ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1642 - (show annotations)
Tue Mar 1 12:48:35 2016 UTC (3 years, 11 months ago) by ph10
File size: 289025 byte(s)
Add valgrind suppressions file to RunTest so that issued that are provoked by 
the SSE2 instruction set when JIT is in use are ignored.
1 ChangeLog for PCRE
2 ------------------
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
7 Version 8.39 xx-xxxxxx-201x
8 ---------------------------
10 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
11 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
12 misbehaved. This bug was found by the LLVM fuzzer.
14 2. Similar to the above, if an isolated \E was present between an item and its
15 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
16 bug was found by the LLVM fuzzer.
18 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
19 working correctly in UCP mode.
21 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
22 was set when the pmatch argument was NULL. It now returns REG_INVARG.
24 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
26 6. An empty \Q\E sequence between an item and its qualifier caused
27 pcre_compile() to misbehave when auto callouts were enabled. This bug was
28 found by the LLVM fuzzer.
30 7. If a pattern that was compiled with PCRE_EXTENDED started with white
31 space or a #-type comment that was followed by (?-x), which turns off
32 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
33 pcre_compile() assumed that (?-x) applied to the whole pattern and
34 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
36 8. An call of pcre_copy_named_substring() for a named substring whose number
37 was greater than the space in the ovector could cause a crash.
39 9. Yet another buffer overflow bug involved duplicate named groups with a
40 group that reset capture numbers (compare 8.38/7 below). Once again, I have
41 just allowed for more memory, even if not needed. (A proper fix is
42 implemented in PCRE2, but it involves a lot of refactoring.)
44 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
45 start of the match to be earlier than the end.
47 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
49 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
50 assertion, caused pcretest to generate incorrect output, and also to read
51 uninitialized memory (detected by ASAN or valgrind).
53 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
54 nested set of parentheses of sufficient size caused an overflow of the
55 compiling workspace (which was diagnosed, but of course is not desirable).
57 14. And yet another buffer overflow bug involving duplicate named groups, this
58 time nested, with a nested back reference. Yet again, I have just allowed
59 for more memory, because anything more needs all the refactoring that has
60 been done for PCRE2. An example pattern that provoked this bug is:
61 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
62 registered as CVE-2016-1283.
64 15. pcretest went into a loop if global matching was requested with an ovector
65 size less than 2. It now gives an error message. This bug was found by
66 afl-fuzz.
68 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
69 ("assertion expected") when (?(?C) was not followed by an opening
70 parenthesis.
72 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
73 actually affect anything, by sheer luck.
75 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
76 static compilation.
78 19. Modified the RunTest script to incorporate a valgrind suppressions file so
79 that certain errors, provoked by the SSE2 instruction set when JIT is used,
80 are ignored.
83 Version 8.38 23-November-2015
84 -----------------------------
86 1. If a group that contained a recursive back reference also contained a
87 forward reference subroutine call followed by a non-forward-reference
88 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
89 compile correct code, leading to undefined behaviour or an internally
90 detected error. This bug was discovered by the LLVM fuzzer.
92 2. Quantification of certain items (e.g. atomic back references) could cause
93 incorrect code to be compiled when recursive forward references were
94 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
95 This bug was discovered by the LLVM fuzzer.
97 3. A repeated conditional group whose condition was a reference by name caused
98 a buffer overflow if there was more than one group with the given name.
99 This bug was discovered by the LLVM fuzzer.
101 4. A recursive back reference by name within a group that had the same name as
102 another group caused a buffer overflow. For example:
103 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
105 5. A forward reference by name to a group whose number is the same as the
106 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
107 a buffer overflow at compile time. This bug was discovered by the LLVM
108 fuzzer.
110 6. A lookbehind assertion within a set of mutually recursive subpatterns could
111 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
113 7. Another buffer overflow bug involved duplicate named groups with a
114 reference between their definition, with a group that reset capture
115 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
116 fixed by always allowing for more memory, even if not needed. (A proper fix
117 is implemented in PCRE2, but it involves more refactoring.)
119 8. There was no check for integer overflow in subroutine calls such as (?123).
121 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
122 being treated as a literal 'l' instead of causing an error.
124 10. There was a buffer overflow if pcre_exec() was called with an ovector of
125 size 1. This bug was found by american fuzzy lop.
127 11. If a non-capturing group containing a conditional group that could match
128 an empty string was repeated, it was not identified as matching an empty
129 string itself. For example: /^(?:(?(1)x|)+)+$()/.
131 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
132 \a and \e in test subject lines.
134 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
135 instead of the EBCDIC value.
137 14. The handling of \c in an EBCDIC environment has been revised so that it is
138 now compatible with the specification in Perl's perlebcdic page.
140 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
141 ASCII/Unicode. This has now been added to the list of characters that are
142 recognized as white space in EBCDIC.
144 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
145 error (correctly) when used outside a class, but did not give an error
146 within a class.
148 17. \h within a class was incorrectly compiled in EBCDIC environments.
150 18. A pattern with an unmatched closing parenthesis that contained a backward
151 assertion which itself contained a forward reference caused buffer
152 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
154 19. JIT should return with error when the compiled pattern requires more stack
155 space than the maximum.
157 20. A possessively repeated conditional group that could match an empty string,
158 for example, /(?(R))*+/, was incorrectly compiled.
160 21. Fix infinite recursion in the JIT compiler when certain patterns such as
161 /(?:|a|){100}x/ are analysed.
163 22. Some patterns with character classes involving [: and \\ were incorrectly
164 compiled and could cause reading from uninitialized memory or an incorrect
165 error diagnosis.
167 23. Pathological patterns containing many nested occurrences of [: caused
168 pcre_compile() to run for a very long time.
170 24. A conditional group with only one branch has an implicit empty alternative
171 branch and must therefore be treated as potentially matching an empty
172 string.
174 25. If (?R was followed by - or + incorrect behaviour happened instead of a
175 diagnostic.
177 26. Arrange to give up on finding the minimum matching length for overly
178 complex patterns.
180 27. Similar to (4) above: in a pattern with duplicated named groups and an
181 occurrence of (?| it is possible for an apparently non-recursive back
182 reference to become recursive if a later named group with the relevant
183 number is encountered. This could lead to a buffer overflow. Wen Guanxing
184 from Venustech ADLAB discovered this bug.
186 28. If pcregrep was given the -q option with -c or -l, or when handling a
187 binary file, it incorrectly wrote output to stdout.
189 29. The JIT compiler did not restore the control verb head in case of *THEN
190 control verbs. This issue was found by Karl Skomski with a custom LLVM
191 fuzzer.
193 30. Error messages for syntax errors following \g and \k were giving inaccurate
194 offsets in the pattern.
196 31. Added a check for integer overflow in conditions (?(<digits>) and
197 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
198 fuzzer.
200 32. Handling recursive references such as (?2) when the reference is to a group
201 later in the pattern uses code that is very hacked about and error-prone.
202 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
203 give an internal error if it is obvious that compiling has gone wrong.
205 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
206 This issue was found by Karl Skomski with a custom LLVM fuzzer.
208 34. The JIT compiler should restore the control chain for empty possessive
209 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
211 35. Match limit check added to JIT recursion. This issue was found by Karl
212 Skomski with a custom LLVM fuzzer.
214 36. Yet another case similar to 27 above has been circumvented by an
215 unconditional allocation of extra memory. This issue is fixed "properly" in
216 PCRE2 by refactoring the way references are handled. Wen Guanxing
217 from Venustech ADLAB discovered this bug.
219 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
220 with a custom LLVM fuzzer.
222 38. Fixed a corner case of range optimization in JIT.
224 39. An incorrect error "overran compiling workspace" was given if there were
225 exactly enough group forward references such that the last one extended
226 into the workspace safety margin. The next one would have expanded the
227 workspace. The test for overflow was not including the safety margin.
229 40. A match limit issue is fixed in JIT which was found by Karl Skomski
230 with a custom LLVM fuzzer.
232 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
233 work under Windows. (Why has it taken so long for anyone to notice?)
235 42. In a character class such as [\W\p{Any}] where both a negative-type escape
236 ("not a word character") and a property escape were present, the property
237 escape was being ignored.
239 43. Fix crash caused by very long (*MARK) or (*THEN) names.
241 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
242 by a single ASCII character in a class item, was incorrectly compiled in
243 UCP mode. The POSIX class got lost, but only if the single character
244 followed it.
246 45. [:punct:] in UCP mode was matching some characters in the range 128-255
247 that should not have been matched.
249 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
250 class, all characters with code points greater than 255 are in the class.
251 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
252 such as \w are turned into Unicode properties), wide characters were not
253 correctly handled, and could fail to match.
256 Version 8.37 28-April-2015
257 --------------------------
259 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
260 for those parentheses to be closed with whatever has been captured so far.
261 However, it was failing to mark any other groups between the hightest
262 capture so far and the currrent group as "unset". Thus, the ovector for
263 those groups contained whatever was previously there. An example is the
264 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
266 2. If an assertion condition was quantified with a minimum of zero (an odd
267 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
269 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
270 unrecognized modifier, a crash could occur.
272 4. An attempt to do global matching in pcretest with a zero-length ovector
273 caused a crash.
275 5. Fixed a memory leak during matching that could occur for a subpattern
276 subroutine call (recursive or otherwise) if the number of captured groups
277 that had to be saved was greater than ten.
279 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
280 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
281 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
283 7. A UTF pattern containing a "not" match of a non-ASCII character and a
284 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
286 8. When a pattern is compiled, it remembers the highest back reference so that
287 when matching, if the ovector is too small, extra memory can be obtained to
288 use instead. A conditional subpattern whose condition is a check on a
289 capture having happened, such as, for example in the pattern
290 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
291 setting the highest backreference number. This mattered only if pcre_exec()
292 was called with an ovector that was too small to hold the capture, and there
293 was no other kind of back reference (a situation which is probably quite
294 rare). The effect of the bug was that the condition was always treated as
295 FALSE when the capture could not be consulted, leading to a incorrect
296 behaviour by pcre_exec(). This bug has been fixed.
298 9. A reference to a duplicated named group (either a back reference or a test
299 for being set in a conditional) that occurred in a part of the pattern where
300 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
301 to be incorrectly calculated, leading to overwriting.
303 10. A mutually recursive set of back references such as (\2)(\1) caused a
304 segfault at study time (while trying to find the minimum matching length).
305 The infinite loop is now broken (with the minimum length unset, that is,
306 zero).
308 11. If an assertion that was used as a condition was quantified with a minimum
309 of zero, matching went wrong. In particular, if the whole group had
310 unlimited repetition and could match an empty string, a segfault was
311 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
312 assertions to be quantified, but not if they are being used as conditions,
313 so the above pattern is faulted by Perl. PCRE has now been changed so that
314 it also rejects such patterns.
316 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
317 failed to allow the zero-repeat case if pcre2_exec() was called with an
318 ovector too small to capture the group.
320 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
321 Red Hat Product Security:
323 (a) A crash if /K and /F were both set with the option to save the compiled
324 pattern.
326 (b) Another crash if the option to print captured substrings in a callout
327 was combined with setting a null ovector, for example \O\C+ as a subject
328 string.
330 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
331 forward reference repeated a large (but limited) number of times within a
332 repeated outer group that has a zero minimum quantifier, caused incorrect
333 code to be compiled, leading to the error "internal error:
334 previously-checked referenced subpattern not found" when an incorrect
335 memory address was read. This bug was reported as "heap overflow",
336 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
337 CVE-2015-2325.
339 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
340 call within a group that also contained a recursive back reference caused
341 incorrect code to be compiled. This bug was reported as "heap overflow",
342 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
343 number CVE-2015-2326.
345 24. Computing the size of the JIT read-only data in advance has been a source
346 of various issues, and new ones are still appear unfortunately. To fix
347 existing and future issues, size computation is eliminated from the code,
348 and replaced by on-demand memory allocation.
350 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
351 adjacent to the end of the range, and the range contained characters with
352 more than one other case, caused incorrect behaviour when compiled in UTF
353 mode. In that example, the range a-j was left out of the class.
355 26. Fix JIT compilation of conditional blocks, which assertion
356 is converted to (*FAIL). E.g: /(?(?!))/.
358 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
359 discovered by the LLVM fuzzer.
361 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
362 when this assertion was used as a condition, for example (?(?!)a|b). In
363 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
364 error about an unsupported item.
366 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
367 possessification code could take exponential time to complete. A recursion
368 depth limit of 1000 has been imposed to limit the resources used by this
369 optimization.
371 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
372 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
373 because \S ensures they are all in the class. The code for doing this was
374 interacting badly with the code for computing the amount of space needed to
375 compile the pattern, leading to a buffer overflow. This bug was discovered
376 by the LLVM fuzzer.
378 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
379 other kinds of group caused stack overflow at compile time. This bug was
380 discovered by the LLVM fuzzer.
382 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
383 between a subroutine call and its quantifier was incorrectly compiled,
384 leading to buffer overflow or other errors. This bug was discovered by the
385 LLVM fuzzer.
387 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
388 assertion after (?(. The code was failing to check the character after
389 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
390 was discovered by the LLVM fuzzer.
392 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
393 a fixed maximum following a group that contains a subroutine reference was
394 incorrectly compiled and could trigger buffer overflow. This bug was
395 discovered by the LLVM fuzzer.
397 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
398 caused a stack overflow instead of the diagnosis of a non-fixed length
399 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
401 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
402 (e.g. /(?<=\Ka)/) could make pcregrep loop.
404 37. There was a similar problem to 36 in pcretest for global matches.
406 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
407 and a subsequent item in the pattern caused a non-match, backtracking over
408 the repeated \X did not stop, but carried on past the start of the subject,
409 causing reference to random memory and/or a segfault. There were also some
410 other cases where backtracking after \C could crash. This set of bugs was
411 discovered by the LLVM fuzzer.
413 39. The function for finding the minimum length of a matching string could take
414 a very long time if mutual recursion was present many times in a pattern,
415 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
416 method has been implemented. This infelicity was discovered by the LLVM
417 fuzzer.
419 40. Static linking against the PCRE library using the pkg-config module was
420 failing on missing pthread symbols.
423 Version 8.36 26-September-2014
424 ------------------------------
426 1. Got rid of some compiler warnings in the C++ modules that were shown up by
427 -Wmissing-field-initializers and -Wunused-parameter.
429 2. The tests for quantifiers being too big (greater than 65535) were being
430 applied after reading the number, and stupidly assuming that integer
431 overflow would give a negative number. The tests are now applied as the
432 numbers are read.
434 3. Tidy code in pcre_exec.c where two branches that used to be different are
435 now the same.
437 4. The JIT compiler did not generate match limit checks for certain
438 bracketed expressions with quantifiers. This may lead to exponential
439 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
440 issue should be resolved now.
442 5. Fixed an issue, which occures when nested alternatives are optimized
443 with table jumps.
445 6. Inserted two casts and changed some ints to size_t in the light of some
446 reported 64-bit compiler warnings (Bugzilla 1477).
448 7. Fixed a bug concerned with zero-minimum possessive groups that could match
449 an empty string, which sometimes were behaving incorrectly in the
450 interpreter (though correctly in the JIT matcher). This pcretest input is
451 an example:
453 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
456 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
457 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
458 for an empty string was breaking the inner loop and carrying on at a lower
459 level, when possessive repeated groups should always return to a higher
460 level as they have no backtrack points in them. The empty string test now
461 occurs at the outer level.
463 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
464 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
466 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
467 doesn't).
469 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
470 to hold a pattern to be incorrectly computed (too small) when there were
471 named back references to duplicated names. This could cause "internal
472 error: code overflow" or "double free or corruption" or other memory
473 handling errors.
475 11. When named subpatterns had the same prefixes, back references could be
476 confused. For example, in this pattern:
478 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
480 the reference to 'Name' was incorrectly treated as a reference to a
481 duplicate name.
483 12. A pattern such as /^s?c/mi8 where the optional character has more than
484 one "other case" was incorrectly compiled such that it would only try to
485 match starting at "c".
487 13. When a pattern starting with \s was studied, VT was not included in the
488 list of possible starting characters; this should have been part of the
489 8.34/18 patch.
491 14. If a character class started [\Qx]... where x is any character, the class
492 was incorrectly terminated at the ].
494 15. If a pattern that started with a caseless match for a character with more
495 than one "other case" was studied, PCRE did not set up the starting code
496 unit bit map for the list of possible characters. Now it does. This is an
497 optimization improvement, not a bug fix.
499 16. The Unicode data tables have been updated to Unicode 7.0.0.
501 17. Fixed a number of memory leaks in pcregrep.
503 18. Avoid a compiler warning (from some compilers) for a function call with
504 a cast that removes "const" from an lvalue by using an intermediate
505 variable (to which the compiler does not object).
507 19. Incorrect code was compiled if a group that contained an internal recursive
508 back reference was optional (had quantifier with a minimum of zero). This
509 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
510 caused segmentation faults because of stack overflows at compile time.
512 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
513 group that is quantified with an indefinite repeat, caused a compile-time
514 loop which used up all the system stack and provoked a segmentation fault.
515 This was not the same bug as 19 above.
517 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
518 Patch by Mike Frysinger.
521 Version 8.35 04-April-2014
522 --------------------------
524 1. A new flag is set, when property checks are present in an XCLASS.
525 When this flag is not set, PCRE can perform certain optimizations
526 such as studying these XCLASS-es.
528 2. The auto-possessification of character sets were improved: a normal
529 and an extended character set can be compared now. Furthermore
530 the JIT compiler optimizes more character set checks.
532 3. Got rid of some compiler warnings for potentially uninitialized variables
533 that show up only when compiled with -O2.
535 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
536 of a match later then the end of the match. The pcretest program was not
537 handling the case sensibly - it was outputting from the start to the next
538 binary zero. It now reports this situation in a message, and outputs the
539 text from the end to the start.
541 5. Fast forward search is improved in JIT. Instead of the first three
542 characters, any three characters with fixed position can be searched.
543 Search order: first, last, middle.
545 6. Improve character range checks in JIT. Characters are read by an inprecise
546 function now, which returns with an unknown value if the character code is
547 above a certain threshold (e.g: 256). The only limitation is that the value
548 must be bigger than the threshold as well. This function is useful when
549 the characters above the threshold are handled in the same way.
551 7. The macros whose names start with RAWUCHAR are placeholders for a future
552 mode in which only the bottom 21 bits of 32-bit data items are used. To
553 make this more memorable for those maintaining the code, the names have
554 been changed to start with UCHAR21, and an extensive comment has been added
555 to their definition.
557 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
558 to the export list in Makefile.am (they were accidentally omitted from the
559 8.34 tarball).
561 9. The informational output from pcretest used the phrase "starting byte set"
562 which is inappropriate for the 16-bit and 32-bit libraries. As the output
563 for "first char" and "need char" really means "non-UTF-char", I've changed
564 "byte" to "char", and slightly reworded the output. The documentation about
565 these values has also been (I hope) clarified.
567 10. Another JIT related optimization: use table jumps for selecting the correct
568 backtracking path, when more than four alternatives are present inside a
569 bracket.
571 11. Empty match is not possible, when the minimum length is greater than zero,
572 and there is no \K in the pattern. JIT should avoid empty match checks in
573 such cases.
575 12. In a caseless character class with UCP support, when a character with more
576 than one alternative case was not the first character of a range, not all
577 the alternative cases were added to the class. For example, s and \x{17f}
578 are both alternative cases for S: the class [RST] was handled correctly,
579 but [R-T] was not.
581 13. The configure.ac file always checked for pthread support when JIT was
582 enabled. This is not used in Windows, so I have put this test inside a
583 check for the presence of windows.h (which was already tested for).
585 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
586 The algorithm provides a way to skip certain starting offsets, and usually
587 faster than linear prefix searches.
589 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
590 as for 'fr_FR' and 'french'. For some reason, however, it then used the
591 Windows-specific input and output files, which have 'french' screwed in.
592 So this could never have worked. One of the problems with locales is that
593 they aren't always the same. I have now updated RunTest so that it checks
594 the output of the locale test (test 3) against three different output
595 files, and it allows the test to pass if any one of them matches. With luck
596 this should make the test pass on some versions of Solaris where it was
597 failing. Because of the uncertainty, the script did not used to stop if
598 test 3 failed; it now does. If further versions of a French locale ever
599 come to light, they can now easily be added.
601 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
602 there was a message during ./configure, but it did not stop. This now
603 provokes an error. The invalid example in README has been corrected.
604 If a value less than the minimum is given, the minimum value has always
605 been used, but now a warning is given.
607 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
608 was a bug in the test system, which is now fixed. Also, the list of various
609 configurations that are tested for each release did not have one with both
610 16/32 bits and --enable-bar-anycrlf. It now does.
612 18. pcretest was missing "-C bsr" for displaying the \R default setting.
614 19. Little endian PowerPC systems are supported now by the JIT compiler.
616 20. The fast forward newline mechanism could enter to an infinite loop on
617 certain invalid UTF-8 input. Although we don't support these cases
618 this issue can be fixed by a performance optimization.
620 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
621 not take account if existing stack usage. There is now a new global
622 variable called pcre_stack_guard that can be set to point to an external
623 function to check stack availability. It is called at the start of
624 processing every parenthesized group.
626 22. A typo in the code meant that in ungreedy mode the max/min qualifier
627 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
628 not match "ab".
630 23. When UTF was disabled, the JIT program reported some incorrect compile
631 errors. These messages are silenced now.
633 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
634 compiler.
636 25. Change all the temporary files used in RunGrepTest to be different to those
637 used by RunTest so that the tests can be run simultaneously, for example by
638 "make -j check".
641 Version 8.34 15-December-2013
642 -----------------------------
644 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
645 executable memory. Patch inspired by Carsten Klein.
647 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
648 this macro is never tested and has no effect, because the work to support
649 coverage involves only compiling and linking options and special targets in
650 the Makefile. The comment in config.h implied that defining the macro would
651 enable coverage support, which is totally false. There was also support for
652 setting this macro in the CMake files (my fault, I just copied it from
653 configure). SUPPORT_GCOV has now been removed.
655 3. Make a small performance improvement in strlen16() and strlen32() in
656 pcretest.
658 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
659 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
660 these cases). There was also one in pcretest.c.
662 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
664 6. In UTF mode, the code for checking whether a group could match an empty
665 string (which is used for indefinitely repeated groups to allow for
666 breaking an infinite loop) was broken when the group contained a repeated
667 negated single-character class with a character that occupied more than one
668 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
669 UTF-8 mode). The effect was undefined: the group might or might not be
670 deemed as matching an empty string, or the program might have crashed.
672 7. The code for checking whether a group could match an empty string was not
673 recognizing that \h, \H, \v, \V, and \R must match a character.
675 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
676 an empty string. If it can, pcretest shows this in its information output.
678 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
679 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
680 matched by pcre_exec() without using JIT:
682 (a) If the rest of the pattern did not match after a maximal run of
683 grapheme clusters, the code for backing up to try with fewer of them
684 did not always back up over a full grapheme when characters that do not
685 have the modifier quality were involved, e.g. Hangul syllables.
687 (b) If the match point in a subject started with modifier character, and
688 there was no match, the code could incorrectly back up beyond the match
689 point, and potentially beyond the first character in the subject,
690 leading to a segfault or an incorrect match result.
692 10. A conditional group with an assertion condition could lead to PCRE
693 recording an incorrect first data item for a match if no other first data
694 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
695 first data item, and therefore matched "ca" after "c" instead of at the
696 start.
698 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
699 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
701 12. The source of pcregrep now includes z/OS-specific code so that it can be
702 compiled for z/OS as part of the special z/OS distribution.
704 13. Added the -T and -TM options to pcretest.
706 14. The code in pcre_compile.c for creating the table of named capturing groups
707 has been refactored. Instead of creating the table dynamically during the
708 actual compiling pass, the information is remembered during the pre-compile
709 pass (on the stack unless there are more than 20 named groups, in which
710 case malloc() is used) and the whole table is created before the actual
711 compile happens. This has simplified the code (it is now nearly 150 lines
712 shorter) and prepared the way for better handling of references to groups
713 with duplicate names.
715 15. A back reference to a named subpattern when there is more than one of the
716 same name now checks them in the order in which they appear in the pattern.
717 The first one that is set is used for the reference. Previously only the
718 first one was inspected. This change makes PCRE more compatible with Perl.
720 16. Unicode character properties were updated from Unicode 6.3.0.
722 17. The compile-time code for auto-possessification has been refactored, based
723 on a patch by Zoltan Herczeg. It now happens after instead of during
724 compilation. The code is cleaner, and more cases are handled. The option
725 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
726 options in pcretest are provided to set it. It can also be set by
727 (*NO_AUTO_POSSESS) at the start of a pattern.
729 18. The character VT has been added to the default ("C" locale) set of
730 characters that match \s and are generally treated as white space,
731 following this same change in Perl 5.18. There is now no difference between
732 "Perl space" and "POSIX space". Whether VT is treated as white space in
733 other locales depends on the locale.
735 19. The code for checking named groups as conditions, either for being set or
736 for being recursed, has been refactored (this is related to 14 and 15
737 above). Processing unduplicated named groups should now be as fast at
738 numerical groups, and processing duplicated groups should be faster than
739 before.
741 20. Two patches to the CMake build system, by Alexander Barkov:
743 (1) Replace the "source" command by "." in CMakeLists.txt because
744 "source" is a bash-ism.
746 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
747 without these the CMake build does not work on Solaris.
749 21. Perl has changed its handling of \8 and \9. If there is no previously
750 encountered capturing group of those numbers, they are treated as the
751 literal characters 8 and 9 instead of a binary zero followed by the
752 literals. PCRE now does the same.
754 22. Following Perl, added \o{} to specify codepoints in octal, making it
755 possible to specify values greater than 0777 and also making them
756 unambiguous.
758 23. Perl now gives an error for missing closing braces after \x{... instead of
759 treating the string as literal. PCRE now does the same.
761 24. RunTest used to grumble if an inappropriate test was selected explicitly,
762 but just skip it when running all tests. This make it awkward to run ranges
763 of tests when one of them was inappropriate. Now it just skips any
764 inappropriate tests, as it always did when running all tests.
766 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
767 character types such as \d or \w, too many callouts were inserted, and the
768 data that they returned was rubbish.
770 26. In UCP mode, \s was not matching two of the characters that Perl matches,
771 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
772 were matched by \h. The code has now been refactored so that the lists of
773 the horizontal and vertical whitespace characters used for \h and \v (which
774 are defined only in one place) are now also used for \s.
776 27. Add JIT support for the 64 bit TileGX architecture.
777 Patch by Jiong Wang (Tilera Corporation).
779 28. Possessive quantifiers for classes (both explicit and automatically
780 generated) now use special opcodes instead of wrapping in ONCE brackets.
782 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
783 (because it's meaningless), this was not happening when PCRE_CASELESS was
784 set. Not wrong, but inefficient.
786 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
787 properties for \w, \d, etc) is present in a test regex. Otherwise if the
788 test contains no characters greater than 255, Perl doesn't realise it
789 should be using Unicode semantics.
791 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
792 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
793 does in Unicode mode.
795 32. Added the "forbid" facility to pcretest so that putting tests into the
796 wrong test files can sometimes be quickly detected.
798 33. There is now a limit (default 250) on the depth of nesting of parentheses.
799 This limit is imposed to control the amount of system stack used at compile
800 time. It can be changed at build time by --with-parens-nest-limit=xxx or
801 the equivalent in CMake.
803 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
804 errors. Perl warns for these when in warning mode, but PCRE has no facility
805 for giving warnings.
807 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
808 However, this was not working for (?!) because it is optimized to (*FAIL),
809 for which PCRE does not allow quantifiers. The optimization is now disabled
810 when a quantifier follows (?!). I can't see any use for this, but it makes
811 things uniform.
813 36. Perl no longer allows group names to start with digits, so I have made this
814 change also in PCRE. It simplifies the code a bit.
816 37. In extended mode, Perl ignores spaces before a + that indicates a
817 possessive quantifier. PCRE allowed a space before the quantifier, but not
818 before the possessive +. It now does.
820 38. The use of \K (reset reported match start) within a repeated possessive
821 group such as (a\Kb)*+ was not working.
823 40. Document that the same character tables must be used at compile time and
824 run time, and that the facility to pass tables to pcre_exec() and
825 pcre_dfa_exec() is for use only with saved/restored patterns.
827 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
828 features for Builds with MSVC:
830 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
831 builds). The .rc files can be used to set FileDescription and many other
832 attributes.
834 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
835 This allows higher-level build scripts which want .pdb files to avoid
836 hard-coding the exact files needed."
838 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
839 mean "start of word" and "end of word", respectively, as a transition aid.
841 43. A minimizing repeat of a class containing codepoints greater than 255 in
842 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
843 compiled to use the heap for recursion.
845 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
846 is configured.
849 Version 8.33 28-May-2013
850 ------------------------
852 1. Added 'U' to some constants that are compared to unsigned integers, to
853 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
854 variables that are added to signed variables, to ensure the result is
855 signed and can be negated.
857 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
858 CMake config files.
860 3. Revise the creation of config.h.generic so that all boolean macros are
861 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
862 overriding via -D on the command line possible.
864 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
865 to unsigned int is reported to make a quite noticeable speed difference in
866 a specific Windows environment. Testing on Linux did also appear to show
867 some benefit (and it is clearly not harmful). Also fixed the definition of
868 Xop which should be unsigned.
870 5. Related to (4), changing the definition of the intermediate variable cc
871 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
872 improvements.
874 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
875 unnecessary spaces.
877 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
878 and later.
880 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
882 9. Optimizing fast_forward_start_bits in JIT.
884 10. Adding support for callouts in JIT, and fixing some issues revealed
885 during this work. Namely:
887 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
889 (b) Minimum length was not checked before the matching is started.
891 11. The value of capture_last that is passed to callouts was incorrect in some
892 cases when there was a capture on one path that was subsequently abandoned
893 after a backtrack. Also, the capture_last value is now reset after a
894 recursion, since all captures are also reset in this case.
896 12. The interpreter no longer returns the "too many substrings" error in the
897 case when an overflowing capture is in a branch that is subsequently
898 abandoned after a backtrack.
900 13. In the pathological case when an offset vector of size 2 is used, pcretest
901 now prints out the matched string after a yield of 0 or 1.
903 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
904 Only supported by the JIT compiler at the moment.
906 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
908 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
909 offset of the starting point of the matching process, provided the offsets
910 vector is large enough.
912 17. The \A escape now records a lookbehind value of 1, though its execution
913 does not actually inspect the previous character. This is to ensure that,
914 in partial multi-segment matching, at least one character from the old
915 segment is retained when a new segment is processed. Otherwise, if there
916 are no lookbehinds in the pattern, \A might match incorrectly at the start
917 of a new segment.
919 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
921 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
922 gives some modest performance improvement in 8-bit mode.
924 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
925 be expressed in certain programming languages using Universal Character
926 Names.
928 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
929 which points out that "non characters" are not "characters that may not
930 appear in Unicode strings" but rather "characters that are reserved for
931 internal use and have only local meaning".
933 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
934 there was a conditional group that depended on an assertion, if the
935 assertion was false, the callout that immediately followed the alternation
936 in the condition was skipped when pcre_exec() was used for matching.
938 23. Allow an explicit callout to be inserted before an assertion that is the
939 condition for a conditional group, for compatibility with automatic
940 callouts, which always insert a callout at this point.
942 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
943 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
945 25. (*PRUNE) is now supported by the JIT compiler.
947 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
949 27. Fix the case where there are two or more SKIPs with arguments that may be
950 ignored.
952 28. (*SKIP) is now supported by the JIT compiler.
954 29. (*THEN) is now supported by the JIT compiler.
956 30. Update RunTest with additional test selector options.
958 31. The way PCRE handles backtracking verbs has been changed in two ways.
960 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
961 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
962 backtracking. In some cases this makes it more Perl-compatible, but Perl's
963 rather obscure rules do not always do the same thing.
965 (2) Previously, backtracking verbs were confined within assertions. This is
966 no longer the case for positive assertions, except for (*ACCEPT). Again,
967 this sometimes improves Perl compatibility, and sometimes does not.
969 32. A number of tests that were in test 2 because Perl did things differently
970 have been moved to test 1, because either Perl or PCRE has changed, and
971 these tests are now compatible.
973 32. Backtracking control verbs are now handled in the same way in JIT and
974 interpreter.
976 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
977 contained a forward subroutine reference caused a compile error.
979 34. Auto-detect and optimize limited repetitions in JIT.
981 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
982 blocking (*UTF) etc.
984 36. In the interpreter, maximizing pattern repetitions for characters and
985 character types now use tail recursion, which reduces stack usage.
987 37. The value of the max lookbehind was not correctly preserved if a compiled
988 and saved regex was reloaded on a host of different endianness.
990 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
991 of the compiled pattern block, expand the flags field from 16 to 32 bits
992 because it was almost full.
994 39. Try madvise first before posix_madvise.
996 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
997 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
998 original change locked out all matches of empty strings. This has been
999 changed so that one match of an empty string per line is recognized.
1000 Subsequent searches on the same line (for colouring or for --only-matching,
1001 for example) do not recognize empty strings.
1003 41. Applied a user patch to fix a number of spelling mistakes in comments.
1005 42. Data lines longer than 65536 caused pcretest to crash.
1007 43. Clarified the data type for length and startoffset arguments for pcre_exec
1008 and pcre_dfa_exec in the function-specific man pages, where they were
1009 explicitly stated to be in bytes, never having been updated. I also added
1010 some clarification to the pcreapi man page.
1012 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1013 a segmentation fault.
1016 Version 8.32 30-November-2012
1017 -----------------------------
1019 1. Improved JIT compiler optimizations for first character search and single
1020 character iterators.
1022 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1023 Patch by Daniel Richard G.
1025 3. Single character iterator optimizations in the JIT compiler.
1027 4. Improved JIT compiler optimizations for character ranges.
1029 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1030 Reported by Giuseppe D'Angelo.
1032 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1033 of a line, was being set incorrectly in cases where .* appeared inside
1034 atomic brackets at the start of a pattern, or where there was a subsequent
1035 *PRUNE or *SKIP.
1037 7. Improved instruction cache flush for POWER/PowerPC.
1038 Patch by Daniel Richard G.
1040 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1041 grep:
1043 (a) There is now no limit to the number of patterns to be matched.
1045 (b) An error is given if a pattern is too long.
1047 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1048 are now supported.
1050 (d) --exclude-from and --include-from (multiple use) have been added.
1052 (e) Exclusions and inclusions now apply to all files and directories, not
1053 just to those obtained from scanning a directory recursively.
1055 (f) Multiple uses of -f and --file-list are now supported.
1057 (g) In a Windows environment, the default for -d has been changed from
1058 "read" (the GNU grep default) to "skip", because otherwise the presence
1059 of a directory in the file list provokes an error.
1061 (h) The documentation has been revised and clarified in places.
1063 9. Improve the matching speed of capturing brackets.
1065 10. Changed the meaning of \X so that it now matches a Unicode extended
1066 grapheme cluster.
1068 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1069 out POSIX threads when JIT support is configured.
1071 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1073 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1074 structure could go wrong in environments where size_t is not the same size
1075 as int.
1077 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1078 set.
1080 15. The EBCDIC support had decayed; later updates to the code had included
1081 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1082 general tidy up of EBCDIC-related issues, and the documentation was also
1083 not quite right. There is now a test that can be run on ASCII systems to
1084 check some of the EBCDIC-related things (but is it not a full test).
1086 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1087 in a small tidy to the code.
1089 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1091 18. If the --only-matching (-o) option in pcregrep is specified multiple
1092 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1093 substrings matched by the 1st and 2nd capturing parentheses. A separating
1094 string can be specified by --om-separator (default empty).
1096 19. Improving the first n character searches.
1098 20. Turn case lists for horizontal and vertical white space into macros so that
1099 they are defined only once.
1101 21. This set of changes together give more compatible Unicode case-folding
1102 behaviour for characters that have more than one other case when UCP
1103 support is available.
1105 (a) The Unicode property table now has offsets into a new table of sets of
1106 three or more characters that are case-equivalent. The MultiStage2.py
1107 script that generates these tables (the pcre_ucd.c file) now scans
1108 CaseFolding.txt instead of UnicodeData.txt for character case
1109 information.
1111 (b) The code for adding characters or ranges of characters to a character
1112 class has been abstracted into a generalized function that also handles
1113 case-independence. In UTF-mode with UCP support, this uses the new data
1114 to handle characters with more than one other case.
1116 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1117 whose other case is greater than 256 are now correctly matched
1118 caselessly. Previously, the high codepoint matched the low one, but not
1119 vice versa.
1121 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1122 of the new class addition function, using character lists defined as
1123 macros alongside the case definitions of 20 above.
1125 (e) Caseless back references now work with characters that have more than
1126 one other case.
1128 (f) General caseless matching of characters with more than one other case
1129 is supported.
1131 22. Unicode character properties were updated from Unicode 6.2.0
1133 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1135 24. Add support for 32-bit character strings, and UTF-32
1137 25. Major JIT compiler update (code refactoring and bugfixing).
1138 Experimental Sparc 32 support is added.
1140 26. Applied a modified version of Daniel Richard G's patch to create
1141 pcre.h.generic and config.h.generic by "make" instead of in the
1142 PrepareRelease script.
1144 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1145 pcre_compile.c when checking for a zero character.
1147 28. Introducing a native interface for JIT. Through this interface, the compiled
1148 machine code can be directly executed. The purpose of this interface is to
1149 provide fast pattern matching, so several sanity checks are not performed.
1150 However, feature tests are still performed. The new interface provides
1151 1.4x speedup compared to the old one.
1153 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1154 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1155 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1157 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1158 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1159 the "old" RFC 2279). Instead, it ended up passing a negative length to
1160 pcre_exec().
1162 31. Add support for GCC's visibility feature to hide internal functions.
1164 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1165 "unknown -C option" after outputting 0 or 1.
1167 33. There is now support for generating a code coverage report for the test
1168 suite in environments where gcc is the compiler and lcov is installed. This
1169 is mainly for the benefit of the developers.
1171 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1172 unaddressable using valgrind annotations, allowing valgrind to detect
1173 invalid memory accesses. This is mainly for the benefit of the developers.
1175 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1177 26. Give configure error if --enable-cpp but no C++ compiler found.
1180 Version 8.31 06-July-2012
1181 -------------------------
1183 1. Fixing a wrong JIT test case and some compiler warnings.
1185 2. Removed a bashism from the RunTest script.
1187 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1188 to unsigned type, result still unsigned" that was given by an MS compiler
1189 on encountering the code "-sizeof(xxx)".
1191 4. Partial matching support is added to the JIT compiler.
1193 5. Fixed several bugs concerned with partial matching of items that consist
1194 of more than one character:
1196 (a) /^(..)\1/ did not partially match "aba" because checking references was
1197 done on an "all or nothing" basis. This also applied to repeated
1198 references.
1200 (b) \R did not give a hard partial match if \r was found at the end of the
1201 subject.
1203 (c) \X did not give a hard partial match after matching one or more
1204 characters at the end of the subject.
1206 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1207 a partial match for the string "\r".
1209 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1210 a partial match for a CR character at the end of the subject string.
1212 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1213 running pcretest, the text "(JIT)" added to the output whenever JIT is
1214 actually used to run the match.
1216 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1217 or /S+[+] with a digit between 1 and 7.
1219 8. OP_NOT now supports any UTF character not just single-byte ones.
1221 9. (*MARK) control verb is now supported by the JIT compiler.
1223 10. The command "./RunTest list" lists the available tests without actually
1224 running any of them. (Because I keep forgetting what they all are.)
1228 12. Applied a (slightly modified) user-supplied patch that improves performance
1229 when the heap is used for recursion (compiled with --disable-stack-for-
1230 recursion). Instead of malloc and free for each heap frame each time a
1231 logical recursion happens, frames are retained on a chain and re-used where
1232 possible. This sometimes gives as much as 30% improvement.
1234 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1235 call.
1237 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1239 15. It is now possible to link pcretest with libedit as an alternative to
1240 libreadline.
1242 16. (*COMMIT) control verb is now supported by the JIT compiler.
1244 17. The Unicode data tables have been updated to Unicode 6.1.0.
1246 18. Added --file-list option to pcregrep.
1248 19. Added binary file support to pcregrep, including the -a, --binary-files,
1249 -I, and --text options.
1251 20. The madvise function is renamed for posix_madvise for QNX compatibility
1252 reasons. Fixed by Giuseppe D'Angelo.
1254 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1256 22. Changed the option for creating symbolic links for 16-bit man pages from
1257 -s to -sf so that re-installing does not cause issues.
1259 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1261 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1262 matches in certain environments (the workspace was not being correctly
1263 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1264 some of the workspace data at the beginning of a restart.
1266 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1267 was not doing so when it should - probably a typo introduced by SVN 528
1268 (change 8.10/14).
1270 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1271 \w+ when the character tables indicated that \x{c4} was a word character.
1272 There were several related cases, all because the tests for doing a table
1273 lookup were testing for characters less than 127 instead of 255.
1275 27. If a pattern contains capturing parentheses that are not used in a match,
1276 their slots in the ovector are set to -1. For those that are higher than
1277 any matched groups, this happens at the end of processing. In the case when
1278 there were back references that the ovector was too small to contain
1279 (causing temporary malloc'd memory to be used during matching), and the
1280 highest capturing number was not used, memory off the end of the ovector
1281 was incorrectly being set to -1. (It was using the size of the temporary
1282 memory instead of the true size.)
1284 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1285 ovector size, it uses memory at the end of the block that it has got.
1287 29. Check for an overlong MARK name and give an error at compile time. The
1288 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1290 30. JIT compiler update.
1292 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1293 Rill for the patch.
1295 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1297 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1299 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1300 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1301 was enabled.
1303 35. Improve JIT code generation for greedy plus quantifier.
1305 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1306 1 to "aa" instead of to an empty string. The bug affected repeated groups
1307 that could potentially match an empty string.
1309 37. Optimizing single character iterators in JIT.
1311 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1312 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1313 codepoints that are too big for the mode are faulted, and in a UTF mode,
1314 disallowed codepoints are also faulted.
1316 39. If PCRE was compiled with UTF support, in three places in the DFA
1317 matcher there was code that should only have been obeyed in UTF mode, but
1318 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1319 processing when bytes with values greater than 127 were present. In 16-bit
1320 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1321 both cases the values are those that cannot be the first data item in a UTF
1322 character. The three items that might have provoked this were recursions,
1323 possessively repeated groups, and atomic groups.
1325 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1326 and pcregrep, because some OS require shared objects to be explicitly
1327 passed to ld, causing the link step to fail if they are not.
1329 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1330 mode, patterns that started with \h* or \R* might be incorrectly matched.
1333 Version 8.30 04-February-2012
1334 -----------------------------
1336 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1337 name is defined in ctype.h.
1339 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1340 only in quite long subpatterns.
1342 3. Removed the function pcre_info(), which has been obsolete and deprecated
1343 since it was replaced by pcre_fullinfo() in February 2000.
1345 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1346 match a (*MARK), and the match failed at the start of the subject, a
1347 reference to memory before the start of the subject could occur. This bug
1348 was introduced by fix 17 of release 8.21.
1350 5. A reference to an unset group with zero minimum repetition was giving
1351 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1352 /(another)?(\1?)test/ matched against "hello world test". This bug was
1353 introduced in release 8.13.
1355 6. Add support for 16-bit character strings (a large amount of work involving
1356 many changes and refactorings).
1358 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1359 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1360 from a file.
1362 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1363 rounding is not applied in this particular case).
1365 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1366 if they appear, or are escaped, in patterns.
1368 10. Get rid of a number of -Wunused-but-set-variable warnings.
1370 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1371 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1372 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1373 also returns the mark "x". This bug applied to capturing parentheses,
1374 non-capturing parentheses, and atomic parentheses. It also applied to some
1375 assertions.
1377 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1378 information out of configure.ac instead of relying on pcre.h.generic, which
1379 is not stored in the repository.
1381 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1382 -lreadline.
1384 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1386 15. Applied Graycode's patch to put the top-level frame on the stack rather
1387 than the heap when not using the stack for recursion. This gives a
1388 performance improvement in many cases when recursion is not deep.
1390 16. Experimental code added to "pcretest -C" to output the stack frame size.
1393 Version 8.21 12-Dec-2011
1394 ------------------------
1396 1. Updating the JIT compiler.
1398 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1399 are added as well.
1401 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1402 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1403 calling _pcre_jit_exec. Some extra comments are added.
1405 4. (*MARK) settings inside atomic groups that do not contain any capturing
1406 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1407 was introduced by change 18 for 8.20.
1409 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1410 ECMA-262 standard.
1412 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1413 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1414 This bug was probably introduced by change 9 of 8.13.
1416 7. While fixing 6 above, I noticed that a number of other items were being
1417 incorrectly rejected as "not fixed length". This arose partly because newer
1418 opcodes had not been added to the fixed-length checking code. I have (a)
1419 corrected the bug and added tests for these items, and (b) arranged for an
1420 error to occur if an unknown opcode is encountered while checking for fixed
1421 length instead of just assuming "not fixed length". The items that were
1422 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1423 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1424 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1426 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1427 being incorrectly compiled and would have given unpredicatble results.
1429 9. A possessively repeated subpattern with minimum repeat count greater than
1430 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1431 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1432 the first (A) could occur when it should not.
1434 10. Add a cast and remove a redundant test from the code.
1436 11. JIT should use pcre_malloc/pcre_free for allocation.
1438 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1439 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1440 is anything other than /usr, -L is still shown).
1442 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1444 14. Perl does not support \N without a following name in a [] class; PCRE now
1445 also gives an error.
1447 15. If a forward reference was repeated with an upper limit of around 2000,
1448 it caused the error "internal error: overran compiling workspace". The
1449 maximum number of forward references (including repeats) was limited by the
1450 internal workspace, and dependent on the LINK_SIZE. The code has been
1451 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1452 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1453 of around 200,000 forward references. While doing this, I also speeded up
1454 the filling in of repeated forward references.
1456 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1457 incorrectly expecting the subject to contain another "a" after the start.
1459 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1460 in the match, the SKIP should be ignored. This was not happening; instead
1461 the SKIP was being treated as NOMATCH. For patterns such as
1462 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1463 tested.
1465 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1466 now much more compatible with Perl, in particular in cases where the result
1467 is a non-match for a non-anchored pattern. For example, if
1468 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1469 "m", where previously it did not return a name. A side effect of this
1470 change is that for partial matches, the last encountered mark name is
1471 returned, as for non matches. A number of tests that were previously not
1472 Perl-compatible have been moved into the Perl-compatible test files. The
1473 refactoring has had the pleasing side effect of removing one argument from
1474 the match() function, thus reducing its stack requirements.
1476 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1477 subsequent uses of /S (without +) incorrectly behaved like /S+.
1479 21. Retrieve executable code size support for the JIT compiler and fixing
1480 some warnings.
1482 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1483 not work when the shorter character appeared right at the end of the
1484 subject string.
1486 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1487 systems.
1489 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1490 output it when the /M option is used in pcretest.
1492 25. The CheckMan script was not being included in the distribution. Also, added
1493 an explicit "perl" to run Perl scripts from the PrepareRelease script
1494 because this is reportedly needed in Windows.
1496 26. If study data was being save in a file and studying had not found a set of
1497 "starts with" bytes for the pattern, the data written to the file (though
1498 never used) was taken from uninitialized memory and so caused valgrind to
1499 complain.
1501 27. Updated RunTest.bat as provided by Sheri Pierce.
1503 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1505 29. Computation of memory usage for the table of capturing group names was
1506 giving an unnecessarily large value.
1509 Version 8.20 21-Oct-2011
1510 ------------------------
1512 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1513 a POSIX class. After further experiments with Perl, which convinced me that
1514 Perl has bugs and confusions, a closing square bracket is no longer allowed
1515 in a POSIX name. This bug also affected patterns with classes that started
1516 with full stops.
1518 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1519 captured substring, but while checking the failing first alternative,
1520 substring 1 is temporarily captured. If the output vector supplied to
1521 pcre_exec() was not big enough for this capture, the yield of the function
1522 was still zero ("insufficient space for captured substrings"). This cannot
1523 be totally fixed without adding another stack variable, which seems a lot
1524 of expense for a edge case. However, I have improved the situation in cases
1525 such as /(a)(b)x|abc/ matched against "abc", where the return code
1526 indicates that fewer than the maximum number of slots in the ovector have
1527 been set.
1529 3. Related to (2) above: when there are more back references in a pattern than
1530 slots in the output vector, pcre_exec() uses temporary memory during
1531 matching, and copies in the captures as far as possible afterwards. It was
1532 using the entire output vector, but this conflicts with the specification
1533 that only 2/3 is used for passing back captured substrings. Now it uses
1534 only the first 2/3, for compatibility. This is, of course, another edge
1535 case.
1537 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1538 main code base, and can be used by building with --enable-jit. When this is
1539 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1540 runtime --no-jit option is given.
1542 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1543 ovector, the return from the function was zero, implying that there were
1544 other matches that did not fit. The correct "exactly full" value is now
1545 returned.
1547 6. If a subpattern that was called recursively or as a subroutine contained
1548 (*PRUNE) or any other control that caused it to give a non-standard return,
1549 invalid errors such as "Error -26 (nested recursion at the same subject
1550 position)" or even infinite loops could occur.
1552 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1553 computing the minimum length on reaching *ACCEPT, and so ended up with the
1554 wrong value of 1 rather than 0. Further investigation indicates that
1555 computing a minimum subject length in the presence of *ACCEPT is difficult
1556 (think back references, subroutine calls), and so I have changed the code
1557 so that no minimum is registered for a pattern that contains *ACCEPT.
1559 8. If (*THEN) was present in the first (true) branch of a conditional group,
1560 it was not handled as intended. [But see 16 below.]
1562 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1563 Sheri Pierce.
1565 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1566 the first byte in a match must be "a".
1568 11. Change 17 for 8.13 increased the recursion depth for patterns like
1569 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1570 pattern contains any instances of (*THEN). If it does not, the old
1571 optimizations are restored. It would be nice to do this on a per-group
1572 basis, but at the moment that is not feasible.
1574 12. In some environments, the output of pcretest -C is CRLF terminated. This
1575 broke RunTest's code that checks for the link size. A single white space
1576 character after the value is now allowed for.
1578 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1579 For "fr", it uses the Windows-specific input and output files.
1581 14. If (*THEN) appeared in a group that was called recursively or as a
1582 subroutine, it did not work as intended. [But see next item.]
1584 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1585 pattern fragments (but not containing any | characters). If A and B are
1586 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1587 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1588 D. In other words, Perl considers parentheses that do not contain any |
1589 characters to be part of a surrounding alternative, whereas PCRE was
1590 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1591 differently. PCRE now behaves in the same way as Perl, except in the case
1592 of subroutine/recursion calls such as (?1) which have in any case always
1593 been different (but PCRE had them first :-).
1595 16. Related to 15 above: Perl does not treat the | in a conditional group as
1596 creating alternatives. Such a group is treated in the same way as an
1597 ordinary group without any | characters when processing (*THEN). PCRE has
1598 been changed to match Perl's behaviour.
1600 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1601 RunGrepTest script failed.
1603 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1604 inevitable for groups that contain captures, but it can lead to a lot of
1605 stack use in large patterns. The old behaviour has been restored for atomic
1606 groups that do not contain any capturing parentheses.
1608 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1609 suppress the check for a minimum subject length at run time. (If it was
1610 given to pcre_exec() or pcre_dfa_exec() it did work.)
1612 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1613 fail to work when decoding hex characters in data strings in EBCDIC
1614 environments.
1616 21. It appears that in at least one Mac OS environment, the isxdigit() function
1617 is implemented as a macro that evaluates to its argument more than once,
1618 contravening the C 90 Standard (I haven't checked a later standard). There
1619 was an instance in pcretest which caused it to go wrong when processing
1620 \x{...} escapes in subject strings. The has been rewritten to avoid using
1621 things like p++ in the argument of isxdigit().
1624 Version 8.13 16-Aug-2011
1625 ------------------------
1627 1. The Unicode data tables have been updated to Unicode 6.0.0.
1629 2. Two minor typos in pcre_internal.h have been fixed.
1631 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1632 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1633 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1635 4. There were a number of related bugs in the code for matching backrefences
1636 caselessly in UTF-8 mode when codes for the characters concerned were
1637 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1638 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1639 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1640 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1641 2-byte code at the end of the subject (it thought there wasn't enough data
1642 left).
1644 5. Comprehensive information about what went wrong is now returned by
1645 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1646 as the output vector has at least 2 elements. The offset of the start of
1647 the failing character and a reason code are placed in the vector.
1649 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1650 now returned is for the first byte of the failing character, instead of the
1651 last byte inspected. This is an incompatible change, but I hope it is small
1652 enough not to be a problem. It makes the returned offset consistent with
1653 pcre_exec() and pcre_dfa_exec().
1655 7. pcretest now gives a text phrase as well as the error number when
1656 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1657 failure, the offset and reason code are output.
1659 8. When \R was used with a maximizing quantifier it failed to skip backwards
1660 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1661 back over a single character (\n). This seems wrong (because it treated the
1662 two characters as a single entity when going forwards), conflicts with the
1663 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1664 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1665 has been changed.
1667 9. Some internal refactoring has changed the processing so that the handling
1668 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1669 time (the PCRE_DOTALL option was changed this way some time ago: version
1670 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1671 which was always a bit of a fudge. It also means that there is one less
1672 argument for the match() function, which reduces its stack requirements
1673 slightly. This change also fixes an incompatibility with Perl: the pattern
1674 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1676 10. More internal refactoring has drastically reduced the number of recursive
1677 calls to match() for possessively repeated groups such as (abc)++ when
1678 using pcre_exec().
1680 11. While implementing 10, a number of bugs in the handling of groups were
1681 discovered and fixed:
1683 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1684 (a|)*(?1) gave a compile-time internal error.
1685 ((a|)+)+ did not notice that the outer group could match an empty string.
1686 (^a|^)+ was not marked as anchored.
1687 (.*a|.*)+ was not marked as matching at start or after a newline.
1689 12. Yet more internal refactoring has removed another argument from the match()
1690 function. Special calls to this function are now indicated by setting a
1691 value in a variable in the "match data" data block.
1693 13. Be more explicit in pcre_study() instead of relying on "default" for
1694 opcodes that mean there is no starting character; this means that when new
1695 ones are added and accidentally left out of pcre_study(), testing should
1696 pick them up.
1698 14. The -s option of pcretest has been documented for ages as being an old
1699 synonym of -m (show memory usage). I have changed it to mean "force study
1700 for every regex", that is, assume /S for every regex. This is similar to -i
1701 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1702 using it. It makes it easier to run collections of tests with and without
1703 study enabled, and thereby test pcre_study() more easily. All the standard
1704 tests are now run with and without -s (but some patterns can be marked as
1705 "never study" - see 20 below).
1707 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1708 restoration of the capturing data to the outer values was not happening
1709 correctly.
1711 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1712 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1713 pattern had matched an empty string, and so incorrectly returned a no
1714 match.
1716 17. There was optimizing code for the last branch of non-capturing parentheses,
1717 and also for the obeyed branch of a conditional subexpression, which used
1718 tail recursion to cut down on stack usage. Unfortunately, now that there is
1719 the possibility of (*THEN) occurring in these branches, tail recursion is
1720 no longer possible because the return has to be checked for (*THEN). These
1721 two optimizations have therefore been removed. [But see 8.20/11 above.]
1723 18. If a pattern containing \R was studied, it was assumed that \R always
1724 matched two bytes, thus causing the minimum subject length to be
1725 incorrectly computed because \R can also match just one byte.
1727 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1728 was incorrectly computed.
1730 20. If /S is present twice on a test pattern in pcretest input, it now
1731 *disables* studying, thereby overriding the use of -s on the command line
1732 (see 14 above). This is necessary for one or two tests to keep the output
1733 identical in both cases.
1735 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1736 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1738 22. When an atomic group that contained a capturing parenthesis was
1739 successfully matched, but the branch in which it appeared failed, the
1740 capturing was not being forgotten if a higher numbered group was later
1741 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1742 group 1 to "a", when in fact it should be unset. This applied to multi-
1743 branched capturing and non-capturing groups, repeated or not, and also to
1744 positive assertions (capturing in negative assertions does not happen
1745 in PCRE) and also to nested atomic groups.
1747 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1748 subject after a captured substring, to make it easier to tell which of a
1749 number of identical substrings has been captured.
1751 24. The way atomic groups are processed by pcre_exec() has been changed so that
1752 if they are repeated, backtracking one repetition now resets captured
1753 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1754 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1755 "aaa". Previously, it would have been "a". As part of this code
1756 refactoring, the way recursive calls are handled has also been changed.
1758 25. If an assertion condition captured any substrings, they were not passed
1759 back unless some other capturing happened later. For example, if
1760 (?(?=(a))a) was matched against "a", no capturing was returned.
1762 26. When studying a pattern that contained subroutine calls or assertions,
1763 the code for finding the minimum length of a possible match was handling
1764 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1765 group 1 called group 2 while simultaneously a separate group 2 called group
1766 1). A stack overflow occurred in this case. I have fixed this by limiting
1767 the recursion depth to 10.
1769 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1770 Fortmann. This supports explicit test numbers on the command line, and has
1771 argument validation and error reporting.
1773 28. An instance of \X with an unlimited repeat could fail if at any point the
1774 first character it looked at was a mark character.
1776 29. Some minor code refactoring concerning Unicode properties and scripts
1777 should reduce the stack requirement of match() slightly.
1779 30. Added the '=' option to pcretest to check the setting of unused capturing
1780 slots at the end of the pattern, which are documented as being -1, but are
1781 not included in the return count.
1783 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1784 compiled something random. Now it gives a compile-time error (as does
1785 Perl).
1787 32. A *MARK encountered during the processing of a positive assertion is now
1788 recorded and passed back (compatible with Perl).
1790 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1791 had alternative anchored branches, the search for a second match in a line
1792 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1793 matched the line "0102" twice. The same bug affected patterns that started
1794 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1795 twice.
1797 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1798 does, and because of capturing effects, quantifying parenthesized
1799 assertions may at times be useful. Quantifiers are now allowed for
1800 parenthesized assertions.
1802 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1804 36. \g was being checked for fancy things in a character class, when it should
1805 just be a literal "g".
1807 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1808 appearance of a nested POSIX class supersedes an apparent external class.
1809 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1810 unescaped square brackets may also appear as part of class names. For
1811 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1812 more like Perl. (But see 8.20/1 above.)
1814 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1815 was because it thought it was \N{name}, which is not supported).
1817 39. Add minix to OS list not supporting the -S option in pcretest.
1819 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1820 cannot analyze patterns in sufficient detail to catch mutual recursions
1821 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1822 subgroup is called recursively as a subpattern for a second time at the
1823 same position in the subject string. In previous releases this might have
1824 been caught by the recursion limit, or it might have run out of stack.
1826 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1827 happen only once. PCRE was, however incorrectly giving a compile time error
1828 "recursive call could loop indefinitely" because it cannot analyze the
1829 pattern in sufficient detail. The compile time test no longer happens when
1830 PCRE is compiling a conditional subpattern, but actual runaway loops are
1831 now caught at runtime (see 40 above).
1833 42. It seems that Perl allows any characters other than a closing parenthesis
1834 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1835 has been changed to be the same.
1837 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1838 as not to get warnings when autogen.sh is called. Also changed
1839 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1841 44. To help people who use pcregrep to scan files containing exceedingly long
1842 lines, the following changes have been made:
1844 (a) The default value of the buffer size parameter has been increased from
1845 8K to 20K. (The actual buffer used is three times this size.)
1847 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1848 PCRE is built.
1850 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1851 to be set at run time.
1853 (d) Numerical values in pcregrep options can be followed by K or M, for
1854 example --buffer-size=50K.
1856 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1857 given and the return code is set to 2.
1859 45. Add a pointer to the latest mark to the callout data block.
1861 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1862 partial match of an empty string instead of no match. This was specific to
1863 the use of ".".
1865 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1866 complete match instead of a partial match. This bug was dependent on both
1867 the PCRE_UTF8 and PCRE_DOTALL options being set.
1869 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1870 starting byte set, because \b was not being ignored.
1873 Version 8.12 15-Jan-2011
1874 ------------------------
1876 1. Fixed some typos in the markup of the man pages, and wrote a script that
1877 checks for such things as part of the documentation building process.
1879 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1880 --match-limit and --recursion-limit options (added for 8.11). In
1881 particular, this made one of the standard tests fail. (The integer value
1882 went into the wrong half of a long int.)
1884 3. If the --colour option was given to pcregrep with -v (invert match), it
1885 did strange things, either producing crazy output, or crashing. It should,
1886 of course, ignore a request for colour when reporting lines that do not
1887 match.
1889 4. Another pcregrep bug caused similar problems if --colour was specified with
1890 -M (multiline) and the pattern match finished with a line ending.
1892 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1893 matched in multiline mode, the following line was shown as part of the
1894 match. This seems wrong, so I have changed it.
1896 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1897 the check for further matches in the same line (so they could be coloured)
1898 to overrun the end of the current line. If another match was found, it was
1899 incorrectly shown (and then shown again when found in the next line).
1901 7. If pcregrep was compiled under Windows, there was a reference to the
1902 function pcregrep_exit() before it was defined. I am assuming this was
1903 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1904 reported by a user. I've moved the definition above the reference.
1907 Version 8.11 10-Dec-2010
1908 ------------------------
1910 1. (*THEN) was not working properly if there were untried alternatives prior
1911 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1912 backtracked to try for "b" instead of moving to the next alternative branch
1913 at the same level (in this case, to look for "c"). The Perl documentation
1914 is clear that when (*THEN) is backtracked onto, it goes to the "next
1915 alternative in the innermost enclosing group".
1917 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1918 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1919 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1920 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1921 (*THEN).
1923 3. If \s appeared in a character class, it removed the VT character from
1924 the class, even if it had been included by some previous item, for example
1925 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1926 of \s, but is part of the POSIX "space" class.)
1928 4. A partial match never returns an empty string (because you can always
1929 match an empty string at the end of the subject); however the checking for
1930 an empty string was starting at the "start of match" point. This has been
1931 changed to the "earliest inspected character" point, because the returned
1932 data for a partial match starts at this character. This means that, for
1933 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1934 (previously it gave "no match").
1936 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1937 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1938 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1939 has an implication that the given string is incomplete (because a partial
1940 match is preferred over a full match). For this reason, these items now
1941 give a partial match in this situation. [Aside: previously, the one case
1942 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1943 match rather than a full match, which was wrong by the old rules, but is
1944 now correct.]
1946 6. There was a bug in the handling of #-introduced comments, recognized when
1947 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1948 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1949 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1950 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1951 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1952 places in pcre_compile().
1954 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1955 comments when looking ahead for named forward references to subpatterns,
1956 the only newline sequence it recognized was NL. It now handles newlines
1957 according to the set newline convention.
1959 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1960 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1961 cater for a lack of strerror(). These oversights have been fixed.
1963 9. Added --match-limit and --recursion-limit to pcregrep.
1965 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1967 11. When the -o option was used, pcregrep was setting a return code of 1, even
1968 when matches were found, and --line-buffered was not being honoured.
1970 12. Added an optional parentheses number to the -o and --only-matching options
1971 of pcregrep.
1973 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1974 can match an empty string. The code to do it in pcretest and pcredemo
1975 needed fixing:
1977 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1978 only one byte after an empty string match just before CRLF (this case
1979 just got forgotten; "any" and "anycrlf" were OK).
1981 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1982 mode when an empty string match preceded an ASCII character followed by
1983 a non-ASCII character. (The code for advancing by one character rather
1984 than one byte was nonsense.)
1986 (c) The pcredemo.c sample program did not have any code at all to handle
1987 the cases when CRLF is a valid newline sequence.
1989 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1990 as a starting offset was within the subject string. There is now a new
1991 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1992 negative or greater than the length of the string. In order to test this,
1993 pcretest is extended to allow the setting of negative starting offsets.
1995 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1996 starting offset points to the beginning of a UTF-8 character was
1997 unnecessarily clumsy. I tidied it up.
1999 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2000 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2002 17. Nobody had reported that the --include_dir option, which was added in
2003 release 7.7 should have been called --include-dir (hyphen, not underscore)
2004 for compatibility with GNU grep. I have changed it to --include-dir, but
2005 left --include_dir as an undocumented synonym, and the same for
2006 --exclude-dir, though that is not available in GNU grep, at least as of
2007 release 2.5.4.
2009 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2010 characters from a string of bytes) have been redefined so as not to use
2011 loops, in order to improve performance in some environments. At the same
2012 time, I abstracted some of the common code into auxiliary macros to save
2013 repetition (this should not affect the compiled code).
2015 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2016 compile-time error is now given if \c is not followed by an ASCII
2017 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2018 different, and any byte value is allowed.)
2020 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2021 START_OPTIMIZE option, which is now allowed at compile time - but just
2022 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2023 to pcregrep and other applications that have no direct access to PCRE
2024 options. The new /Y option in pcretest sets this option when calling
2025 pcre_compile().
2027 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2028 back references. Groups containing recursive back references were forced to
2029 be atomic by that change, but in the case of named groups, the amount of
2030 memory required was incorrectly computed, leading to "Failed: internal
2031 error: code overflow". This has been fixed.
2033 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2034 pcretest.c, to avoid build problems in some Borland environments.
2037 Version 8.10 25-Jun-2010
2038 ------------------------
2040 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2041 THEN.
2043 2. (*ACCEPT) was not working when inside an atomic group.
2045 3. Inside a character class, \B is treated as a literal by default, but
2046 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2047 causes the error). The code is unchanged, but I tidied the documentation.
2049 4. Inside a character class, PCRE always treated \R and \X as literals,
2050 whereas Perl faults them if its -w option is set. I have changed PCRE so
2051 that it faults them when PCRE_EXTRA is set.
2053 5. Added support for \N, which always matches any character other than
2054 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2056 6. When compiling pcregrep with newer versions of gcc which may have
2057 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2058 declared with attribute warn_unused_result" were given. Just casting the
2059 result to (void) does not stop the warnings; a more elaborate fudge is
2060 needed. I've used a macro to implement this.
2062 7. Minor change to pcretest.c to avoid a compiler warning.
2064 8. Added four artifical Unicode properties to help with an option to make
2065 \s etc use properties (see next item). The new properties are: Xan
2066 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2068 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2069 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2070 this option. Modified pcretest to add /W to test this facility. Added
2071 REG_UCP to make it available via the POSIX interface.
2073 10. Added --line-buffered to pcregrep.
2075 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2076 studied, and the match started with a letter with a code point greater than
2077 127 whose first byte was different to the first byte of the other case of
2078 the letter, the other case of this starting letter was not recognized
2079 (#976).
2081 12. If a pattern that was studied started with a repeated Unicode property
2082 test, for example, \p{Nd}+, there was the theoretical possibility of
2083 setting up an incorrect bitmap of starting bytes, but fortunately it could
2084 not have actually happened in practice until change 8 above was made (it
2085 added property types that matched character-matching opcodes).
2087 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2088 possible starting bytes for non-anchored patterns.
2090 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2091 \R, and also a number of cases that involve Unicode properties, both
2092 explicit and implicit when PCRE_UCP is set.
2094 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2095 input, it could crash or give wrong results if characters with values
2096 greater than 0xc0 were present in the subject string. (Detail: it assumed
2097 UTF-8 input when processing these items.)
2099 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2100 size_t is 64-bit (#991).
2102 17. Added a check for running out of memory when PCRE is compiled with
2103 --disable-stack-for-recursion (#990).
2105 18. If the last data line in a file for pcretest does not have a newline on
2106 the end, a newline was missing in the output.
2108 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2109 less than 128) in its various bitmaps. However, there is a facility for
2110 generating tables according to the current locale when PCRE is compiled. It
2111 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2112 characters, are recognized by isspace() and therefore were getting set in
2113 these tables, and indeed these tables seem to approximate to ISO 8859. This
2114 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2115 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2116 which of course cannot start UTF-8 characters. I have changed the code so
2117 that only real ASCII characters (less than 128) and the correct starting
2118 bytes for UTF-8 encodings are set for characters greater than 127 when in
2119 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2120 altogether.)
2122 20. Added the /T option to pcretest so as to be able to run tests with non-
2123 standard character tables, thus making it possible to include the tests
2124 used for 19 above in the standard set of tests.
2126 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2127 reference to a subpattern the other side of a comment that contains an
2128 opening parenthesis caused either an internal compiling error, or a
2129 reference to the wrong subpattern.
2132 Version 8.02 19-Mar-2010
2133 ------------------------
2135 1. The Unicode data tables have been updated to Unicode 5.2.0.
2137 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2138 configured.
2140 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2141 original author of that file, following a query about its status.
2143 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2144 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2146 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2147 quantifier applied to a forward-referencing subroutine call, could compile
2148 incorrect code or give the error "internal error: previously-checked
2149 referenced subpattern not found".
2151 6. Both MS Visual Studio and Symbian OS have problems with initializing
2152 variables to point to external functions. For these systems, therefore,
2153 pcre_malloc etc. are now initialized to local functions that call the
2154 relevant global functions.
2156 7. There were two entries missing in the vectors called coptable and poptable
2157 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2158 I've fixed the data, and added a kludgy way of testing at compile time that
2159 the lengths are correct (equal to the number of opcodes).
2161 8. Following on from 7, I added a similar kludge to check the length of the
2162 eint vector in pcreposix.c.
2164 9. Error texts for pcre_compile() are held as one long string to avoid too
2165 much relocation at load time. To find a text, the string is searched,
2166 counting zeros. There was no check for running off the end of the string,
2167 which could happen if a new error number was added without updating the
2168 string.
2170 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2172 11. \K was not working if it appeared in an atomic group or in a group that
2173 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2174 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2175 the assertion is positive, but not if it is negative.
2177 12. Change 11 fortuitously reduced the size of the stack frame used in the
2178 "match()" function of pcre_exec.c by one pointer. Forthcoming
2179 implementation of support for (*MARK) will need an extra pointer on the
2180 stack; I have reserved it now, so that the stack frame size does not
2181 decrease.
2183 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2184 item in branch that calls a recursion is a subroutine call - as in the
2185 second branch in the above example - was incorrectly given the compile-
2186 time error "recursive call could loop indefinitely" because pcre_compile()
2187 was not correctly checking the subroutine for matching a non-empty string.
2189 14. The checks for overrunning compiling workspace could trigger after an
2190 overrun had occurred. This is a "should never occur" error, but it can be
2191 triggered by pathological patterns such as hundreds of nested parentheses.
2192 The checks now trigger 100 bytes before the end of the workspace.
2194 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2197 Version 8.01 19-Jan-2010
2198 ------------------------
2200 1. If a pattern contained a conditional subpattern with only one branch (in
2201 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2202 computed the wrong minimum data length (which is of course zero for such
2203 subpatterns). This could cause incorrect "no match" results.
2205 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2206 the pattern is reset in the first branch, pcre_compile() failed with
2207 "internal error: code overflow at offset...". This happened only when
2208 the reset was to the original external option setting. (An optimization
2209 abstracts leading options settings into an external setting, which was the
2210 cause of this.)
2212 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2213 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2214 assertion pattern did not match (meaning that the assertion was true), it
2215 was incorrectly treated as false if the SKIP had been reached during the
2216 matching. This also applied to assertions used as conditions.
2218 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2219 assertion subpattern, including such a pattern used as a condition,
2220 unpredictable results occurred, instead of the error return
2223 5. The C++ GlobalReplace function was not working like Perl for the special
2224 situation when an empty string is matched. It now does the fancy magic
2225 stuff that is necessary.
2227 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2228 removed. (These were left over from very, very early versions of PCRE.)
2230 7. Some cosmetic changes to the code to make life easier when compiling it
2231 as part of something else:
2233 (a) Change DEBUG to PCRE_DEBUG.
2235 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2236 called "current" as "current_branch", to prevent a collision with the
2237 Linux macro when compiled as a kernel module.
2239 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2240 prevent a collision with the Linux macro when compiled as a kernel
2241 module.
2243 8. In pcre_compile() there are some checks for integer overflows that used to
2244 cast potentially large values to (double). This has been changed to that
2245 when building, a check for int64_t is made, and if it is found, it is used
2246 instead, thus avoiding the use of floating point arithmetic. (There is no
2247 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2248 double.
2250 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2251 2005 (difference between two addresses compared to an unsigned value).
2253 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2254 custom one, because of the following reported problem in Windows:
2256 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2257 under Win32.
2258 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2259 therefore missing the function definition.
2260 - The compiler thus generates a "C" signature for the test function.
2261 - The linker fails to find the "C" function.
2262 - PCRE fails to configure if asked to do so against libbz2.
2264 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2265 messages were output:
2267 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2268 rerunning libtoolize, to keep the correct libtool macros in-tree.
2269 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2271 I have done both of these things.
2273 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2274 most of the time, it *can* run out if it is given a pattern that contains a
2275 runaway infinite recursion. I updated the discussion in the pcrestack man
2276 page.
2278 13. Now that we have gone to the x.xx style of version numbers, the minor
2279 version may start with zero. Using 08 or 09 is a bad idea because users
2280 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2281 interpreted as invalid octal numbers. I've updated the previous comment in
2282 configure.ac, and also added a check that gives an error if 08 or 09 are
2283 used.
2285 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2286 causing partial matching to fail when the end of the subject matched \W
2287 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2289 15. There were some discrepancies between the declarations in pcre_internal.h
2290 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2291 their definitions. The declarations used "const uschar *" and the
2292 definitions used USPTR. Even though USPTR is normally defined as "const
2293 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2294 reported that: "This difference in casting confuses some C++ compilers, for
2295 example, SunCC recognizes above declarations as different functions and
2296 generates broken code for hbpcre." I have changed the declarations to use
2297 USPTR.
2299 16. GNU libtool is named differently on some systems. The autogen.sh script now
2300 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2301 (FreeBSD).
2303 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2304 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2305 comment: "Figure out how to create a longlong from a string: strtoll and
2306 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2307 instance, but it only takes 2 args instead of 3!"
2309 18. A subtle bug concerned with back references has been fixed by a change of
2310 specification, with a corresponding code fix. A pattern such as
2311 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2312 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2313 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2314 same bug. Such groups have to be quantified to be useful, or contained
2315 inside another quantified group. (If there's no repetition, the reference
2316 can never match.) The problem arises because, having left the group and
2317 moved on to the rest of the pattern, a later failure that backtracks into
2318 the group uses the captured value from the final iteration of the group
2319 rather than the correct earlier one. I have fixed this in PCRE by forcing
2320 any group that contains a reference to itself to be an atomic group; that
2321 is, there cannot be any backtracking into it once it has completed. This is
2322 similar to recursive and subroutine calls.
2325 Version 8.00 19-Oct-09
2326 ----------------------
2328 1. The table for translating pcre_compile() error codes into POSIX error codes
2329 was out-of-date, and there was no check on the pcre_compile() error code
2330 being within the table. This could lead to an OK return being given in
2331 error.
2333 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2334 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2335 in a Windows environment.
2337 3. The pcregrep --count option prints the count for each file even when it is
2338 zero, as does GNU grep. However, pcregrep was also printing all files when
2339 --files-with-matches was added. Now, when both options are given, it prints
2340 counts only for those files that have at least one match. (GNU grep just
2341 prints the file name in this circumstance, but including the count seems
2342 more useful - otherwise, why use --count?) Also ensured that the
2343 combination -clh just lists non-zero counts, with no names.
2345 4. The long form of the pcregrep -F option was incorrectly implemented as
2346 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2347 but it seems right to fix it, and I didn't think it was worth preserving
2348 the old behaviour.
2350 5. The command line items --regex=pattern and --regexp=pattern were not
2351 recognized by pcregrep, which required --regex pattern or --regexp pattern
2352 (with a space rather than an '='). The man page documented the '=' forms,
2353 which are compatible with GNU grep; these now work.
2355 6. No libpcreposix.pc file was created for pkg-config; there was just
2356 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2358 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2359 when UCP support is not needed, by modifying the Python script that
2360 generates it from Unicode data files. This should not matter if the module
2361 is correctly used as a library, but I received one complaint about 50K of
2362 unwanted data. My guess is that the person linked everything into his
2363 program rather than using a library. Anyway, it does no harm.
2365 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2366 was a minimum greater than 1 for a wide character in a possessive
2367 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2368 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2369 character. Chaos in the form of incorrect output or a compiling loop could
2370 result.
2372 9. The restrictions on what a pattern can contain when partial matching is
2373 requested for pcre_exec() have been removed. All patterns can now be
2374 partially matched by this function. In addition, if there are at least two
2375 slots in the offset vector, the offset of the earliest inspected character
2376 for the match and the offset of the end of the subject are set in them when
2377 PCRE_ERROR_PARTIAL is returned.
2379 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2380 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2381 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2382 and may be more useful for multi-segment matching.
2384 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2385 used to be given if ever the end of the subject was reached; now it is
2386 given only if matching could not proceed because another character was
2387 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2388 string "Z", which now yields "no match" instead of "partial match". In the
2389 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2390 final character ended with (*FAIL).
2392 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2393 if the pattern had a "must contain" character that was already found in the
2394 earlier partial match, unless partial matching was again requested. For
2395 example, with the pattern /dog.(body)?/, the "must contain" character is
2396 "g". If the first part-match was for the string "dog", restarting with
2397 "sbody" failed. This bug has been fixed.
2399 13. The string returned by pcre_dfa_exec() after a partial match has been
2400 changed so that it starts at the first inspected character rather than the
2401 first character of the match. This makes a difference only if the pattern
2402 starts with a lookbehind assertion or \b or \B (\K is not supported by
2403 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2404 matching functions compatible, and I think it's the right thing to do.
2406 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2407 so that the demonstration program is easily available in environments where
2408 PCRE has not been installed from source.
2410 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2411 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2412 library.
2414 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2415 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2416 is not the first non-POSIX option to be added. Clearly some people find
2417 these options useful.
2419 17. If a caller to the POSIX matching function regexec() passes a non-zero
2420 value for nmatch with a NULL value for pmatch, the value of
2421 nmatch is forced to zero.
2423 18. RunGrepTest did not have a test for the availability of the -u option of
2424 the diff command, as RunTest does. It now checks in the same way as
2425 RunTest, and also checks for the -b option.
2427 19. If an odd number of negated classes containing just a single character
2428 interposed, within parentheses, between a forward reference to a named
2429 subpattern and the definition of the subpattern, compilation crashed with
2430 an internal error, complaining that it could not find the referenced
2431 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2432 [The bug was that it was starting one character too far in when skipping
2433 over the character class, thus treating the ] as data rather than
2434 terminating the class. This meant it could skip too much.]
2436 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2437 /g option in pcretest when the pattern contains \K, which makes it possible
2438 to have an empty string match not at the start, even when the pattern is
2439 anchored. Updated pcretest and pcredemo to use this option.
2441 21. If the maximum number of capturing subpatterns in a recursion was greater
2442 than the maximum at the outer level, the higher number was returned, but
2443 with unset values at the outer level. The correct (outer level) value is
2444 now given.
2446 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2447 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2448 make it do so. The string so far is captured, making this feature
2449 compatible with Perl.
2451 23. The tests have been re-organized, adding tests 11 and 12, to make it
2452 possible to check the Perl 5.10 features against Perl 5.10.
2454 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2455 pattern matches a fixed length string. PCRE did not allow this; now it
2456 does. Neither allows recursion.
2458 25. I finally figured out how to implement a request to provide the minimum
2459 length of subject string that was needed in order to match a given pattern.
2460 (It was back references and recursion that I had previously got hung up
2461 on.) This code has now been added to pcre_study(); it finds a lower bound
2462 to the length of subject needed. It is not necessarily the greatest lower
2463 bound, but using it to avoid searching strings that are too short does give
2464 some useful speed-ups. The value is available to calling programs via
2465 pcre_fullinfo().
2467 26. While implementing 25, I discovered to my embarrassment that pcretest had
2468 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2469 study optimizations had never been tested with that matching function.
2470 Oops. What is worse, even when it was passed study data, there was a bug in
2471 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2472 were also very few tests of studied patterns with pcre_dfa_exec().
2474 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2475 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2476 on the other side of the coin, they are no longer allowed to have different
2477 names, because these cannot be distinguished in PCRE, and this has caused
2478 confusion. (This is a difference from Perl.)
2480 28. When duplicate subpattern names are present (necessarily with different
2481 numbers, as required by 27 above), and a test is made by name in a
2482 conditional pattern, either for a subpattern having been matched, or for
2483 recursion in such a pattern, all the associated numbered subpatterns are
2484 tested, and the overall condition is true if the condition is true for any
2485 one of them. This is the way Perl works, and is also more like the way
2486 testing by number works.
2489 Version 7.9 11-Apr-09
2490 ---------------------
2492 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2493 (pcretest), all targets were linked against these libraries. This included
2494 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2495 libraries. This caused unwanted dependencies to be created. This problem
2496 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2497 pcretest is linked with readline.
2499 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2500 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2501 moved outside it again, because FALSE and TRUE are already defined in AIX,
2502 but BOOL is not.
2504 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2505 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2507 4. The pcregrep documentation said spaces were inserted as well as colons (or
2508 hyphens) following file names and line numbers when outputting matching
2509 lines. This is not true; no spaces are inserted. I have also clarified the
2510 wording for the --colour (or --color) option.
2512 5. In pcregrep, when --colour was used with -o, the list of matching strings
2513 was not coloured; this is different to GNU grep, so I have changed it to be
2514 the same.
2516 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2517 each matching line was coloured. Now it goes on to look for further matches
2518 of any of the test patterns, which is the same behaviour as GNU grep.
2520 7. A pattern that could match an empty string could cause pcregrep to loop; it
2521 doesn't make sense to accept an empty string match in pcregrep, so I have
2522 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2523 seems to be how GNU grep behaves. [But see later change 40 for release
2524 8.33.]
2526 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2527 start or after a newline", because the conditional assertion was not being
2528 correctly handled. The rule now is that both the assertion and what follows
2529 in the first alternative must satisfy the test.
2531 9. If auto-callout was enabled in a pattern with a conditional group whose
2532 condition was an assertion, PCRE could crash during matching, both with
2533 pcre_exec() and pcre_dfa_exec().
2535 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2536 used for matching.
2538 11. Unicode property support in character classes was not working for
2539 characters (bytes) greater than 127 when not in UTF-8 mode.
2541 12. Added the -M command line option to pcretest.
2543 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2545 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2547 16. Added comments and documentation about mis-use of no_arg in the C++
2548 wrapper.
2550 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2551 from Martin Jerabek that uses macro names for all relevant character and
2552 string constants.
2554 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2555 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2556 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2557 these, but not everybody uses configure.
2559 19. A conditional group that had only one branch was not being correctly
2560 recognized as an item that could match an empty string. This meant that an
2561 enclosing group might also not be so recognized, causing infinite looping
2562 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2563 with the subject "ab", where knowledge that the repeated group can match
2564 nothing is needed in order to break the loop.
2566 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2567 exec(), but without supplying a callout function, matching went wrong.
2569 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2570 leak if the size of the offset vector was greater than 30. When the vector
2571 is smaller, the saved offsets during recursion go onto a local stack
2572 vector, but for larger vectors malloc() is used. It was failing to free
2573 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2574 error, in fact).
2576 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2577 heapframe that is used only when UTF-8 support is enabled. This caused no
2578 problem, but was untidy.
2580 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2581 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2582 included within another project.
2584 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2585 slightly modified by me:
2587 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2588 not building pcregrep.
2590 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2591 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2593 25. Forward references, both numeric and by name, in patterns that made use of
2594 duplicate group numbers, could behave incorrectly or give incorrect errors,
2595 because when scanning forward to find the reference group, PCRE was not
2596 taking into account the duplicate group numbers. A pattern such as
2597 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2599 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2600 the feature of a custom pointer more persuasive (as requested by a user).
2602 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2603 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2604 pre-defined.
2606 28. Added support for (*UTF8) at the start of a pattern.
2608 29. Arrange for flags added by the "release type" setting in CMake to be shown
2609 in the configuration summary.
2612 Version 7.8 05-Sep-08
2613 ---------------------
2615 1. Replaced UCP searching code with optimized version as implemented for Ad
2616 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2617 stage table and inline lookup instead of a function, giving speed ups of 2
2618 to 5 times on some simple patterns that I tested. Permission was given to
2619 distribute the MultiStage2.py script that generates the tables (it's not in
2620 the tarball, but is in the Subversion repository).
2622 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2623 scripts.
2625 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2626 a group with a zero qualifier. The result of the study could be incorrect,
2627 or the function might crash, depending on the pattern.
2629 4. Caseless matching was not working for non-ASCII characters in back
2630 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2631 It now works when Unicode Property Support is available.
2633 5. In pcretest, an escape such as \x{de} in the data was always generating
2634 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2635 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2636 truncation.
2638 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2640 7. Added two (int) casts to pcregrep when printing the difference of two
2641 pointers, in case they are 64-bit values.
2643 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2644 test 2 if it fails.
2646 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2647 and a #define of that name to empty if it is not externally set. This is to
2648 allow users of MSVC to set it if necessary.
2650 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2651 the convenience functions in the pcre_get.c source file.
2653 11. An option change at the start of a pattern that had top-level alternatives
2654 could cause overwriting and/or a crash. This command provoked a crash in
2655 some environments:
2657 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2659 This potential security problem was recorded as CVE-2008-2371.
2661 12. For a pattern where the match had to start at the beginning or immediately
2662 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2663 pcre_dfa_exec() could read past the end of the passed subject if there was
2664 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2665 pcretest so that it places the subject at the end of its malloc-ed buffer.
2667 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2668 exec() might read past the end of the data buffer in UTF-8 mode.
2670 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2671 the data contained the byte 0x85 as part of a UTF-8 character within its
2672 first line. This applied both to normal and DFA matching.
2674 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2675 /^[^d]*?$/8 failed to match "abc".
2677 16. Added a missing copyright notice to pcrecpp_internal.h.
2679 17. Make it more clear in the documentation that values returned from
2680 pcre_exec() in ovector are byte offsets, not character counts.
2682 18. Tidied a few places to stop certain compilers from issuing warnings.
2684 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2685 supplied by Stefan Weber. I made a further small update for 7.8 because
2686 there is a change of source arrangements: the pcre_searchfuncs.c module is
2687 replaced by pcre_ucd.c.
2690 Version 7.7 07-May-08
2691 ---------------------
2693 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2694 a string to a long long, pretend we don't even have a long long." This is
2695 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2697 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2698 pre-7.6 versions, which defined a global no_arg variable instead of putting
2699 it in the RE class. (See also #8 below.)
2701 3. Remove a line of dead code, identified by coverity and reported by Nuno
2702 Lopes.
2704 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2706 (1) The include/exclude patterns were being applied to the whole pathnames
2707 of files, instead of just to the final components.
2709 (2) If there was more than one level of directory, the subdirectories were
2710 skipped unless they satisfied the include/exclude conditions. This is
2711 inconsistent with GNU grep (and could even be seen as contrary to the
2712 pcregrep specification - which I improved to make it absolutely clear).
2713 The action now is always to scan all levels of directory, and just
2714 apply the include/exclude patterns to regular files.
2716 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2717 --exclude_dir in the tests to avoid scanning .svn directories.
2719 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2720 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2721 doesn't support NULs in patterns.
2723 7. Added some missing "const"s to declarations of static tables in
2724 pcre_compile.c and pcre_dfa_exec.c.
2726 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2727 caused by fix #2 above. (Subsequently also a second patch to fix the
2728 first patch. And a third patch - this was a messy problem.)
2730 9. Applied Craig's patch to remove the use of push_back().
2732 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2733 matching function regexec().
2735 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2736 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2737 references. PCRE supports relative numbers with this syntax (I don't think
2738 Oniguruma does).
2740 12. Previously, a group with a zero repeat such as (...){0} was completely
2741 omitted from the compiled regex. However, this means that if the group
2742 was called as a subroutine from elsewhere in the pattern, things went wrong
2743 (an internal error was given). Such groups are now left in the compiled
2744 pattern, with a new opcode that causes them to be skipped at execution
2745 time.
2747 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2748 to the way PCRE behaves:
2750 (a) A lone ] character is dis-allowed (Perl treats it as data).
2752 (b) A back reference to an unmatched subpattern matches an empty string
2753 (Perl fails the current match path).
2755 (c) A data ] in a character class must be notated as \] because if the
2756 first data character in a class is ], it defines an empty class. (In
2757 Perl it is not possible to have an empty class.) The empty class []
2758 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2759 The negative empty class [^] matches any one character, independently
2760 of the DOTALL setting.
2762 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2763 non-existent subpattern following a character class starting with ']' and
2764 containing () gave an internal compiling error instead of "reference to
2765 non-existent subpattern". Fortunately, when the pattern did exist, the
2766 compiled code was correct. (When scanning forwards to check for the
2767 existence of the subpattern, it was treating the data ']' as terminating
2768 the class, so got the count wrong. When actually compiling, the reference
2769 was subsequently set up correctly.)
2771 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2772 it was being rejected as not supported by pcre_dfa_exec(), even though
2773 other assertions are supported. I have made pcre_dfa_exec() support
2774 (*FAIL).
2776 16. The implementation of 13c above involved the invention of a new opcode,
2777 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2778 cannot be changed at match time, I realized I could make a small
2779 improvement to matching performance by compiling OP_ALLANY instead of
2780 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2781 on the OP_ANY path.
2783 17. Compiling pcretest on Windows with readline support failed without the
2784 following two fixes: (1) Make the unistd.h include conditional on
2785 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2787 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2788 ncurses library to be included for pcretest when ReadLine support is
2789 requested, but also to allow for it to be overridden. This patch came from
2790 Daniel Bergström.
2792 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2793 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2794 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2795 spotting this.
2798 Version 7.6 28-Jan-08
2799 ---------------------
2801 1. A character class containing a very large number of characters with
2802 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2803 overflow.
2805 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2806 HAVE_LONG_LONG is not defined.
2808 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2809 bring them up to date and include new features. This patch includes:
2811 - Fixed PH's badly added libz and libbz2 support.
2812 - Fixed a problem with static linking.
2813 - Added pcredemo. [But later removed - see 7 below.]
2814 - Fixed dftables problem and added an option.
2815 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2817 - Added readline support for pcretest.
2818 - Added an listing of the option settings after cmake has run.
2820 4. A user submitted a patch to Makefile that makes it easy to create
2821 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2822 Makefile.am that cause it to include this special target, without
2823 affecting anything else. Note that the same mingw target plus all
2824 the other distribution libraries and programs are now supported
2825 when configuring with CMake (see 6 below) instead of with
2826 Configure/Make.
2828 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2829 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2830 exported in the Windows port". It has not yet been confirmed that the patch
2831 solves the problem, but it does no harm.
2833 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2834 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2835 with CMake, and also correct the comment about stack recursion.
2837 7. Remove the automatic building of pcredemo from the ./configure system and
2838 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2839 of a program that users should build themselves after PCRE is installed, so
2840 building it automatically is not really right. What is more, it gave
2841 trouble in some build environments.
2843 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2846 Version 7.5 10-Jan-08
2847 ---------------------
2849 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2850 values in parens when parsing an RE using the C++ wrapper."
2852 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2853 Characters greater than 255 were excluded from the class instead of being
2854 included.
2856 3. The same bug as (2) above applied to negated POSIX classes such as
2857 [:^space:].
2859 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2860 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2861 I have changed it.
2863 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2864 first named subpattern) and a construct such as (?&a) would reference the
2865 first named subpattern whose name started with "a" (in other words, the
2866 length check was missing). Both these problems are fixed. "Subpattern name
2867 expected" is now given for (?&) (a zero-length name), and this patch also
2868 makes it give the same error for \k'' (previously it complained that that
2869 was a reference to a non-existent subpattern).
2871 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2872 this is right because (?- can be followed by option settings as well as by
2873 digits. I have, however, made the messages clearer.
2875 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2876 than the number used in the conditional) now cause a compile-time error.
2877 This is actually not compatible with Perl, which accepts such patterns, but
2878 treats the conditional as always being FALSE (as PCRE used to), but it
2879 seems to me that giving a diagnostic is better.
2881 8. Change "alphameric" to the more common word "alphanumeric" in comments
2882 and messages.
2884 9. Fix two occurrences of "backslash" in comments that should have been
2885 "backspace".
2887 10. Remove two redundant lines of code that can never be obeyed (their function
2888 was moved elsewhere).
2890 11. The program that makes PCRE's Unicode character property table had a bug
2891 which caused it to generate incorrect table entries for sequences of
2892 characters that have the same character type, but are in different scripts.
2893 It amalgamated them into a single range, with the script of the first of
2894 them. In other words, some characters were in the wrong script. There were
2895 thirteen such cases, affecting characters in the following ranges:
2897 U+002b0 - U+002c1
2898 U+0060c - U+0060d
2899 U+0061e - U+00612
2900 U+0064b - U+0065e
2901 U+0074d - U+0076d
2902 U+01800 - U+01805
2903 U+01d00 - U+01d77
2904 U+01d9b - U+01dbf
2905 U+0200b - U+0200f
2906 U+030fc - U+030fe
2907 U+03260 - U+0327f
2908 U+0fb46 - U+0fbb1
2909 U+10450 - U+1049d
2911 12. The -o option (show only the matching part of a line) for pcregrep was not
2912 compatible with GNU grep in that, if there was more than one match in a
2913 line, it showed only the first of them. It now behaves in the same way as
2914 GNU grep.
2916 13. If the -o and -v options were combined for pcregrep, it printed a blank
2917 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2918 does the same. The return code can be used to tell if there were any
2919 non-matching lines.
2921 14. Added --file-offsets and --line-offsets to pcregrep.
2923 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2924 infinitely looping recursion. The bug was that positive lookaheads were not
2925 being skipped when checking for a possible empty match (negative lookaheads
2926 and both kinds of lookbehind were skipped).
2928 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2929 inclusion of <windows.h> to before rather than after the definition of
2930 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2932 17. Specifying a possessive quantifier with a specific limit for a Unicode
2933 character property caused pcre_compile() to compile bad code, which led at
2934 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2935 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2936 caused the error; without that there was no problem.
2938 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2940 19. Added --enable-pcretest-libreadline.
2942 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2943 RE::GlobalReplace(). As a result, the number of replacements returned was
2944 double what it should be. I removed one of the increments, but Craig sent a
2945 later patch that removed the other one (the right fix) and added unit tests
2946 that check the return values (which was not done before).
2948 21. Several CMake things:
2950 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2951 the names libpcre and libpcreposix, not just pcre and pcreposix.
2953 (2) The above change means that pcretest and pcregrep are now correctly
2954 linked with the newly-built libraries, not previously installed ones.
2958 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2959 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2960 UTF-8 newline character). The key issue is that the pattern starts .*;
2961 this means that the match must be either at the beginning, or after a
2962 newline. The bug was in the code for advancing after a failed match and
2963 checking that the new position followed a newline. It was not taking
2964 account of UTF-8 characters correctly.
2966 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2967 character classes. PCRE was not treating the sequence [:...:] as a
2968 character class unless the ... were all letters. Perl, however, seems to
2969 allow any characters between [: and :], though of course it rejects as
2970 unknown any "names" that contain non-letters, because all the known class
2971 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2972 for example, whereas PCRE did not - it did not recognize a POSIX character
2973 class. This seemed a bit dangerous, so the code has been changed to be
2974 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2975 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2976 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2977 Perl does, and where it didn't before.
2979 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2980 Windows environments %n is disabled by default.
2983 Version 7.4 21-Sep-07
2984 ---------------------
2986 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2987 means that a class such as [\s] counted as "explicit reference to CR or
2988 LF". That isn't really right - the whole point of the change was to try to
2989 help when there was an actual mention of one of the two characters. So now
2990 the change happens only if \r or \n (or a literal CR or LF) character is
2991 encountered.
2993 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2994 of both had grown to the point where there were only 3 bits left.
2995 Fortunately, there was spare space in the data structure, and so I have
2996 moved the internal flags into a new 16-bit field to free up more option
2997 bits.
2999 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3000 but did not set the internal JCHANGED flag - either of these is enough to
3001 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3002 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3003 start sets both bits.
3005 4. Added options (at build time, compile time, exec time) to change \R from
3006 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3008 5. doc/pcresyntax.html was missing from the distribution.
3010 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3011 compatibility, even though it is no longer used.
3013 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3014 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3015 windows.h file is present (where different names are used). [This was
3016 reversed later after testing - see 16 below.]
3018 8. Changed all #include <config.h> to #include "config.h". There were also
3019 some further <pcre.h> cases that I changed to "pcre.h".
3021 9. When pcregrep was used with the --colour option, it missed the line ending
3022 sequence off the lines that it output.
3024 10. It was pointed out to me that arrays of string pointers cause lots of
3025 relocations when a shared library is dynamically loaded. A technique of
3026 using a single long string with a table of offsets can drastically reduce
3027 these. I have refactored PCRE in four places to do this. The result is
3028 dramatic:
3030 Originally: 290
3031 After changing UCP table: 187
3032 After changing error message table: 43
3033 After changing table of "verbs" 36
3034 After changing table of Posix names 22
3036 Thanks to the folks working on Gregex for glib for this insight.
3038 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3039 unicode-properties was also set.
3041 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3043 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3044 checked only for CRLF.
3046 14. Added casts to pcretest.c to avoid compiler warnings.
3048 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3050 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3051 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3052 entirely. This removes changes made in 7 above.
3054 17. The CMake files have been updated, and there is now more information about
3055 building with CMake in the NON-UNIX-USE document.
3058 Version 7.3 28-Aug-07
3059 ---------------------
3061 1. In the rejigging of the build system that eventually resulted in 7.1, the
3062 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3063 brackets there is not right, since it causes compilers to look for an
3064 installed pcre.h, not the version that is in the source that is being
3065 compiled (which of course may be different). I have changed it back to:
3067 #include "pcre.h"
3069 I have a vague recollection that the change was concerned with compiling in
3070 different directories, but in the new build system, that is taken care of
3071 by the VPATH setting the Makefile.
3073 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3074 when the subject happened to end in the byte 0x85 (e.g. if the last
3075 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3076 characters but of course it shouldn't be taken as a newline when it is part
3077 of another character. The bug was that, for an unlimited repeat of . in
3078 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3079 characters when looking for a newline.
3081 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3083 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3084 in debug output.
3086 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3087 long printing in the pcrecpp unittest when running under MinGW.
3089 6. ESC_K was left out of the EBCDIC table.
3091 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3092 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3093 limit also applies to "virtual nesting" when a pattern is recursive, and in
3094 this case 1000 isn't so big. I have been able to remove this limit at the
3095 expense of backing off one optimization in certain circumstances. Normally,
3096 when pcre_exec() would call its internal match() function recursively and
3097 immediately return the result unconditionally, it uses a "tail recursion"
3098 feature to save stack. However, when a subpattern that can match an empty
3099 string has an unlimited repetition quantifier, it no longer makes this
3100 optimization. That gives it a stack frame in which to save the data for
3101 checking that an empty string has been matched. Previously this was taken
3102 from the 1000-entry workspace that had been reserved. So now there is no
3103 explicit limit, but more stack is used.
3105 8. Applied Daniel's patches to solve problems with the import/export magic
3106 syntax that is required for Windows, and which was going wrong for the
3107 pcreposix and pcrecpp parts of the library. These were overlooked when this
3108 problem was solved for the main library.
3110 9. There were some crude static tests to avoid integer overflow when computing
3111 the size of patterns that contain repeated groups with explicit upper
3112 limits. As the maximum quantifier is 65535, the maximum group length was
3113 set at 30,000 so that the product of these two numbers did not overflow a
3114 32-bit integer. However, it turns out that people want to use groups that
3115 are longer than 30,000 bytes (though not repeat them that many times).
3116 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3117 made it possible to implement the integer overflow checks in a much more
3118 dynamic way, which I have now done. The artificial limitation on group
3119 length has been removed - we now have only the limit on the total length of
3120 the compiled pattern, which depends on the LINK_SIZE setting.
3122 10. Fixed a bug in the documentation for get/copy named substring when
3123 duplicate names are permitted. If none of the named substrings are set, the
3124 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3125 empty string.
3127 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3128 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3129 because the ] is interpreted as the first data character and the
3130 terminating ] is not found. PCRE has been made compatible with Perl in this
3131 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3132 cause memory overwriting.
3134 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3135 string has been matched (to stop an infinite loop). It was not recognizing
3136 a conditional subpattern that could match an empty string if that
3137 subpattern was within another subpattern. For example, it looped when
3138 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3139 condition was not nested. This bug has been fixed.
3141 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3142 past the start of the subject in the presence of bytes with the top bit
3143 set, for example "\x8aBCD".
3145 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3146 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3148 14. Optimized (?!) to (*FAIL).
3150 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3151 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3152 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3153 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3154 does: it's just the validity check that is more restrictive.
3156 16. Inserted checks for integer overflows during escape sequence (backslash)
3157 processing, and also fixed erroneous offset values for syntax errors during
3158 backslash processing.
3160 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3161 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3163 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3164 caused an overrun.
3166 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3167 something other than just ASCII characters) inside a group that had an
3168 unlimited repeat caused a loop at compile time (while checking to see
3169 whether the group could match an empty string).
3171 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3172 [\P{Any}] did so. (Error in the code for printing property names.)
3174 21. An orphan \E inside a character class could cause a crash.
3176 22. A repeated capturing bracket such as (A)? could cause a wild memory
3177 reference during compilation.
3179 23. There are several functions in pcre_compile() that scan along a compiled
3180 expression for various reasons (e.g. to see if it's fixed length for look
3181 behind). There were bugs in these functions when a repeated \p or \P was
3182 present in the pattern. These operators have additional parameters compared
3183 with \d, etc, and these were not being taken into account when moving along
3184 the compiled data. Specifically:
3186 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3187 length.
3189 (b) An item such as \pL+ within a repeated group could cause crashes or
3190 loops.
3192 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3193 "reference to non-existent subpattern" error.
3195 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3197 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3198 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3200 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3202 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3203 character were causing crashes (broken optimization).
3205 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3206 \p or \P) caused a compile-time loop.
3208 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3209 break. For example, the unstudied pattern [\r\n]A does not match the string
3210 "\r\nA" because change 7.0/46 below moves the current point on by two
3211 characters after failing to match at the start. However, the pattern \nA
3212 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3213 the same is true. There doesn't seem any very clean way out of this, but
3214 what I have chosen to do makes the common cases work: PCRE now takes note
3215 of whether there can be an explicit match for \r or \n anywhere in the
3216 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3217 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3218 pattern has explicit CR or LF references.
3220 29. Added (*CR) etc for changing newline setting at start of pattern.
3223 Version 7.2 19-Jun-07
3224 ---------------------
3226 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3227 which is apparently normally available under Windows.
3229 2. Re-jig the pcregrep tests with different newline settings in an attempt
3230 to make them independent of the local environment's newline setting.
3232 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3234 4. Some of the "internals" tests were previously cut out when the link size
3235 was not 2, because the output contained actual offsets. The recent new
3236 "Z" feature of pcretest means that these can be cut out, making the tests
3237 usable with all link sizes.
3239 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3240 stack recursion. This gives a massive performance boost under BSD, but just
3241 a small improvement under Linux. However, it saves one field in the frame
3242 in all cases.
3244 6. Added more features from the forthcoming Perl 5.10:
3246 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3247 recursion call. It refers to the nth most recently opened parentheses.
3249 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3250 to be opened parentheses.
3252 (c) Conditions that refer to capturing parentheses can be specified
3253 relatively, for example, (?(-2)... or (?(+3)...
3255 (d) \K resets the start of the current match so that everything before
3256 is not part of it.
3258 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3260 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3261 reference syntax.
3263 (g) (?| introduces a group in which the numbering of parentheses in each
3264 alternative starts with the same number.
3266 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3268 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3271 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3272 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3273 for detecting groups that can match an empty string.
3275 9. A pattern with a very large number of alternatives (more than several
3276 hundred) was running out of internal workspace during the pre-compile
3277 phase, where pcre_compile() figures out how much memory will be needed. A
3278 bit of new cunning has reduced the workspace needed for groups with
3279 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3280 workspace instead of running out of the 4096 that are available.
3282 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3284 11. Applied patch from Google to remove an optimization that didn't quite work.
3285 The report of the bug said:
3287 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3288 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3289 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3291 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3292 it matched the wrong number of bytes.
3295 Version 7.1 24-Apr-07
3296 ---------------------
3298 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3299 that is more "standard", making use of automake and other Autotools. There
3300 is some re-arrangement of the files and adjustment of comments consequent
3301 on this.
3303 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3304 for recursive directory scanning broke on some systems because the files
3305 are not scanned in any specific order and on different systems the order
3306 was different. A call to "sort" has been inserted into RunGrepTest for the
3307 approprate test as a short-term fix. In the longer term there may be an
3308 alternative.
3310 3. I had an email from Eric Raymond about problems translating some of PCRE's
3311 man pages to HTML (despite the fact that I distribute HTML pages, some
3312 people do their own conversions for various reasons). The problems
3313 concerned the use of low-level troff macros .br and .in. I have therefore
3314 removed all such uses from the man pages (some were redundant, some could
3315 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3316 HTML has been updated to handle .nf/.fi and to complain if it encounters
3317 .br or .in.
3319 4. Updated comments in configure.ac that get placed in config.h.in and also
3320 arranged for config.h to be included in the distribution, with the name
3321 config.h.generic, for the benefit of those who have to compile without
3322 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3324 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3325 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3326 makevp.bat for the current PCRE, using the additional files
3327 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3329 6. A Windows user reported a minor discrepancy with test 2, which turned out
3330 to be caused by a trailing space on an input line that had got lost in his
3331 copy. The trailing space was an accident, so I've just removed it.
3333 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3334 that is needed.
3336 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3337 as "const" (a) because they are and (b) because it helps the PHP
3338 maintainers who have recently made a script to detect big data structures
3339 in the php code that should be moved to the .rodata section. I remembered
3340 to update Builducptable as well, so it won't revert if ucptable.h is ever
3341 re-created.
3343 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3344 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3345 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3346 support is not required. This saves 1.5-2K of code, which is important in
3347 some applications.
3349 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3350 so as not to refer to the tables, even though these functions will never be
3351 called when UTF-8 support is disabled. Otherwise there are problems with a
3352 shared library.
3354 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3356 (a) It was defining its arguments as char * instead of void *.
3358 (b) It was assuming that all moves were upwards in memory; this was true
3359 a long time ago when I wrote it, but is no longer the case.
3361 The emulated memove() is provided for those environments that have neither
3362 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3363 is clearly not the case, as these two bugs were recently reported.
3365 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3366 and Detrail to create the HTML documentation, the .txt form of the man
3367 pages, and it removes trailing spaces from listed files. It also creates
3368 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3369 case, it wraps all the #defines with #ifndefs. This script should be run
3370 before "make dist".
3372 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3373 with Unicode property support.
3375 (a) For a maximizing quantifier, if the two different cases of the
3376 character were of different lengths in their UTF-8 codings (there are
3377 some cases like this - I found 11), and the matching function had to
3378 back up over a mixture of the two cases, it incorrectly assumed they
3379 were both the same length.
3381 (b) When PCRE was configured to use the heap rather than the stack for
3382 recursion during matching, it was not correctly preserving the data for
3383 the other case of a UTF-8 character when checking ahead for a match
3384 while processing a minimizing repeat. If the check also involved
3385 matching a wide character, but failed, corruption could cause an
3386 erroneous result when trying to check for a repeat of the original
3387 character.
3389 13. Some tidying changes to the testing mechanism:
3391 (a) The RunTest script now detects the internal link size and whether there
3392 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3393 values substituted by "configure". (The RunGrepTest script already did
3394 this for UTF-8.) The configure.ac script no longer substitutes the
3395 relevant variables.
3397 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3398 with length and offset values. This means that the output is different
3399 for different internal link sizes. Test 2 is skipped for link sizes
3400 other than 2 because of this, bypassing the problem. Unfortunately,
3401 there was also a test in test 3 (the locale tests) that used /B and
3402 failed for link sizes other than 2. Rather than cut the whole test out,
3403 I have added a new /Z option to pcretest that replaces the length and
3404 offset values with spaces. This is now used to make test 3 independent
3405 of link size. (Test 2 will be tidied up later.)
3407 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3408 segmentation fault instead of returning the appropriate error message.
3410 15. In multiline mode when the newline sequence was set to "any", the pattern
3411 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3412 This doesn't seem right; it now treats the CRLF combination as the line
3413 ending, and so does not match in that case. It's only a pattern such as ^$
3414 that would hit this one: something like ^ABC$ would have failed after \r
3415 and then tried again after \r\n.
3417 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3418 in an attempt to make files that differ only in their line terminators
3419 compare equal. This works on Linux.
3421 17. Under certain error circumstances pcregrep might try to free random memory
3422 as it exited. This is now fixed, thanks to valgrind.
3424 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3425 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3426 was because its rules for how to advance for /g after matching an empty
3427 string at the end of a line did not allow for this case. They now check for
3428 it specially.
3430 20. pcretest is supposed to handle patterns and data of any length, by
3431 extending its buffers when necessary. It was getting this wrong when the
3432 buffer for a data line had to be extended.
3434 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3435 CRLF as a newline sequence.
3437 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3438 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3439 I have nevertheless tidied it up.
3441 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3443 24. Added a man page for pcre-config.
3446 Version 7.0 19-Dec-06
3447 ---------------------
3449 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3450 moving to gcc 4.1.1.
3452 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3453 sys/time.h, which is documented as needed for this function. It doesn't
3454 seem to matter on Linux, but it showed up on some releases of OS X.
3456 3. It seems that there are systems where bytes whose values are greater than
3457 127 match isprint() in the "C" locale. The "C" locale should be the
3458 default when a C program starts up. In most systems, only ASCII printing
3459 characters match isprint(). This difference caused the output from pcretest
3460 to vary, making some of the tests fail. I have changed pcretest so that:
3462 (a) When it is outputting text in the compiled version of a pattern, bytes
3463 other than 32-126 are always shown as hex escapes.
3465 (b) When it is outputting text that is a matched part of a subject string,
3466 it does the same, unless a different locale has been set for the match
3467 (using the /L modifier). In this case, it uses isprint() to decide.
3469 4. Fixed a major bug that caused incorrect computation of the amount of memory
3470 required for a compiled pattern when options that changed within the
3471 pattern affected the logic of the preliminary scan that determines the
3472 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3473 that the computed length was too small. The symptoms of this bug were
3474 either the PCRE error "internal error: code overflow" from pcre_compile(),
3475 or a glibc crash with a message such as "pcretest: free(): invalid next
3476 size (fast)". Examples of patterns that provoked this bug (shown in
3477 pcretest format) are:
3479 /(?-x: )/x
3480 /(?x)(?-x: \s*#\s*)/
3481 /((?i)[\x{c0}])/8
3482 /(?i:[\x{c0}])/8
3484 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3485 is now done differently.
3487 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3488 wrapper classes; (b) implement a new function in the C++ scanner that is
3489 more efficient than the old way of doing things because it avoids levels of
3490 recursion in the regex matching; (c) add a paragraph to the documentation
3491 for the FullMatch() function.
3493 6. The escape sequence \n was being treated as whatever was defined as
3494 "newline". Not only was this contrary to the documentation, which states
3495 that \n is character 10 (hex 0A), but it also went horribly wrong when
3496 "newline" was defined as CRLF. This has been fixed.
3498 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3499 was being set to -1 for the "end of line" case (supposedly a value that no
3500 character can have). Though this value is never used (the check for end of
3501 line is "zero bytes in current character"), it caused compiler complaints.
3502 I've changed it to 0xffffffff.
3504 8. In pcre_version.c, the version string was being built by a sequence of
3505 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3506 string (as it is for production releases) called a macro with an empty
3507 argument. The C standard says the result of this is undefined. The gcc
3508 compiler treats it as an empty string (which was what was wanted) but it is
3509 reported that Visual C gives an error. The source has been hacked around to
3510 avoid this problem.
3512 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3513 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3514 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3515 of them did).
3517 10. Originally, pcretest opened its input and output without "b"; then I was
3518 told that "b" was needed in some environments, so it was added for release
3519 5.0 to both the input and output. (It makes no difference on Unix-like
3520 systems.) Later I was told that it is wrong for the input on Windows. I've
3521 now abstracted the modes into two macros, to make it easier to fiddle with
3522 them, and removed "b" from the input mode under Windows.
3524 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3526 12. Added -help and --help to pcretest as an official way of being reminded
3527 of the options.
3529 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3530 and pcrecpp.cc because they annoy compilers at high warning levels.
3532 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3534 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3535 scripts are not C programs :-) and which was not noticed because it works
3536 on Linux.
3538 16. pcretest is supposed to handle any length of pattern and data line (as one
3539 line or as a continued sequence of lines) by extending its input buffer if
3540 necessary. This feature was broken for very long pattern lines, leading to
3541 a string of junk being passed to pcre_compile() if the pattern was longer
3542 than about 50K.
3544 17. I have done a major re-factoring of the way pcre_compile() computes the
3545 amount of memory needed for a compiled pattern. Previously, there was code
3546 that made a preliminary scan of the pattern in order to do this. That was
3547 OK when PCRE was new, but as the facilities have expanded, it has become
3548 harder and harder to keep it in step with the real compile phase, and there
3549 have been a number of bugs (see for example, 4 above). I have now found a
3550 cunning way of running the real compile function in a "fake" mode that
3551 enables it to compute how much memory it would need, while actually only
3552 ever using a few hundred bytes of working memory and without too many
3553 tests of the mode. This should make future maintenance and development
3554 easier. A side effect of this work is that the limit of 200 on the nesting
3555 depth of parentheses has been removed (though this was never a serious
3556 limitation, I suspect). However, there is a downside: pcre_compile() now
3557 runs more slowly than before (30% or more, depending on the pattern). I
3558 hope this isn't a big issue. There is no effect on runtime performance.
3560 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3561 newline (only possible for the last line of a file) and it was a
3562 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3564 19. Added additional timing features to pcretest. (1) The -tm option now times
3565 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3566 separate command line item, by a number that specifies the number of
3567 repeats to use when timing. The default is 50000; this gives better
3568 precision, but takes uncomfortably long for very large patterns.
3570 20. Extended pcre_study() to be more clever in cases where a branch of a
3571 subpattern has no definite first character. For example, (a*|b*)[cd] would
3572 previously give no result from pcre_study(). Now it recognizes that the
3573 first character must be a, b, c, or d.
3575 21. There was an incorrect error "recursive call could loop indefinitely" if
3576 a subpattern (or the entire pattern) that was being tested for matching an
3577 empty string contained only one non-empty item after a nested subpattern.
3578 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3579 incorrectly, because the \d was being skipped in the check.
3581 22. The pcretest program now has a new pattern option /B and a command line
3582 option -b, which is equivalent to adding /B to every pattern. This causes
3583 it to show the compiled bytecode, without the additional information that
3584 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3585 is the same as /B/I).
3587 23. A new optimization is now able automatically to treat some sequences such
3588 as a*b as a*+b. More specifically, if something simple (such as a character
3589 or a simple class like \d) has an unlimited quantifier, and is followed by
3590 something that cannot possibly match the quantified thing, the quantifier
3591 is automatically "possessified".
3593 24. A recursive reference to a subpattern whose number was greater than 39
3594 went wrong under certain circumstances in UTF-8 mode. This bug could also
3595 have affected the operation of pcre_study().
3597 25. Realized that a little bit of performance could be had by replacing
3598 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3600 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3602 27. Possessive quantifiers such as a++ were previously implemented by turning
3603 them into atomic groups such as ($>a+). Now they have their own opcodes,
3604 which improves performance. This includes the automatically created ones
3605 from 23 above.
3607 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3608 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3609 the first matched character to be a colon. This applied both to named and
3610 numbered groups.
3612 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3614 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3615 building PCRE on Windows easier, so I have included it in the distribution.
3617 31. There is now a check in pcretest against a ridiculously large number being
3618 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3619 loop, the loop is abandoned.
3621 32. Forward references to subpatterns in conditions such as (?(2)...) where
3622 subpattern 2 is defined later cause pcre_compile() to search forwards in
3623 the pattern for the relevant set of parentheses. This search went wrong
3624 when there were unescaped parentheses in a character class, parentheses
3625 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3627 33. "Subroutine" calls and backreferences were previously restricted to
3628 referencing subpatterns earlier in the regex. This restriction has now
3629 been removed.
3631 34. Added a number of extra features that are going to be in Perl 5.10. On the
3632 whole, these are just syntactic alternatives for features that PCRE had
3633 previously implemented using the Python syntax or my own invention. The
3634 other formats are all retained for compatibility.
3636 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3637 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3638 also .NET compatible.
3640 (b) A recursion or subroutine call to a named group can now be defined as
3641 (?&name) as well as (?P>name).
3643 (c) A backreference to a named group can now be defined as \k<name> or
3644 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3645 5.10, are also .NET compatible.
3647 (d) A conditional reference to a named group can now use the syntax
3648 (?(<name>) or (?('name') as well as (?(name).
3650 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3651 groups (named and numbered) that are never evaluated inline, but can be
3652 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3653 is always false. There may be only one alternative in such a group.
3655 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3656 as the simple (?(R). The condition is true only if the most recent
3657 recursion is that of the given number or name. It does not search out
3658 through the entire recursion stack.
3660 (g) The escape \gN or \g{N} has been added, where N is a positive or
3661 negative number, specifying an absolute or relative reference.
3663 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3664 some "unreachable code" warnings.
3666 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3667 things, this adds five new scripts.
3669 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3670 There were also incompatibilities regarding the handling of \Q..\E inside
3671 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3672 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3674 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3675 matches an empty string, and forcibly breaks the loop. There were bugs in
3676 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3677 against aaaa the result was just "a" rather than "aaaa", for example. Two
3678 separate and independent bugs (that affected different cases) have been
3679 fixed.
3681 39. Refactored the code to abolish the use of different opcodes for small
3682 capturing bracket numbers. This is a tidy that I avoided doing when I
3683 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3684 The new approach is not only tidier, it makes it possible to reduce the
3685 memory needed to fix the previous bug (38).
3687 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3688 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3689 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3690 mode.
3692 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3693 report.
3695 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3696 copy construction and assignment in the C++ wrapper.
3698 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3699 couple of bugs that could have given wrong results in the "--newline=crlf"
3700 case.
3702 44. Added a number of casts and did some reorganization of signed/unsigned int
3703 variables following suggestions from Dair Grant. Also renamed the variable
3704 "this" as "item" because it is a C++ keyword.
3706 45. Arranged for dftables to add
3708 #include "pcre_internal.h"
3710 to pcre_chartables.c because without it, gcc 4.x may remove the array
3711 definition from the final binary if PCRE is built into a static library and
3712 dead code stripping is activated.
3714 46. For an unanchored pattern, if a match attempt fails at the start of a
3715 newline sequence, and the newline setting is CRLF or ANY, and the next two
3716 characters are CRLF, advance by two characters instead of one.
3719 Version 6.7 04-Jul-06
3720 ---------------------
3722 1. In order to handle tests when input lines are enormously long, pcretest has
3723 been re-factored so that it automatically extends its buffers when
3724 necessary. The code is crude, but this _is_ just a test program. The
3725 default size has been increased from 32K to 50K.
3727 2. The code in pcre_study() was using the value of the re argument before
3728 testing it for NULL. (Of course, in any sensible call of the function, it
3729 won't be NULL.)
3731 3. The memmove() emulation function in pcre_internal.h, which is used on
3732 systems that lack both memmove() and bcopy() - that is, hardly ever -
3733 was missing a "static" storage class specifier.
3735 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3736 containing an extended class (one that cannot be represented by a bitmap
3737 because it contains high-valued characters or Unicode property items, e.g.
3738 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3739 pattern, but PCRE should not loop if you do not (it no longer does).
3740 [Detail: two cases were found: (a) a repeated subpattern containing an
3741 extended class; (b) a recursive reference to a subpattern that followed a
3742 previous extended class. It wasn't skipping over the extended class
3743 correctly when UTF-8 mode was not set.]
3745 5. A negated single-character class was not being recognized as fixed-length
3746 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3747 compile error "lookbehind assertion is not fixed length".
3749 6. The RunPerlTest auxiliary script was showing an unexpected difference
3750 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3751 write a Perl script that can interpret lines of an input file either as
3752 byte characters or as UTF-8, which is what "perltest" was being required to
3753 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3754 can't do is switch easily at run time between having the "use utf8;" pragma
3755 or not. In the end, I fudged it by using the RunPerlTest script to insert
3756 "use utf8;" explicitly for the UTF-8 tests.
3758 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3759 the end of the subject string, contrary to the documentation and to what
3760 Perl does. This was true of both matching functions. Now it matches only at
3761 the start of the subject and immediately after *internal* newlines.
3763 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3764 a pointer to an int instead of a pointer to an unsigned long int. This
3765 caused problems on 64-bit systems.
3767 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3768 instance of the 'standard' template library not being so standard".
3770 10. There was no check on the number of named subpatterns nor the maximum
3771 length of a subpattern name. The product of these values is used to compute
3772 the size of the memory block for a compiled pattern. By supplying a very
3773 long subpattern name and a large number of named subpatterns, the size
3774 computation could be caused to overflow. This is now prevented by limiting
3775 the length of names to 32 characters, and the number of named subpatterns
3776 to 10,000.
3778 11. Subpatterns that are repeated with specific counts have to be replicated in
3779 the compiled pattern. The size of memory for this was computed from the
3780 length of the subpattern and the repeat count. The latter is limited to
3781 65535, but there was no limit on the former, meaning that integer overflow
3782 could in principle occur. The compiled length of a repeated subpattern is
3783 now limited to 30,000 bytes in order to prevent this.
3785 12. Added the optional facility to have named substrings with the same name.
3787 13. Added the ability to use a named substring as a condition, using the
3788 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3789 are numbers (not recommended). Forward references are permitted.
3791 14. Added forward references in named backreferences (if you see what I mean).
3793 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3794 pattern could run off the end of the subject. For example, the pattern
3795 "(?s)(.{1,5})"8 did this with the subject "ab".
3797 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3798 PCRE_CASELESS was set when matching characters that were quantified with ?
3799 or *.
3801 17. A character class other than a single negated character that had a minimum
3802 but no maximum quantifier - for example [ab]{6,} - was not handled
3803 correctly by pce_dfa_exec(). It would match only one character.
3805 18. A valid (though odd) pattern that looked like a POSIX character
3806 class but used an invalid character after [ (for example [[,abc,]]) caused
3807 pcre_compile() to give the error "Failed: internal error: code overflow" or
3808 in some cases to crash with a glibc free() error. This could even happen if
3809 the pattern terminated after [[ but there just happened to be a sequence of
3810 letters, a binary zero, and a closing ] in the memory that followed.
3812 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3813 over the years. Originally (before any Unicode support), just the bottom 8
3814 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3815 output from "man perlunicode" includes this:
3817 The regular expression compiler produces polymorphic opcodes. That
3818 is, the pattern adapts to the data and automatically switches to
3819 the Unicode character scheme when presented with Unicode data--or
3820 instead uses a traditional byte scheme when presented with byte
3821 data.
3823 Sadly, a wide octal escape does not cause a switch, and in a string with
3824 no other multibyte characters, these octal escapes are treated as before.
3825 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3826 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3827 Unicode string.
3829 I have not perpetrated such confusion in PCRE. Up till now, it took just
3830 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3831 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3832 translate to the appropriate multibyte character.
3834 29. Applied some refactoring to reduce the number of warnings from Microsoft
3835 and Borland compilers. This has included removing the fudge introduced
3836 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3837 a warning about an unused variable.
3839 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3840 characters since release 4.0, because Perl (from release 5.004) does not.
3841 [Or at least, is documented not to: some releases seem to be in conflict
3842 with the documentation.] However, when a pattern was studied with
3843 pcre_study() and all its branches started with \s, PCRE still included VT
3844 as a possible starting character. Of course, this did no harm; it just
3845 caused an unnecessary match attempt.
3847 22. Removed a now-redundant internal flag bit that recorded the fact that case
3848 dependency changed within the pattern. This was once needed for "required
3849 byte" processing, but is no longer used. This recovers a now-scarce options
3850 bit. Also moved the least significant internal flag bit to the most-
3851 significant bit of the word, which was not previously used (hangover from
3852 the days when it was an int rather than a uint) to free up another bit for
3853 the future.
3855 23. Added support for CRLF line endings as well as CR and LF. As well as the
3856 default being selectable at build time, it can now be changed at runtime
3857 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3858 specify that it is scanning data with non-default line endings.
3860 24. Changed the definition of CXXLINK to make it agree with the definition of
3861 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3863 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3864 recursions. This makes a big different to stack usage for some patterns.
3866 26. If a subpattern containing a named recursion or subroutine reference such
3867 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3868 the space required for the compiled pattern went wrong and gave too small a
3869 value. Depending on the environment, this could lead to "Failed: internal
3870 error: code overflow at offset 49" or "glibc detected double free or
3871 corruption" errors.
3873 27. Applied patches from Google (a) to support the new newline modes and (b) to
3874 advance over multibyte UTF-8 characters in GlobalReplace.
3876 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3877 difference for some implementation of PCRE in some Windows version.
3879 29. Added some extra testing facilities to pcretest:
3881 \q<number> in a data line sets the "match limit" value
3882 \Q<number> in a data line sets the "match recursion limt" value
3883 -S <number> sets the stack size, where <number> is in megabytes
3885 The -S option isn't available for Windows.
3888 Version 6.6 06-Feb-06
3889 ---------------------
3891 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3892 in pcreposix.h. I have copied the definition from pcre.h.
3894 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3895 because pcre.h is no longer a built file.
3897 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3898 not normally included in the compiled code.
3901 Version 6.5 01-Feb-06
3902 ---------------------
3904 1. When using the partial match feature with pcre_dfa_exec(), it was not
3905 anchoring the second and subsequent partial matches at the new starting
3906 point. This could lead to incorrect results. For example, with the pattern
3907 /1234/, partially matching against "123" and then "a4" gave a match.
3909 2. Changes to pcregrep:
3911 (a) All non-match returns from pcre_exec() were being treated as failures
3912 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3913 error message is output. Some extra information is given for the
3915 probably the only errors that are likely to be caused by users (by
3916 specifying a regex that has nested indefinite repeats, for instance).
3917 If there are more than 20 of these errors, pcregrep is abandoned.
3919 (b) A binary zero was treated as data while matching, but terminated the
3920 output line if it was written out. This has been fixed: binary zeroes
3921 are now no different to any other data bytes.
3923 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3924 used to set a locale for matching. The --locale=xxxx long option has
3925 been added (no short equivalent) to specify a locale explicitly on the
3926 pcregrep command, overriding the environment variables.
3928 (d) When -B was used with -n, some line numbers in the output were one less
3929 than they should have been.
3931 (e) Added the -o (--only-matching) option.
3933 (f) If -A or -C was used with -c (count only), some lines of context were
3934 accidentally printed for the final match.
3936 (g) Added the -H (--with-filename) option.
3938 (h) The combination of options -rh failed to suppress file names for files
3939 that were found from directory arguments.
3941 (i) Added the -D (--devices) and -d (--directories) options.
3943 (j) Added the -F (--fixed-strings) option.
3945 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3947 (l) Added the --colo(u)r option.
3949 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3950 is not present by default.
3952 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3953 items such as (?R) or (?1), when the recursion could match a number of
3954 alternatives. If it matched one of the alternatives, but subsequently,
3955 outside the recursion, there was a failure, the code tried to back up into
3956 the recursion. However, because of the way PCRE is implemented, this is not
3957 possible, and the result was an incorrect result from the match.
3959 In order to prevent this happening, the specification of recursion has
3960 been changed so that all such subpatterns are automatically treated as
3961 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3963 4. I had overlooked the fact that, in some locales, there are characters for
3964 which isalpha() is true but neither isupper() nor islower() are true. In
3965 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3966 and ordfeminine) are like this. This affected the treatment of \w and \W
3967 when they appeared in character classes, but not when they appeared outside
3968 a character class. The bit map for "word" characters is now created
3969 separately from the results of isalnum() instead of just taking it from the
3970 upper, lower, and digit maps. (Plus the underscore character, of course.)
3972 5. The above bug also affected the handling of POSIX character classes such as
3973 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3974 permanent tables. Instead, the bit maps for such a class were previously
3975 created as the appropriate unions of the upper, lower, and digit bitmaps.
3976 Now they are created by subtraction from the [[:word:]] class, which has
3977 its own bitmap.
3979 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3980 It is created by subtracting the vertical space characters (\x09, \x0a,
3981 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3982 subtraction was done in the overall bitmap for a character class, meaning
3983 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3984 be recognized. This bug has been fixed.
3986 7. Patches from the folks at Google:
3988 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3989 real life, but is still worth protecting against".
3991 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3992 regular expressions".
3994 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3995 have it.
3997 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3998 "configure" and the latter not, in order to fix a problem somebody had
3999 with compiling the Arg class on HP-UX.
4001 (e) Improve the error-handling of the C++ wrapper a little bit.
4003 (f) New tests for checking recursion limiting.
4005 8. The pcre_memmove() function, which is used only if the environment does not
4006 have a standard memmove() function (and is therefore rarely compiled),
4007 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4008 returning a result (though PCRE never actually uses the result).
4010 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4011 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4012 returned instead of calling malloc() with an overflowing number that would
4013 most likely cause subsequent chaos.
4015 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4017 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4018 with this option is matched, the nmatch and pmatch options of regexec() are
4019 ignored.
4021 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4022 provided in case anyone wants to the the POSIX interface with UTF-8
4023 strings.
4025 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4026 C++ linking (needed for some HP-UX environments).
4028 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4029 (unused parameter) and in the pcre_printint() function (omitted "default"
4030 switch label when the default is to do nothing).
4032 15. Added some code to make it possible, when PCRE is compiled as a C++
4033 library, to replace subject pointers for pcre_exec() with a smart pointer
4034 class, thus making it possible to process discontinuous strings.
4036 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4037 much the same function. They were added by different people who were trying
4038 to make PCRE easy to compile on non-Unix systems. It has been suggested
4039 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4040 for compiling on Windows systems. I have therefore replaced it with
4041 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4042 defaults to "extern" for C or "extern C" for C++, which works fine on
4043 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4044 SCOPE with something explicit in config.h. In addition:
4046 (a) pcreposix.h still had just "extern" instead of either of these macros;
4047 I have replaced it with PCRE_DATA_SCOPE.
4049 (b) Functions such as _pcre_xclass(), which are internal to the library,
4050 but external in the C sense, all had PCRE_EXPORT in their definitions.
4051 This is apparently wrong for the Windows case, so I have removed it.
4052 (It makes no difference on Unix-like systems.)
4054 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4055 of recursive calls to match(). This is different to MATCH_LIMIT because
4056 that limits the total number of calls to match(), not all of which increase
4057 the depth of recursion. Limiting the recursion depth limits the amount of
4058 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4059 when PCRE is compiled, and changed at run time. A patch from Google adds
4060 this functionality to the C++ interface.
4062 18. Changes to the handling of Unicode character properties:
4064 (a) Updated the table to Unicode 4.1.0.
4066 (b) Recognize characters that are not in the table as "Cn" (undefined).
4068 (c) I revised the way the table is implemented to a much improved format
4069 which includes recognition of ranges. It now supports the ranges that
4070 are defined in UnicodeData.txt, and it also amalgamates other
4071 characters into ranges. This has reduced the number of entries in the
4072 table from around 16,000 to around 3,000, thus reducing its size
4073 considerably. I realized I did not need to use a tree structure after
4074 all - a binary chop search is just as efficient. Having reduced the
4075 number of entries, I extended their size from 6 bytes to 8 bytes to
4076 allow for more data.
4078 (d) Added support for Unicode script names via properties such as \p{Han}.
4080 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4081 matching that character.
4083 20. When matching a repeated Unicode property with a minimum greater than zero,
4084 (for example \pL{2,}), PCRE could look past the end of the subject if it
4085 reached it while seeking the minimum number of characters. This could
4086 happen only if some of the characters were more than one byte long, because
4087 there is a check for at least the minimum number of bytes.
4089 21. Refactored the implementation of \p and \P so as to be more general, to
4090 allow for more different types of property in future. This has changed the
4091 compiled form incompatibly. Anybody with saved compiled patterns that use
4092 \p or \P will have to recompile them.
4094 22. Added "Any" and "L&" to the supported property types.
4096 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4097 but give a compile time error if the value is greater than 0xff.
4099 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4100 accidentally not being installed or uninstalled.
4102 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4103 made were to insert the current release number. This seemed silly, because
4104 it made things harder for people building PCRE on systems that don't run
4105 "configure". I have turned pcre.h into a distributed file, no longer built
4106 by "configure", with the version identification directly included. There is
4107 no longer a pcre.h.in file.
4109 However, this change necessitated a change to the pcre-config script as
4110 well. It is built from pcre-config.in, and one of the substitutions was the
4111 release number. I have updated configure.ac so that ./configure now finds
4112 the release number by grepping pcre.h.
4114 26. Added the ability to run the tests under valgrind.
4117 Version 6.4 05-Sep-05
4118 ---------------------
4120 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4121 "--" to be printed when multiple files were scanned, even when none of the
4122 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4123 consider it to be a bug, and have restored the previous behaviour.
4125 2. A couple of code tidies to get rid of compiler warnings.
4127 3. The pcretest program used to cheat by referring to symbols in the library
4128 whose names begin with _pcre_. These are internal symbols that are not
4129 really supposed to be visible externally, and in some environments it is
4130 possible to suppress them. The cheating is now confined to including
4131 certain files from the library's source, which is a bit cleaner.
4133 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4134 file's purpose clearer.
4136 5. Reorganized pcre_ucp_findchar().
4139 Version 6.3 15-Aug-05
4140 ---------------------
4142 1. The file libpcre.pc.in did not have general read permission in the tarball.
4144 2. There were some problems when building without C++ support:
4146 (a) If C++ support was not built, "make install" and "make test" still
4147 tried to test it.
4149 (b) There were problems when the value of CXX was explicitly set. Some
4150 changes have been made to try to fix these, and ...
4152 (c) --disable-cpp can now be used to explicitly disable C++ support.
4154 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4155 backslash in a target when C++ was disabled. This confuses some
4156 versions of "make", apparently. Using an intermediate variable solves
4157 this. (Same for CPP_LOBJ.)
4159 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4160 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4161 necessary on certain architectures.
4163 4. Added a setting of -export-symbols-regex to the link command to remove
4164 those symbols that are exported in the C sense, but actually are local
4165 within the library, and not documented. Their names all begin with
4166 "_pcre_". This is not a perfect job, because (a) we have to except some
4167 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4168 available (and never for static libraries). I have made a note to try to
4169 find a way round (a) in the future.
4172 Version 6.2 01-Aug-05
4173 ---------------------
4175 1. There was no test for integer overflow of quantifier values. A construction
4176 such as {1111111111111111} would give undefined results. What is worse, if
4177 a minimum quantifier for a parenthesized subpattern overflowed and became
4178 negative, the calculation of the memory size went wrong. This could have
4179 led to memory overwriting.
4181 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4183 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4184 operating environments where this matters.
4186 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4187 PCRE options from within the C++ wrapper.
4189 5. Named capturing subpatterns were not being correctly counted when a pattern
4190 was compiled. This caused two problems: (a) If there were more than 100
4191 such subpatterns, the calculation of the memory needed for the whole
4192 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4193 back references of the form \12, where the number was greater than 9, were
4194 not recognized as back references, even though there were sufficient
4195 previous subpatterns.
4197 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4198 versions of gcc, e.g. 2.95.4.
4201 Version 6.1 21-Jun-05
4202 ---------------------
4204 1. There was one reference to the variable "posix" in pcretest.c that was not
4205 surrounded by "#if !defined NOPOSIX".
4207 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4208 the cross-check on the old pcre_info() function, for the benefit of the
4209 cut-down version of PCRE that is currently imported into Exim.
4211 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4212 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4213 patterns that start (?i) but I don't think that matters. The use of (?i) is
4214 just an example; this all applies to the other options as well.
4216 4. Since libtool seems to echo the compile commands it is issuing, the output
4217 from "make" can be reduced a bit by putting "@" in front of each libtool
4218 compile command.
4220 5. Patch from the folks at Google for configure.in to be a bit more thorough
4221 in checking for a suitable C++ installation before trying to compile the
4222 C++ stuff. This should fix a reported problem when a compiler was present,
4223 but no suitable headers.
4225 6. The man pages all had just "PCRE" as their title. I have changed them to
4226 be the relevant file name. I have also arranged that these names are
4227 retained in the file doc/pcre.txt, which is a concatenation in text format
4228 of all the man pages except the little individual ones for each function.
4230 7. The NON-UNIX-USE file had not been updated for the different set of source
4231 files that come with release 6. I also added a few comments about the C++
4232 wrapper.
4235 Version 6.0 07-Jun-05
4236 ---------------------
4238 1. Some minor internal re-organization to help with my DFA experiments.
4240 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4241 didn't matter for the library itself when fully configured, but did matter
4242 when compiling without UCP support, or within Exim, where the ucp files are
4243 not imported.
4245 3. Refactoring of the library code to split up the various functions into
4246 different source modules. The addition of the new DFA matching code (see
4247 below) to a single monolithic source would have made it really too
4248 unwieldy, quite apart from causing all the code to be include in a
4249 statically linked application, when only some functions are used. This is
4250 relevant even without the DFA addition now that patterns can be compiled in
4251 one application and matched in another.
4253 The downside of splitting up is that there have to be some external
4254 functions and data tables that are used internally in different modules of
4255 the library but which are not part of the API. These have all had their
4256 names changed to start with "_pcre_" so that they are unlikely to clash
4257 with other external names.
4259 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4260 a different (DFA) algorithm. Although it is slower than the original
4261 function, it does have some advantages for certain types of matching
4262 problem.
4264 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4265 including restarting after a partial match.
4267 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4268 defined when compiling for Windows was sent to me. I have put it into the
4269 code, though I have no means of testing or verifying it.
4271 7. Added the pcre_refcount() auxiliary function.
4273 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4274 match before or at the first newline in the subject string. In pcretest,
4275 the /f option on a pattern can be used to set this.
4277 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4278 would behave wrongly. This has been present in PCRE since release 4.0.
4280 10. A number of changes to the pcregrep command:
4282 (a) Refactored how -x works; insert ^(...)$ instead of setting
4283 PCRE_ANCHORED and checking the length, in preparation for adding
4284 something similar for -w.
4286 (b) Added the -w (match as a word) option.
4288 (c) Refactored the way lines are read and buffered so as to have more
4289 than one at a time available.
4291 (d) Implemented a pcregrep test script.
4293 (e) Added the -M (multiline match) option. This allows patterns to match
4294 over several lines of the subject. The buffering ensures that at least
4295 8K, or the rest of the document (whichever is the shorter) is available
4296 for matching (and similarly the previous 8K for lookbehind assertions).
4298 (f) Changed the --help output so that it now says
4300 -w, --word-regex(p)
4302 instead of two lines, one with "regex" and the other with "regexp"
4303 because that confused at least one person since the short forms are the
4304 same. (This required a bit of code, as the output is generated
4305 automatically from a table. It wasn't just a text change.)
4307 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4308 option but starts with a hyphen. Could be a pattern or a path name
4309 starting with a hyphen, for instance.
4311 (h) "-" can be given as a file name to represent stdin.
4313 (i) When file names are being printed, "(standard input)" is used for
4314 the standard input, for compatibility with GNU grep. Previously
4315 "<stdin>" was used.
4317 (j) The option --label=xxx can be used to supply a name to be used for
4318 stdin when file names are being printed. There is no short form.
4320 (k) Re-factored the options decoding logic because we are going to add
4321 two more options that take data. Such options can now be given in four
4322 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4324 (l) Added the -A, -B, and -C options for requesting that lines of context
4325 around matches be printed.
4327 (m) Added the -L option to print the names of files that do not contain
4328 any matching lines, that is, the complement of -l.
4330 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4331 continue to scan other files.
4333 (o) The -s option was incorrectly implemented. For compatibility with other
4334 greps, it now suppresses the error message for a non-existent or non-
4335 accessible file (but not the return code). There is a new option called
4336 -q that suppresses the output of matching lines, which was what -s was
4337 previously doing.
4339 (p) Added --include and --exclude options to specify files for inclusion
4340 and exclusion when recursing.
4342 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4343 Hopefully, it now does.
4345 12. Missing cast in pcre_study().
4347 13. Added an "uninstall" target to the makefile.
4349 14. Replaced "extern" in the function prototypes in Makefile.in with
4350 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4351 world, but is set differently for Windows.
4353 15. Added a second compiling function called pcre_compile2(). The only
4354 difference is that it has an extra argument, which is a pointer to an
4355 integer error code. When there is a compile-time failure, this is set
4356 non-zero, in addition to the error test pointer being set to point to an
4357 error message. The new argument may be NULL if no error number is required
4358 (but then you may as well call pcre_compile(), which is now just a
4359 wrapper). This facility is provided because some applications need a
4360 numeric error indication, but it has also enabled me to tidy up the way
4361 compile-time errors are handled in the POSIX wrapper.
4363 16. Added VPATH=.libs to the makefile; this should help when building with one
4364 prefix path and installing with another. (Or so I'm told by someone who
4365 knows more about this stuff than I do.)
4367 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4368 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4369 match everything, including newlines. This is not POSIX-compatible, but
4370 somebody wanted the feature. From pcretest it can be activated by using
4371 both the P and the s flags.
4373 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4375 19. libpcre.pc was being incorrectly installed as executable.
4377 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4378 it now also looks for '\r' so that it will work unmodified on Windows.
4380 21. Added Google's contributed C++ wrapper to the distribution.
4382 22. Added some untidy missing memory free() calls in pcretest, to keep
4383 Electric Fence happy when testing.
4387 Version 5.0 13-Sep-04
4388 ---------------------
4390 1. Internal change: literal characters are no longer packed up into items
4391 containing multiple characters in a single byte-string. Each character
4392 is now matched using a separate opcode. However, there may be more than one
4393 byte in the character in UTF-8 mode.
4395 2. The pcre_callout_block structure has two new fields: pattern_position and
4396 next_item_length. These contain the offset in the pattern to the next match
4397 item, and its length, respectively.
4399 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4400 insertion of callouts before each pattern item. Added the /C option to
4401 pcretest to make use of this.
4403 4. On the advice of a Windows user, the lines
4405 #if defined(_WIN32) || defined(WIN32)
4406 _setmode( _fileno( stdout ), 0x8000 );
4407 #endif /* defined(_WIN32) || defined(WIN32) */
4409 have been added to the source of pcretest. This apparently does useful
4410 magic in relation to line terminators.
4412 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4413 for the benefit of those environments where the "b" makes a difference.
4415 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4416 to know about it. I have put a hack into configure.in that adds in code
4417 to set GCC=yes if CC=icc. This seems to end up at a point in the
4418 generated configure script that is early enough to affect the setting of
4419 compiler options, which is what is needed, but I have no means of testing
4420 whether it really works. (The user who reported this had patched the
4421 generated configure script, which of course I cannot do.)
4423 LATER: After change 22 below (new libtool files), the configure script
4424 seems to know about icc (and also ecc). Therefore, I have commented out
4425 this hack in configure.in.
4427 7. Added support for pkg-config (2 patches were sent in).
4429 8. Negated POSIX character classes that used a combination of internal tables
4430 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4431 [[:^ascii]]. Typically, they would match almost any characters. The other
4432 POSIX classes were not broken in this way.
4434 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4435 to find the match, as PCRE was deluded into thinking that the match had to
4436 start at the start point or following a newline. The same bug applied to
4437 patterns with negative forward assertions or any backward assertions
4438 preceding ".*" at the start, unless the pattern required a fixed first
4439 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4441 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4442 starting at the last subject character, bytes beyond the end of the subject
4443 string were read.
4445 11. Renamed the variable "class" as "classbits" to make life easier for C++
4446 users. (Previously there was a macro definition, but it apparently wasn't
4447 enough.)
4449 12. Added the new field "tables" to the extra data so that tables can be passed
4450 in at exec time, or the internal tables can be re-selected. This allows
4451 a compiled regex to be saved and re-used at a later time by a different
4452 program that might have everything at different addresses.
4454 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4455 -R library as well as a -L library.
4457 14. The debugging options of pcretest (-d on the command line or D on a
4458 pattern) showed incorrect output for anything following an extended class
4459 that contained multibyte characters and which was followed by a quantifier.
4461 15. Added optional support for general category Unicode character properties
4462 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4463 support. It adds about 90K to the size of the library. The meanings of the
4464 inbuilt class escapes such as \d and \s have NOT been changed.
4466 16. Updated pcredemo.c to include calls to free() to release the memory for the
4467 compiled pattern.
4469 17. The generated file chartables.c was being created in the source directory
4470 instead of in the building directory. This caused the build to fail if the
4471 source directory was different from the building directory, and was
4472 read-only.
4474 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4475 file. No doubt somebody will tell me if they don't make sense... Also added
4476 Dan Mooney's comments about building on OpenVMS.
4478 19. Added support for partial matching via the PCRE_PARTIAL option for
4479 pcre_exec() and the \P data escape in pcretest.
4481 20. Extended pcretest with 3 new pattern features:
4483 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4484 write the compiled pattern to the file whose name is "rest-of-line".
4485 This is a straight binary dump of the data, with the saved pointer to
4486 the character tables forced to be NULL. The study data, if any, is
4487 written too. After writing, pcretest reads a new pattern.
4489 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4490 compiled pattern from the given file. There must not be any
4491 occurrences of "<" in the file name (pretty unlikely); if there are,
4492 pcretest will instead treat the initial "<" as a pattern delimiter.
4493 After reading in the pattern, pcretest goes on to read data lines as
4494 usual.
4496 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4497 and 16-bit fields in a compiled pattern, to simulate a pattern that
4498 was compiled on a host of opposite endianness.
4500 21. The pcre-exec() function can now cope with patterns that were compiled on
4501 hosts of opposite endianness, with this restriction:
4503 As for any compiled expression that is saved and used later, the tables
4504 pointer field cannot be preserved; the extra_data field in the arguments
4505 to pcre_exec() should be used to pass in a tables address if a value
4506 other than the default internal tables were used at compile time.
4508 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4509 now diagnosed as an error. Previously, most of the time, a negative number
4510 would have been treated as zero, but if in addition "ovector" was passed as
4511 NULL, a crash could occur.
4513 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4514 new versions from the libtool 1.5 distribution (the last one is a copy of
4515 a file called libtool.m4). This seems to have fixed the need to patch
4516 "configure" to support Darwin 1.3 (which I used to do). However, I still
4517 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4518 workstation).
4520 24. Changed the PCRE licence to be the more standard "BSD" licence.
4523 Version 4.5 01-Dec-03
4524 ---------------------
4526 1. There has been some re-arrangement of the code for the match() function so
4527 that it can be compiled in a version that does not call itself recursively.
4528 Instead, it keeps those local variables that need separate instances for
4529 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4530 needs to "recurse". Keeping track of where control must go is done by means
4531 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4532 hide most of the details from the main code, and operates only if
4533 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4534 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4535 operating.
4537 To make it easier for callers to provide specially tailored get/free
4538 functions for this usage, two new functions, pcre_stack_malloc, and
4539 pcre_stack_free, are used. They are always called in strict stacking order,
4540 and the size of block requested is always the same.
4542 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4543 PCRE has been compiled to use the stack or the heap for recursion. The
4544 -C option of pcretest uses this to show which version is compiled.
4546 A new data escape \S, is added to pcretest; it causes the amounts of store
4547 obtained and freed by both kinds of malloc/free at match time to be added
4548 to the output.
4550 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4551 what's available on my current Linux desktop machine.
4553 3. When matching a UTF-8 string, the test for a valid string at the start has
4554 been extended. If start_offset is not zero, PCRE now checks that it points
4555 to a byte that is the start of a UTF-8 character. If not, it returns
4556 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4557 this is necessary because there may be backward assertions in the pattern.
4558 When matching the same subject several times, it may save resources to use
4559 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4561 4. The code for checking the validity of UTF-8 strings has been tightened so
4562 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4563 containing "overlong sequences".
4565 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4566 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4567 should have been "&", but it just so happened that all the cases this let
4568 through by mistake were picked up later in the function.
4570 6. I had used a variable called "isblank" - this is a C99 function, causing
4571 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4573 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4574 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4575 so the version is shown; (c) stop "make test" echoing "./RunTest".
4577 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4579 9. The prototype for memmove() for systems that don't have it was using
4580 size_t, but the inclusion of the header that defines size_t was later. I've
4581 moved the #includes for the C headers earlier to avoid this.
4583 10. Added some adjustments to the code to make it easier to compiler on certain
4584 special systems:
4586 (a) Some "const" qualifiers were missing.
4587 (b) Added the macro EXPORT before all exported functions; by default this
4588 is defined to be empty.
4589 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4590 that it reads its output file name as an argument instead of writing
4591 to the standard output and assuming this can be redirected.
4593 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4594 class containing characters with values greater than 255, PCRE compilation
4595 went into a loop.
4597 12. A recursive reference to a subpattern that was within another subpattern
4598 that had a minimum quantifier of zero caused PCRE to crash. For example,
4599 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4600 recursion. If the recursively-called subpattern itself had a zero repeat,
4601 that was OK.
4603 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4604 buffer into which it was copied (for escape processing) was still set at
4605 1024, so long lines caused crashes.
4607 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4608 "internal error: code overflow...". This applied to any character class
4609 that was followed by a possessive quantifier.
4611 15. Modified the Makefile to add libpcre.la as a prerequisite for
4612 libpcreposix.la because I was told this is needed for a parallel build to
4613 work.
4615 16. If a pattern that contained .* following optional items at the start was
4616 studied, the wrong optimizing data was generated, leading to matching
4617 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4618 matching string must start with a or b or c. The correct conclusion for
4619 this pattern is that a match can start with any character.
4622 Version 4.4 13-Aug-03
4623 ---------------------
4625 1. In UTF-8 mode, a character class containing characters with values between
4626 127 and 255 was not handled correctly if the compiled pattern was studied.
4627 In fixing this, I have also improved the studying algorithm for such
4628 classes (slightly).
4630 2. Three internal functions had redundant arguments passed to them. Removal
4631 might give a very teeny performance improvement.
4633 3. Documentation bug: the value of the capture_top field in a callout is *one
4634 more than* the number of the hightest numbered captured substring.
4636 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4637 in incorrectly linking with a previously installed version. They now link
4638 explicitly with libpcre.la.
4640 5. configure.in no longer needs to recognize Cygwin specially.
4642 6. A problem in pcre.in for Windows platforms is fixed.
4644 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4645 pcretest, it used to include the size of the study block as part of its
4646 output. Unfortunately, the structure contains a field that has a different
4647 size on different hardware architectures. This meant that the tests that
4648 showed this size failed. As the block is currently always of a fixed size,
4649 this information isn't actually particularly useful in pcretest output, so
4650 I have just removed it.
4652 8. Three pre-processor statements accidentally did not start in column 1.
4653 Sadly, there are *still* compilers around that complain, even though
4654 standard C has not required this for well over a decade. Sigh.
4656 9. In pcretest, the code for checking callouts passed small integers in the
4657 callout_data field, which is a void * field. However, some picky compilers
4658 complained about the casts involved for this on 64-bit systems. Now
4659 pcretest passes the address of the small integer instead, which should get
4660 rid of the warnings.
4662 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4663 both compile and run time, and gives an error if an invalid UTF-8 sequence
4664 is found. There is a option for disabling this check in cases where the
4665 string is known to be correct and/or the maximum performance is wanted.
4667 11. In response to a bug report, I changed one line in Makefile.in from
4669 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4670 to
4671 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4673 to look similar to other lines, but I have no way of telling whether this
4674 is the right thing to do, as I do not use Windows. No doubt I'll get told
4675 if it's wrong...
4678 Version 4.3 21-May-03
4679 ---------------------
4681 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4682 Makefile.
4684 2. Some refactoring to improve the quality of the code:
4686 (i) The utf8_table... variables are now declared "const".
4688 (ii) The code for \cx, which used the "case flipping" table to upper case
4689 lower case letters, now just substracts 32. This is ASCII-specific,
4690 but the whole concept of \cx is ASCII-specific, so it seems
4691 reasonable.
4693 (iii) PCRE was using its character types table to recognize decimal and
4694 hexadecimal digits in the pattern. This is silly, because it handles
4695 only 0-9, a-f, and A-F, but the character types table is locale-
4696 specific, which means strange things might happen. A private
4697 table is now used for this - though it costs 256 bytes, a table is
4698 much faster than multiple explicit tests. Of course, the standard
4699 character types table is still used for matching digits in subject
4700 strings against \d.
4702 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4703 ending in _t are). So I've renamed it as ESC_tee.
4705 3. The first argument for regexec() in the POSIX wrapper should have been
4706 defined as "const".
4708 4. Changed pcretest to use malloc() for its buffers so that they can be
4709 Electric Fenced for debugging.
4711 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4712 to read one or more bytes before the start of the subject string. Often this
4713 had no effect on PCRE's behaviour, but in some circumstances it could
4714 provoke a segmentation fault.
4716 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4717 to try to read one or more bytes before the start of the subject string.
4719 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4720 UTF-8 support could misbehave in various ways if the subject string
4721 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4722 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4723 back over UTF-8 characters.)
4726 Version 4.2 14-Apr-03
4727 ---------------------
4729 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4731 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4732 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4733 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4734 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4735 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4737 Note: automatic setting of the BUILD variables is not yet working
4738 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4739 compile-time but not at link-time
4740 [LINK]: use for linking executables only
4741 make different versions for Windows and non-Windows
4742 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4743 libraries
4744 [LINK_FOR_BUILD]: new variable
4745 [OBJEXT]: use throughout
4746 [EXEEXT]: use throughout
4747 <winshared>: new target
4748 <wininstall>: new target
4749 <dftables.o>: use native compiler
4750 <dftables>: use native linker
4751 <install>: handle Windows platform correctly
4752 <clean>: ditto
4753 <check>: ditto
4754 copy DLL to top builddir before testing
4756 As part of these changes, -no-undefined was removed again. This was reported
4757 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4758 in any case.
4760 3. Some tidies to get rid of compiler warnings:
4762 . In the match_data structure, match_limit was an unsigned long int, whereas
4763 match_call_count was an int. I've made them both unsigned long ints.
4765 . In pcretest the fact that a const uschar * doesn't automatically cast to
4766 a void * provoked a warning.
4768 . Turning on some more compiler warnings threw up some "shadow" variables
4769 and a few more missing casts.
4771 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4772 option, a class that contained a single character with a value between 128
4773 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4775 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4776 option, a class that contained several characters, but with at least one
4777 whose value was between 128 and 255 caused PCRE to crash.
4780 Version 4.1 12-Mar-03
4781 ---------------------
4783 1. Compiling with gcc -pedantic found a couple of places where casts were
4784 needed, and a string in dftables.c that was longer than standard compilers are
4785 required to support.
4787 2. Compiling with Sun's compiler found a few more places where the code could
4788 be tidied up in order to avoid warnings.
4790 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4791 first of these names is deprecated in the latest Autoconf in favour of the name
4792 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4793 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4794 analogy I have changed it to CFLAGS_FOR_BUILD.
4796 4. Added -no-undefined to the linking command in the Makefile, because this is
4797 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4798 linking step for the pcreposix library.
4800 5. PCRE was failing to diagnose the case of two named groups with the same
4801 name.
4803 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4804 literal character that is needed in the subject for a match, and scans along to
4805 ensure that it is present before embarking on the full matching process. This
4806 saves time in cases of nested unlimited repeats that are never going to match.
4807 Problem: the scan can take a lot of time if the subject is very long (e.g.
4808 megabytes), thus penalizing straightforward matches. It is now done only if the
4809 amount of subject to be scanned is less than 1000 bytes.
4811 7. A lesser problem with the same optimization is that it was recording the
4812 first character of an anchored pattern as "needed", thus provoking a search
4813 right along the subject, even when the first match of the pattern was going to
4814 fail. The "needed" character is now not set for anchored patterns, unless it
4815 follows something in the pattern that is of non-fixed length. Thus, it still
4816 fulfils its original purpose of finding quick non-matches in cases of nested
4817 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4820 Version 4.0 17-Feb-03
4821 ---------------------
4823 1. If a comment in an extended regex that started immediately after a meta-item
4824 extended to the end of string, PCRE compiled incorrect data. This could lead to
4825 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4827 2. Moved to autoconf 2.53 and libtool 1.4.2.
4829 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4830 the special perltest8 script is no longer needed - all the tests can be run
4831 from a single perltest script.
4833 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4834 by \s. It has now been removed in PCRE. This means it isn't recognized as
4835 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4836 class [:space:] *does* include VT, thereby creating a mess.
4838 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4839 space and tab.
4841 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4842 its new features into the main test script, reducing the number of scripts.
4844 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4845 were backward compatible, and made the (?i) apply to the whole pattern, as if
4846 /i were given. Now it behaves more logically, and applies the option setting
4847 only to what follows. PCRE has been changed to follow suit. However, if it
4848 finds options settings right at the start of the pattern, it extracts them into
4849 the global options, as before. Thus, they show up in the info data.
4851 8. Added support for the \Q...\E escape sequence. Characters in between are
4852 treated as literals. This is slightly different from Perl in that $ and @ are
4853 also handled as literals inside the quotes. In Perl, they will cause variable
4854 interpolation. Note the following examples:
4856 Pattern PCRE matches Perl matches
4858 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4859 \Qabc\$xyz\E abc\$xyz abc\$xyz
4860 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4862 For compatibility with Perl, \Q...\E sequences are recognized inside character
4863 classes as well as outside them.
4865 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4866 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4867 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4868 signed/unsigned warnings.
4870 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4871 option for pcretest, so I've replaced it by a simple function that does just
4872 that job.
4874 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4875 "pcregrep -".
4877 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4878 Java package. This provides some syntactic sugar for simple cases of what my
4879 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4880 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4881 item, you can use this simplified notation. Note that only makes sense with
4882 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4883 greediness, whatever the setting of the PCRE_UNGREEDY option.
4885 13. A change of greediness default within a pattern was not taking effect at
4886 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4887 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4888 was abstracted outside.
4890 14. PCRE now supports the \G assertion. It is true when the current matching
4891 position is at the start point of the match. This differs from \A when the
4892 starting offset is non-zero. Used with the /g option of pcretest (or similar
4893 code), it works in the same way as it does for Perl's /g option. If all
4894 alternatives of a regex begin with \G, the expression is anchored to the start
4895 match position, and the "anchored" flag is set in the compiled expression.
4897 15. Some bugs concerning the handling of certain option changes within patterns
4898 have been fixed. These applied to options other than (?ims). For example,
4899 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4900 been the other way round. Some of this was related to change 7 above.
4902 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4903 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4904 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4905 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4907 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4908 mode. Unlike ".", it always matches newline, whatever the setting of
4909 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4910 assertions. Perl allows it, but it doesn't (in general) work because it can't
4911 calculate the length of the lookbehind. At least, that's the case for Perl
4912 5.8.0 - I've been told they are going to document that it doesn't work in
4913 future.
4915 18. Added an error diagnosis for escapes that PCRE does not support: these are
4916 \L, \l, \N, \P, \p, \U, \u, and \X.
4918 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4919 reading past the end of the pattern in cases such as /[abcd/.
4921 20. PCRE was getting more memory than necessary for patterns with classes that
4922 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4924 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4925 compiling PCRE for use with Virtual Pascal.
4927 22. Small fix to the Makefile to make it work properly if the build is done
4928 outside the source tree.
4930 23. Added a new extension: a condition to go with recursion. If a conditional
4931 subpattern starts with (?(R) the "true" branch is used if recursion has
4932 happened, whereas the "false" branch is used only at the top level.
4934 24. When there was a very long string of literal characters (over 255 bytes
4935 without UTF support, over 250 bytes with UTF support), the computation of how
4936 much memory was required could be incorrect, leading to segfaults or other
4937 strange effects.
4939 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4940 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4941 there was a subsequent back reference to those brackets. This meant that, for
4942 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4943 possible to check for precisely this case. All we can do is abandon the
4944 optimization if .* occurs inside capturing brackets when there are any back
4945 references whatsoever. (See below for a better fix that came later.)
4947 26. The handling of the optimization for finding the first character of a
4948 non-anchored pattern, and for finding a character that is required later in the
4949 match were failing in some cases. This didn't break the matching; it just
4950 failed to optimize when it could. The way this is done has been re-implemented.
4952 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4954 28. Added a new feature that provides some of the functionality that Perl
4955 provides with (?{...}). The facility is termed a "callout". The way it is done
4956 in PCRE is for the caller to provide an optional function, by setting
4957 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4958 global variable. By default it is unset, which disables all calling out. To get
4959 the function called, the regex must include (?C) at appropriate points. This
4960 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4961 This provides a means of identifying different callout points. When PCRE
4962 reaches such a point in the regex, if pcre_callout has been set, the external
4963 function is called. It is provided with data in a structure called
4964 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4965 matching continues; if it returns a non-zero value, the match at the current
4966 point fails. However, backtracking will occur if possible. [This was changed
4967 later and other features added - see item 49 below.]
4969 29. pcretest is upgraded to test the callout functionality. It provides a
4970 callout function that displays information. By default, it shows the start of
4971 the match and the current position in the text. There are some new data escapes
4972 to vary what happens:
4974 \C+ in addition, show current contents of captured substrings
4975 \C- do not supply a callout function
4976 \C!n return 1 when callout number n is reached
4977 \C!n!m return 1 when callout number n is reached for the mth time
4979 30. If pcregrep was called with the -l option and just a single file name, it
4980 output "<stdin>" if a match was found, instead of the file name.
4982 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4983 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4984 pcre_exec(). This saves a malloc/free per call. The default value of
4985 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4986 when configuring.
4988 32. The default maximum size of a compiled pattern is 64K. There have been a
4989 few cases of people hitting this limit. The code now uses macros to handle the
4990 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4991 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4992 configuring. Tests 2 and 5 work only with 2-byte links because they output
4993 debugging information about compiled patterns.
4995 33. Internal code re-arrangements:
4997 (a) Moved the debugging function for printing out a compiled regex into
4998 its own source file (printint.c) and used #include to pull it into
4999 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5000 separate copies.
5002 (b) Defined the list of op-code names for debugging as a macro in
5003 internal.h so that it is next to the definition of the opcodes.
5005 (c) Defined a table of op-code lengths for simpler skipping along compiled
5006 code. This is again a macro in internal.h so that it is next to the
5007 definition of the opcodes.
5009 34. Added support for recursive calls to individual subpatterns, along the
5010 lines of Robin Houston's patch (but implemented somewhat differently).
5012 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5013 allow it to read and process whole directories in Win32. This code was
5014 contributed by Lionel Fourquaux; it has not been tested by me.
5016 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5017 used to name a group. Names consist of alphanumerics and underscores, and must
5018 be unique. Back references use the syntax (?P=name) and recursive calls use
5019 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5020 numbers. The function pcre_fullinfo() can be used after compilation to extract
5021 a name/number map. There are three relevant calls:
5023 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5024 PCRE_INFO_NAMECOUNT yields the number of entries
5025 PCRE_INFO_NAMETABLE yields a pointer to the map.
5027 The map is a vector of fixed-size entries. The size of each entry depends on
5028 the length of the longest name used. The first two bytes of each entry are the
5029 group number, most significant byte first. There follows the corresponding
5030 name, zero terminated. The names are in alphabetical order.
5032 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5033 case instead of 255. Making it the same both with and without UTF-8 support
5034 means that the same test output works with both.
5036 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5037 calling malloc() with a zero argument.
5039 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5040 optimization. I've improved things by keeping a bitmap of backreferences with
5041 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5042 fact referenced, the optimization can be applied. It is unlikely that a
5043 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5044 the match to follow \n) will appear inside brackets with a number greater than
5045 31, but if it does, any back reference > 31 suppresses the optimization.
5047 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5048 of disabling numbered capturing parentheses. Any opening parenthesis that is
5049 not followed by ? behaves as if it were followed by ?: but named parentheses
5050 can still be used for capturing (and they will acquire numbers in the usual
5051 way).
5053 41. Redesigned the return codes from the match() function into yes/no/error so
5054 that errors can be passed back from deep inside the nested calls. A malloc
5055 failure while inside a recursive subpattern call now causes the
5056 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5058 42. It is now possible to set a limit on the number of times the match()
5059 function is called in a call to pcre_exec(). This facility makes it possible to
5060 limit the amount of recursion and backtracking, though not in a directly
5061 obvious way, because the match() function is used in a number of different
5062 circumstances. The count starts from zero for each position in the subject
5063 string (for non-anchored patterns). The default limit is, for compatibility, a
5064 large number, namely 10 000 000. You can change this in two ways:
5066 (a) When configuring PCRE before making, you can use --with-match-limit=n
5067 to set a default value for the compiled library.
5069 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5070 a different value is set. See 45 below.
5072 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5074 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5075 of things that can be changed at compile time. The first argument specifies
5076 what is wanted and the second points to where the information is to be placed.
5077 The current list of available information is:
5081 The output is an integer that is set to one if UTF-8 support is available;
5082 otherwise it is set to zero.
5086 The output is an integer that it set to the value of the code that is used for
5087 newline. It is either LF (10) or CR (13).
5091 The output is an integer that contains the number of bytes used for internal
5092 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5096 The output is an integer that contains the threshold above which the POSIX
5097 interface uses malloc() for output vectors. See item 31 above.
5101 The output is an unsigned integer that contains the default limit of the number
5102 of match() calls in a pcre_exec() execution. See 42 above.
5104 44. pcretest has been upgraded by the addition of the -C option. This causes it
5105 to extract all the available output from the new pcre_config() function, and to
5106 output it. The program then exits immediately.
5108 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5109 order to support additional features. One way would have been to define
5110 pcre_exec2() (for example) with extra arguments, but this would not have been
5111 extensible, and would also have required all calls to the original function to
5112 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5113 is used for passing in "extra" data from pcre_study().
5115 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5116 contains the following fields:
5118 flags a bitmap indicating which of the following fields are set
5119 study_data opaque data from pcre_study()
5120 match_limit a way of specifying a limit on match() calls for a specific
5121 call to pcre_exec()
5122 callout_data data for callouts (see 49 below)
5124 The flag bits are also defined in pcre.h, and are
5130 The pcre_study() function now returns one of these new pcre_extra blocks, with
5131 the actual study data pointed to by the study_data field, and the
5132 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5133 before. That is, this change is entirely upwards-compatible and requires no
5134 change to existing code.
5136 If you want to pass in additional data to pcre_exec(), you can either place it
5137 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5138 block.
5140 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5141 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5142 times with different match limits, until it finds the minimum value needed for
5143 pcre_exec() to complete. The value is then output. This can be instructive; for
5144 most simple matches the number is quite small, but for pathological cases it
5145 gets very large very quickly.
5147 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5148 returns the size of the data block pointed to by the study_data field in a
5149 pcre_extra block, that is, the value that was passed as the argument to
5150 pcre_malloc() when PCRE was getting memory in which to place the information
5151 created by pcre_study(). The fourth argument should point to a size_t variable.
5152 pcretest has been extended so that this information is shown after a successful
5153 pcre_study() call when information about the compiled regex is being displayed.
5155 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5156 because what follows is always an absolute path. (Later: it turns out that this
5157 is more than cosmetic for MinGW, because it doesn't like empty path
5158 components.)
5160 49. Some changes have been made to the callout feature (see 28 above):
5162 (i) A callout function now has three choices for what it returns:
5164 0 => success, carry on matching
5165 > 0 => failure at this point, but backtrack if possible
5166 < 0 => serious error, return this value from pcre_exec()
5168 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5169 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5170 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5171 use by callout functions. It will never be used by PCRE itself.
5173 (ii) The pcre_extra structure (see 45 above) has a void * field called
5174 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5175 pcre_callout_block structure has a field of the same name. The contents of
5176 the field passed in the pcre_extra structure are passed to the callout
5177 function in the corresponding field in the callout block. This makes it
5178 easier to use the same callout-containing regex from multiple threads. For
5179 testing, the pcretest program has a new data escape
5181 \C*n pass the number n (may be negative) as callout_data
5183 If the callout function in pcretest receives a non-zero value as
5184 callout_data, it returns that value.
5186 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5187 there were some redundant $(CFLAGS) in commands that are now specified as
5188 $(LINK), which already includes $(CFLAGS).
5190 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5191 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5192 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5193 one-byte characters throughout. Note that case-insensitive matching applies
5194 only to characters whose values are less than 256. PCRE doesn't support the
5195 notion of cases for higher-valued characters.
5197 (i) A character class whose characters are all within 0-255 is handled as
5198 a bit map, and the map is inverted for negative classes. Previously, a
5199 character > 255 always failed to match such a class; however it should
5200 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5202 (ii) A negated character class with a single character < 255 is coded as
5203 "not this character" (OP_NOT). This wasn't working properly when the test
5204 character was multibyte, either singly or repeated.
5206 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5207 mode, for example: \x{100}{2,3}.
5209 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5210 singly or repeated) now correctly test multibyte characters. However,
5211 PCRE doesn't recognize any characters with values greater than 255 as
5212 digits, spaces, or word characters. Such characters always match \D, \S,
5213 and \W, and never match \d, \s, or \w.
5215 (v) Classes may now contain characters and character ranges with values
5216 greater than 255. For example: [ab\x{100}-\x{400}].
5218 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5219 PCRE in UTF-8 mode.
5221 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5222 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5223 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5224 value.)
5226 53. The single man page has become too large. I have therefore split it up into
5227 a number of separate man pages. These also give rise to individual HTML pages;
5228 these are now put in a separate directory, and there is an index.html page that
5229 lists them all. Some hyperlinking between the pages has been installed.
5231 54. Added convenience functions for handling named capturing parentheses.
5233 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5234 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5235 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5236 are faulted.
5238 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5239 calling configure. These values are used when compiling the dftables.c program
5240 which is run to generate the source of the default character tables. They
5241 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5242 you will need to set these values.
5244 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5247 Version 3.9 02-Jan-02
5248 ---------------------
5250 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5252 2. If --disable-static was given, the building process failed when trying to
5253 build pcretest and pcregrep. (For some reason it was using libtool to compile
5254 them, which is not right, as they aren't part of the library.)
5257 Version 3.8 18-Dec-01
5258 ---------------------
5260 1. The experimental UTF-8 code was completely screwed up. It was packing the
5261 bytes in the wrong order. How dumb can you get?
5264 Version 3.7 29-Oct-01
5265 ---------------------
5267 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5268 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5269 this didn't happen under Solaris 8, where I normally test things.
5271 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5272 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5273 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5274 this fix an hour or so after the initial 3.7 release.)
5277 Version 3.6 23-Oct-01
5278 ---------------------
5280 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5281 offsets passed as NULL with zero offset count.
5283 2. The config.guess and config.sub files had not been updated when I moved to
5284 the latest autoconf.
5287 Version 3.5 15-Aug-01
5288 ---------------------
5290 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5291 had been forgotten.
5293 2. By using declared but undefined structures, we can avoid using "void"
5294 definitions in pcre.h while keeping the internal definitions of the structures
5295 private.
5297 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5298 user point of view, this means that both static and shared libraries are built
5299 by default, but this can be individually controlled. More of the work of
5300 handling this static/shared cases is now inside libtool instead of PCRE's make
5301 file.
5303 4. The pcretest utility is now installed along with pcregrep because it is
5304 useful for users (to test regexs) and by doing this, it automatically gets
5305 relinked by libtool. The documentation has been turned into a man page, so
5306 there are now .1, .txt, and .html versions in /doc.
5308 5. Upgrades to pcregrep:
5309 (i) Added long-form option names like gnu grep.
5310 (ii) Added --help to list all options with an explanatory phrase.
5311 (iii) Added -r, --recursive to recurse into sub-directories.
5312 (iv) Added -f, --file to read patterns from a file.
5314 6. pcre_exec() was referring to its "code" argument before testing that
5315 argument for NULL (and giving an error if it was NULL).
5317 7. Upgraded Makefile.in to allow for compiling in a different directory from
5318 the source directory.
5320 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5321 options bits, the pointer it was passed was to an int instead of to an unsigned
5322 long int. This mattered only on 64-bit systems.
5324 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5325 generated) instead of pcre.in, which it its source. Also made the same change
5326 in several of the .c files.
5328 10. A new release of gcc defines printf() as a macro, which broke pcretest
5329 because it had an ifdef in the middle of a string argument for printf(). Fixed
5330 by using separate calls to printf().
5332 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5333 script, to force use of CR or LF instead of \n in the source. On non-Unix
5334 systems, the value can be set in config.h.
5336 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5337 absolute limit. Changed the text of the error message to make this clear, and
5338 likewise updated the man page.
5340 13. The limit of 99 on the number of capturing subpatterns has been removed.
5341 The new limit is 65535, which I hope will not be a "real" limit.
5344 Version 3.4 22-Aug-00
5345 ---------------------
5347 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5349 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5352 Version 3.3 01-Aug-00
5353 ---------------------
5355 1. If an octal character was given, but the value was greater than \377, it
5356 was not getting masked to the least significant bits, as documented. This could
5357 lead to crashes in some systems.
5359 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5360 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5362 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5363 These just pass their arguments on to (pcre_free)(), but they are provided
5364 because some uses of PCRE bind it to non-C systems that can call its functions,
5365 but cannot call free() or pcre_free() directly.
5367 4. Add "make test" as a synonym for "make check". Corrected some comments in
5368 the Makefile.
5370 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5371 Makefile.
5373 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5374 command called pgrep for grepping around the active processes.
5376 7. Added the beginnings of support for UTF-8 character strings.
5378 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5379 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5380 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5381 out for the ar command.)
5384 Version 3.2 12-May-00
5385 ---------------------
5387 This is purely a bug fixing release.
5389 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5390 of ZA. This was just one example of several cases that could provoke this bug,
5391 which was introduced by change 9 of version 2.00. The code for breaking
5392 infinite loops after an iteration that matches an empty string was't working
5393 correctly.
5395 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5396 when matched against abbab (for example). After matching an empty string, it
5397 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5398 caused it to match further down the string than it should.
5400 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5401 was there because it doesn't seem to be needed, and it causes trouble on some
5402 systems, as it is not a Standard C header. It has been removed.
5404 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5405 were reported on the Macintosh. The changes were from
5407 while ((c = *(++ptr)) != 0 && c != '\n');
5408 to
5409 while ((c = *(++ptr)) != 0 && c != '\n') ;
5411 Totally extraordinary, but if that's what it takes...
5413 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5414 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5415 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5416 assumes the way PCRE uses memmove() (always moving upwards).
5418 6. PCRE is being used in one environment where strchr() is not available. There
5419 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5420 faster code anyway.
5423 Version 3.1 09-Feb-00
5424 ---------------------
5426 The only change in this release is the fixing of some bugs in Makefile.in for
5427 the "install" target:
5429 (1) It was failing to install pcreposix.h.
5431 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5434 Version 3.0 01-Feb-00
5435 ---------------------
5437 1. Add support for the /+ modifier to perltest (to output $` like it does in
5438 pcretest).
5440 2. Add support for the /g modifier to perltest.
5442 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5443 matches null strings.
5445 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5446 pattern. Perl treats empty patterns specially - it reuses the most recent
5447 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5448 effect.
5450 5. The POSIX interface was broken in that it was just handing over the POSIX
5451 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5452 required a bigger vector, with some working space on the end. This means that
5453 the POSIX wrapper now has to get and free some memory, and copy the results.
5455 6. Added some simple autoconf support, placing the test data and the
5456 documentation in separate directories, re-organizing some of the
5457 information files, and making it build pcre-config (a GNU standard). Also added
5458 libtool support for building PCRE as a shared library, which is now the
5459 default.
5461 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5462 09 are not valid octal constants. Single digits will be used for minor values
5463 less than 10.
5465 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5466 existing programs that set these in the POSIX interface can use PCRE without
5467 modification.
5469 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5470 return all that pcre_info() returns, plus additional data. The pcre_info()
5471 function is retained for compatibility, but is considered to be obsolete.
5473 10. Added experimental recursion feature (?R) to handle one common case that
5474 Perl 5.6 will be able to do with (?p{...}).
5476 11. Added support for POSIX character classes like [:alpha:], which Perl is
5477 adopting.
5480 Version 2.08 31-Aug-99
5481 ----------------------
5483 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5484 trying to match at the startoffset position, but instead was moving forward to
5485 the next newline as if a previous match had failed.
5487 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5488 and could get into a loop if a null string was matched other than at the start
5489 of the subject.
5491 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5492 be distinguished at compile time, and for completeness also added PCRE_DATE.
5494 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5495 in GnuWin32 environments.
5498 Version 2.07 29-Jul-99
5499 ----------------------
5501 1. The documentation is now supplied in plain text form and HTML as well as in
5502 the form of man page sources.
5504 2. C++ compilers don't like assigning (void *) values to other pointer types.
5505 In particular this affects malloc(). Although there is no problem in Standard
5506 C, I've put in casts to keep C++ compilers happy.
5508 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5509 should be (const char *).
5511 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5512 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5513 However, I haven't made this a standard facility. The documentation doesn't
5514 mention it, and the Makefile doesn't support it.
5516 5. The Makefile now contains an "install" target, with editable destinations at
5517 the top of the file. The pcretest program is not installed.
5519 6. pgrep -V now gives the PCRE version number and date.
5521 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5522 causing the entire string to be ignored, instead of just the last character.
5524 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5525 non-matching string, it can take a very, very long time, even for strings of
5526 quite modest length, because of the nested recursion. PCRE now does better in
5527 some of these cases. It does this by remembering the last required literal
5528 character in the pattern, and pre-searching the subject to ensure it is present
5529 before running the real match. In other words, it applies a heuristic to detect
5530 some types of certain failure quickly, and in the above example, if presented
5531 with a string that has no trailing " it gives "no match" very quickly.
5533 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5534 other alternatives are tried instead.
5537 Version 2.06 09-Jun-99
5538 ----------------------
5540 1. Change pcretest's output for amount of store used to show just the code
5541 space, because the remainder (the data block) varies in size between 32-bit and
5542 64-bit systems.
5544 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5545 start matching at. This allows lookbehinds to work when searching for multiple
5546 occurrences in a string.
5548 3. Added additional options to pcretest for testing multiple occurrences:
5550 /+ outputs the rest of the string that follows a match
5551 /g loops for multiple occurrences, using the new startoffset argument
5552 /G loops for multiple occurrences by passing an incremented pointer
5554 4. PCRE wasn't doing the "first character" optimization for patterns starting
5555 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5556 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5557 the letter 'x'. On long subject strings, this gives a significant speed-up.
5560 Version 2.05 21-Apr-99
5561 ----------------------
5563 1. Changed the type of magic_number from int to long int so that it works
5564 properly on 16-bit systems.
5566 2. Fixed a bug which caused patterns starting with .* not to work correctly
5567 when the subject string contained newline characters. PCRE was assuming
5568 anchoring for such patterns in all cases, which is not correct because .* will
5569 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5570 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5571 must be retried after every newline in the subject.
5574 Version 2.04 18-Feb-99
5575 ----------------------
5577 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5578 computation of the store needed to hold the pattern was incorrect (too large).
5579 If such patterns were nested a few deep, this could multiply and become a real
5580 problem.
5582 2. Added /M option to pcretest to show the memory requirement of a specific
5583 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5585 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5586 compiled in such a way that the backtracking after subsequent failure was
5587 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5588 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5591 Version 2.03 02-Feb-99
5592 ----------------------
5594 1. Fixed typo and small mistake in man page.
5596 2. Added 4th condition (GPL supersedes if conflict) and created separate
5597 LICENCE file containing the conditions.
5599 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5600 Perl, that is the internal \ allows the delimiter to be included in the
5601 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5602 the final delimiter, add \ to the end of the pattern (to test the error).
5604 4. Added the convenience functions for extracting substrings after a successful
5605 match. Updated pcretest to make it able to test these functions.
5608 Version 2.02 14-Jan-99
5609 ----------------------
5611 1. Initialized the working variables associated with each extraction so that
5612 their saving and restoring doesn't refer to uninitialized store.
5614 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5615 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5616 fix the problem.
5618 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5619 calls, and wasn't printing the correct value for compiling calls. Increased the
5620 default value of LOOPREPEAT, and the number of significant figures in the
5621 times.
5623 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5625 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5626 a building problem on Windows NT with a FAT file system.
5629 Version 2.01 21-Oct-98
5630 ----------------------
5632 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5633 to character tables built by pcre_maketables() in the current locale. If NULL
5634 is passed, the default tables are used.
5637 Version 2.00 24-Sep-98
5638 ----------------------
5640 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5641 it any more.
5643 2. Allow quantification of (?>) groups, and make it work correctly.
5645 3. The first character computation wasn't working for (?>) groups.
5647 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5648 end of the subject) and add 5.005's \z, which really does match only at the
5649 very end of the subject.
5651 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5653 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5654 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5655 localized options. All options to pcre_study() were also removed.
5657 7. Add other new features from 5.005:
5659 $(?<= positive lookbehind
5660 $(?<! negative lookbehind
5661 (?imsx-imsx) added the unsetting capability
5662 such a setting is global if at outer level; local otherwise
5663 (?imsx-imsx:) non-capturing groups with option setting
5664 (?(cond)re|re) conditional pattern matching
5666 A backreference to itself in a repeated group matches the previous
5667 captured string.
5669 8. General tidying up of studying (both automatic and via "study")
5670 consequential on the addition of new assertions.
5672 9. As in 5.005, unlimited repeated groups that could match an empty substring
5673 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5674 runtime if any iteration does actually match an empty substring.
5676 10. Include the RunTest script in the distribution.
5678 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5679 discrepancies, some of which were old and were also with respect to 5.004. They
5680 have now been fixed.
5683 Version 1.09 28-Apr-98
5684 ----------------------
5686 1. A negated single character class followed by a quantifier with a minimum
5687 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5688 program crashes, or just wrong answers. This did not apply to negated classes
5689 containing more than one character, or to minima other than one.
5692 Version 1.08 27-Mar-98
5693 ----------------------
5695 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5697 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5698 latter must appear before anything that relies on it in the pattern.
5701 Version 1.07 16-Feb-98
5702 ----------------------
5704 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5705 repeat of a potentially empty string).
5708 Version 1.06 23-Jan-98
5709 ----------------------
5711 1. Added Markus Oberhumer's little patches for C++.
5713 2. Literal strings longer than 255 characters were broken.
5716 Version 1.05 23-Dec-97
5717 ----------------------
5719 1. Negated character classes containing more than one character were failing if
5720 PCRE_CASELESS was set at run time.
5723 Version 1.04 19-Dec-97
5724 ----------------------
5726 1. Corrected the man page, where some "const" qualifiers had been omitted.
5728 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5729 input syntax.
5731 3. Fixed memory leak which occurred when a regex with back references was
5732 matched with an offsets vector that wasn't big enough. The temporary memory
5733 that is used in this case wasn't being freed if the match failed.
5735 4. Tidied pcretest to ensure it frees memory that it gets.
5737 5. Temporary memory was being obtained in the case where the passed offsets
5738 vector was exactly big enough.
5740 6. Corrected definition of offsetof() from change 5 below.
5742 7. I had screwed up change 6 below and broken the rules for the use of
5743 setjmp(). Now fixed.
5746 Version 1.03 18-Dec-97
5747 ----------------------
5749 1. A erroneous regex with a missing opening parenthesis was correctly
5750 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5751 on some systems.
5753 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5754 it was reported that one broken compiler failed on the former because "code" is
5755 also an independent variable.
5757 3. The erroneous regex a[]b caused an array overrun reference.
5759 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5760 fail on data ending with that character. (It was going on too far, and checking
5761 the next character, typically a binary zero.) This was specific to the
5762 optimized code for single-character negative classes.
5764 5. Added a contributed patch from the TIN world which does the following:
5766 + Add an undef for memmove, in case the the system defines a macro for it.
5768 + Add a definition of offsetof(), in case there isn't one. (I don't know
5769 the reason behind this - offsetof() is part of the ANSI standard - but
5770 it does no harm).
5772 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5773 most of the places where whitespace preceded '#'. I have given up and
5774 allowed the remaining 2 cases to be at the margin.
5776 + Rename some variables in pcre to eliminate shadowing. This seems very
5777 pedantic, but does no harm, of course.
5779 6. Moved the call to setjmp() into its own function, to get rid of warnings
5780 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5782 7. Constructs such as \d{8,} were compiling into the equivalent of
5783 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5784 outcome, but in this particular case used more store than had been allocated,
5785 which caused the bug to be discovered because it threw up an internal error.
5787 8. The debugging code in both pcre and pcretest for outputting the compiled
5788 form of a regex was going wrong in the case of back references followed by
5789 curly-bracketed repeats.
5792 Version 1.02 12-Dec-97
5793 ----------------------
5795 1. Typos in pcre.3 and comments in the source fixed.
5797 2. Applied a contributed patch to get rid of places where it used to remove
5798 'const' from variables, and fixed some signed/unsigned and uninitialized
5799 variable warnings.
5801 3. Added the "runtest" target to Makefile.
5803 4. Set default compiler flag to -O2 rather than just -O.
5806 Version 1.01 19-Nov-97
5807 ----------------------
5809 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5810 like /([ab]*)*/, that is, for classes with more than one character in them.
5812 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5813 as /((?>a*))*/ (a PCRE_EXTRA facility).
5816 Version 1.00 18-Nov-97
5817 ----------------------
5819 1. Added compile-time macros to support systems such as SunOS4 which don't have
5820 memmove() or strerror() but have other things that can be used instead.
5822 2. Arranged that "make clean" removes the executables.
5825 Version 0.99 27-Oct-97
5826 ----------------------
5828 1. Fixed bug in code for optimizing classes with only one character. It was
5829 initializing a 32-byte map regardless, which could cause it to run off the end
5830 of the memory it had got.
5832 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5835 Version 0.98 22-Oct-97
5836 ----------------------
5838 1. Fixed bug in code for handling temporary memory usage when there are more
5839 back references than supplied space in the ovector. This could cause segfaults.
5842 Version 0.97 21-Oct-97
5843 ----------------------
5845 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5847 2. Optimized negated single characters not to use a bit map.
5849 3. Brought error texts together as macro definitions; clarified some of them;
5850 fixed one that was wrong - it said "range out of order" when it meant "invalid
5851 escape sequence".
5853 4. Changed some char * arguments to const char *.
5855 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5857 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5858 pcretest.
5861 Version 0.96 16-Oct-97
5862 ----------------------
5864 1. Added a simple "pgrep" utility to the distribution.
5866 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5867 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5868 where "ddd" means "one or more decimal digits".
5870 3. Fixed serious bug. If a pattern had a back reference, but the call to
5871 pcre_exec() didn't supply a large enough ovector to record the related
5872 identifying subpattern, the match always failed. PCRE now remembers the number
5873 of the largest back reference, and gets some temporary memory in which to save
5874 the offsets during matching if necessary, in order to ensure that
5875 backreferences always work.
5877 4. Increased the compatibility with Perl in a number of ways:
5879 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5880 to request this handling. The option can be set at compile or exec time.
5882 (b) $ matches before a terminating newline by default; an option
5883 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5884 mode). The option can be set at compile or exec time.
5886 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5887 the same as Perl's. If the decimal number it represents is less than 10
5888 or there aren't that many previous left capturing parentheses, an octal
5889 escape is read. Inside a character class, it's always an octal escape,
5890 even if it is a single digit.
5892 (d) An escaped but undefined alphabetic character is taken as a literal,
5893 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5894 escapes.
5896 (e) {0} is now permitted. (The previous item is removed from the compiled
5897 pattern).
5899 5. Changed all the names of code files so that the basic parts are no longer
5900 than 10 characters, and abolished the teeny "globals.c" file.
5902 6. Changed the handling of character classes; they are now done with a 32-byte
5903 bit map always.
5905 7. Added the -d and /D options to pcretest to make it possible to look at the
5906 internals of compilation without having to recompile pcre.
5909 Version 0.95 23-Sep-97
5910 ----------------------
5912 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5913 \x20 at the start of a run of normal characters. These were being treated as
5914 real characters, instead of the source characters being re-checked.
5917 Version 0.94 18-Sep-97
5918 ----------------------
5920 1. The functions are now thread-safe, with the caveat that the global variables
5921 containing pointers to malloc() and free() or alternative functions are the
5922 same for all threads.
5924 2. Get pcre_study() to generate a bitmap of initial characters for non-
5925 anchored patterns when this is possible, and use it if passed to pcre_exec().
5928 Version 0.93 15-Sep-97
5929 ----------------------
5931 1. /(b)|(:+)/ was computing an incorrect first character.
5933 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5934 but not actually doing anything yet.
5936 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5937 as Perl does (e.g. [-az] or [az-]).
5939 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5940 all possible positions.
5942 5. Split up into different modules to avoid including unneeded functions in a
5943 compiled binary. However, compile and exec are still in one module. The "study"
5944 function is split off.
5946 6. The character tables are now in a separate module whose source is generated
5947 by an auxiliary program - but can then be edited by hand if required. There are
5948 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5949 toupper() in the code.
5951 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5952 make them global. Abolish the function for setting them, as the caller can now
5953 set them directly.
5956 Version 0.92 11-Sep-97
5957 ----------------------
5959 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5960 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5962 2. Caseless matching was not working in character classes if the characters in
5963 the pattern were in upper case.
5965 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5967 4. Make PCRE_ANCHORED public and accept as a compile option.
5969 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5970 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5971 pass them.
5973 6. Give an error if bad option bits passed at compile or run time.
5975 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5976 pcretest to cause it to pass that flag.
5978 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5979 options, and the first character, if set.
5981 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5984 Version 0.91 10-Sep-97
5985 ----------------------
5987 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5988 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5990 2. PCRE was looping on encountering an indefinitely repeated back reference to
5991 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5992 Perl does - treats the match as successful.
5994 ****


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

  ViewVC Help
Powered by ViewVC 1.1.5