code.
6014 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
6015 make them global. Abolish the function for setting them, as the caller can now
6016 set them directly.
6019 Version 0.92 11-Sep-97
6020 ----------------------
6022 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
6023 (e.g. /a{1,3}/) was broken (I mis-optimized it).
6025 2. Caseless matching was not working in character classes if the characters in
6026 the pattern were in upper case.
6028 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
6030 4. Make PCRE_ANCHORED public and accept as a compile option.
6032 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
6033 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
6034 pass them.
6036 6. Give an error if bad option bits passed at compile or run time.
6038 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
6039 pcretest to cause it to pass that flag.
6041 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
6042 options, and the first character, if set.
6044 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
6047 Version 0.91 10-Sep-97
6048 ----------------------
6050 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
6051 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
6053 2. PCRE was looping on encountering an indefinitely repeated back reference to
6054 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what